Atlantic Hurricane Track Versions

So far I’ve located three slightly different versions of the Atlantic track data:

Here are a few boring comments on these data sets for anyone who’s working on them.

The first two versions have the same number of lines and only 35 of 13085 lines differ. None of the differing lines record wind speeds, but appear to contain codes for landfalls and things like that, codes which I haven’t been using so far. Wind speeds are in knots.

The third dataset seems to be the most official. I haven’t fully reconciled the differences, but here’s what I’ve noticed so far. Wind speeds in the third data set are recorded in mph and kph. In spot checks, the mph and kph speeds are each calculated by multiplying the speed in knots by the appropriate conversion factor and then rounding to the nearest 5. So there’s not a lot of precision in the measurement.

In all but 3 storms, the number of measurements matches. Two storms – 1166- Barry and 1360 (Ophelia -2005) – have 1 (2 – Ophelia) extra reading at low wind speeds and the difference is irrelevant. For the other hurricane (Andrew- 1992), there was a material difference and it was very hard to figure out why there was a difference. Eventually I figured out that that the updated version had not recorded measurements in 6 hour intervals, but had included 4 or 5 measurements at intermediate times during very high wind speed intervals – contrary to the practice in the data set. One can easily work around the difference, but if you were assuming (reasonably) that the data was coming in 6-hour intervals, it would lead to an error.


  1. Bob K
    Posted Jan 18, 2007 at 1:50 AM | Permalink


    Just want to make you aware there are a couple storms from the 1850’s where the data seems to have been fudged.

    storm 5: 09/13/1851 all 16 plots at 32.5N,73.5W 50kts
    storm 38: 09/22/1857 all 20 plots at 32.5N, 73.5W 70kts

  2. David Smith
    Posted Jan 18, 2007 at 5:54 AM | Permalink

    Minor footnote:

    All three databases include subtropical storms mixed in amongst the hurricanes and tropical storms. This occurs in the 1970s and later. There are maybe six to ten such storms over 30 years.

    The subtropical storms are labeled but, if one doesn’t know they are there, they can inadvertently get included in storm counts.

    Subtropical storms are part tropical and part “regular”. Dr Curry recommends excluding them in a storm count. Sometimes they accidentally get included in some studies, including Holland/Webster and, if I remember correctly, Emanuel 2005.

  3. David Smith
    Posted Jan 18, 2007 at 6:25 AM | Permalink

    RE #2 Clarification: I think that Dr Curry actually said that she knows of no study that includes subtropical cyclones, which I interpret to mean that they should not be used.

    By the way, there is a swirl in the far Atlantic this morning. It is very small and weak. However, don’t be stunned if the NHC folks watch it and decide, based on satellite images, that is has subtropical or tropical characteristics for a day or so and they thus incorporate another trash system into the modern database.

  4. Steve McIntyre
    Posted Jan 18, 2007 at 9:11 AM | Permalink

    #3. In doing my own counts, I’ve tended to use wind speeds recorded in the Track data rather than the classifications and can replicate most counts by choosing restrictions appropriately.

  5. David Smith
    Posted Jan 18, 2007 at 10:07 AM | Permalink

    Warning: minor comment on a dull topic!

    Actually, “subtropical storms” are unrelated to windspeed. They are not tropical cyclones.

    In years past they were probably ignored, if they were detected at all (and probably were not).

    A Wikipedia writeup is here .

    I count 19 subtropical storms in the databases over the last 35 years, which is 0.5 per year.
    I saw none prior to 1970.

    If one includes subtropical storms in a climatological exercise, then one is inflating the modern storm count by over 0.5 storms a year (versus earlier decades). That’s not huge, but it’s big enough to be worth noting and something which (in my opinion) should be noted in papers. Once again, it’s comparing apples to oranges.

  6. Posted Jan 18, 2007 at 7:48 PM | Permalink

    Looks to me like Holland/Webster have been doing more than a little bit of lurking at CA and Prometheus, judging by this presentation:

  7. MarkR
    Posted Jan 18, 2007 at 8:33 PM | Permalink

    #6 Their slide one claiming to show a 150% increase between regime 1 and 3, is a complete mis-reading of the data/graph. What the graph says to me is that the storm frequency oscillates around regime 2. and it’s a copy of one willis produced, but no acknowledgement.

  8. Posted Jan 18, 2007 at 9:05 PM | Permalink

    Hey, at least they are now considering the possibility that storm counts weren’t always precisely counted! They are now also hinting that they are aware of possible physical processes for why the ratio of landfalls could change. Though they tell us they will explain later — still, nice change! (Elsner who they cite, discusses the role of the NAO.)

    It will be interesting to read the details of how they estimate the undercount. (I really can’t back the analysis method out of the power point presentation.)

    (Hhhmmm…If they are lurking, should I mention they use the term “Nyquist frequency” incorrectly on page 15 of “the dreadful paper”? Cuz’ they really might want to edit that sentence before … Well, they can look up “Nyquist frequency” and decide for themselves! 🙂 )

  9. Steve McIntyre
    Posted Jan 18, 2007 at 9:20 PM | Permalink

    #6. they never mentioned the east-west issue prior to my raising it here …

  10. Posted Jan 18, 2007 at 9:33 PM | Permalink

    .. Their figure showing landfall proportions on slide 6 is [like] Figure 2 in my draft paper. …

  11. David Smith
    Posted Jan 18, 2007 at 10:09 PM | Permalink

    RE #6 Steve M, you may want to consider #6 for a new thread.

    I’m glad HW is looking at the data they use.

    It will take me a while, and some red wine, to understand the details of their presentation. In advance of that, here is some data for your consideration.

    Below are the counts of tropical cyclones which did not come within 100km of land, grouped in 10-year buckets.

    Plot these and see if a familiar shape emerges.

    1905-1914: 12 tropical cyclones
    1915-1924: 11
    1925-1934: 14
    1935-1944: 12
    1945-1954: 24
    1955-1964: 21
    1965-1974: 31 (40)
    1975-1984: 41 (49)
    1985-1994: 37 (38)
    1995-2004: 50 (53)

    (the number in parentheses include the non-tropical cyclones known as “subtropical storms”, in case HW still use subtropical storms in their cyclone counts)

    Since these always-at-sea tropical cyclones did not make landfall or brush close to land, they would have been detected by ship, aircraft or satellite (with a few exceptions). Consider that

    * Ships have been around, but are historically sparse outside the major commercial shipping lanes.

    * Aircraft patrolled part of the western Atlantic (Bermuda-Antilles-Miami triangle) from roughly 1948-1959, as best as I can tell, but not the vast regions of the eastern Atlantic.

    * Satellites became geosync, with nightime vision, in the 1970s, the timeframe at which their reliability for finding distant systems became robust (my opinion)

    My contention is that the hockeystick shape is mainly due to improvement in detection. Ship detection is not robust at all, aircraft was limited, early satellites were a big improvement and modern satellites a great improvement

    HW’s contention (I believe) is that the hockeystick is mostly real and due to AGW. (I scanned the presentation for a physical mechanism that would explain the near-steady increase in entirely-at-sea storm counts that didn’t follow the AMO while near-land storms followed the AMO, but saw none.)

    A remarkable thing is that, if one backs these entirely-at-sea storms from the total tropical cyclone counts, such that only the robust near-land storms are used, one gets a pretty oscillation (AMO?) that rises little over the last 100 years.

  12. David Smith
    Posted Mar 24, 2007 at 8:42 AM | Permalink

    Within the next few weeks we’ll get the predictions for 2007 Atlantic storms, which I”ll collect and post. These should include the Europeans (timing unknown), Gray/Klotzbach (in two weeks), several semi-private entities (already coming in), Michael Mann (15+ storms), Steve M and anyone else who wants to join in.

    If you have a prediction, please post and I’ll add it in. This is all for fun.

    Here’s mine, and my reasoning:

    First, I view the Atlantic as having active and inactive decadal modes (the “AMO”). In these eras, there are active, normal and inactive seasons. So, my prediction will be a combo of mode and year, for example “active mode, normal season”. The year 2007 was “active mode, inactive season”.

    In the active mode the average number of storms is about 9 (using modern detection techniques) while in the active mode it is about 13 storms. I then adjust this for whether an upcoming season looks active or inactive.

    Here goes:

    Mode –

    I use both persistence (AMO active phases appear to last about three decades, and we’re approaching the middle of the current AMO) and the average of the last three season’s ASO sea level pressure (SLP) in the western Atlantic (to confirm that we’re still in an active mode).

    Here is the plot of western tropical Atlantic SLP ( link ). It shows the switch in 1995 from inactive to active mode, and it shows that recent years have maintained the active mode (we’re not close to the end of the active mode). Interestingly, it also indicates that the prior switch from active to inactive was gradual – perhaps active phases are worst in their first decade, and then gradually weaken. That would be nice.

    Conclusion: active mode, which means an average season would bring 12 to 15 storms.

    Season –

    First, the famous El Nino / La Nina. which affects wind shear and (probably) atmospheric stability in the Atlantic. We appear to be headed towards a weak La Nina, which gives seasonal conditions favorable for storms

    Second is the January-March atmospheric pressure in the Atlantic between Nova Scotia and Bermuda. There is a pretty good correlation between this and the seasonal activity in active modes. I take this as an early indicator of Atlantic pressure and wind patterns later in the year. This indicator points to an active season.

    Third is the SST in the tropical mid and eastern Atlantic. Excessive warmth is associated with a northward shift and improved phasing of critical parts of the ITCZ, which is favorable to storm development. Excessive warmth also exposes seedlings to the important +26C water earlier-than-normal, giving them an earlier start in their growth. The current NOAA forecast for SST in this region during August-October shows (gasp) near-normal conditions. This indicator points towards a normal season.

    Fourth is the SST south of the ITCZ. Below-normal temperatures there create a contrast and affect wind strength and direction, affecting the seedlings). The NOAA forecast is for slightly below-normal temperatures, creating a more-favorable contrast. This indicator points towards a slightly-active season.

    Those are my main indicators, made from a classical approach. My conclusion is that this will be an active season nested in an active mode, giving 15 to 18 Atlantic storms. We shall see.

  13. John Norris
    Posted Mar 24, 2007 at 10:07 AM | Permalink

    Since it is amateur hour here at CA, and I most certainly qualify:

    2007 Atlantic Storms: 10
    2007 Hurricanes: 5

    Why, because my amateur climate/weather analysis says that that is about the long term average. And it would really upset a lot of AGW experts who think they are above amateur status if we only had another average year, and an amateur got it right by saying that it was going to be average.

  14. David Smith
    Posted Mar 24, 2007 at 10:38 AM | Permalink

    Re #13 Excellent! John, you’re booked for 10.

  15. Paul Linsay
    Posted Mar 25, 2007 at 6:23 AM | Permalink

    OK, I have a proprietary interest in this so my predictions are

    2007 Atlantic Storms 8.7+-3.0, 95% probability of 13 or fewer
    2007 Atlantic Hurricanes 5.3 +- 2.3, 95% probability of 9 or fewer

    Uping the ante, mean time between storms

    2007 Atlantic Storms 11.6 days, 95% within 36 days of each other
    2007 Atlantic Hurricanes 15.1 days, 95% within 33 days of each other

  16. Paul Linsay
    Posted Mar 25, 2007 at 10:13 AM | Permalink

    Whoops, that should have been

    mean time between storms

    2007 Atlantic Storms 11.6 days, 95% within 35 days of each other
    2007 Atlantic Hurricanes 15.1 days, 95% within 45 days of each other

  17. David Smith
    Posted Mar 25, 2007 at 7:30 PM | Permalink

    Paul, you’re booked for 6 to 12 during 2007, close to 50-year average of about 10 storms per season.

  18. Steve Sadlov
    Posted Mar 26, 2007 at 11:47 AM | Permalink

    My “funny money” says 9 Storms and 4 ‘canes. No Cat 5s, one borderline Cat 3/4 the rest wimpy 1s or 2s.

  19. Posted Mar 26, 2007 at 12:37 PM | Permalink

    I’ll go with 10 Atlantic Storms, as

    P(4 \leq AS \leq 15)=0.95

    I have also much better conditional prediction, under condition x. The prediction and explanation of x will be available in 2008.

  20. David Smith
    Posted Mar 26, 2007 at 2:55 PM | Permalink

    Re #18, #19 I’ve booked your forecasts. If you nail it, you’ll be known as a “Wizard of The Winds”.

    The latest sea surface temperature (SST) forecast from NOAA shows SST at, or slightly below, normal in the mid-tropics development region this storm season. Subsurface water supports that call. If SST dominates, then we should have a near-normal season, with around 10 storms.

    If anyone knows when and where I can find the European forecast, please post. Gray/Klotzbach’s is April 4.

  21. David Smith
    Posted Mar 28, 2007 at 5:52 PM | Permalink

    AccuWeather’s preliminary forecast / promotion is here .

    They missed their forecast pretty badly in 2006, but you’d never know that from reading this 2007 promo.

  22. David Smith
    Posted Apr 3, 2007 at 2:41 PM | Permalink

    Gray/Klotzbach’s April forecast is here .Their forecast for 2007 is for 17 named storms. That makes Michael Mann (15+ storms), me (15-18), Gray/Klotzbach (17) and TSR (17) bedfellows, and a crowded one at that.

    Jeff Masters at Weather Underground has an interesting writeup on the lack of skill of April forecasts, including a spreadsheet from Gray/Klotzbach in which they show that their April forecasts have no “skill”, while their later ones do.

  23. David Smith
    Posted Apr 13, 2007 at 5:09 PM | Permalink

    Here are the current forecasts, rounded to the nearest full storm:

    Paul Linsay: 9 Atlantic storms in 2007
    Steve Sadlov: 9
    UC: 10
    John Norris: 10
    Private forecast firm: 14
    Michael Mann: 15+
    David Smith: 16
    Gray/Klotzbach: 17
    Tropical Storm Risk, Inc: 17

    It’s shaping up like a bimodal distribution.

    As best as I can tell, the Europeans (EUROSIP) will provide their 2007 forecast around May 15

  24. David Smith
    Posted May 8, 2007 at 11:12 AM | Permalink

    Tropical Storm Risk, Inc (TSR) has slightly lowered its forecast for 2007. Their current (May) forecast is for 16 storms, while it had been 17 storms.

    For the western Pacific they forecast a near-normal 2007 with 27 storms.

    The link to their forecasts is here .

    Still no word from the Europeans (EUROSIP). My hope is that they issue a public forecast on or about June 1.

  25. Steve Sadlov
    Posted May 8, 2007 at 3:08 PM | Permalink

    The backpeddling has begun. I’m stickin’ with 9. Note the cold core low, with Aleutian characteristics, spinning off of the SE US. (!) Of course it’s only weather, but it is interesting that this highly unusual storm is there, at the time of great change in ocean thermal characteristics, mostly trending toward cooling in both the Pacific and Atlantic.

  26. David Smith
    Posted May 8, 2007 at 5:34 PM | Permalink

    Re #25 Steve S, that storm is an example of something which, pre-1970, would have been basically ignored. But nowadays, thanks to satellites, buoys, much-improved computer models and 20 million retirees watching the Weather Channel, it gets press coverage.

    Until recently the National Hurricane Center seemed eager to classify any small twist in the atmosphere as a “tropical storm” (see the 2006 twist they studied for months and finally decided that, for less than a day, it was “tropical” and so it counted in the records). That swells the storm count compared to earlier years. Now, though, with all the apocalyptic professors using those apple-to-oranges storm counts to criticize the NHC, Landsea, etc, the NHC may be less eager to classify anything that’s warm and wiggles as tropical.

  27. Mark T.
    Posted May 8, 2007 at 5:44 PM | Permalink

    That swells the storm count compared to earlier years. Now, though, with all the apocalyptic professors using those apple-to-oranges storm counts to criticize the NHC, Landsea, etc, the NHC may be less eager to classify anything that’s warm and wiggles as tropical.

    Isn’t that what Landsea has apparently recently shown to be true? At least, hasn’t he come to the conclusion that the major uptick in ‘cane count, other than the multi-decadal oscillation, is due to increased ability to detect storms in the first place? In general, I think this appears in other historic records: namely, as your ability to detect anything improves, you tend to detect more of them.


  28. Steve Sadlov
    Posted May 9, 2007 at 10:00 AM | Permalink

    RE: #26 – Woe unto us. Our wager has been compromised. Have you heard the news? NOAA has crossed a rubicon, crossed earlier by the Europeans. They are now naming cold core Lows. The very low I was alluding to is now a named storm!

  29. Steve Sadlov
    Posted May 9, 2007 at 10:17 AM | Permalink

    Question of the day. Is it a tropical cyclone, or, a retrograde nor’easter at a record low latitude?

  30. Steve Sadlov
    Posted May 9, 2007 at 10:21 AM | Permalink

    By the way, the energy of that storm, when it passed over California a week ago, resulted in record low snow levels for so late in the season – ~3500 feet here in NoCal.

  31. Steve Sadlov
    Posted May 9, 2007 at 1:32 PM | Permalink

    As expected, the flurry of “early start to hurricane season” blather has begun in the main stream media and on the net.

    Earrrrrrrrrrliest hurrrrrrrican seasonnnnnnn everrrrrrrrrr!

  32. David Smith
    Posted May 9, 2007 at 1:44 PM | Permalink

    RE #28 The more meaningful tropical information is this, from Jeff Masters:

    Longest period without a tropical cyclone ends
    Andrea’s formation brings to a close the longest period on record globally without a tropical cyclone. The last advisory issued on a tropical cyclone this year was at 06 GMT on April 6th, for Tropical Cyclone Cliff in the Southern Hemisphere. Today’s 15 GMT advisory on Subtropical Storm Andrea ends the record longest period without a tropical cyclone at 33.4 days, besting the old record of 31.5 days set mid-April to mid-May in 1984. Reliable records of global tropical cyclone numbers go back to the beginning of the satellite era, about 1970.

    Andrea is a subtropical cyclone, which means it lacks the key features of a tropical system. It does not have a warm center, it does not have outflow, it is situated over water that is too cool for tropical development, there are no thunderstorms near the center, it is wrapped in dry air, and so forth. This afternoon the showers are shrinking and the center is getting hard to define.

    These weak hybrid systems are really an outgrowth of the satellite era – prior to good satellite coverage, they were either not seen or no one really cared.

    Regarding the record, Andrea should not count as the end point for the no-storm period, as it is not tropical.

  33. Steve Sadlov
    Posted May 10, 2007 at 10:06 AM | Permalink

    RE: #32 – Further to this, I am going to write a proposed operational definition for a named tropical cyclone that should be counted toward a given annual named storm count. It’s sorely needed, in light of the “Andrea” charade.

  34. David Smith
    Posted May 13, 2007 at 7:16 PM | Permalink

    At this link is a spreadsheet which contains two charts. Click on the spreadsheet icon on the left side (one day I’ll learn how to do this in a better way).

    Chart One shows the tropical cyclone time series for the Northern Indian Ocean (NIO). The data is suspect, as storms may have been missed in the earlier period and the definition of storm may have shifted. Nevertheless, there is a dramatic change in cyclone frequency in….. you guessed it, 1976-77, the intriguing time of global climate shift.

    Chart Two shows the sea surface temperature (SST) of the Northern Indian Ocean and the annual cyclone count. Both are smoothed values. Note that the NIO SST was trendless until……….you guessed it, 1976-77, after which SST began to rise. And, note that when SST rose, storm frequency dropped. That’s rather the opposite of what some suggest happens in the Atlantic.

    Another piece of the 1976-77 climate shift puzzle.

  35. Paul Linsay
    Posted May 13, 2007 at 8:47 PM | Permalink

    #34, David. I’ve been (very slowly) analyzing all the tropical cyclone data. The NIO struck me as very weird since the drop is from a mean of 15 per year to a mean of 5 per year. None of the other basins show anything like that. I’d be very suspicious of the data, either before or after the drop, maybe both. Have you looked at SST in the other basins?

  36. Bob Koss
    Posted May 13, 2007 at 10:34 PM | Permalink


    The S. Hemisphere Basin is also responsible for the same area. I don’t know how they divide responsibility. You can see from these charts showing all storms vs. just those with wind that many of the early years there was no wind-speed associated with them. Makes it hard to put any reliability in the early years of data.

  37. David Smith
    Posted May 14, 2007 at 10:02 AM | Permalink

    Re #35 Paul, I agree that the Indian Ocean data needs to be viewed with caution. I think that the storm count after about 1970 is reasonably good, as satellite coverage should detect the presence of storms. Pre-1970 is a different matter, as the storm reports would rely on ships and landfall data. The tendency, though, would probably be to underreport earlier storms, which would mean that there may have been even more storms in the early years.

    What makes me place some credence in the mid-1970s shift is that other aspects of the Indian Ocean troposphere also changed in 76-77. My guess is that some upper air features relocated a bit, creating more wind shear over the ocean. it’s something to explore.

  38. David Smith
    Posted May 14, 2007 at 10:06 AM | Permalink

    Re #36 You’re right, the Indian Ocean early wind speed data is full of gaps and (my opinion) just about useless until the newer generation of satellites in the early 1980s. That is somewhat different from the storm count issue, as the earlier satellites could see storms, but not well enough to consistently and reliably determine intensity.

  39. Steve Sadlov
    Posted May 14, 2007 at 10:16 AM | Permalink

    Ho-hum …..

    What will be interesting will be to see just how long the cold fronts continue to dive far to the south, thereby disrupting the tropical regime in the Gulf, Northern Carribean and Atlantic around the tropic of Cancer. Although we may be going into a La Nina, the Jet Stream continues its non zonal behavior east of the Rockies.

  40. Steve Sadlov
    Posted May 14, 2007 at 1:53 PM | Permalink



    This area of low pressure was part of cold front that moved off the East coast last week, and cut itself off from the main flow, remaining essentially stationary since then. Numerous advisories and warnings for high winds and seas were issued by local National Weather Service offices for it. Early Wednesday, hurricane hunter aircraft found that, as a result of the low remaining over the warmer gulf stream waters, the low took on some tropical characteristics, including thunderstorms near the circulation center and an outflow in the upper atmosphere.

    This resulted in the system being upgraded to a sub-tropical storm, and the National Hurricane Center initiated advisories. Because a sub-tropical storm can have nearly identical impacts of a pure tropical storm, and there is always the possibility of any sub-tropical storm becoming purely tropical, it is given a name.


    Dennis Feltgen
    Acting Public Affairs Officer
    National Hurricane Center
    Miami, Florida


    Any comments?

  41. David Smith
    Posted May 14, 2007 at 2:01 PM | Permalink

    RE #39 Steve S., I’m going to see if the Auditblog feature, offered at one time on CA, is still available and would work for weather-related commentary (= not climate!). That way we can avoid cluttering ClimateAudit with short-term meteorological commentary. If something new is posted on a “weather” Auditblog then a short sentence on CA Unthreaded could let anyone interested know that.

    Meanwhile, I’ll note that May 2007 has above-average potential for tropical Atlantic (Western Caribbean) activity. And, even though a May tropical storm would not be due to SST, I’m sure we’d still hear AGW doom-and-gloom.

  42. Jonathan Schafer
    Posted May 14, 2007 at 2:35 PM | Permalink

    Throwing in my $.02.

    14 Named storms (will include 3 “sub-tropical” type)
    7 Hurricanes (1 CAT 5, 1 CAT 4, 2 CAT 3, 1 CAT 2, 2 CAT 1)
    5 Storms make landfall – 3 hurricane, 2 TS
    3 of the storms hit on the gulf coast (2 TS, 1 Cat 3 Hurricane (TX/LA border, MS, and FLA)
    2 hit the east coast (NC and SC) (1 TS, 1 Cat1 Hurricane)

    Factors included in this forecast: Onset of La Nina with lower shear, increased preciptation in Sahel area of Africa which should lower saharan dust levels over the Atlantic, leading to a warmup of SST’s, estimated position of Bermuda High should steer more storms into the GOM this year, but a late season shift will shut off the Gulf and allow for several storms to hit the East Coast, and last but not least, my “interpretation” of various weather patterns which means absolutely nothing from a scientific basis. If I’m way off, I promise to apologize…

  43. David Smith
    Posted May 14, 2007 at 3:51 PM | Permalink

    RE #42 Jonathan, you’re in for 14 named storms. If you get close then you’re a wizard. If you’re off, blame Saharan dust or George Bush.

    I do hope we get the European (EUROSIP) forecast soon.

  44. David Smith
    Posted May 14, 2007 at 4:11 PM | Permalink

    Re #40 Yes.

    First, let me say that I admire the work of the US National Hurricane Center. They do a very good job, especially considering all the uncertainties they deal with.

    Second, they rightly err on the side of caution, as their pronouncements affect lives and they would rather cry wolf unnecessarily than to be silent about something that becomes deadly. In the case of Andrea, it had strong (non-tropical) winds and was close to shore, so they erred on the side of caution.

    Having said all that, though, the words offered above, by the Public Affairs Officer, are a bit of a stretch. The initial storm discussion is here , which discusses their reasoning for naming it. Outflow was almost non-existent and the thunderstorms were still some distance east of the center.

    What is needed is a totally different name for these weak non-tropical hybrid systems, a name which does not include the word “tropical”.

  45. Steve Sadlov
    Posted May 14, 2007 at 4:40 PM | Permalink

    RE: #41 – That link has the NWS scorecard for TCs.

  46. Steve Sadlov
    Posted May 14, 2007 at 4:42 PM | Permalink

    RE: #45 – My bad, that link is only the discussion. Link with scorecards is one level up from that.

  47. David Smith
    Posted May 14, 2007 at 6:57 PM | Permalink

    In looking for enlightenment on the decline in Northern Indian Ocean tropical cyclones I saw this citation (from CO2 Science Magazine):

    Singh, O.P. 2001. Long term trends in the frequency of monsoonal cyclonic disturbances over the north Indian ocean. Mausam 52: 655-658.

    I don’t have access to that paper, but a related paper states that:

    “Singh (2001) investigated the long term trends in the frequency of cyclonic disturbances over the Bay of Bengal and the Arabian Sea using 100-year (1890-1999) data and found significant decreasing trends.”

    I’ll keep looking for an elaboration, to learn if the mid-1970s decline was real or a data artifact.

  48. Steve Sadlov
    Posted May 22, 2007 at 4:43 PM | Permalink

    Snnnnnoorrrrre …..

    I’ll still call for 9 (with the proviso that we don’t cheat like the NHS and count named mid latitude cyclones with limited to no outflow and cold cores).

  49. David Smith
    Posted May 22, 2007 at 6:50 PM | Permalink

    Here are the bets to-date. The numbers apply to tropical systems only, so the recent subtropical storm Andrea does not count. Where a range was given, I took a midpoint.

    Group One is clustered around the 1950-2005 average of 10 storms –

    Paul Linsay: 9
    Steve Sadlov: 9
    UC: 10
    John Norris: 10

    Group Two is clustered around the 1995-2006 average of 15 storms –

    Unnamed private forecast firm: 14
    Accuweather: 14
    US National Hurricane Center: 15
    Michael Mann: 15+
    Tropical Storm Risk, Inc: 16
    Gray/Klotzbach: 17
    David Smith: 17

    Group Three is peerless

    EUROSIP European forecast (“Cette prévision est le première secret jusqu’ aux fins de saison. Nous ne sommes pas factices.”)

    Michael Mann, Bill Gray and I are the three at the high end of the prediction range. Strange bedfellows, indeed.

  50. Jonathan Schafer
    Posted May 22, 2007 at 9:14 PM | Permalink


    I didn’t see my name listed in Group Two, unless I’m “Unnamed private forecast firm”, lol.

    Jonathan Schafer

  51. Jonathan Schafer
    Posted May 22, 2007 at 9:23 PM | Permalink

    I found this/a> to be interesting.

    I have not read the original paper. I still believe that Saharan dust played a major factor in the lack of storms compared to forecasts. Maybe not for the reasons stated in the original paper though. As I recall, it was more a case of the storms ingesting the dry, dust-laden air into the developing circulation that short-circuited the development, not the reduction in SST. SST as a significant measurement is a bit overrated, IMO anyway. High SST but shallow depth will not lead to more storms based on what I’ve read. Slightly lower SST but more warmth at depth is a more compelling factor.

    What I read earlier this year is that the Sahel region has had higher than normal rainfall. This should reduce the amount of dust over the Atlantic during peak TC season. Assuming the anomolous, persistant LP trough that existed for much of the season over the SW Atlantic/NE Caribbean is absent this year, I still think the higher forecasts will be more accurate.

    We should know shortly.

  52. Jonathan Schafer
    Posted May 22, 2007 at 9:24 PM | Permalink

    Oops, sorry about the closing tag on my link. Not sure what happened. The whole post was not meant to be a link.

  53. John Norris
    Posted May 22, 2007 at 9:31 PM | Permalink

    I’m stick’n with 10. That high priced Group Two doesn’t scare me. Opening day is just around the corner, no?

    By the way, since you are not likely to award cash money to the winner, I think you need to put together a trophy for the inaugural David Smith Atlantic Tropical Storm Count Open; or perhaps a virtual digital version that you can post up at the end of the season.

  54. John G. Bell
    Posted May 22, 2007 at 10:03 PM | Permalink

    I’ll go 8 +-4 with 50% confidence :). How much confidence does the US National Hurricane Center have in 15? Less than 5%? To be at the 95% level, what range do they give?

  55. David Smith
    Posted May 23, 2007 at 6:28 AM | Permalink

    Re #50 Oops, my mistake. I’ll repost the list next week, once Gray/Klotzbach have issued their final pre-season update.

    Re #53 We’ll have a Certificate of Accomplishment for the person or people who come closest to the 2007 actual number. I envision a pair of dice being part of that artwork.
    Those of us who are waaaay wrong can, as our consolation prize, publish an hurricane paper in Science magazine.

  56. Steve Sadlov
    Posted May 23, 2007 at 10:21 AM | Permalink

    RE: #54 – Interestingly, the NHS are prog’ing a mere 5 ‘canes vs my wager there will be a mere 4.

  57. David Smith
    Posted May 27, 2007 at 6:48 PM | Permalink

    Here are the entries, four days before the official start of the Atlantic season. As before, I took a midpoint of any ranges offered.

    John G. Bell: 8
    Paul Linsay: 9
    Steve Sadlov: 9
    1950-2000 average: 9.6
    uc: 10
    John Norris: 10

    IWIC: 13
    Accuweather: 14
    Jonathan Schafer: 14
    Private forecast firm: 14
    1995-2006 average: 14.5
    Michael Mann: 15+
    TSR: 16
    Gray/Klotzbach: 17
    David Smith: 17

    Meteo France: ???

    There’s still time for new entrants.

    Gray/Klotzbach will issue their final revision this coming week. I expect them to adjust their forecast down slightly, which would leave me and Michael Mann as the two extremists.

  58. Bob Koss
    Posted May 27, 2007 at 9:21 PM | Permalink


    I guess I might as well jump in with a WAG.

    Put me down for 12 named, 5 of them hurricanes.

    I’d go a little lower, but I’m inclined to think they’ll stretch the justification for a name.

  59. John Baltutis
    Posted May 28, 2007 at 1:06 AM | Permalink

    I’ll go with eleven.

  60. DeWitt Payne
    Posted May 28, 2007 at 2:43 AM | Permalink

    Fifteen named, nine hurricanes.

  61. David Smith
    Posted May 29, 2007 at 4:52 AM | Permalink

    RE #58,59,60 Gentlemen, you’re in! I’ll update the list Thursday, when Gray/Klotzbach issue their start-of-season update.

    On the topic of Atlantic storms, I wrote a few notes about an old time series plot I came across this weekend. The notes are on the CA auditblog located here .

    Posted May 29, 2007 at 7:53 AM | Permalink

    Being the Swedish foremost TS/TH expert…just
    joking… if I were, tropical weather knowledge in
    Sweden would be almost non-existant…Anyhow
    14 named 7 TH. Andrea not counted…

  63. John A
    Posted May 29, 2007 at 8:35 AM | Permalink

    7 named hurricanes, 2 will reach the US mainland. One will hit Galveston, TX.

  64. Bob Koss
    Posted May 29, 2007 at 9:36 AM | Permalink


    Do you have the NOAA prediction from last week? It’s here.

  65. Steve Sadlov
    Posted May 29, 2007 at 11:06 AM | Permalink

    More boredom. The impacts of multiple mid latitude weather systems continue to move along the Westerly steering currents and thereby prevent anything serious in the Atlantic and its adjuncts.

    There is a named (real) TS in the East Pacific – Alvin. Make that “Allllll-VIN!” 😉

  66. David Smith
    Posted May 31, 2007 at 5:14 AM | Permalink

    Gray/Klotzbach will issue their final pre-season forecast this morning. If anyone else wants to enter or adjust their forecast, today’s the day.

    On a side note, there’s a seedling near Cuba today that may turn into the first tropical system of 2007. Later today I’ll post (on the auditblog) about some aspects of this disturbance, to try to illustrate how these things work and why sea surface temperature is but one of many factors involved.

  67. David Smith
    Posted May 31, 2007 at 9:36 AM | Permalink

    Gray/Klotzbach’s final pre-season forecast, located here , appears unchanged from their April issue.

    Which is good, because it gives Michael Mann and me some company out on our near-hyper-season limb.

  68. Steve Sadlov
    Posted May 31, 2007 at 10:58 AM | Permalink

    David Smith – Where is the link to the auditblog?

  69. Ken Fritsch
    Posted May 31, 2007 at 11:14 AM | Permalink

    David Smith, my late entry into the named storm contest will be the average of the Gray/Klotzbach and Meteo France predicted counts and will change during the season with this same average. Since my models show much uncertainty, I’ll have to go with the “experts”.

  70. David Smith
    Posted May 31, 2007 at 11:25 AM | Permalink

    Re #69 Steve S, the seedling article is not written yet so there’s no link at the moment. In the meanwhile, here is a link to the just-issued note from the US hurricane people on this seedling.

    What I plan to do this evening on the auditblog is to illustrate some of the atmospheric phenomena that play a role in these storms and which affect their climatology.

  71. Steve Sadlov
    Posted May 31, 2007 at 2:42 PM | Permalink

    RE: #71 – Dave, interesting update. Those mid latitude steering currents are really keeping the weather from becoming truly tropical down to a very low latitude given the time of year. Easterly waves are staying quite far to the south. Of course, it is quite early. It will be interesting to see what’s happening in August and September.

  72. David Smith
    Posted May 31, 2007 at 6:34 PM | Permalink

    It’s 00Z June 1, so here are the contest entries.

    Entries offering ranges were reduced to the midpoints of those ranges. Subtropical storms, like Andrea, aren’t tropical systems so they aren’t counted as tropical (except in peer-reviewed articles in Science and Nature magazines). There are further rules, but I won’t release them because you’ll just try to find something wrong with them 🙂

    John A.: 7, including one that hits my town (ouch)
    John G. Bell: 8
    Paul Linsay: 9
    Steve Sadlov: 9
    1950-1994 average: 9
    uc: 10
    John Norris: 10
    John Baltutis: 11
    Bob Koss: 12
    IWIC: 13
    Accuweather: 14
    Jonathan Schafer: 14
    IW (private forecast firm): 14
    Steffan Lindstrom (on behalf of Swedish Hurricane Centre): 14
    1995-2006 average: 14.5
    DeWitt Payne: 15
    US National Hurricane Center: 15
    Michael Mann: 15+
    TSR: 16
    Gray/Klotzbach: 17
    David Smith: 17

    Ken Fritsch: ((GK+MF)/2) + homogenisation adjustment
    Meteo France: ????

    If I have missed someone, please post.

  73. David Smith
    Posted Jun 1, 2007 at 3:06 PM | Permalink

    RE #72 Steve S, that tropical disturbance didn’t work as an illustration, so I didn’t do the writeup.

    But, it did develop into the first Atlantic tropical storm (see here ) which should bring much-needed rain to bender in Florida.

  74. Steve Sadlov
    Posted Jun 1, 2007 at 3:39 PM | Permalink

    RE: #74 – A tropical storm with an occluded front attached to it? While the NHS has a stronger case with this one than they did with Andrea, it still has lacked certain things:
    1) Not a result of an Easterly Wave
    2) Never recurved – has traveled nearly straight line, SW to NE
    3) Did not start to move toward US until it was subsumed into mid latitude flows and patterns – in other words, until after it was inarguably a typical warmish core midlatitude low (like the kinds of warm core cyclonic storms we get here in California which originate near Hawaii – these are never, never, never, called TS)

    This is going to be a frustrating season. Need to check what Sobel has to say about this.

  75. Bill F
    Posted Jun 1, 2007 at 3:46 PM | Permalink

    Given the NHC’s apparent proclivity to name any area of cloudiness a storm this season (for very thinly disguised reasons), I am going to see if I can break off that extreme limb and predict 18.

    Posted Jun 1, 2007 at 4:07 PM | Permalink

    Thank you for upgrading me to a Cat..purr… 5 Hurricane…
    I think IⳭ heading for Tennesee…Al here we go…
    one minor thing only: My first name is Staffan! Steffan
    is only used!? in double-names like Curt-Steffan (Giesecke
    a former top official of Employers Association SAF former
    abbreviation too…My cat Bull will be 17 hopefully in some
    hours…SMHI World May Weather claimed “Andrea”(2007)
    was “tropical” No way… You see compared to these guys I really
    seem to be the”foremost TS/TH expert” to quote myself!!!LOL
    Mark Twain wasn⳴ dark enough about “Homo Sapiens”
    PS David have you ever thought of Alⳳ AIT the smoke on the
    DVD swirls clockwise Isn⳴ it supposed to symbolise “Katrina”
    of 2005?????? Compare Heidi Cullen…Enough!DS

  77. Steve Sadlov
    Posted Jun 1, 2007 at 4:15 PM | Permalink

    RE: #77 – —–> Mark Twain wasn⳴ dark enough about “Homo Sapiens”

    Hear, here!

  78. David Smith
    Posted Jun 1, 2007 at 4:16 PM | Permalink

    From a climatological perspective the interesting thing is that, in earlier years, people may have never known of Barry’s existence. Thanks to modern satellites, automated buoys, doppler radar and aircraft a weak borderline system like Barry was detected.

    Even when Barry passes onto land it will be little more than a short-lived breezy thunderstorm in a couple of shoreline counties. Prior to say, the 1930s, that weather would have probably been ignored and never named.

    Therein lies a problem when early periods are compared to modern times. It’s like comparing apples to oranges.

  79. Steve Sadlov
    Posted Jun 1, 2007 at 4:21 PM | Permalink

  80. Steve Sadlov
    Posted Jun 1, 2007 at 4:31 PM | Permalink

    The NHS just jumped the shark (at least they did in my book):

    I’m sorry, that is not a TS.

  81. jae
    Posted Jun 1, 2007 at 5:55 PM | Permalink

    73, David: put me down for 12.

  82. David Smith
    Posted Jun 1, 2007 at 6:09 PM | Permalink

    Re #77 My apology, Staffan! To make ammends, I will also cite you as “Svensk Kung av Cyklon” (I hope that is correct).

  83. David Smith
    Posted Jun 1, 2007 at 7:10 PM | Permalink

    A short blog entry on Barry ( link ).

    jae, you’re in for 12.

  84. Steve Sadlov
    Posted Jun 1, 2007 at 7:30 PM | Permalink

    David I posted a comment at your blog. Related note, check out the IR link I have a couple posts up from here in this thread. The cold front is well defined all the way down to the area offshore from Belize! That is some amazingly low latitude penetration of a cP airmass especially for so late in the year. It’s even further south than the fronts that swept past us in the Bahamas in early Feb.

  85. David Smith
    Posted Jun 1, 2007 at 7:50 PM | Permalink

    Re #85 Steve, I’m honored by your posting a comment. The auditblog lets me play God with a comment and choose to accept, delete or edit.

    And maybe I can find your IP address…I’ll ask mike or gavin (sigh) for guidance

    Posted Jun 1, 2007 at 8:19 PM | Permalink

    ….Just woke up weⳲe 8h ahead now whereⳳ the server
    situated?? Well David I don⳴ know about your Swedish
    teacher but she or he is a little bit rusty perhaps..
    Perhaps “Kung av Cyklon” should be “Orkankungen”
    You know the nazis used the gas named Cyklon B to kill
    jews, gypsies and others during WW2…There is also
    apparently a Swedish company in Malung named “Cyklon”
    Everything weather- and climate-related is very sexy
    right now…GWS…
    But my question about Goreⳳ smoke Katrina clockwise
    you don⳴ need a satellite to see it he wrote with
    a subtle sardonic? smile on his lips… 73ⳳ man!

    Posted Jun 1, 2007 at 9:15 PM | Permalink

    #87 Addendum
    There are other Swedish companies in market research
    etc…named “Cyklon”etc…Steve Sadlov`s link to NOAA
    shows that Barry is just a film composer who married
    Jane Birkin when she was 19 and they got a daughter
    ….Apparently I⳶e got to start a blog of my own
    to end all polito- culture-etc allusions…
    Actually ever since 2005 I⳶e tried to educate people
    about detection and different abilities between 1933
    and 2005 to detect tropical systems…I haveⳮt
    mentioned buoys and drop-sonds etc too much for
    “Joanne or Joe Public” in the street, “Svensson”
    BTW Dave and/or Steve S I saved a film(there is
    smart DL software for streaming…) about
    NHC reconnaisance aircraft showing dropping of
    dropsondes. Do you know if they get stuck and
    they have to snip them…?? Or how does it work
    If always cutting the sondes and if they also
    had a temperature sensor we would get a picture
    of the sea temperature all the way down. It must
    have an ultra longwave transmitter then CMIIW …

  88. David Smith
    Posted Jun 1, 2007 at 9:37 PM | Permalink

    Re #88 Dropsondes use radio transmission to send data to the aircraft. They have a GPS receiver, so they know their (almost) exact location at any point, which is a big improvement over early versions. Neat tools.

    Al Gore’s Katrina is indeed spinning backwards for a Northern Hemisphere storm. If his backwards storm had hit New Orleans then that city would have likely stayed dry, thanks to the winds moving the water away from the city.

    Posted Jun 1, 2007 at 10:07 PM | Permalink

    Lurking on other threads for a while…But
    what about my idea of all-levels sea temperatures
    measuring right under a TS/TH? Would it add something to our knowledge??
    And has anybody succeded in public to ask Al Gore
    about “AIT-Katrina” spinning backwards…would be
    really low ….

  90. Ken Fritsch
    Posted Jun 2, 2007 at 11:15 AM | Permalink

    Re #85 Steve, I’m honored by your posting a comment. The auditblog lets me play God with a comment and choose to accept, delete or edit.

    I anticipate visiting your auditblog for background information on NATL tropical storm development this season — of course, you knew that without my commenting.

  91. David Smith
    Posted Jun 2, 2007 at 3:15 PM | Permalink

    Re #91 Ken, I plan to use the auditblog to post various images and show how the professionals use things like QUIKSCAT to evaluate tropical seedlings and storms. Auditblog is quite user-friendly and robust for komputerklutz like me.

  92. David Smith
    Posted Jun 3, 2007 at 11:14 AM | Permalink

    Re #75, #76 Here’s a discussion, a bit technical, where someone else expresses astonishment that Barry was considered a tropical storm.


    I think the US National Hurricane Center feels under pressure to name anything that spins as a tropical cyclone, due in part to the Weather Channel (a popular US cable station) sensationalizing everything and getting viewers riled up when disturbances aren’t immediately called storms.

  93. John G. Bell
    Posted Jun 3, 2007 at 12:22 PM | Permalink

    I think you are right. I heard a reporter, on NPR I think, say that he was told that the a storm was named on the first day of the season so that it would motivate people to buy emergency supplies. It is good to get people ready but it kind of makes the named storm count useless.


  94. K
    Posted Jun 3, 2007 at 2:27 PM | Permalink

    #94: How near?

    Not very near! Barry hit the coast as a mild storm – the paper said 20-30 MPH and put a lot of desired rain into a somewhat parched Florida.

    But things were better yesterday. Barry was to pick up force in the warm waters near Tampa Bay.

    In June the USNHC* has had it their way on two of three days. Not a bad percentage.

    *Actually I think they do a good job and should issue too many warnings rather than too few. It is the media that wants terrific hurricane stories.

  95. Ken Fritsch
    Posted Jun 3, 2007 at 4:50 PM | Permalink

    How should Barry be handled/considered in scoring AGW good outcomes versus bad outcomes? It would appear that Barry can add to the count of named storms the increased frequency of which some would want to attribute to AGW and in general this would be considered a bad outcome. On the other hand, its less than destructive forces were beneficial to mitigating the Florida and Georgia droughts.

    How about this version for the media? Barry originated from less than significantly elevated SSTs and thus is more in tune with the old time pre-AGW storms that were milder in intensity as thermodynamically predicted (without mention of quantifying the differential magnitude) and thus a better example of the good old days of tropical storms and before AGW made everything bad. I should copy Dr. Curry with this spin just in case she is asked to comment on Barry.

  96. David Smith
    Posted Jun 3, 2007 at 7:43 PM | Permalink

    Storms like Barry and Andrea inflate the modern storm numbers, making it misleading to compare recent storm numbers with those before the modern satellite era (1970s and later). That inflation of storm count plays into the hands of Emanuel, Mann, Webster and Holland, all of whom have published time series comparing today’s inflated storm counts with those as far back as the 1860s.

    The highest landfall wind I could find for Barry was 28 mph, which is like a garden-variety Great Lakes storm.

    The single best way to characterize a hurricane, or season, is its ACE (Accumulated Cyclone Energy) index, which is a function of count, duration and intensity. Storms like Barry have an extremely low ACE, almost as if they never existed, which effectively purges them from the records. But, ACE is an unfamilar concept for the public whereas storm count is familiar, and therein lies the importance of storm count.

    I don’t think we’ve heard the last from the authors of 2005. Hurricanes are drama-filled, destructive and deadly, aspects which have traction with the American public. Hurricanes are too valuable an issue to be left alone by the catastrophists, regardless of the weakness of their case.

  97. Bill F
    Posted Jun 3, 2007 at 8:53 PM | Permalink

    Storms like Barry and Andrea inflate the modern storm numbers

    Which is why I put my prediction at 18. If they named the first two cloudbursts of the season, I doubt they will not name a few more later in the season. I would have predicted about 13-14, but with their obviously liberal naming practices so far, I am sure they numbers will be high not matter what happens, because that is what the models predict will happen. If the numbers don’t match the model outputat the end of the season, they will go back after the season and add some “adjustments” to pick up a few extra storms that were not called storms. Their rationale will be that the measurements at the time failed to accurately show what the models were able to retroactively construct.

    Posted Jun 4, 2007 at 6:01 AM | Permalink


    David and Bill…

    Just a little question:Is our contest only about
    NHC officially named storms and in that case,
    I am already disqualified as I stated
    “Andrea not counted” in my entry!! I think we what see here without
    being a boxing expert a parallel of the boxing world… I think there are at least
    5 global boxing organisations with different ratings???
    With NOAA “devouring” NHC…Do we know anything about
    the internal discussions on NHC?? Todayⳳ NHC
    forecaster: Rubin-Oster Googled it: ONE hit
    Have they begun using pseudonyms??
    More of my disillusion about naming TS etc in thread
    #10 Posts 106 and some of the following …

  99. Ken Fritsch
    Posted Jun 4, 2007 at 10:28 AM | Permalink

    David Smith, I know sometime ago you noted that Webster and Holland used subtropical storms in their storm counts that I believe were assumed to include only tropical storms. Have there been any further developments on this point (of contention)? Evidently from this season’s experiences a subtropical storm can be given a name and thus enter the count of named storms. Is there a count that “officially” separates subtropical and tropical named storms?

    I agree with you that using the ACE index alleviates a lot of these subjective calls. Maybe your contest should have used ACE. If my predictions on counts go astray I will be looking for Dr. Mann to argue my case for the appropriate adjustments.

  100. Steve Sadlov
    Posted Jun 4, 2007 at 10:32 AM | Permalink

    RE: #97 – January 1996. San Francisco. A deep low with occluded front dips far to the south, sweeping just past the eastern shore of the island of Hawaii, due east of Hilo. From there, it cuts a northeastward course. The center of the Low makes a direct hit on the Golden Gate. As it reaches shore, sustained winds are in excess of 65 MPH, with gusts to 90MPH. As a result, downed trees, horrible damage to the Golden Gate Park conservatory of flowers, power outages, flash floods …. the usual and expected outcomes for a storm such as this. Of course, the storm was not named. If we get another one some winter, it will not be named. Why? Because it’s the Pacific, not the Gulf or Atlantic. Ah, the “logic” of the NHC. (Not!)

  101. aurbo
    Posted Jun 4, 2007 at 12:31 PM | Permalink

    The first incidence I recall of storms being named was a fictional one. The storm was named Maria and served as the anthropomorphic title character in the novel Storm by George R. Stewart. It was published in 1941 and chronicles the history of a North Pacific cyclone that moved into the Pacific Northwest, then redeveloped east of the Rockies in a 5-day assault on the US wreaking considerable havoc along the way. I remember the book well. It likely contributed to my life-long involvement in meteorology first as an avocation as a child, and then at the professional level for the past fifty years.

    The sequential naming of Hurricanes began in 1951 using the phonetic alphabet. Prior to that, memorable hurricanes were occasionally given the name of the Saint’s Day that the storm gained notoriety when it struck a Caribbean island. At other times the storm might have given the name of a sailing vessel that was particularly affected. The use of female names began in 1953 Then, perhaps due to criticism from feminists, the use of alternating male and female names was instituted in 1979.

    Exactly when the avid use of names for wimps like Andrea and Barry was instituted is less well defined.

  102. David Smith
    Posted Jun 4, 2007 at 12:59 PM | Permalink

    Actually, in 2002 the NHC began naming subtropical storms. Before 2002 they received numbers or Greek alphabet names. I think they should have stuck with numbers.

    Ken, I noticed that Gray/Klotzbach make a point of removing subtropical storms from their storm counts. Their May 31 release says,

    Additional Note

    Subtropical storm Andrea formed off the southeast coast of the United States on May 9. Since Andrea was never classified as a tropical storm by the National Hurricane Center, it will not be counted as a named storm in our seasonal statistics.

    Whether the 2005 authors will remove them from future counts is unknown.

    There is another, even bigger, issue with modern (satellite-derived) records, in that they detect periods of momentary strength in weak cyclones which, in earlier days, were often missed. The result is an increase in very weak, short-lived storms in the recent records.

    If our 2007 forecasts develop divergence (from reality) problems then we’ll have a contest to see who can come up with the most most creative “adjustment” to reality.

  103. Steve Sadlov
    Posted Jun 4, 2007 at 1:34 PM | Permalink

    RE: #93 – After reading that, I am utterly disgusted. Up is down, down is up …. for the Love of Big Brother …

  104. Steve Sadlov
    Posted Jun 5, 2007 at 12:25 PM | Permalink

    RE: #93 – I am now a member of that blog. My handle is SteveS1954.

  105. Steve Sadlov
    Posted Jun 5, 2007 at 1:25 PM | Permalink


  106. MarkW
    Posted Jun 5, 2007 at 2:36 PM | Permalink

    Obviously it’s being caused by global warming.

  107. Steve Sadlov
    Posted Jun 5, 2007 at 2:43 PM | Permalink

    David Smith and I had some discussion about “something” shifting in 1976 – 1977. There was some discussion about it having to do with the “dipole” of the Far Western Pacific Warm Pool and one in the Indian Ocean. The unusual cyclone threatening Oman sent me to Google wondering what the history of Arabian Sea cyclones has been. The “something” shifting, which may or may not be related, could be related to PDO or something even higher than that in the “hierachy” of “oscillator” sources.

  108. Bill F
    Posted Jun 5, 2007 at 4:10 PM | Permalink

    Steve S,

    If you really want to look at something that will make you scratch your chin with regard to PDO shifts, check out this paper:

    The Bz component of the Interplanetary Magnetic Field is one of the things that determines how much solar flux reaches the Earth’s atmosphere. Currently, when the Bz is strong and favors a southerly orientation, if can “link up with” or partially cancel out earth’s magnetic field, producing strong auroras at the poles as solar flux is allowed to more easily reach the earth’s atmosphere. The traditional thinking had been that Bz fluctuated back and forth randomly around zero and balanced out. The paper shows that there are prolonged periods of time (several years or more long) where one orientation of the Bz is favored over the other. If you assume that oscillations on the scale of the PDO and AMO have to be driven by changes in energy between the poles and the tropics, what better way to explain a shift than a prolonged shift in how much solar flux reaches the poles due to the orientation of the IMF? Such a phenomenon could also cause the oscillation theorized by Svensmark that results in Greenland and Antarctica trending in opposite directions with respect to temperature.

    I am hoping as some more of the satellite projects designed to monitor the relationship between the sun and the earth’s magentic field mature, that more definitive data regarding oscillations in solar flux at the poles will become available.

  109. Steve Sadlov
    Posted Jun 5, 2007 at 4:19 PM | Permalink

    RE: #109 – It is indeed intriguing. The sun’s own field lines may interact with the earth’s “dynamo” system, predisposing one polarity over the other. Also, the configuration of the plates may have a role. Who knows what innate things about the core may also predispose polarity.

  110. Steve Sadlov
    Posted Jun 6, 2007 at 5:04 PM | Permalink

    ZZZZZZZZZZZZZZZzzzzzzzz …….

  111. Bill F
    Posted Jun 8, 2007 at 5:06 PM | Permalink

    Hey not so fast Steve…that bit of cloudiness over florida has some movement to it that somebody could call a clockwise rotation. I also wouldn’t be surprised if there might be a backyard weather station somewhere that could measure a gust of wind greater than 35 mph in one of those seabreeze thunderstorms. By my estimation, there is about a 50/50 chance NHC will decide to give this thing a name…

  112. David Smith
    Posted Jun 8, 2007 at 6:41 PM | Permalink

    I won’t say the NHC is trigger-happy in their search for storms but this afternoon a hurricane hunter aircraft buzzed my coffee cup. I guess when I stirred the cup it made the steaming vapors spin… Tropical Storm Starbucks…

    Bad jokes aside, the NHC works hard and does a very good job. I admire their work. But, they did blow it on Andrea and probably on Barry. I hope they’ll correct those in the end-of-season reanalysis so that they don’t clutter the climatology.

    Winds are favorable for an active June.

  113. Bill F
    Posted Jun 9, 2007 at 11:07 AM | Permalink

    I agree. I spent nearly 72 hours huddled in a Cozumel hotel bathroom in the middle of Hurricane Wilma 2 years ago, so I have great appreciation for anybody that voluntarily flies into the middle of something like that on a routine basis. They are undoubtedly hard working, dedicated professionals, but they need to make sure that they apply their own stated criteria to naming these storms, as they can’t help but be aware of how intense the scrutiny of the numbers is going to be if we have the active season everybody expects. Having storms like Andrea and Barry in there that probably don’t belong is going to needlessly add to the rancor of the scientific analysis of the season when it is over.

  114. Steve Sadlov
    Posted Jun 12, 2007 at 4:01 PM | Permalink

    Snnooooooorrrrrreeeee ……pppppphhhhhhhh …..

    So long as the southern branch of the split jet continues to pump mP and cP out over the Gulf, and so long as the northern branch continues to keep the digging trough in place along the East Coast, and so long as the jet and Azores/Bermuda High remain at latitudes more typical of April …. the sleepy nature of things will continue. But oh, I know, I must wait until Aug-Sep before declaring anything. 😉

  115. Steve Sadlov
    Posted Jun 14, 2007 at 3:09 PM | Permalink

    RE: Inflated named storm count via naming of mid latitude cyclones, cold fronts, outflow boundaries, sea breeze fronts, etc ….. LOL!

    I refuse to let this sink into Wayback Machine-land:

  116. Steve Sadlov
    Posted Jun 19, 2007 at 10:42 AM | Permalink

    Yawn ………

    We are now approaching the true “prime time” window for tropical cyclones. An overwhelming majority of them occur between the summer solstice and a week or two after the fall equinox. The 100 days from this Thursday onward will make or break the various forecasts for this season.

    As of today …. still an unbelievable winter or spring like NW flow pinning down truely tropical weather against the north shore of South America. Meanwhile over the open Atlantic, the jet stream dips again to the south, pushing the Azores High so far south that Easterly waves coming from the Sahel have no moisture in them. I’m still gunning for 9 named TCs (excluding the two bogus ones so far) and 5 ‘canes.

  117. Steve Sadlov
    Posted Jun 19, 2007 at 11:41 AM | Permalink

    oops, I meant 4 ‘canes.

  118. David Smith
    Posted Jun 20, 2007 at 12:49 PM | Permalink

    Well, we finally have a dynamical (computer model) forecast of 2007 Atlantic hurricane activity, from UK met, located here . Weather Underground also has a nice writeup:

    This is the first year that the UK Met Office has issued a forecast of hurricane season activity, so we don’t have any previous years to evaluate their forecasts. The results of their experimental forecasts issued for the 1987-2002 seasons are scheduled to be published later this year in the peer-reviewed journal Geophysical Research Letters. The UK Met Office claims that their forecast out-performed the forecasts made for the 2005 and 2006 Atlantic hurricane season issued by the other major seasonal forecast groups. I have high hopes for the UK Met Office forecast, since it is based on a promising new method–running a dynamical seasonal prediction computer model of the global atmosphere-ocean system. The Dr. Bill Gray/CSU forecast is based on statistical patterns of hurricane activity observed from past years. These statistical techniques do not work very well when the atmosphere behaves in ways it has not behaved in the past. The UK Met Office forecast avoids this problem by using a global computer forecast model–the GloSea model (short for GLObal SEAsonal model). GloSea is based on the HadCM3 model–one of the leading climate models used to formulate the influential UN Intergovernmental Panel on Climate Change (IPCC) report. GloSea subdivides the atmosphere into a 3-dimensional grid 3.75° in longitude, 2.5° in latitude (277.5 km), and 19 levels in the vertical. This atmospheric model is coupled to an ocean model of even higher resolution. The initial state of the atmosphere and ocean as of June 1, 2007 were fed into the model, and the mathematical equations governing the motions of the atmosphere and ocean were solved at each grid point every few minutes, progressing out in time until the end of November (yes, this takes a colossal amount of computer power!) It’s well-known that slight errors in specifying the initial state of the atmosphere can cause large errors in the forecast. This “sensitivity to initial conditions” is taken into account by making many model runs, each with a slight variation in the starting conditions which reflect the uncertainty in the initial state. This generates an “ensemble” of forecasts and the final forecast is created by analyzing all the member forecasts of this ensemble. Forty ensemble members were generated for this year’s UK Met Office forecast. The researchers counted how many tropical storms formed during the six months the model ran to arrive at their forecast of ten named storms for the remainder of this hurricane season. Of course, the exact timing and location of these ten storms are bound to differ from what the model predicts, since one cannot make accurate forecasts of this nature so far in advance.

    The grid used by GloSea is fine enough to see hurricanes form, but is too coarse to properly handle important features of these storms. This lack of resolution results in the model not generating the right number of storms. This discrepancy is corrected by looking back at time for the years 1987-2002, and coming up with correction factors (i.e., “fudge” factors) that give a reasonable forecast. This year’s GloSea forecast shows a cooling trend in the tropical Atlantic sea surface temperature (SST) compared to what we’ve seen in recent years, and is a major reason why the UK Met Office forecast is so much lower than the other seasonal Atlantic forecasts. I believe that the GloSea model has high enough resolution to do as good a job as the other seasonal hurricane forecasts this year, but it’s hard to make an informed judgment until their research results are published. The GloSea forecast is based on sound science, though, and does call into question whether or not the other seasonal forecasts are forecasting unrealistically high levels of hurricane activity in the Atlantic this year. I think that is probably the case, and a better forecast can be made by averaging together the four models into a consensus forecast. Consensus forecasts are difficult to beat, and the consensus of the CSU, NOAA, TSR, and UK Met Office forecasts yields a prediction of 13 more named storms this year, for a total of 15.

    The future of seasonal hurricane forecasts
    The future of seasonal hurricane forecasts using global dynamical computer models is bright. A group using the European Center for Medium Range Weather Forecasting (ECWMF) model is also experimenting with some promising techniques using that model. Models like the GloSea and ECMWF will only get better as increased computer power and better understanding of the atmosphere are incorporated, necessitating less use of “fudge” factors based on historical hurricane patterns. If human-caused climate change amplifies in coming decades, statistical seasonal hurricane forecasts like the CSU’s may be limited in how much they can be improved, since the atmosphere may move into new patterns very unlike what we’ve seen in the past 100 years. It is my expectation that ten years from now, seasonal hurricane forecasts based on global computer models such as the UK Met Office’s GloSea will regularly out-perform the statistical forecasts issued by CSU.

    Basically they forecast 10 tropical cyclones, based on counting the storms generated by their computer model run through November 30.

    Interesting. More comments later.

  119. Steve Sadlov
    Posted Jun 20, 2007 at 2:46 PM | Permalink

    So, the pro-“killer-AGW” model, used by the IPCC, which assumes warrrrrmmmmming oceans and mmmoooorrrre TCs in the 21st century, says 10 TC’s? Hmmmm …. maybe my forecast of 9 is high? LOL ……. of course their model does acknowledge aspects of reality, for example the real world cooling of Atlantic SST in the short term.

  120. Bob Koss
    Posted Jun 20, 2007 at 4:48 PM | Permalink

    They made hind-cast predictions for 1987-2002? Why those particular years and not include 2003-2004? Why skip two years and then refer to 2005-2006 hindcasts? As far as I’m concerned, they get a credibility black mark.

    Sea level cell area is more than 40,000 sq. miles at 20N. Not exactly a small area. That could almost contain an entire small storm.

  121. David Smith
    Posted Jun 20, 2007 at 7:37 PM | Permalink

    The UK Met approach is interesting.

    As I understand it, they use June 1 as their starting point and run their global weather model for June thru November. But, since they don’t perfectly know all atmospheric conditions on June 1, they run a total of 40 variations (an “ensemble”) of June 1 conditions, each with slightly different June 1 initial conditions than the others. They then average the number of tropical cyclones formed by the 40 ensemble members to arrive at a raw forecast.

    But, they’ve found that this raw average number is usually wrong. So, they apply a “fudge factor” to the computer-generated number to arrive at their seasonal prediction.

    The fudge factor comes from examination of prior seasons, using the same 40-member ensemble approach. I wonder if they also have different fudge factors for different broad conditions (El Nino years, high NAO years, pre-1995 AMO years, etc).

    This dynamical (global weather model) approach is what Judith Curry was enthusiastic about last year, thinking it will outperform the analog methods used by Gray/Klotzbach, NOAA and others. We shall see.

    My experience in watching the GFS and ECMWF model outputs are that they give pretty good hints about overall conditions up to a week in advance, but they veer widely off the mark after a week. They are also poor at wind shear conditions, where relativve locations of atmospheric features matter a lot.

    For fun, I plan to track and discuss a weather feature called an easterly wave as it crosses the Caribbean into the Gulf of Mexico the next several days. It has a slight chance of developing into a tropical cyclone early next week in the Gulf. This will be a chance to illustrate, using satellite images and maps, ideas like ITCZ, wind shear, co-location of vorticity, dry air, etc.

    And for clarification, the UK Met forecast is for 10 tropical cyclones after June 20. Since Barry occurred on June 1 (illegitimate though it may be) the UK met forecast will be recorded in our contest as 11 storms in 2007.

    Posted Jun 21, 2007 at 5:10 AM | Permalink


    Bob! In the case CMIIW but hurricane “Vince” 2005 off Madeira didn⳴ even
    require that much space …If Iⳤ cleaned my living-room its eye would
    have had a nice time and been watching TV there for an hour or so…

  123. Steve Sadlov
    Posted Jun 21, 2007 at 9:57 AM | Permalink

    RE: #122 – I thought Barry was not going to be counted? Counting it makes no sense, since it’s formation mechanism was not an Easterly wave.

  124. David Smith
    Posted Jun 21, 2007 at 10:09 AM | Permalink

    Re #124 It depends ( on whether it can help me achieve my forecast of 18 storms) 🙂

    There’s an area of disturbed weather in the western Caribbean today which I’ll track on the auditblog, as indicated on #122. I think it will be a good one to use as an example.

  125. Posted Jun 21, 2007 at 2:43 PM | Permalink

    Its already hurricane season. We should be prepared anytime. I think it is recommended that we provide ourselves a checklist for the preparation of any incoming storms. I found this checklist a while ago: Click Here

  126. Steve Sadlov
    Posted Jun 21, 2007 at 4:22 PM | Permalink

    Baker Roofing? Great, now we’ve gotting spamming general contractors. Go away!

  127. David Smith
    Posted Jun 21, 2007 at 9:24 PM | Permalink

    Comments on a tropical disturbance south of Cuba are here .

  128. Steve Sadlov
    Posted Jun 26, 2007 at 11:26 AM | Permalink

    (North American satire)NHC, I’d like you to meet the Maytag repair man. (/North American satire).

  129. cbone
    Posted Jun 26, 2007 at 12:22 PM | Permalink

    Latest NOAA ENSO briefing. It looks like ENSO neutral continues to be the meme… that would favor a more average season.

    Click to access enso_evolution-status-fcsts-web.pdf

  130. Steve Sadlov
    Posted Jun 26, 2007 at 12:28 PM | Permalink

    RE: #130 – I would not be at all surprised if the weather / climate community were to experience a “forget everything you think you know about ENSO” moment over the next couple of years. The reason I say this is, almost everything in the modern lexicon of true science knowledge (as opposed to Chilean fishermen’s yarns) about ENSO has been accumulated during a PDO positive phase. We really know (in the utmost scientific sense) next to nothing about ENSO under a PDO negative regime.

  131. Steve Sadlov
    Posted Jun 26, 2007 at 12:32 PM | Permalink

    …… Although obviously there are proxy data and anecdotal / indirect instrument records implying something about pre-1976. But in the strictest sense of tying together, real time, current ENSO and PDO conditions, with multi observation band satellite obs, etc, our knowledge base began around the time of the LANDSAT era. …..

  132. Steve Sadlov
    Posted Jun 29, 2007 at 11:05 AM | Permalink

    More boredom …..

    Posting early as I will be gettin’ out of Dodge for the next 10 days, in search of real summer weather. Where I can walk in a t-shirt at 10PM. LOL …..

    Notice the damping effect of the ongoing aggressive Sahel/Sahara dust. Cold front will reach the Bahamas mid week. Is it summer anywhere? 😉

  133. Dave Dardinger
    Posted Jun 29, 2007 at 12:17 PM | Permalink

    re: #133

    Try Phoenix. It’s 110+ here as far as the eye can see. Supposed to be 115 some next week. Normally this drops down sometime in July when the monsoon hits with its higher humidity. But this year, who knows?

  134. Posted Jun 29, 2007 at 6:19 PM | Permalink

    David and others can stop speculating about the statistical status of TS Barry now.

    Here is the official re-analysis:

    Click to access TCR-AL022007_Barry.pdf

    They find it was indeed a tropical storm for a while, and the observations in the report confirm that diagnosis:

    “Barry was embedded within strong southwesterly shear ahead of a mid-latitude trough and the cloud pattern had features of both a tropical and a subtropical cyclone. However, data from the Air Force Reconnaissance plane during the afternoon of 1 June indicated that the area of strongest winds was very close to the center, of the order of 5 to 10 n mi. This structure is typical of tropical cyclones, and because organized convection developed near the center during that period, Barry has been classified as a tropical cyclone. There was a flight level (1500 ft) peak wind of 67 knots in the northwest portion of the circulation reported at 2058 UTC 1 June. These winds were not representative of the intensity of the cyclone since they were short-lived and appeared to be associated with a convective downburst.”

  135. David Smith
    Posted Jun 29, 2007 at 9:10 PM | Permalink

    RE #135 Good find, Carl, especially since it’s quite early for reanalysis reports to appear. It’s a judgment call, like in a sports game, and I’ll stand by the NHC conclusion.

    The question to be asked is, if this was say 1940, would a weak, brief wisp of a storm like Barry have been detected? The plain answer is no. Barry was detected solely by the extensive surveillance of modern satellites and recon flights.

    The modern record is cluttered by such marginal systems (I estimate about 10% of the reported storms of the last decade or two were such arguable storms, including #10 of 2006, which required several months of cloud reanalysis to even find its 18 hours of existence). Before the 1970s these weak storms were generally missed.

    So, we can’t compare modern storm counts with pre-modern counts. Yet, certain authors do this without even blushing. Sad.

  136. Judith Curry
    Posted Jun 30, 2007 at 7:43 AM | Permalink

    David, a quick comment on the accuracy of the historical north atlantic hurricane counts. There are a number of sources of errors, some which would lead to undercounting and others to overcounting:

    1) storms in the open ocean prior to the 1940’s, that were simply missed by ships. this has been estimated by Knutsen and Vecchi in a forthcoming paper that examined in detail the ship track locations, to be 1-2.5 (the underestimation varying in time, larger errors in the 1850’s)
    2) double counting prior to 1900. If you look at the actual tracks, you see that some tracks are parallel, close together, and with overlapping dates. we are in the process of documenting this and will submit to the HURDAT committee
    3) spurious inclusion of subtropical storms in the database, which would contribute to overcounting particularly prior to 1950
    4) misclassification of weak storms: not capturing the weak storms from about 1950-2003, and random misclassifications prior to 1950 (not capturing weak storms, as well as overclassifying a tropical depression)

    Note, these errors are not systematic, there is no reason to believe that the overcounting errors are smaller than the undercounting errors. The comparison that Landsea did between 1933 (previous record) and 2005 is an interesting case, which raises the main issue that it seems the climateauditors have focused on. The tracks in 1933 were all in the western part of the atlantic, whereas there was a substantial number of tracks in 2005 in the eastern atlantic. Two factors account for this (undercounting was probably at most a 10%, 2 storm issue for 1933). The first is that the North Atlantic warm pool (defined as greater than 28C confined mainly to gulf and caribbean during 1933, whereas it now extends all the way to Africa. Storms are forming much further east as a result. Further, in 2005 the subtropical anticyclone was weak, and storms could easily travel northward in the mid atlantic. Weather maps in the 1930’s in the mid atlantic are of little use, but the 1933 tracks are consistent with a stronger anticyclone that steered the storms towards the U.S. and Caribbean (although this steering factor was less important in 1933 owing to the storms forming in the western atlantic).

    In spite of the errors in the earlier part of the record (cumulatively, this probably reflects random errors that may be as large as 20% in any one year), it seems that the main modes of climate variability (ENSO, AMO) can still be discerned from this data record.

    There are a number of papers that are forthcoming on this general issue. This number of atlantic tropical cyclones back to 1851 remains a useful data set. anyone using the data set should mention the uncertainties in the earlier part of the data set (in the review process, i recommend that a paper be rejected if it does not mention these uncertainties).

    The North Atlantic intensity data prior to 1970 is still subject to great debate, there is some hope that the intensity data from the 1950’s and 1960’s can be reliably sorted out, but this hasn’t happened yet.

  137. Steve McIntyre
    Posted Jun 30, 2007 at 9:48 AM | Permalink

    #137. Speaking of review processes, Judith, some that I presume is one of your associates wrote a disgraceful review of a submission that I and a coauthor made to GRL. (I presume this by the citations in the review). The two reviews were comic and represented much of what is wrong with climate science. One reviewer said that the results were already well-known and accepted in the literature and that the article should be rejected on that count. The other reviewer said that the results were all wrong and even “fraudulent” if you can imagine such a term being used. The editor rejected the article based on this “consensus”. Obviously one review completely contradicted the other – they could both be wrong, but they could not both be right.

  138. David Smith
    Posted Jun 30, 2007 at 10:02 AM | Permalink

    Hello, Judith, good to hear from you. Your comments raise some interesting and engaging points. At the moment I’m reading Mooney’s book (for a critique), which occupies my free time, but I plan to visit several of the points you raise.

    Steve M, perhaps you could start Judith’s comments (#137) as a thread in a few weeks when the storm season becomes a more-visible subject. Even though CA is suffering from “hurricane fatigue” there may still be some interesting aspects to be explored, especially if statistics can be used to examine old vs recent patterns in the available data with regards to Judith’s points.

  139. GMF
    Posted Jun 30, 2007 at 11:08 AM | Permalink


    Yet another example of consensus in climate science. If there is any point of agreement (both reviewers recommending rejection) that is a consensus – all else (totally contradictory reasons) is irrelevant.

    Honestly, if three IPCC authors sat down to lunch and they all had the tuna on rye, the headlines would read “IPCC meeting reaches Consensus on the Future”.

  140. Kenneth Fritsch
    Posted Jun 30, 2007 at 11:09 AM | Permalink

    Anyone care to explain the following long term trends in NATL named storms and hurricanes?

    Named storms ===> Significantly increasing trend.

    Hurricanes ===> Increasing trend but to a significantly lesser extent.

    Landfall events ===> No trend.

  141. Judith Curry
    Posted Jun 30, 2007 at 1:40 PM | Permalink

    Re #138: Steve, I suspect that you are referring to the Pielke and McIntyre paper that was submitted/rejected? I know for a fact that Webster did not review this paper (and I certainly didn’t); I mainly heard about this paper and its rejection via Pielke’s posts on the tropical listserv. If this was submitted to GRL (I think it was?) we have been frustrated by the fact that we are virtually never asked to review any of the hurricane/global warming papers submitted to GRL, for which we would have been able to identify key flaws and make suggestions for improvements. Peer review is a bit of a crap shoot in terms of who you draw for reviewers, we have certainly received our share of reviews that didn’t make any sense, to put it charitably. Most editors honor requests by authors not to submit a paper to a few specific reviewers if a credible reason is given. Peer review isn’t perfect, but important contributions eventually manage to get published somewhere if the author perseveres.

  142. Judith Curry
    Posted Jun 30, 2007 at 2:18 PM | Permalink

    Re #141: Kenneth, re the long term trend in the existing NATL data base. If one accepts that the recent uptick is a result of AMO + global warming, why isn’t the global warming signal (increase since 1995 relative to previous active period .ca 1950s) apparent in hurricane # and most especially in U.S. landfall #?

    The discrimination between hurricane and tropical storm prior to the 1950’s is arguably part of the overall uncertainty in intensity determination. Particularly in the earlier part of the record, there was typically insufficient information to distinguish between tropical storms and hurricanes.

    If we look at the data since 1950, when the discrimination between TS and hurricane should be credible, the recent decade does show an increase in hurricanes relative to the 1950’s, but the increase is not as great as that seen in total tropical cyclones. We have hypothesized that the signal from global warming in the north atlantic manifests itself in a change in the distribution of tropical cyclone intensities: more tropical storms, and more cat 4+5. this remains difficult to sort out owing to the debate over intensities of major hurricanes prior to 1970 (the landsea 1993 paper), but hopefully that issue will be sorted out soon

    with regards to landfall numbers, i think the explanation is more straightforward. As the north atlantic warms, not only does the average sea surface temperature change, but the gradients change (this issue is discussed at length in forthcoming paper by Kossin which can be found at

    Click to access Kossin_Vimont_BAMS_2007.pdf

    The Atlantic warm pool (>28C) now extends to the coast of africa, and tropical cyclones are forming more to the east. The changes in SST gradients have changed the atmospheric dynamics in the region and so the steering currents. We are working on sorting out the landfall locations as a function of different climate regimes (AMO, ENSO, NAO) in the context of the changing SST gradients (AMM plus the east west gradient)
    For more info on the atlantic warm pool issue, see peter websters spring AGU presentation

    Click to access Webster.2007.CharneyLecture.AGU.pdf

    In any event there is much work to be done to sort all of this out, but there are plausible data quality and physical arguments that could explain this.

  143. Judith Curry
    Posted Jun 30, 2007 at 2:22 PM | Permalink

    one final point to raise re hurricane tracks. One of my students, Mark Jelinek, has been adopting your “audit” strategy regarding the HURDAT database, continuing to identify ongoing undocumented changes to the data base. I am collecting all of the things that we have found so far and send to the HURDAT group, will post this maybe next week when I have time to sort through it all. The HURDAT data is a moving target, and there is much going on that is not being documented. The data auditing is an important issue, and kudos to steve and climateaudit for pushing hard on this issue.

  144. Judith Curry
    Posted Jun 30, 2007 at 2:25 PM | Permalink

    Re Chris Mooney’s book, look forward to David’s review. Here is a copy of a review i posted at (the fine print says it takes 5-7 days to appear)

    To provide a frame of reference for this review, I and my colleagues Peter Webster and Greg Holland are among the scientists that are featured prominently in Storm World. Our involvement in the issue of hurricanes and global warming began when we published an article in Science shortly before the landfall of Hurricane Rita, where we reported a doubling of the number of category 4 and 5 hurricanes globally since 1970. When Chris Mooney first approached me with his idea for writing a book on this topic, I was somewhat skeptical. I couldn’t see how this could be accomplished given the rapid changes in the science (I was worried the book would be outdated before it was published), the complexities of the technical aspects of the subject, a concern about how the individual scientists would be treated and portrayed, and a concern that the political aspects of the issue would be handled in a partisan way. Over the course of the past year and a half, it became apparent that Mooney was researching this issue extremely thoroughly and was developing a good grasp of both the history and technical aspects of the subject. Upon finally reading the book, I can only say Storm World has far exceeded any hope or expectation that I could have had for a book on this subject. The book is surprisingly rich in technical detail, and Mooney has grasped the nuances of the breadth of scientific arguments and uncertainties. He provides a fascinating history with rich insights into the current controversy. The individual scientists are portrayed accurately as well as sympathetically and colorfully. The political aspects are treated in an insightful and nonpartisan manner. I am most impressed by the fresh insights provided by this book, which besides being a “good read,” Storm World is an important and timely contribution that deserves careful consideration in the dialogue and debate on hurricane policy in the U.S. Storm World is science journalism at its absolute best.

  145. Posted Jun 30, 2007 at 2:30 PM | Permalink

    Dear Dr. McIntyre,

    If one of my colleagues reviewed your paper I am not sure who it was. It was not me as I make it a rule to sign papers that I suggest be rejected or re-submitted to major revision. Perhaps that is why I have a diminishing number of friends. But I do remember Pielke Jr. posting the paper on the Tropical Storms list-serve. I stated my surprise on that list-serve at that time that the paper seemed to have the same statistical flaw of the Emanuel papers that you had criticized in that the end points of the time series appeared to be held constant before smoothing. But it was a quick read.

    I have never been sure what the agenda of has been, or if it does have one at all. But I think you believe that there should be some integrity in the statistics used in climate analyses. And I applaud that wholeheartedly. So it was a bit of a surprise to find that perhaps there may have been some failing in the analysis in the P-M paper.

    Has the paper been resubmitted with corrected statistics?

    Like Judith, I think that the peer review process is useful process especially as the journals allow you to suggest reviewers and the names of those who perhaps should not review a paper. Also, new ideas are the ones that are the hardest to get published and perhaps that is the way it should be. If a new paradigm is proposed then it should have the greatest scrutiny before it is allowed to replace the old. I guess that is Kuhn at work!

    Keep up the good work and resubmit the article.

    Peter W

  146. Posted Jun 30, 2007 at 2:42 PM | Permalink

    One other point. I think that the field can argue forever about whether or not there has been an undercounting of 1 or 2 storms in 1920, 1943 or 1885. And perhaps we can never know. But I must say that this “climate arithmetic” should not replace basic science and enquiry. I am afraid that it has with all of the acrimony of the “defense of the last paradign”. Perhaps it is time to move on. The dual interaction of climate and tropical cyclones has rested on the Gray (1968) paper. And a great paper it was at the time. But with large advances in geophysical fluid dynamics do these relationships found by Gray 40 years ago still hold? And if they do, is there a physical basis for them? I gave the Charney Lecture at the spring AGU meeting in Acapulco and chose to talk on this issue. If anyone is interested, you can find the talk at and click on “presentations’. I will leave it up for another week or so. Comments are welcome on the dynamics and the statistics, of course.


  147. bernie
    Posted Jun 30, 2007 at 6:58 PM | Permalink

    I for one appreciate Judith’s and Peter’s civilized and informed contributions to the discussion. Side by side they also present an interesting, if somewhat nuanced, contrast in the need for establishing a solid empirical foundation for building models that can predict the future. Perhaps I will have to buy and work the Mooney book after all.
    Steve, is the paper you referred to posted here?

  148. David Smith
    Posted Jun 30, 2007 at 7:39 PM | Permalink

    Over the next week I hope to finish reading Mooney’s book and comment on Judith’s post and Peter Webster’s presentation. I’m like a kid in a candy store!

    I did put Mooney down and scanned Webster’s recent presentation. My initial take is that it is good, thought-provoking science (though I have problems with some of the early portions on Michaels, Hoyos and SST, which I’ll cover on CA). It is void of what I call “AGW sound-bite science” and it is kind to Bill Gray which, considering all the acrimony, had to be a very tough thing to do.

  149. John Norris
    Posted Jun 30, 2007 at 9:13 PM | Permalink

    re David Smith #73

    I didn’t see a Curry or Webster entry in the 2007 tropical storm contest. Surely with all that research and resulting TS knowledge they would be favorites. Any 2007 TS predictions in any of their papers?

  150. David Smith
    Posted Jun 30, 2007 at 9:55 PM | Permalink

    Re #150 John, I’ve seen nothing. Perhaps like Meteo France, “Un homme sage prévoit après le fait”.

  151. Judith Curry
    Posted Jul 1, 2007 at 2:06 PM | Permalink

    re #150: we are not making seasonal hurricane forecasts. we are working on a decadal forecast, and also ways to improve the predictability of the genesis of individual hurricanes. I stand my comments of about a year ago that the coupled climate models being run in seasonal mode by the European centers (UKMO, ECMWF, METEO FRANCE) should give better forecasts that the statistical forecasts being provided by Gray, NOAA, etc. Unfortunately, only UKMO has made their forecasts public this year (they are predicting below the 10 year average), and an anectdotal statement from ECMWF said that they are not predicting an overly active season, i interpret this to be lower than the noaa/gray forecasts. my assessment of the 2005 season (without an actual forecast): looks like we are headed into a week La Nina situation. SSTs are somewhat cool (certainly relative to 2005) in the MDR and we don’t see the massively warm deep Gulf Loop current this season. The Saharan Air Layer (dust issue) seems typical this year, with episodic dust outbreaks, this should die down in the next month, so we are not seeing the massive dust influence that we saw in 2006. The subtropical anticyclone is relatively strong, so i doubt that we will see the situation in 2005 where many storms formed in the east atlantic and swung north (more likely to see storms moving into the caribbean/gulf and up the east coast of the U.S.)

  152. bernie
    Posted Jul 1, 2007 at 2:34 PM | Permalink

    In my experience reviews on Amazon normally appear within 24 hours. Yours is still not up – currently there is but a single review.

    P.S. I think it is smart to stay out of the yearly prediction business. The competition here reflects the general skepticism at CA about the claimed precision of the GCMs. IMO whatever is happening with climate, based on current documented trends, will be shifting gradually on a global basis. Since the mechanisms generating hurricanes are at least regional if not global, trends will emerge overtime with existing variability largely continuing. I would be interested if your own work suggests some kind of tipping point where the number and intensity of hurricanes would increase dramatically.

  153. Posted Jul 1, 2007 at 3:10 PM | Permalink

    There is a paper in the pipeline by the ECMWF group that looks at the perfomance of their coupled ocean-atmopshere climate model during the last 25 years or so. Compares their skill wiith the Gray scheme during this period over the 30 or so year period. I remember that they explain about 50% of the variance (seasonal correlation numbers of storms numbers about 0.75) compared to Gray 15-20% of the variance (correlations 0.45). It is somewhere in the GRL system under revision.

    To answer Mr. Norris’s question (#150) that given “all that research and resulting TS knowledge” and why I am not involved in your competition, the answer is simple. “…all that research and resulting TS knowledge…” has led me to follow the track record! So, if I were silly enough to become involved in a TS-Atlantic competion, I would side with the Europeans and their 10-12 storms. I have too much fun forecasting Bangldesh floods and also Indian ocean hurricanes! ( and The latter refer to TS risk out to 10 days so that may not interest you. The flood forecasts are in three bands 1-10 days, 20-30 days (currently under revision) and seasonal 1-6 month. Something a little different for you to chew over.

    Peter W

  154. Posted Jul 1, 2007 at 3:34 PM | Permalink

    Yesterday was the first time I had revisited your blog in something like 6 months. But Judith Curry pointed out a submission by Professor Pielke Jr. of a month or so ago where he stated: “…..Looks to me like Holland/Webster have been doing more than a little bit of lurking at CA and Prometheus, judging by this presentation…..”. [This] presentation at the AMS meeting [was] based on a paper that had been accepted some months before in the PRoySoc. …

  155. David Smith
    Posted Jul 1, 2007 at 4:07 PM | Permalink

    Peter Webster recently posted his Charney lecture Powerpoint (see #147). It addresses aspects of tropical cyclones and climate.

    It’s a good presentation. There are some new points to ponder, such as estimations of heat transport by tropical cyclones and (for those who like meteorology) a section on genesis. It contains no bombshell claims and there’s no sound-bite science. There should be no controversy.

    The presentation is more conceptual than historical data-driven and is weighted towards meteorological concepts, so it may be of minimal interest to most CA readers.

    Towards the front is an acknowledgment of the role of Bill Gray in developing the science, which is a nice touch.

    I almost always have questions and comments, and this is no exception 🙂 . This of course is a Powerpoint and unaccompanied by words, so I haved missed some context and clarification. Nevertheless, here are my thoughts as I looked through the slides:

    Slides 13 to 16. For background, Michaels et al discusses the relationship between the SST beneath a hurricane and its intensity, with circa 28C being a threshold for major hurricanes. Michaels shows that if 28C+ SST is not present, then it is quite unlikely that a storm will achieve major status.
    Slide 14 states (red box) that, “many major storms, especially in the NATL, appear to form at SSTs less than 28C” and slide 16 shows dots of major hurricanes outside the 28C isotherm. The problem is that those dots are where the seedlings first attained tropical cyclone status and are not where the storms transformed into major hurricanes. Since Michaels is concerned with the latter (where the storms attained major status). so the dots on slide # 16 seem meaningless.
    A alternate plot, of the dots where the storm attained major hurricane status, shows a concentration in 40W to 50W, with only three cases of major status east of 40W. These revised dots are much closer, and often inside, the 28C isotherm.
    Given that, I see no evidence for bullet (ii) on slide 16.

    Slides 17 and 18. Slide 17 shows the eastward progression of the 28C isotherm over the 20’th Century, implying progressive oceanic warming. However, this graphic is sensitive to the exact half-decades chosen, because of the effect of AMO (see the Curry/Webster plot of SST on slide 24). For example, move the years backwards by a decade (1910-15;1950-55; 1990-95) and see if the latter part of the progression holds – it doesn’t.
    A better, apples-to-apples comparison would be to compare SST in similar parts of the AMO cycle. Same for slide 18.

    Slides 21 and 22. These show rather progressive warming of the world’s tropical oceans. I’m trying to reconcile these plots with this presentation , which shows changes in Pacific and Indian Ocean Warm Pool areas. I realize they are measuring somewhat different things and timeframes, but what I see is a Pacific warm area with little/no expansion over the 20’th Century and an Indian Ocean warm area which varies greatly in what may be a natural cycle. The Atlantic also has cyclical behavior, so I wonder how much of the oceanic warming shown on the slides is due to the Atlantic and Indian Ocean cycles being in-phase at this point in time.

    Slide 24. I think that a plot (upper left corner) of tropical SST and storm count which extend back to 1855 should have error bars. Errors in storm count is a well-discussed issue and tropical SST may be especially problematic, given that it was a relatively untraveled region over much of the period.

    Slides 26 to 29: Ooyama looks like the better criteria.

    Slide 30: I can’t tell what the red dots signify or the time period. I also wonder how accurate/precise is the climatology shading.

    Slides 31 to 35: Good section on tropical atmospheric stability in a warming environment. I wonder if the atmospheric profiles generated by this modeling are consistent with output from the GCMs.

    Slide 38: There are a couple of side issues with regards to Hoyos which I’ll mention here. One, quite minor and inconsequential, is that I never understood why the SST plots in Webster (2005) and Hoyos (2006) are not completely identical. The second, of more importance, is that I wonder what Hoyos’ conclusions would look like if Kossin’s recent results on intensity were used. Would the third bullet on slide 38 still hold? Would the revised Hoyos show that, actually, little has changed in the overall tropical atmosphere in recent decades?

    Slides 43 to 69: These are interesting meteorological slides on storm genesis. I grasp the significance of maybe 70% of the content and will work towards a higher percentage.

    Slides 70 to 78: These I recommend as worth a look by everyone, as they estimate the role of tropical cyclones in moving heat from the tropics towards the poles.To move so much heat towards the poles with such (relatively) small amount of cirrus output makes me wonder about some of Lindzen’s ideas on tropical cirrus behavior.

    OK, back to Mooney. I’m keenly interested in his description of the events of 2005 but I don’t want to jump ahead.

  156. David Smith
    Posted Jul 1, 2007 at 4:19 PM | Permalink

    Our contest is a spoof. The forecasts generally fell into one of two categories: 1950-1995 climatology (9 or 10 storms) or the 1996-2005 average (14 or 15 storms), which is kind of interesting.

  157. Gerald Browning
    Posted Jul 1, 2007 at 4:32 PM | Permalink


    There is a flaw in the logic of your last question and a simple answer that anyone can see if they think about it for more than a second.
    If you want to state someone’s science is weak, you might mention those who used flawed data and methodology on hurricane counts related to AGW.
    Give me a break.


  158. David Smith
    Posted Jul 1, 2007 at 4:44 PM | Permalink

    Re #141 Bernie, I think much of the explanation lies in changes in detection (of both existence and intensity) over the decades, for reasons discussed at length here. I am not sure that all the parties involved realize how much of an (educated) guessing game occurs on intensities, and thus classifications, even today. A read of storm reanalysis reports, available at the NHC website, can illuminate the varying and conflicting data that the forecasters deal with.

    I’ll offer some detail later, some of it a rehash, and try to do it in the order of Judith’s points.

    For now, back to Mooney.

  159. Posted Jul 1, 2007 at 4:59 PM | Permalink

    Dr. Browning

    This is the reason I tend to avoid blogs in all flavors. Please remember that I was commenting on the comments of someone that makes a living on the interpretation of other peoples intent. So where is the non-sequitor? The paper was weak, amen! Other papers are weak as well but do I need to list those in the context of a flawed paper? Note that the PM paper uses the same (“flawed”) data used in a number of other studies.

    Judith tells me that blogs are only useful when the comments are worthwhile. The trick is not to respond to the silly and the biased. Not in my nature so the best thing I can do is to take another hiatus and get back to what I do best.

    Peter W

  160. Posted Jul 1, 2007 at 4:59 PM | Permalink

    Dr. Browning

    This is the reason I tend to avoid blogs in all flavors. Please remember that I was commenting on the comments of someone that makes a living on the interpretation of other peoples intent. So where is the non-sequitor? The paper was weak, amen! Other papers are weak as well but do I need to list those in the context of a flawed paper? Note that the PM paper uses the same (“flawed”) data used in a number of other studies.

    Judith tells me that blogs are only useful when the comments are worthwhile. The trick is not to respond to the silly and the biased. Not in my nature so the best thing I can do is to take another hiatus and get back to what I do best.

    Peter W

  161. Kenneth Fritsch
    Posted Jul 1, 2007 at 5:14 PM | Permalink

    This is the reason I tend to avoid blogs in all flavors.

    I agree. Some people should avoid blogs and it has been my experience that it is those who tend to get caught up in the heat of the moment. If you are not contributing new information from your specialty or arguing the salient points in a technical discussion and merely responding to real or imagined personal attacks, you are better served not wasting time at a blog.

  162. Kenneth Fritsch
    Posted Jul 1, 2007 at 5:22 PM | Permalink

    Our contest is a spoof.

    David Smith, thanks for pointing to the obvious. The selections were made, I thought, in a non-serious manner that someone would make for a non-monetary related Super Bowl pick. I think my strong male ego could withstand being wrong and wrong big time ‘€” and particularly since my selections were based on other group’ predictions. Well maybe I do have an ego problem.

    Posted Jul 1, 2007 at 6:28 PM | Permalink

    Peter! Just eyeballing the 20 first slides of your
    Acapulco presentation it seems that the area that
    could generate TS/TH has increased more than the
    actual number of TS/TH! Am I right?

  164. Kenneth Fritsch
    Posted Jul 1, 2007 at 6:33 PM | Permalink

    Re: #143

    Thanks, Judith, for your reply to my queries about the relative long term trends of NATL located named storms (NS), hurricanes (H) and landfall events (LE) and the trends going from significantly increasing for NS to increasing for (H), but significantly less than for NS, and finally having no trend for LE. The simple explanation for these relationships is that the trends follow the ease with which these categories could have been found and categorized over the long term time period of interest. I think an alternative and most likely much more complex explanation has to be clearly and unequivocally made to the replace this simple one.
    The other point, of course, that is immediately evident is that it is primarily the land fall events that cause damage and would, I assume, go to the top of the list for explanations and analyses with regards to the no trend in frequency phenomena and a better understanding of any potential changes in intensity. Otherwise, I have the same problem with the scientific attention given to this topic as I do for the man looking for his keys under the street lamp.

  165. Gerald Browning
    Posted Jul 1, 2007 at 10:48 PM | Permalink

    Peter (#161),

    I think people like you avoid blogs like this because in an open forum
    all of your illogical statements can be discussed and scrutinized instead of “reviewed” by your peers (buddies). I started to post on Steve’s blog because the points I wanted to make could be made here in an open forum and judged by their mathematical relevance, not by the flawed peer review system.

    I have backed up all of my statements with mathematics (e.g. see Exponential Growth thread). If you can find a flaw in the mathematics, feel free to do so. After a considerable and lengthy discussion with Jimy Dudhia at NCAR (one of Holland’s lackeys), it came out in the end that he admitted that the numerical models are a tuning nightmare and refused to run two simple tests to show the serious flaws in numerical models based on the hydrostatic or nonhydrostatic equations.

    Before you call anyone’s work weak, how about someone look at your best piece of work in more detail. Do you have a suggestion as to which piece that might be?


  166. John A
    Posted Jul 2, 2007 at 1:08 AM | Permalink

    Don’t forget the John A forecast for the Atlantic:

    6 named hurricanes + or – 3

    You heard it here first.

  167. David Smith
    Posted Jul 2, 2007 at 5:12 AM | Permalink

    RE #163, #167

    The other thing about the contest (but tell no one, it’s secret) is that I plan to subtly adjust the storm count and rules as we go along, to assure that my prediction is closest.

    He who keeps score, wins. 🙂

  168. David Smith
    Posted Jul 2, 2007 at 5:16 AM | Permalink

    By the way, I’m halfway through the Mooney book and so far, so good. He’s a talented writer who (so far) bends over backwards to avoid painting anyone as a villian or hero(ine).

    I’ll be offline until I finish reading it.

  169. Pat Frank
    Posted Jul 2, 2007 at 4:02 PM | Permalink

    #146 — “Also, new ideas are the ones that are the hardest to get published and perhaps that is the way it should be. If a new paradigm is proposed then it should have the greatest scrutiny before it is allowed to replace the old. I guess that is Kuhn at work!

    Better theoretical frameworks — not paradigms — get accepted because they explain more of the observables and at a deeper level of theory than did the previous ones. Typically, a new improved explanation can make sense of data that for years have been lying unexplained and usually (for that reason) ignored in the literature.

    When a new explanation from theory does that, it is typically accepted fairly quickly. Kuhn was basically wrong about science. It’s not a political process, nor a consensual one, nor a generational one.

    Those intellectual traditions that show political preferences or that follow generational fashions are ipso facto identifiable as not science.

  170. Bob Koss
    Posted Jul 3, 2007 at 11:27 AM | Permalink

    5,000 year hurricane record.

  171. Steve McIntyre
    Posted Jul 3, 2007 at 11:32 AM | Permalink


    I have never been sure what the agenda of [SM- org] has been, or if it does have one at all. But I think you believe that there should be some integrity in the statistics used in climate analyses.

    Peter, the mere fact that you think is possible that does not have an agenda (or that, if it does, you do not find its agenda obvious) is something that will surprise many people and that I find quite gratifying. While I find it very difficult to locate climate science arguments that are convincing and while I find many advocates to be promotional, I don’t dismiss the issues on those grounds, although some readers do. I’ve had business experience with mining promoters and just because someone is a promoter and not particularly reliable, doesn’t mean that his property is worthless.

    However, you’re quite right that one of its objectives is the use of valid statistical procedures and I have no objection to people applying such standards to anything that I do. Despite much damning evidence in connection with various studies that I’ve criticized, I try to be careful in the language that I use and I was surprised at the angriness of the GRL reviewer. This type of outburst usually says more about the character of the reviewer than the paper being reviewed.

  172. Bill F
    Posted Jul 3, 2007 at 5:56 PM | Permalink

    FWIW, I appreciate both Dr. Webster (even if he is back on hiatus) and Dr Curry for taking some time to post their thoughts here. I think if you look past some of the off-topic accusations and such that are common to any open forum on the net on almost any topic, you will find that most people here do have a very earnest desire to explore the science and learn more about what they don’t know. I think “audit” is a great one word title to describe the “agenda” here. I haven’t seen any evidence to suggest that Steve M. has gone after anybody or accused anybody of wrong doing without first trying very hard to get their input and assistance in understanding what they have done in their research. Those who deal with Steve and Climate Audit on a respectful level and try to help explain why and how they came to their conclusions are generally rewarded with a return of that same level of respect. Those who stonewall or treat this community with disdain or disrespect earn the same in return.

  173. David Smith
    Posted Jul 3, 2007 at 9:33 PM | Permalink

    Ken asked in #141 about Atlantic trends and what might be driving them. To offer a bit more detail on my opinion I’ve revised an auditblog article ( link ) from May and added one more plot.

    The plot at that post breaks storms into one of three categories:
    * those that affected the US or Canada, where detection and record-keeping has been rather good
    * those that affected Latin America or The Bahamas, where record-keeping may have been spotty prior to the late 1800s
    * those storms that stayed entirely at sea

    The US/Canadian category shows little or no increase.
    The Latin American storm count shows little change after the 1880s
    The At-Sea storms showed little change until WW2, after which a rise began as aircraft, crude satellites, modern satellites/quikscat and then Doppler Radar and buoys came into use.

    It’s also interesting that some oscillation, apparently consistent with the AMO, appears.

    The second plot removes an odd type of storm which I believe is a child of modern detection methods. These odd storms are very weak (40 knots or less) and exist for less than a day. Weak, short-lived storms have winds that cover only a small area, because their wind field has not had time to expand. They are quite hard to detect by the occasional ship and none appear in the older (pre-WW2) record. But, thanks to things like Quikscat and Doppler radar, they are now detected and recorded.

    The second plot shows even less increase, if any at all, except in the entirely-at-sea category.

    Comments and suggestions appreciated.

  174. Roger Pielke Jr.
    Posted Jul 4, 2007 at 3:58 AM | Permalink

    … my mother always used to say that if you can’t say anything nice, don’t say anything at all, it is good advice). ..

    Best regards,

    Roger Pielke, Jr.

  175. bernie
    Posted Jul 4, 2007 at 7:45 AM | Permalink

    The above interesting interaction, emphatically illustrates the point made throughout history (e.g., Newton vs Leibnitz, Franklin vs. Crick/Watson) that doing science has many of the same dynamics as making sausages, i.e., messy and blood drenched (metaphorically, of course). With science, however, everybody pretends that scientists are dispassionate, objective, and even unworldly. Clearly they are not, nor should we expect them to be. The most interpersonally dysfuntional meetings I have ever particpated in were at the NSF – nobody appeared to listen to each other and all came with their minds apparently wide shut! It was quite eye-opening.

  176. Judith Curry
    Posted Jul 4, 2007 at 8:40 AM | Permalink

    …As i posted at climateaudit some months ago on this issue (probably earlier on this thread), Greg Holland’s analysis on east vs west atlantic storms was motivated by a talk that Chris Landsea did on this general topic many months before your analysis. Chris Landsea has now published something in EOS on this topic. Greg Holland to my knowledge has never been to climateaudit and in any event he tends to disdain blogs in general.

    The idea of looking at east vs west genesis tracks probably first belongs to Landsea (in any event, he was apparently the first person to publish on this, see his EOS article). Ideas aren’t copyrighted, graphs published in refereed journal publications are. Presentations made at scientific meetings aren’t copyrighted. If someone listens to talk at a conference and is then motivated to do further analyses of their own on this general topic, well I would say that this what scientific conferences and presentations are intended to accomplish. When I get an idea to do a scientific analysis, it typically does not come from from a vacuum, but it is stimulated in part by my consideration of analyses that other scientists have done. I then reference any relevant papers or conference presentations in a scientific paper. In the acknowledgements, I would typically mention conversations with individuals that may have been influential and/or helpful in working out the science. In my experience, both Webster and Holland have been generous in including other scientists as coauthors and in mentioning helpful contributions in the acknowledgements. It appears that they did not regard anything that you did as influencing what they were doing, and in fact Webster stated in his earlier post that he was critical of what you did.

    Your own analysis (that you submitted for publication) seems to have been heavily influenced by the exchange going on in the blogosphere, particularly on climateaudit. anything posted on climateaudit is not copyrighted (although individuals choosing to copyright something on a blog could probably do so by a statement accompanying the graph that this is copyrighted). My own posts in the blogosphere are intended to stimulate other people to think about these issues, and I assume that this is SteveM’s intent in hosting this blog. I have been giving away alot of ideas here for “free” (you can judge for yourselves as to whether anything I have posted here is of actual value to the climateauditors or other lurkers). I choose not to get involved in the realtime analysis of data in the blogosphere (i believe this has rather annoyed Margo at times) largely for the reason that I choose to do my own scientific analyses in the traditional copyrighted manner of the peer reviewed journals.

    The blogosphere potentially opens up an important new venue for moving the science forward on highly relevant and rapidly evolving scientific topics. The dialogue has often been interesting and provocative nevertheless, and I have found many posts to provide food for thought and have helped sharpen my critical thinking skills on the subject. I haven’t yet seen anything here on climateaudit on the hurricane topic that I have found to be of scientific value other than general chastisements on the issue of statistical significance testing. This chastisement is well taken (by me anyways). If it ever turns out that I write a paper where my ideas are influenced in an important way by the dialogue on climateaudit, I would include climateaudit in the acknowledgements. At one point I even invited Bender to coauthor a paper with me…

  177. Roger Pielke, Jr.
    Posted Jul 4, 2007 at 9:06 AM | Permalink


    … the advantage of posting pre-publication papers on the internet is to create a record for exactly this type of situation (unfortunately). …

    [It is incorrect] that I circulated the Pielke-McIntyre paper on the TS list…

    One great irony in this is that a basis for rejecting the Pielke/McIntyre paper (as reviewed by [someone who appeared to be associated with Holland and/or Webster]) was that the subject of “landfall proportions” and “easting” had already been discussed by Webster/Holland in their 2007 AMS presentation, so the claim was that our work was not new (ahem).

    I am sure the readers and posters here will especially welcome this comment of yours: “I haven’t yet seen anything here on climateaudit on the hurricane topic that I have found to be of scientific value …” So why do you keep coming back?

  178. David Smith
    Posted Jul 4, 2007 at 9:26 AM | Permalink

    Must be a full moon.

    Something unpleasant is also happening at the US National Hurricane Center this weekend. One of the people with inside contacts posts at this link , which should be interesting reading over the next several days. Jeff Masters also plans to post on the same subject later today. I’ll provide links as things develop.

  179. Kenneth Fritsch
    Posted Jul 4, 2007 at 12:37 PM | Permalink

    Re: #174

    Comments and suggestions appreciated.

    I enjoy reading your analyses, David Smith, and find them more insightful than I get from Drs. Webster and Curry here (although I must read in detail the two papers linked by Judith before completely passing judgment on what she has offered recently). I suspect, that as Judith and Peter indicate, some people do better writing peer reviewed articles than holding forth in the blogosphere. That situation is somewhat disappointing to me since I believe I have gotten a good insight into NATL TS and hurricane frequencies from the postings here at CA, but would like to hear in simple language what the “increasing NATL storm frequency with SST” school of thought has to say.

    I have referenced my basic understanding with credits given to many posters here at CA in past posts and will repeat that understanding without credits here. The long term trends for NATL for TS, hurricanes and landfall events (where none is shown) can be explained most simply by an increasing ability over time to detect and categorize these events. By removing the trends and cyclical components of these events one can use a Poisson distribution with a time invariant mean to describe their past frequencies as indicated by a chi square goodness of fit tests.

  180. David Smith
    Posted Jul 4, 2007 at 3:46 PM | Permalink

    Re #180 Ken I calculated the correlation coefficient for entirely-at-sea Atlantic storms and Atlantic tropical SST. It’s sometimes suggested that higher SST are leading to more storms, so I thought I’d check the data. I double-smoothed it to increase my chances of a strong correlation.

    The correlation coefficient for those two series (for 1955-2005) is 0.004


    I’ll narrow this check to the eastern Atlantic (east of 60W).

  181. Posted Jul 4, 2007 at 3:49 PM | Permalink

    One thing I find surprising about the debate on global warming and Atlantic hurricanes is the lack of discussion on the North Atlantic oscillation (NAO). A weak preseason NAO phase tends to favor tracks that follow lines of latitude. In contrast a strong NAO phase tends to favor tracks that cross latitudes (hurricanes that, in general, get steered away from the U.S. coast). We speculate the reason for this is related to the position and strength of the subtropical high pressure system during the season.

    Interestingly, the NAO was in a positive phase for much of the 1970s and 1980s with historic highs in the early 1990s and speculation about a link to global warming has been made. Osborn et al. (1999) show that the NAO from the 1960s to early 1990s is outside the range of earlier variability in the instrumental record and also outside the range of variability simulated using UK Hadley Centre’s numerical model. Thus with greater warmth and perhaps more Atlantic hurricanes it is possible that the threat to the United States as defined by the probability of a strike will remain relatively constant rather than increase.

    In fact there is some evidence for this in the historical record of U.S. hurricane counts which show no long term trend but a tendency for a smaller ratio of landfall counts to basin-wide counts. The differential influence of improvements in observing technologies on landfall and total counts tends to confound attempts to understand this tendency as noted in Elsner and Kara (1999). Moreover, conditional on the phase of the NAO, there are statistically significant positive relationships between Atlantic sea-surface temperature (SST) and both U.S. hurricane counts (Elsner and Jagger 2006) and insured losses (Jagger et al. 2007).

    Originally posted at

  182. steven mosher
    Posted Jul 4, 2007 at 4:30 PM | Permalink

    Re 178.

    RPjr. I like your work, but if you are going to pee in the pool and raise the tempature,
    you should probably hang around awhile.

  183. Steve McIntyre
    Posted Jul 4, 2007 at 4:44 PM | Permalink

    #182. Dr Elsner, thanks for posting here.

    To others, JAmes Elsner is a prominent authority on hurricanes.

  184. Posted Jul 5, 2007 at 5:50 AM | Permalink

    Re 175 and etc.


    Seeing we are sharing emails, here is the one you sent to me. Appended below. Note “co-authorship” suggestion. Just to set the record straight.
    Yes, I erred. i should have been honest in what I thought of the paper. Even though I was busy I thought that to use that as an excuse rather than what I really thought about the paper and the offer (I don’t allow my name to be on papers I have nothing t do with). ..

    Peter W

    X-Sieve: CMU Sieve 2.2
    Date: Thu, 21 Dec 2006 09:07:07 -0700 (MST)
    From: Pielke Roger A
    To: peter webster
    Subject: comment for your reaction or co-authorship
    X-Greylist: Message not sent from an IPv4 address, not delayed by milter-
    greylist-2.0.2 ( []); Thu, 21 Dec 2006 09:07:07 -0700
    X-GT-Spam-Rating: (7%)
    X-GT-Spam-Details: No antispam rules were triggered by this message
    X-GT-AVAS-Version:, Antispam-Engine:, Antispam-Data:


    The attached is a very short analysis that was motivated by your recent
    paper, but which expresses an analysis that I have been mulling for some

    To be clear, it is _not_ a critique of your work. It seeks to contribute
    to your “hurricane artimetic” exercises by distilling some implications
    that seem thus far not to have been discussed in the literature. In fact,
    the analysis uses the exact same assumptions and data that you use in your
    work. It provides no insights on theory or attribution.

    I am sharing this with you to get your feedback, but also to ask if you
    would be interested in co-authoring. Note that I have shared this draft
    with several other members of our community for comment.

    I do ask that you don’t pass it on to anyone else at this point as it is a
    draft and I am uncertain what I will do with it. It is a serious,
    sceintific effort to discuss an aspect of this issue that has yet to be

    Let me know if you are interested in looking at it or possibly

    Best regards,


  185. Roger Pielke Jr.
    Posted Jul 5, 2007 at 7:17 AM | Permalink

    Thanks Peter! That email of mine would indeed seem to set the record straight, and especially the part where I observe of the paper that I shared with you,

    “It seeks to contribute to your “hurricane arithmetic” exercises by distilling some implications that seem thus far not to have been discussed in the literature. . . It is a serious, scientific effort to discuss an aspect of this issue that has yet to be discussed.”

    .. , are you ever going to share “what you really think” about the paper, rather than leaving is shrouded in mystery? … and here is the link:

    … well, let’s hear the science!

  186. Judith Curry
    Posted Jul 5, 2007 at 9:05 AM | Permalink

    For anyone interested in the issue of landfalling proportion, here is the relevant statement from Holland/Webster presentation:

    “The high proportion of landfalling storms in the pre-satellite era has been used to imply poor observations of cyclones in the eastern Atlantic (e.g. Solow and Moore, 2002, Landsea 2006).”

    In googling “landfall proportion” i came across some interesting links that i hadn’t found before:

    Click to access Tavener.pdf

    Click to access 0701176v2.pdf

    Click to access 60HUActivityRates_whitepaper.pdf

  187. John Lang
    Posted Jul 5, 2007 at 9:18 AM | Permalink

    Even though the actual hurricane numbers show multi-decadal variation and no clear trend overall, it is pretty clear that Holland, Webster and Curry will continue pushing their view that North Atlantic hurricane numbers are increasing sharply and their intensity is increasing sharply.

    They will continue arguing that this trend is caused by increasing ocean temperatures which is likely due to global warming even though there is no evidence for this at all.

    Despite appearing civil on the surface, they will continue to use their passive-agressive-argumentative-condescending-questioning the competetence of the work of anyone else in the hurricane research field approach that is evident in the above posts. I imagine this extends to the anonymous peer-review process as well.

    I don’t believe it would be productive to collaborate with them Roger and I would continue trying to get the actual data and the actual analysis published despite the above approach of the above.

  188. Kenneth Fritsch
    Posted Jul 5, 2007 at 10:44 AM | Permalink

    James Elsner in post # 182 does what I prefer when someone with professional knowledge joins the discussion in that he gives his personal summary of the important points in the debate.

    Judith Curry lists some links to papers she has googled but fails to give any of her summarized views and points on the topic. I went to these links of Judith’s and extracted below what I thought best summarized what they had to offer.

    Peter Webster’s posts here have not (as yet) seriously broached any scientific issues.

    First link:

    Preliminary results (as yet unpublished) suggest that the hypothesis that the probability of storms making landfall doesn’t change in time cannot be rejected.

    Second link:

    it is not acceptable to wait until the effects of the trend are well understood before commenting on the possible implications consider scientific evidence on climate change with regards to parameter setting Lloyds ICA guidance 2006

    Third link:

    We have compared the likely performance of direct and indirect methods for predicting landfalling hurri-cane numbers from SST. The direct method is based on building a linear regression model directly from SST to landfalling hurricane numbers. The indirect method is based on building a regression model from SST to basin numbers, and then predicting landfalling numbers from basin numbers using a constant proportion…

    .. Which method should we then use in practice? If we had to chose one method, our results seem to imply that we should choose the indirect method, since it is more accurate. The simulation results suggest, however, that the performance of the two methods is likely to be very close for the values of the parameters appropriate for hurricanes in the real world. Given the possibility to use two methods we would use both, as alterative points of view.

    Ideally we would also be able to solve the more realistic model analytically, as we have done for the linear-normal case. We are working on that.

    Fourth link:

    Atlantic hurricane activity has remained persistently high since 1995 (except for El Nià±o years).
    In 2004 and 2005, the high activity in the basin translated into U.S. landfall and the highest loss
    years ever experienced by the insurance industry. With strong evidence that higher than average
    activity rates are likely to persist for at least a decade, it is no longer appropriate to employ a longterm historical baseline for characterizing medium-term activity rates in hurricane catastrophe (Cat) models. Acknowledging that the long-term historical baseline is no longer the best measure of
    current activity also means it is necessary to be explicit about the intended time horizon of Cat
    model activity rate projections.

    Fifth link:

    This dissertation examines the historical record of hurricanes and tropical storms in
    the Atlantic Basin to determine the eventual landfall probability for the United States
    coastline based on the complete tracks of the storms. A spatial dimension is added so that the entire basin is evaluated to determine which storms in all portions of the basin ultimately strike the United States. A tessellation of 3,375 hexagons are systematically evaluated and eventual landfall probabilities are calculated for all storms passing through each hexagon. Probabilities are calculated and mapped for each of twelve states and regions from Texas to Maine. The maps show spatial areas that contribute storms to each of the twelve states and regions. Additionally, an average length of time until landfall is calculated for the entire Atlantic Basin based on the complete period of record. This highlights regions of the Atlantic Basin lying outside of the maximum forecast period ‘€” up to 15 days prior to potential landfall.

    Sixth link:

    We have investigated whether there is a statistical relationship between MDR SST and the number of hurricanes making landfall in the US. In previous work we’ve seen a strong relationship between MDR SST and the number of Atlantic basin hurricanes, and so our a priori assumption is that there must be some relationship for landfall numbers as well. Our analysis, however, finds only a weak relationship for total hurricane numbers and no relationship at all for intense hurricane numbers.

    Why could this be? We see two possible explanations:

    ‘€¢ There is a physical relationship between SST and landfalling hurricane numbers, but this relation-ship is mostly obscured by the signal-to-noise ratio, which is very poor because there are so few landfalling hurricanes.
    ‘€¢ Even though there is a strong physical relationship between SST and basin hurricane numbers,
    there is only a weak physical relationship between SST and landfalling hurricane numbers. The
    effects of SST conspire to change the proportion of hurricanes that make landfall in such a way that the effects that are seen in the basin numbers almost disappear when we consider the landfalls. For instance, higher SSTs may mean higher numbers of hurricanes in the basin (on average), but they may also mean a lower proportion making landfall, and these two effects may combine in such a way that the actual number making landfall remains the same.
    And of course reality may be a combination of these two effects.

  189. David Smith
    Posted Jul 5, 2007 at 11:36 AM | Permalink

    Re #189 Kenneth, I briefly read the sixth link at lunch, which explores whether higher Atlantic SST means more landfalling hurricanes (their answer was a qualified “no”). It would have been interesting if they had taken two additional steps: (1) use detrended SST in their analysis and (2) use SST contrast (MDR SST vs tropical South Atlantic SST).

    Had they checked those, they may have gotten statistically significant results (my guess). Such results would reflect the action of natural factors (AMO types) on hurricane frequency.

  190. David Smith
    Posted Jul 8, 2007 at 11:39 AM | Permalink

    Re #137

    2) double counting prior to 1900. If you look at the actual tracks, you see that some tracks are parallel, close together, and with overlapping dates. we are in the process of documenting this and will submit to the HURDAT committee

    Judith, I went through the 19’th century storm tracks given at the Unisys website. I looked for situations as described, where it is possible that a storm got counted twice. What I found was one year (1887) where two storms look suspiciously alike and two other years where a double-count is less likely but plausible.

    For the two 1887 storms to actually be just one storm counted twice requires a misdating by 5 days, while the other two require both mislocation and some amount of misdating.

    If all three involve duplicates and result in three storms being removed from the database then the 19’th century storm count gets reduced by 0.06 storms per year, which is small potatoes compared to the

    1-2.5 (the underestimation varying in time, larger errors in the 1850’s)

    for undetected storms mentioned in item #1.

  191. David Smith
    Posted Jul 9, 2007 at 8:45 AM | Permalink

    Post #137 looks at possible undercounting and overcounting of Atlantic tropical cyclones. My goal is to look at order of magnitude of the possible errors, and see if they do offset. In this post I take a look at item #3:

    3) spurious inclusion of subtropical storms in the database, which would contribute to overcounting particularly prior to 1950

    A couple of easy-to-read references on subtropical storms are here and here . It’s important to avoid mixing subtropical and tropical systems in studies, because their origins and characteristios are different. But, separation can sometimes be difficult and sometimes studies, including well-known ones, inadvertently mix the two.

    For my purposes subtropical storms are formed above 20N and do not reach hurricane intensity. Tropical storms are mostly formed at lower latitudes and obviously can reach hurricane intensity. The bases for this distinction can be found in the references.

    Do historical records (ship reports) provide enough information to allow reanalysis? Generally no, though subtropical storms have their strongest winds separated from their lowest pressure, while the strongest winds in tropical cyclones are found at the center. It might be possible in some cases to use this distinction to separate the two. In most cases we need satellites and aircraft recon to distinguish between the two.

    Is there a way to estimate an upper limit on how many subtropical systems might be in the records? Possibly. For 1970 to the present, the US NHC includes in its records those subtropical storms which had characteristics close to those of tropical storms and which could reasonably be misidentified (my assumption) as tropical in nature. If we divide the number of subtropical storms (1970-2006) by the number of overall storms which might be subtropical (those that form north of 20N but do not reach hurricane intensity) we get about 20% of the group which might reasonably be misidentified.

    Then we take that 20% estimate and apply it to the storms in earlier periods, to see what kind of impact a misidentification might have:

    For 1901-1950 I count 87 storms in the pool (formed north of 20N but did not reach hurricane intensity). Twenty percent of 87 over 50 years is about 0.35 storms per year that could be misclassified. For 1851-1900 I get about 0.25 storms per year.

    For the period 1970-2006 the rate is 0.5 to 0.6 storms per year. For 1951-1969, the NHC record seems to be clear of subtropical storms, as the inclusion into the NHC records began about 1970 after years of discussion and study (see reference).

    Now, I note that my methodology for estimating the pool of possibilities (those north of 20N and which did not become hurricanes) is probably broader than it needs to be, as earlier storm tracks often did not go back into the deep tropics (below 20N) and, for reasons I won’t get into here, the Gulf of Mexico percentage is overstated. But, our the purposes here I leave those alone.

    What does all this mean? Well, the main message is that the effects of misidentification appear to be small. I see no problem, for study purposes, of using scenarios where an adjustment (reduction) of around 0.3 storms per year is applied to pre-1950 periods. But, people doing the studies also need to be sure to exclude the 0.5 or so subtropical storms per year inadvertently included in some studies for 1970-2005. Apples to apples is the goal, as best as can be done.

    Next, the misclassification question.

  192. Kenneth Fritsch
    Posted Jul 9, 2007 at 10:05 AM | Permalink

    Re: #s 191 and 192

    Good work again, David Smith — and you are so damn polite.

    But, separation can sometimes be difficult and sometimes studies, including well-known ones, inadvertently mix the two.

  193. Steve Sadlov
    Posted Jul 9, 2007 at 11:51 AM | Permalink

    RE: #135 – There is still an element of subjective interpretation. Let us face the facts. The NHC really have to tap dance and bend and form the data to be able to portray Barry as a TS. If Barry was a TS, then so too were innumerable cyclones which have impinged on the West Coast of the US. A bit a problem here. Opens a real can of worms.

  194. David Smith
    Posted Jul 9, 2007 at 12:34 PM | Permalink

    Re #193


    Ken, my view is that the advocates, with the possible exception of Emanuel, now acknowledge that hurricane science is not settled. They seem to be avoiding the bold public statements and sound-bite science of a couple of years ago and I hope that continues.

    If it does, if they avoid the media and ideological posturing while they work out the science, then internet irritants like us will fade away. I for one am considerably more interested in water vapor and ocean circulation and a variety of measurement topics than I am in subtropical storms and HURDAT revisions and would rather spend my free time on those.

    I’m not totally convinced, however, that they won’t give in to the temptation to make an AGW/hurricane statement to the press the next time a bad storm hits Miami and the BBC calls. I just don’t know, so I try to stay up to date on their possible angles of argument.

  195. David Smith
    Posted Jul 9, 2007 at 12:37 PM | Permalink

    RE #194 Steve, at times I wonder if the current NHC director and his apparent meltdown is somehow involved in how Barry was classified. I know that sounds bizarre but it’s a bizarre world right now at the NHC.

  196. Steve Sadlov
    Posted Jul 9, 2007 at 1:01 PM | Permalink

    RE: #196 – Until we get more facts about the meltdown, I would certainly not discount it. This would actually be a good new topic – namely, just what is the operational definition of a TC? Some initial thoughts to frame the discussion – there is a continuum of cyclonic behavior with a “pure” TC, which started as an Easterly Wave and went straight into “classic” TC mode, at one end of the spectrum. At the other end, are “pure” polar jet originated cold core systems. The problem area of course is in the transition zone. Here in Cali, we’ve gotten “polar” storms which dipped way down by Hawaii, and hit our coast with overt eye structures and siginificant characteristics of TCs. Same thing happens when a low trundles along the jet stream, then gets cut off over Mexico, then dawdles off of Yucatan, retrogrades as if it were Easterly Wave originated, wobbles a bit, then heads north east. An agreed operational definition, perhaps excluding storms lacking a certain “pedigree,” would simplify classification and determination of whether or not to name / count as a TC. The time is right for it.

  197. David Smith
    Posted Jul 9, 2007 at 3:22 PM | Permalink

    I offer five charts and three comments, to help illustrate part of the measurement problem.

    The charts are from US National Hurricane Center final reports on individual storms. It may take a minute or two of looking at a chart before the meaning becomes apparent.

    The chart shows the change in maximum winds of a storm over the life of the storm. Each of the little dots ot symbols represent, at a point in time, the windspeed estimate from a satellite method, aircraft method, ship, land, buoy and sometimes radar. The heavy black line is the best-guess by an analyst, weighing the different pieces of information, of the true maximum windspeed.

    The charts are for two entirely at-sea storms, two famous landfalling storms and a recent (2006) storm.




    Ernesto (2006)


    1. Notice how, at a given point in time, a forecaster is faced with varying and conflicting indications of windspeed. He or she must make a best-guess and pick just one. It’s part objective and part subjective even with the best modern tools at hand.

    2. Notice how the windspeed varies over time, and the critical importance of taking samples throughout the life of the storm if one is to capture the maximum strength of the storm. If one has only one or two data points, which is common pre-1950 especially for at-sea storms, then one has probably missed the peak and thus underestimeted the storm strength. It’s a basic principle of sampling.

    3. Notice how different techniques give different estimates and imagine that, as often happens, techniques come in and out of favor, and get adjusted, over time. That creates the possibility of apples to oranges comparisons and spurious trend changes.

    4. Improved sampling tends to increase the reports of tropical storms and hurricanes, consistent with what we see in the records, as the peaks are captured. For example, the number of reported storms whose life is less than 24 hours has grown notably as sampling has grown.

  198. Steve Sadlov
    Posted Jul 9, 2007 at 4:03 PM | Permalink

    RE: #198 – In retrospect, designating Irene and Ernesto as hurricanes may have been questionable. Only the satellite figures are responsible for pulling the best guess curve above the threshold. There seems to be a systemic high bias with satellite data. I’d personally go with the drop sondes.

  199. Steve Sadlov
    Posted Jul 9, 2007 at 5:19 PM | Permalink

    This will surely put you to sleep:

    Even the normally hyperactive, longer season Pacific has been fairly quiet thus far.

  200. David Smith
    Posted Jul 9, 2007 at 5:39 PM | Permalink

    RE #199 There was controversy over naming Ernesto a hurricane last year because the supporting data was rather weak. In the final report the poor writer had to do some dancing to support the 65 knot estimate and I suspect that his/her nose grew an inch or two, like Pinocchio. I suspect the naming happened because the storm was approaching land (Haiti) and the NHC erred on the side of caution, which is their obligation.

    Re #200 Patience, grasshopper.

    In the worth a look category is the real-time satellite animation of the current Pacific typhoon ( link ). The clouds stretch the equivalent of Toronto to Miami. The dark colors at the center are cloud tops probably at -80C (likely into the tropopause). Headed towards Japan.

  201. Steve Sadlov
    Posted Jul 9, 2007 at 7:06 PM | Permalink

    RE: #201 – RE: The typhoon – looks to me like it may strike the Philippines. Worst case scenario would wobble it just north of them and take it in for a direct hit on Hong Kong. Say, didn’t they make a TV show about that? ….. and a CA specific note, those cloud tops …. gee, I wonder if that sort of mechanism is factored into the GCMs (he asked naively)?

  202. Steve Sadlov
    Posted Jul 9, 2007 at 7:08 PM | Permalink

    BTW – I was in South China for a couple typhoons last year. Typhoons are to South China as summer Cold Fronts are to the Midwestern US. Dime a dozen (but mostly below Cat 3 on landfall).

  203. David Smith
    Posted Jul 14, 2007 at 12:53 PM | Permalink

    It’s mid-July and time for a contest update. If I have messed up or missed your forecast, please let me know. As before, forecasts involving ranges were converted into midpoints, though I’ll restate the ranges at the end of the season.

    John A.: 7
    John G. Bell: 8
    Paul Linsay: 9
    Steve Sadlov: 9
    1950-1994 average: 9
    uc: 10
    John Norris: 10
    John Baltutis: 11
    UK Met: 11
    Bob Koss: 12
    jae: 12
    IWIC: 13
    Accuweather: 14
    Jonathan Schafer: 14
    IW (private forecast firm): 14
    Steffan Lindstrom (on behalf of Swedish Hurricane Centre): 14
    1995-2006 average: 14.5
    DeWitt Payne: 15
    US National Hurricane Center: 15
    Michael Mann: 15+
    TSR: 16
    Gray/Klotzbach: 17
    David Smith: 17
    Bill F: 18
    Ken Fritsch: ((GK+MF)/2)
    Meteo France: secret

    As of mid-July, we have had only one named storm, and a dubious one at that. Slow start.

    For the period 1995-2006 (the recent active phase of the AMO), seasons with zero or one storm by July 15 averaged 13.5 storms for the entire season. The highest was 2001, with 15 storms while the lowest season had 12 storms. Over the last 100 years the most-active season that started slowly was 1969, with 17 storms.

    It’s beginning to look bleak for the Apocalyptics, including me and Michael Mann. However, the average CA participant forecast is 11.8 storms, so things are looking promising for the group.

    (Of course, what will count is the adjusted end-of-season number, as the raw count will need some adjustments and homogenisations.)

  204. steven mosher
    Posted Jul 14, 2007 at 2:59 PM | Permalink

    Oh I forgot to put in my guess. I guess 13

  205. David Smith
    Posted Jul 14, 2007 at 6:56 PM | Permalink

    RE #205 An astute prognostication, Steven, and remarkable timing 🙂

  206. steven mosher
    Posted Jul 14, 2007 at 8:05 PM | Permalink

    I was gunna guess 13.5, but figured you guys would whine when I want to count the first
    storm as a .5

  207. David Smith
    Posted Jul 14, 2007 at 9:00 PM | Permalink

    Re #207 But the whining would be nothing compared to the howls you’ll hear when Michael Mann and I hockey-stick the data at the end of the season so as to validate our now-unlikely forecasts.

    Just as with global temperature, “He who keeps score, wins!”

    Seriously, those closest to the final number (and subtropical storms don’t count) will get a certificate of wizardry, suitable for framing. That includes the UK Met should they win, who were sporting enough to make their estimate public at the start of the season.

  208. TAC
    Posted Jul 14, 2007 at 9:25 PM | Permalink

    If steve mosher isn’t going to take, I’ll place my bets on the average, 13.5.

  209. Philip B
    Posted Jul 15, 2007 at 5:03 AM | Permalink

    FYI, there has been a marked reduction in cyclones around Australia over the last 40 years.

    BTW, the NOAA launches a new hurricane outlook product tomorrow.

  210. steven mosher
    Posted Jul 15, 2007 at 8:25 AM | Permalink

    Here is my Handy dandy AGW prediction.

    GW will cause one of the following:

    1. more frequent storms OR
    2. More intense storms OR
    3. More landfalling storms OR
    4. More intense landfalling storms OR
    5. A freakishly strong storm

    Some year in the future!

    I like bets you can’t lose.

  211. Kenneth Fritsch
    Posted Jul 15, 2007 at 10:21 AM | Permalink

    Ken Fritsch: ((GK+MF)/2)
    Meteo France: secret

    David Smith it appears as time goes on in the hurricane season that in the absence of a Meteo France prediction I might have to assume their number as zero and use 1/2 of the Gray/Klotzbach predicted number. I will let you know how my thinking evolves on this issue as the season proceeds.

  212. MarkW
    Posted Jul 15, 2007 at 6:50 PM | Permalink

    I make it a point to never release my predictions until the season is over.
    Amazingly enough, my record is perfect.

  213. David Smith
    Posted Jul 15, 2007 at 7:43 PM | Permalink

    Re #212, 213 Spoken with wisdom. Reminds me of a movie: The Sting.

  214. steven mosher
    Posted Jul 15, 2007 at 9:41 PM | Permalink

    no past posting allowed at this roulette table. My bet was clearly made while the bal was still in play

    Now, what is the tie breaker? landfalls or Landfall proportion ( Tm RPjr)

  215. David Smith
    Posted Jul 16, 2007 at 5:04 AM | Permalink

    Well, one tie-breaker option is arm wresting. Another possibility would be to guess, er, forecast the November GISS Northern Hemisphere temperature anomaly. That tiebreaker would occur immediately after the end of the season and would qualify the winner as a tropical wizard, with a minor in crisp fall days.

  216. Steve Sadlov
    Posted Jul 16, 2007 at 10:16 AM | Permalink

    RE: #204 – David S – RE: “For the period 1995-2006 (the recent active phase of the AMO), seasons with zero or one storm by July 15 averaged 13.5 storms for the entire season. The highest was 2001, with 15 storms while the lowest season had 12 storms.”

    What were these stats for the past 100 years? (for what little the quality of that record period is worth ….. LOL!)

  217. David Smith
    Posted Jul 16, 2007 at 12:05 PM | Permalink

    Re #217 Since 1945 it is normal to get 1 named storm by August 1 and then another 9 by the end of the season. We’ve had one storm (sort of) and so chances are, based on long-term averages, we have 9 or so to go.

    Two notes: one, in the active phase of the AMO the annual numbers are higher, and we’re in the active phase, so the chances are that the season total will be greater than 9.

    Second, the first two months of the season really don’t say much about how the rest of the year will go, only that the chances of a hyper-active season are near-zero.

  218. Steve Sadlov
    Posted Jul 16, 2007 at 12:52 PM | Permalink

    This year is a test of the AMO theory.

  219. David Smith
    Posted Jul 16, 2007 at 3:42 PM | Permalink

    The eastern Pacific saw its first hurricane of 2007 today. This is the fifth-latest date for a first hurricane over the last 40 or so years. A slow start.

    For the Atlantic, Jeff Masters posted his two-week outlook today ( link )

  220. steven mosher
    Posted Jul 16, 2007 at 3:55 PM | Permalink

    SteveS. I am kinda slow. AMO theory? Can you spell it out for me..

  221. jae
    Posted Jul 16, 2007 at 4:45 PM | Permalink

    From Chris Landsea/NOAA:

    While only about 60% of the Atlantic tropical storms and minor hurricanes ( Saffir-Simpson Scale categories 1 and 2) originate from easterly waves, nearly 85% of the intense (or major) hurricanes have their origins as easterly waves (Landsea 1993). It is suggested, though, that nearly all of the tropical cyclones that occur in the Eastern Pacific Ocean can also be traced back to Africa (Avila and Pasch 1995).

    It is currently completely unknown how easterly waves change from year to year in both intensity and location and how these might relate to the activity in the Atlantic (and East Pacific).

    So we know that the easterly waves cause hurricanes, but we don’t know much about the easterly waves. Ergo, we still don’t know much about the formation of hurricanes. But we can certainly still blame them on AGW.

  222. steven mosher
    Posted Jul 16, 2007 at 5:56 PM | Permalink

    RE 222.

    Hits Jae with AGW cave man logic stick.

    Wind Strong; Sea Warm; Socks wet. AGW is to blame.

  223. Steve Sadlov
    Posted Jul 16, 2007 at 7:05 PM | Permalink

    RE: #221 – The AMO theory is the theory that positive phase AMO begets, in the absence of significant shear (e.g. El Nino) active seasons in the Atlantic and its adjuncts. A competing or clarifying theory to counter it may be that one must also consider other factors such as PDO and the state of the Indian Ocean as well, and that the simple AMO plus ENSO forumulation is insufficient. This year, we have AMO positive, and ENSO neutral to negative. According the AMO theory it should be more active than the 100 year mean. My bet, of 9 named storms, was a specific bet against the AMO theory and for a more complex set of factors.

  224. John Lang
    Posted Jul 17, 2007 at 8:21 AM | Permalink

    For you hurricane watchers, the following NOAA site has the most up-to-date SST anomaly information.

    The July 16th map shows mild La Nina conditions, cool southern ocean conditions, cool eastern Atlantic-hurricane-formation-region conditions but the Gulf of Mexico and the Carribean are warming up. Overall, it appears that ocean SSTs are below normal right now.

    But one should keep in mind that ocean temps change fairly rapidly on weekly scale so the last 2-month animation is good to look at too.

  225. Jonathan Schafer
    Posted Jul 17, 2007 at 10:06 AM | Permalink


    I’m a bit dispairing in my forecast. Reading Dr. Master’s link confirmed what I had seen…Namely, that the Sahel region had for the second year in a row received above average precipitation. My assumption, thus proved wrong so far, was that the amount of dust over teh Atlantic would be reduced this year, leading to higher Central Atlantic SST’s than has actually occurred this year. That was one of the inputs into my forecast. Well, we haven’t hit August yet, so we’ll see what happens. It’s not Cape Verde season yet anyway, and it’s that persistant trough of lowpressure in the Carribbean that seams to be preventing anything in the Carribbean and GOM from forming right now anyway. Too much wind shear from the sub-tropical jet. We’ll see if the GFS forecast holds true and the trough dissipates, thus lowering the shear over the region.

  226. Steve Sadlov
    Posted Jul 17, 2007 at 11:28 AM | Permalink

    RE: #226 – Try this on for size. Cold SSTs in formative areas. PDO shift to negative overwhelming the effects of La Nina and AMO. Indian Ocean temperature regime incurring some sort of major shift. Results in easterlies in excess of normal velocity over Sahel and southern Sahara. Dust wipes out TDs. Plus, as you mentioned, due to PDO shift and Indian Ocean shift, split jets are in abundance with southerly branches dipping into tropics. One additional effect of all this is that ITCZ gets narrowed and once you move only a few degrees north or south away from it, convection dissipates.

  227. Steve Sadlov
    Posted Jul 17, 2007 at 2:20 PM | Permalink

    Ho-hum ……

    Still, the seemingly endless African dust …..

  228. David Smith
    Posted Jul 17, 2007 at 9:22 PM | Permalink

    A few thoughts on an Atlantic storm measurement problem are here on a CA auditblog.

    The bottom line is that the Atlantic storm record for recent decades contains a type of storm (weak, short-lived tropical storm) which is absent from the records in earlier decades. This is due, I believe, to changes in storm detection and monitoring.

  229. David Smith
    Posted Jul 19, 2007 at 10:41 AM | Permalink

    Off-topic: I’ll be in New Orleans this weekend and will bring back some photos and stories on how the city is really doing in its recovery.

  230. SteveSadlov
    Posted Jul 24, 2007 at 11:30 AM | Permalink



  231. David Smith
    Posted Jul 24, 2007 at 2:32 PM | Permalink

    Kenneth Trenberth has weighed in with an article on Superhurricanes in Scientific American.

    I have not read the article yet but Jeff Masters has, and reports on it at his Weather Underground blog:

    Even Masters, who is solidly in the AGW camp, uses the word “hype” to describe much of it. Even his section on the “good science” is rather puny.

    Masters has called it as he sees it, in public and in clear language. I wonder if other hurricane specialists will do the same.

  232. David Smith
    Posted Jul 24, 2007 at 2:37 PM | Permalink

    Re #232 Oops, the man’s name is Kevin, not Kenneth. Sorry.

    This would make a good CA topic except that there’s probably very little data in the article.

  233. Steve McIntyre
    Posted Jul 24, 2007 at 3:02 PM | Permalink

    By this time in 2005, there had been 2 hurricanes with peak wind 135 knots and 140 knots respectively (and 42 hurricane quarter-days).

    All the hurricanes in 2005 reminded of the vorticies that you see behind an outboard motor boat moving along a lake – I’m not saying that there’s any validity to this comparison – just that it reminded me of it and the reason that my guess last year for a low season was based on the lack of early vortices roiling up the Atlantic. Given that July 2007 has also been quiet, my guess is that 2007 will be more similar to 2006 than to 2005.

  234. Kenneth Fritsch
    Posted Jul 24, 2007 at 5:45 PM | Permalink

    Re: #232

    Masters says:

    “Both theory and computer models predict a 3-5% increase in hurricane winds per degree C increase in tropical SSTs, and there is concern that the actual increase may be much more than this”.

    I am not much good at prediciting hurricanes and tropical storms but I am getting quite proficient at anticipating what the hot new climatology publications will be:

    Climate models that will predict significantly more intense hurricanes with increases in SST than previously supposed and predicted.

    Hypercanes will be discovered from past warm climates using sedimentation as proxies.

    Chris Mooney hinted at this in “Storm” and quotes Kerry Emanuel on both of these subjects. Hypercanes in the past and the future would make great marketing points for selling AGW mitigation actions.

  235. steven mosher
    Posted Jul 24, 2007 at 6:40 PM | Permalink

    235 Mann and Gavin already hinted at Hypercanes.

    Think of it as an analog to “warming n the pipeline”

    All those 2006 Hurricanes that couldnt form and transform STT into kinetic energy.. are just
    waiting in the wings.. so we might see HYPERCANES someday. Keeps fear alive.
    We might be Jupiter someday! My fear is that we will be Uranus.

    Enviro eschatolgy: supervolcano, mega tsunami, killer comet, hypercane, mega flood,
    super bird flu, mass extinction, grey goo, brain eating halatosis, dick Cheney.
    ( is that like a prince albert?) All on discovery channel. I swear every time I watch one of these disater things I expect a climatologist to come out and ask for tithing to fit Global warming.

    As an agnostic I preferred the christian eschatology. Tammy faye baker, rest her soul,
    was much more appealing than Dr Heidi Cullen.

    Now I will go look and see if Gore’s numerology translatess to 666, after TOBS adjustment and homogeneity rigging.

  236. tetris
    Posted Jul 24, 2007 at 7:18 PM | Permalink

    Re 231 abd 236
    Hype Canes, indeed… If the 2007 season turns out to be a repeat of 2006 or, heavens forbid, end with a lower cane count, no doubt a way will be found of linking the outcome to AGW as well, along the lines of what’s going on in the UK right now.

  237. David Smith
    Posted Jul 24, 2007 at 8:06 PM | Permalink

    Hypercanes is an idea floated by Kerry Emanuel in the mid-1990s. I recall reading about them years ago, perhaps one of those doctor-lobby magazines like Discover or Time. It’s a bit stale.

    Kevin Trenberth would get more traction with an article on hyper-tornadoes (700 mph winds, you can hear the roar 8,000 miles away, one could suck up all of Lake Michigan including the Edmund Fitzgerald). Or, as an alternative, a writeup on how Big Foot is getting bigger and meaner due to global warming.

  238. mccall
    Posted Jul 24, 2007 at 10:06 PM | Permalink

    It’s looking like we’ll have to wait til next year at least, for a hypercane:

  239. Gerhard H.W.
    Posted Jul 25, 2007 at 3:06 AM | Permalink

    David, please count /me in:

    12 named storms (including 3 bogus ones), 4 canes and one major

  240. David Smith
    Posted Jul 25, 2007 at 5:04 AM | Permalink

    RE #240 You’re in. We’ll have a special August 1 revision opportunity, with awards for the best start-of-season forecast and for the best start-of-August forecast. I’ll post a note on this.

    Re #238 Hypercanes is a stale story with little scare value so I suggest Trenberth use an alternate topic. My suggestion is Big Foot (who is also known as Sasquatch). Here is the story line:

    * Big Foot and Ms. Big Foot live happily in the woods until global warming
    * The warmth keeps them awake at night and they spend extra time making Little Big Foots (Feet?). It’s a runaway procreation situation.
    * The lack of sleep and extra Foots to feed make the family grumpy.
    * The extra Little Big Foots are spreading across the globe, coming soon to the woods near you.
    * Be afraid, be very afraid.

  241. Jonathan Schafer
    Posted Jul 25, 2007 at 9:23 AM | Permalink


    I think they should focus on how the bigfeet have to walk 60 miles uphill both ways in order to find food and that they are dying off due to changes in their environment. At one time, there were thousands of these creatures roaming free, but now, due to environmental changes caused by man, there are only a few left in the wild, and were recently added to the endangered species list in hopes of protecting them.

  242. Kenneth Fritsch
    Posted Jul 25, 2007 at 9:57 AM | Permalink

    Kevin Trenberth would get more traction with an article on hyper-tornadoes (700 mph winds, you can hear the roar 8,000 miles away, one could suck up all of Lake Michigan including the Edmund Fitzgerald).

    David Smith, I truly wanted to believe your Big Foot scenario, but since the Edmund Fitzgerald sank in Lake Superior I have reasons to doubt. Hypercanes, in my estimation, are just what the doctor ordered for getting the public’s attention and directing it towards the potential super adverse effects of AGW. Maybe Dr. Curry needs to get involved as she seems better attuned to getting public attention than some of her colleagues who appear less versed in the art of public discourse.

  243. David Smith
    Posted Jul 25, 2007 at 10:58 AM | Permalink

    Re #243 🙂

    I caught my error last night about a nanosecond after pressing Submit, but then decided if it’s in Scientific American, why does it have to be right?

    Perhaps it was a hypercane that sunk it.

    This evening I’ll post some photos I took in New Orleans last weekend. The city is forever changed, like Galveston Texas after its 1900 hurricane.

  244. SteveSadlov
    Posted Jul 25, 2007 at 1:16 PM | Permalink

    RE: “Tammy faye baker, rest her soul, was much more appealing than Dr Heidi Cullen.”

    You made me laugh so hard I almost busted a gut.

  245. steven mosher
    Posted Jul 25, 2007 at 1:44 PM | Permalink


    I was especially fond of that rant. By chance the other day while channel surfing I switched between Larry King, who had the racooned eyed Baker featured, and the weather channel which has the ectomorphic Cullen featured.

    I hope that were not drinking when you read my post. A nose enema is most unpleasent.

  246. Sam Urbinto
    Posted Jul 25, 2007 at 5:35 PM | Permalink

    How many words does it take to say “We had no data so we guessed using models.”?

    Ah, here it is:

    The extended reconstructed sea surface temperature (ERSST) was constructed using the most recently available International Comprehensive Ocean-Atmosphere Data Set (ICOADS) SST data and improved statistical methods that allow stable reconstruction using sparse data.


    high-frequency SST anomalies are reconstructed by fitting to a set of spatial modes

  247. D. Patterson
    Posted Jul 26, 2007 at 7:30 AM | Permalink

    And the beat goes on…”Worst of Atlantic hurricane season still to come By Jim Loney; Wed Jul 25, 10:47 AM ET. See the NOAA and other hurricane predictions in today’s news story at:

    One of the comments claims the El Nino in the Pacific will not suppress tropical storm and hurricane formation this year in the way it did last year. So, they are still forecasting a higher than normal season of tropical storm and hurricane activity in the Atlantic.

  248. SteveSadlov
    Posted Jul 26, 2007 at 11:38 AM | Permalink

    RE: #248 – Most are in denial about the mostly cool summer being experienced in the NH. The lone outposts of summer heat this year are normal and expected – The South Asian – African High Pressure belt and the smaller one in the southern half of the southwestern interior of North America. There has been sproacic heat in SE Asia inclusive of parts of China. There have been outbreaks of the Saharan heat into SE Europe. But by and large, this is a cool summer, especially, and critically, in the food growing areas. This cooling has apparently shut down the hurricane factories of the NH.

  249. Dave Dardinger
    Posted Jul 26, 2007 at 12:09 PM | Permalink

    re: #249

    …summer heat this year are normal and expected – The South Asian – African High Pressure belt and the smaller one in the southern half of the southwestern interior of North America.

    The 7 day forcast for here in Phoenix shows the highs under 100 every day which is pretty rare, I’d think, for the last of July – beginning of August. We did have it hot earlier in the month, however. But the top highs of 115 aren’t that rare in summer in the valley.

  250. Bill F
    Posted Jul 26, 2007 at 4:56 PM | Permalink

    Anybody want to give odds on whether the NHC will hang a name on the mass of clouds in the gulf around the high and call it a displaced SH tropical cyclone? If so, will they use a name from the SH list or the Atlantic NH list?

  251. Joe B
    Posted Jul 27, 2007 at 7:28 AM | Permalink

    Bill Gray is in the WSJ today talking about Hurricanes and the Atlantic’s circulation:


  252. SteveSadlov
    Posted Jul 27, 2007 at 1:03 PM | Permalink

    RE: #251 – impressive impact from a cold front that fits in more with Spring or late Fall than Summer …. so long as there are splits / deep southern loops in the polar jet like that, this season shall remain a dud.

  253. SteveSadlov
    Posted Jul 27, 2007 at 1:07 PM | Permalink

    RE: #252 – I continue my low ball forecast of 9 named storms. Here’s why I disagree with Gray. While Gray’s positive AMO + La Nina = more storms theory is a good first approximation, it has a notable blind spot. It fails to account for the ways that other things like the PDO and Indian Ocean’s indices can do things like send mid latitude weather deep into the tropics in the Western Hemisphere, NH summer, and, exacerbate African dust. Since these later two things are happening, and since PDO looks to have turned negative, and the Indian Ocean is all wierded out, I am sticking to my guns.

  254. David Smith
    Posted Jul 27, 2007 at 7:27 PM | Permalink

    Last weekend I traveled to New Orleans to see how the city has done since Hurricane Katrina. Some of my photographs and comments are given on the CA auditblog:

    New Orleans Notebook

  255. Kenneth Fritsch
    Posted Jul 28, 2007 at 10:21 AM | Permalink

    David Smith, that was a well written article on the status of New Orleans and the pictures (quality and quantity) were a bonus and something most frequently lacking in a newspaper and even magazine article.

    Even though some of my friends have been trial lawyers, I could not resist chuckling at your comment here:

    Recently a scandal developed when formaldehyde vapors were detected in several trailers, fumes known to attract trial lawyers.

  256. David Smith
    Posted Jul 28, 2007 at 10:32 AM | Permalink

    We’re approaching August 1, which marks the traditional start of the active period ( see this chart for how activity changes over the summer. Slope = activity.)

    Historically, the June/July activity has little relationship to the August/September/October activity. However we can say that the odds of a hyper-active total season (15+ storms) are very low when the first two months are weak like this year.

    So far wind shear has been near-normal, the Caribbean TUTT has been near-normal, SST are near-normal and African dust/dry air has been somewhat above normal. These traditional indicators have been, as Steve Sadlov says, ho-hum.

    What has impressed me, and Steve Sadlov too, has been the lack of strong seedlings in the ITCZ. (I believe this is also what Steve M talks in his analogy of swirls (vorticity) in a boat wake.) The ITCZ worldwide has been rather inactive for much of the summer, especially in the Pacific. The Atlantic ITCZ has been a dud, too, compared to recent years.

    So, we come to August 1, and the time when the professional visionaries offer a revision of their June 1 forecast. That opportunity is open to CA participants, too. We’ll keep our start-of-season contest and simply add a second August 1 contest.

    My guess (er, forecast) is that we’ll see 12 to 15 total storms (rounded to 14) this season, which a revision downward from my 15 to 18 storm forecast made back in April. I continue to believe that things are in place for an active season, with the mystery being what has happened to the storm seedlings.

    We’ll close this revised-forecast opportunity on August 3.

  257. David Smith
    Posted Jul 28, 2007 at 2:33 PM | Permalink

    Re #256 Ken one thing that makes me pessimistic about New Orleans is the number of small businesspeople who are making escape plans. They are currently financially bound to the town but want out. They are the ones who could bring energy, innovation and organization to New Orleans but many have mentally departed the city and their bodies and assets will follow over the next five to ten years.

  258. David Smith
    Posted Jul 29, 2007 at 10:23 AM | Permalink

    So far in 2007 the western Pacific has seen five tropical cyclones. This is below normal, with only about 10% of the years since 1970 showing five or fewer storms by the end of July. In those slow-start years the seasonal totals averaged around 20 storms, which is about 7 below normal.

    Now things can change in a heartbeat and storm seasons don’t always stick to the calendar. But the odds are the western Pacific, which is the most active region on the globe, will have a below-normal year.

    The odd thing about the western Pacific has been the relative lack of seedlings, similar to the Atlantic.

    The two other NH (Northern Hemisphere) cyclone regions (Indian Ocean and eastern Pacific) have been near-normal so far.

    In summary, it looks like 2007 is shaping up as a near-normal year in the Northern Hemisphere, which follows a near-normal storm season in the Southern Hemisphere.

    We’ll have a much clearer picture about the NH in about six weeks.

  259. Jonathan Schafer
    Posted Jul 29, 2007 at 7:09 PM | Permalink

    Anyone seen this yet?

    Study blames climate change for increase in Atlantic hurricanes

    Holland and Webster have a new study out. No surprise what their conclusion is, or the ringing endorsement from Kerry Emmanual.

    Landsea says technology changes leading to greater discovery of hurricanes is the real reason. I doubt there’s anything new in the report.

  260. Gerhard H.W.
    Posted Jul 30, 2007 at 8:58 AM | Permalink

    Re #260 To my opinion, compared to Grays WSJ today statement (#252), many Climatologis(mis)ts are playing a new type of the Will Rogers Game: Climatemandering

    .oO(Has anyone said cherry picking behind me?)

  261. Posted Jul 30, 2007 at 10:21 AM | Permalink

    Holland and Webster have a new study out.

    I’ve seen that new paper ( ) before ( )

  262. Bob Koss
    Posted Jul 30, 2007 at 10:25 AM | Permalink

    The observation quality was discussed a couple months ago, and in my opinion they’re comparing apples and oranges. Storm count can’t be as valid as using track level information. There’s a huge difference in sample size. A storm may have lasted 5 days or 15 days, but using tracks allows you to tell how long the weather was bad.

    Here’s a comment with some graphs that break things down by track.
    It’s easy to see how the quality has changed from noting the wind speed when the storms were discovered and when they were lost. Further down is a comment with bar graphs showing where the observation quality has improved. It’s the middle of the ocean.

  263. Kenneth Fritsch
    Posted Jul 30, 2007 at 12:08 PM | Permalink

    Re: #263

    The observation quality was discussed a couple months ago, and in my opinion they’re comparing apples and oranges. Storm count can’t be as valid as using track level information. There’s a huge difference in sample size. A storm may have lasted 5 days or 15 days, but using tracks allows you to tell how long the weather was bad.

    That the authors are claiming no more than a broad brush approach in their analysis is enough to make one question the results and conclusions of this paper, but I find that even more telling is that the authors fail to discuss in any detail the evidence for a long term improving detection and categorizing of tropical storms and hurricanes: from tracks as you have indicated in your post, the variation in detection over time with location as discussed by Steve M, a number of other indicators as noted in various posts by David Smith, the constancy of the land falling events, and increasing trends rates with category over time as the difficulty for detection for that category increases. Combine this with the failure to discuss in detail Kossin’s reanalysis showing since the 1980s that frequencies and intensities of tropical storms and hurricanes has not increased worldwide. That one can find part of one of many tropical storm regions in the world where the SST correlates well with a storm characteristic for the whole region would come as no surprise to anyone who has done any data snooping looking for such relationships and having to find rationalizations for them later.

  264. David Smith
    Posted Jul 30, 2007 at 3:52 PM | Permalink

    Re #260

    Here are two charts which I hope illustrate several points in the debate. I will try to stay general:

    Chart 1

    The time series shows Atlantic storms broken into two detection categories –

    1. Those which came within 100 miles of land (including islands) and thus could be detected from land.
    2. Those which stayed well at sea and never came within 100 miles of land, requiring detection by ships until perhaps the 1930s and 1940s, after which time aircraft and then satellites could detect them.

    Chart 1 shows that the land-detected storms have a cyclical pattern (AMO-related) with a modest peak-to-peak rise over 140 years.

    The ocean-detected storms show little change until the advent of aircraft and especially satellites, after which time they steadily rose.

    That’s the storm-detection method aspect.

    There is a second aspect of the debate which has to do with continuous monitoring and with improvments in intensity-measurement tools (aircraft, satellite, radar, buoys). Storms strengthen and weaken, sometimes quickly. In older times these intensity changes were likely missed. Today, they are captured. In older times brief surges in strength to minimal storm strength for short periods were often missed. Today, they are not missed.

    Below is a chart in which I removed storms which existed for 18 hours or less. These very brief, weak storms are captured today but were likely missed prior to satellites, aircraft, radar, etc.

    Chart #2

    This shows that, when these brief storms are removed, the land-detected storms show little peak-to-peak trend over the last 140 years, and what little increase there is may be due to 1870s record-keeping or may be due to AGW. The ocean-detected storms are what increased.

    Anyway, with regards to the Webster/Holland news release, I liked two quotes today –

    From Margie Kieper at Weather Underground, the leading tropical weather website:

    An extraordinarily large number of news media were advanced a preview of a paper due to clear the publication embargo today, providing a new twist on analysis of North Atlantic tropical cyclone activity in relation to SST, and it seems all of them carried a story on it. Most tout a sensational headline about global warming increasing hurricanes

    Yes, a big PR campaign.

    On the content, here’s Chris Landsea of the National Hurricane Center:

    The work, he said, is “sloppy science that neglects the fact that better monitoring by satellites allows us to observe storms and hurricanes that were simply missed earlier. The doubling in the number of storms and hurricanes in 100 years that they found in their paper is just an artifact of technology, not climate change.”

    And who paid for this?

    The research was funded by the National Science Foundation.


  265. TCO
    Posted Jul 30, 2007 at 10:53 PM | Permalink

    138 and 175 (McIntyre and Pielke): These comments about review battles are off-topic for the thread. Confine them to Unthreaded or start a new thread. “Jumping” on Curry et al is not appropriate when off topic to the thread.

  266. Bob Koss
    Posted Jul 30, 2007 at 11:50 PM | Permalink


    Interesting way to look at the data.

    How did you arrive at the 18 hour figure? I assume the storms you removed from chart #2 are Frieda, Amelia, Edouard, Gabrielle, and Kyle between 1977-1996. The shortest of those is six tracks of data. All located in the Gulf or Carribean.

    Your chart doesn’t included 2005, but two storms that year had only five and six tracks of data.

    During the 1851-1900 period 31 storms had only one track of data. Many of those being Cat. 1 or higher far from land.

  267. Gerhard H.W.
    Posted Jul 31, 2007 at 7:20 AM | Permalink

    Watch out for the next bogo-cane

  268. John G. Bell
    Posted Jul 31, 2007 at 8:19 AM | Permalink

    Does “TS” Chantal have any rotation at all? National Bogocane Center? Come on! Perhaps I really am looking at the wrong thing…

  269. John G. Bell
    Posted Jul 31, 2007 at 8:32 AM | Permalink

    OK, now I see it. Not very tropical.

  270. Jonathan Schafer
    Posted Jul 31, 2007 at 8:51 AM | Permalink

    This is the reason why I will not modify my count for the season. They seem bound and determined to include every little thing to the count. If this isn’t an example of the undercounting that would have existed in the years before satellite and/or plane observation, then maybe nothing is.

    David, I’m not going to revise my count, although maybe I should…upward. The NCH seems hell bent on counting nearly every puff of clouds, no matter how short lived and lacking in so many standard tropical characteristics. Despite the slow start, I’m actually wondering if I should increase my count, based on the NHC’s propensity for naming nearly everything.

  271. Bob Koss
    Posted Jul 31, 2007 at 9:05 AM | Permalink

    LOL. I guess they’re making sure they stay ahead of the curve this year. Don’t want to do like last year and leave a storm without a name.

    There was one in July last year that formed at roughly the same location which they didn’t even bother to name or put out advisories on. They went back after the season and classified it a tropical storm. That brought the total for the season up to ten. Couldn’t hind-cast a name because date order and name order wouldn’t match.

  272. SteveSadlov
    Posted Jul 31, 2007 at 9:48 AM | Permalink

    RE: #268 – In the world of Green newspeak, Noreasters are Tropical Storms. The man said, thou shalt have one named storm prior to July’s expiry …. and thus, it came to pass.

  273. Kenneth Fritsch
    Posted Jul 31, 2007 at 9:58 AM | Permalink

    Re #265

    David Smith, I had a difficult time reading the bluish lines on your charts. I knew they were there but at first glance only could see the decadal frequecies at sea. That is important information so I would hope you would consider redoing the charts.

    Holland and Webster commented in their recent paper that more storms could have been sited in the early history because ships would not have independent information to avoid them as they do now. I think when one is entirely convinced that there has to be a significant link between SST and TS and hurricanes one can ratinalize just about anything to explain the data.

  274. Bill F
    Posted Jul 31, 2007 at 10:15 AM | Permalink

    Since August 1 is rapidly approaching, I guess I had better update my prediction. I had said 18, just because it was the equivalent of bidding $1 more than everybody else on the Price is Right and I hoped to claim all territory north of 17 for myself in the event we had a repeat of 2005. Also, given the early start in the Bogocane derby, I was actually surprised not to see them hang a name on the high pressure over the GOM last week since it had clouds and a rotation (albeit clockwise).

    Given the conditions thus far, I would be tempted to lower my prediction to 10 named and 3 major. However, I think NHC/NOAA is dead set on trying to make sure they don’t end up with their predictions way off ala last year, so I think they are going to make every effort to name anything they can from here on out. Therefore, I predict 12 named and 4 major, with at least one of the named storms coming in mid to late November with a non-tropical origin, and at least one “major” being called such based on satellite values alone despite no dropsonde data indicating anything close to Class 3 winds.

  275. SteveSadlov
    Posted Jul 31, 2007 at 10:28 AM | Permalink

    RE: #275 – I am hopeful that the current axe swinging at the NHC will result in those padding the count getting fired soon. That’s the basis of my highly risky bet of 9 named storms.

  276. Bill F
    Posted Jul 31, 2007 at 10:45 AM | Permalink

    I was expecting the same Steve, but the naming of Chantal doesn’t give me any confidence in that regard…

  277. David Smith
    Posted Jul 31, 2007 at 12:27 PM | Permalink

    RE #267 Hello Bob. Thanks for your post for several reasons, the first of which is that it brought to light an error I made: The weak storm defintion is for those which are of tropical storm strength for 24 or fewer hours, not the 18 or fewer hours I typed.

    This makes no difference in the post but it’s good to correct mistakes, substantive or not – thanks.

    Now, the question is, what is the basis for 24 hours (or 18 hours)? There is usually a sequence of events when a tropical depression reaches storm strength. First the central pressure falls. Then the windspeed increases to storm strength in a small area of the storm, usually on the eastern side in squalls. Then that area (the windfield) expands. This takes time, with time varying by storm. I think a general rule is about six hours for the pressure fall to translate into storm winds and then a half-day to a day, or longer, for the windfield to expand.

    My contention is that the near-continuous recon, satellite and radar will pick up this intensification process in the early part of the sequence, when the actual area of storm-force winds is quite small and unlikely to be detected by a ship (unless it just happens to be in the right place at the right time). By 24 hours, the windfield has expanded and a reasonable expectation can be made for ship detection, even though that is problematic if there are few ships around.

    The use of a duration (be it 12,18, 24 or 30 hours) is somewhat arbitrary, but using 18 or 24 hours seems reasonable to me, as it is midrange. The better method would be to review the HURDAT records and extract those storms which relied solely on modern intensity estimates (satellite, radar, recon, etc) with no surface observation (ship or land observer). These would be the true “modern detection method only” storms.

    I’m longwinded in this, I see, but I hope it answers the question.

    To compare our methods, I list Bret, Gert, Jose, Lee and Alpha in 2005 as being 24 hour or less tropical storms. I count one observation (listed every six hours) as constituting a six-hour period.

  278. David Smith
    Posted Jul 31, 2007 at 12:43 PM | Permalink

    Re #274 Hello Ken, I’ll redo the charts and make things more visible. I experiemtn with things – some work and some don’t.

    I haven’t seen the Holland/Webster paper. If i understand your extract, I can’t understand their reasoning on undercount. Seems like there are two cases:

    1. A storm is detected and a warning goes out, making other ships flee.
    2. A storm is not detected until it wallops an unsuspecting ship

    In both cases a storm is detected, so the storm count is the same.

    One could probably make a case that unsuspecting ships in pre-wireless days sometimes were sunk by storms (as happened to the Spanish treasure ships) and no report was ever made, which means that early seasons were possibly underreported.

    I’m curious – did the paper use error bars on their storm count?

  279. David Smith
    Posted Jul 31, 2007 at 1:08 PM | Permalink

    The current QUIKSCAT map for the globe is here . (This is near real-time, so it changes.)

    Click on the reddish area east of New York – that’s Chantal. Ignore the black wind barbs (rain-contaminated). I think what you’ll see are few barbs of 35 knots or above (tropical storm strength). It would be hard for a ship to detect this one, especially in earlier years of lower ship coverage. And how would a ship captain know that he was in a tropical storm versus simply a front or an area of thunderstorms?

    Thanks to satellite appearance this sytem is considered a tropical storm, though its life is expected to be only 18 hours before it becomes extratropical. Pre-satellite, it’s an unlikely detection and, often, who cares as it is an inconsequential system.

    Get ready for Dean and Erin if those Caribbean and Gulf thunderstorms hint at a swirl, or if the satellite spots the steam above your coffee cup.

  280. Kenneth Fritsch
    Posted Jul 31, 2007 at 1:19 PM | Permalink

    David Smith, this excerpt from the paper here is to what I was referring and the authors rationale goes something like this: One could obtain better information about storms in earlier times because ships not being warned away would approach closer. Of course implicit in this view is that the warning system used later could not provide as detailed information as a ship (which could while fighting to survive the storm), but was simply a grossly operating system that had capabilities to do little more than direct the ships away from the storms. For such a rationale to have any hard basis, would require more than a leap in faith.

    Balancing the early observational problems was a tendency for ships to provide better information on tropical cyclones, as they were less apt to avoid storms being without the excellent warning system that is now available (Holland 1981; Mann & Emanuel 2006).

    Error bars are only implied in terms of brush strokes. Without further explanation (but using a “carefully examined” data record) they make, what appears to be, rather arbitrary conclusions about passed missed storms. All in all their arguments seem to me quite weak.

    In summary, we consider that the veracity of the NATL tropical storm
    database is sufficient to enable the broad brush analysis that we undertake in this
    study. Prior to 1945, we concentrate on the total number of tropical cyclones,
    irrespective of intensity. After 1945, we extend the analysis and consider a broad
    categorization of total number of cyclones, minor hurricanes (categories 1 and 2)
    and major hurricanes (categories 3’€”5)

    Questions have been raised over the quality of the NATL data even for such a
    broad brush accounting. For example, a recent study by Landsea et al. (2006)
    claimed that long-term trends in tropical cyclone numbers and characteristics
    cannot be determined owing to the poor quality of the database in the NATL,
    even after the incorporation of satellite data into the database. We have carefully
    examined the data record and considered as yet unpublished analyses by other
    investigators. Our conclusion is that the number of earlier missed storms most
    likely lies between 1 and 3 per year prior to 1900, less than 2 in the early
    nineteenth century and dropping off to essentially zero by 1960. The conclusion
    by Landsea (2007) of much higher numbers of missing storms is considered to be
    based on a false premise of an assumed constancy of landfalling storms ratio (Mann
    et al. submitted a,b; Holland in press). Landsea et al. (2006) also state unequivocally
    that there is no trend in anyNorth Atlantic tropical stormcharacteristics (frequency
    or intensity) after 1960. This is at odds with several other studies (Emanuel 2005;
    Webster et al. 2005; Curry et al. 2006), which found a strong trend similar to that in
    figure 1. To remove the upward trend since 1960 would require the implausible
    conclusion that up to 5 tropical cyclones were missed prior to 1995.

  281. Posted Jul 31, 2007 at 1:32 PM | Permalink

    Those refs were not in the draft version:

    Mann, M. E., Emanuel, K. A., Holland, G. J. & Webster, P. J. Submitted a. Atlantic Tropical Cyclone Revisited.

    Mann, M. E., Sabbatelli, T. H., & Neu, U. Submitted b. Evidence for a modest undercount bias in early historical Atlantic tropical cyclone counts.

  282. MarkW
    Posted Jul 31, 2007 at 1:51 PM | Permalink

    David Smith,

    two other scenarios

    3) An undetected storm wallops an unsuspecting ship, but either before or after the brief period in which that storm was tropical strength, so it never gets recorded as a tropical storm.
    4) An undetected storm wallops no ships, so remains undetected.

  283. SteveSadlov
    Posted Jul 31, 2007 at 2:11 PM | Permalink

    RE: #280 – Unless it originated as an Easterly Wave which became a TD while having a westward component of movement, it is not a TS, in my book (this hints at my eventual operational definition of TCs, which is meant to cut through all the …. crud, and create a more stable criterion for naming or not naming. My Op Def will cull out all features which have arisen via mid latitude processes).

  284. Bill F
    Posted Jul 31, 2007 at 2:23 PM | Permalink

    The whole idea of ship counts being lower recently due to better warnings is a complete red herring. If ships were warned away from a TS, then the TS was already known and cannot be “undercounted” because a ship didn’t see it. Storms like Chantal would be recorded by ships as noreasters prior to the satellite technology and overzealous NHC staff that we have today.

  285. John Baltutis
    Posted Jul 31, 2007 at 3:17 PM | Permalink

    On Holland/Webster paper:

    NOAA debunks report linking hurricanes to climate change.

  286. Bill F
    Posted Jul 31, 2007 at 3:26 PM | Permalink

    I believe I saw elsewhere (maybe in another comment posted here) that Chris Landsea is the “unnamed” NHC representative who called it “sloppy science”.

  287. Jaye
    Posted Jul 31, 2007 at 3:31 PM | Permalink

    I saw an article about it on MSNBC but they emphasized the false premise of warming causing an increase in storms and then in a subtitle said that it was called sloppy science by a “US Official”.

  288. David Smith
    Posted Jul 31, 2007 at 3:57 PM | Permalink

    Re #281 Wow

    The issue since 1960 is not whether cloud systems were entirely missed but whether their brief surges in intensity were captured. We now have advanced-generation satellites, doppler radar, a buoy network and improvements in recon to help detect these intensity surges. Those came about between 1960 and today, with much of the improvement occurring by the early 1980s.

    I offer an example – Hurricane Cindy of 2005. Here is an excerpt from the reanalysis report, explaining how and why the NHC decided that Cindy was a huricane instead of a tropical storm:

    A detailed post-storm analysis of Doppler velocity data from the NOAA National Weather Service (NWS) Slidell, Louisiana WSR-88D Doppler radar (KLIX), however, indicates Cindy was slightly (5 kt) and briefly stronger ‘€” a hurricane with 65-kt winds. The radar indicated a narrow but relatively lengthy swath of spotty Doppler velocities of at least 71 kt aloft in the eastern semicircle of Cindy’s circulation (Fig. 5 and 6). These winds were detected as early as 2330 UTC 5 July at a distance of at least 120 n mi south of the radar site and continued to just inland of the southeastern Louisiana coast a few hours later. The swath of wind speeds depicted should not be construed as being continuous in both time and space.

    That’s technical talk, but what it says is that a later check of doppler radar (an advanced type of velocity-measuring radar) records found an narrow, intermittent area of 70kt winds at cloud level which, if adjusted down to the surface, would equate to minimal hurricane strength (65kts) for six hours.

    What did ships observe? Here’s the ship observation report for Cindy. The maximum sustained wind is the next-to-last number:

    Table 2. Selected ship and buoy reports with winds of at least 34 kt for Hurricane Cindy, 3-7 July 2005.

    Date/Time (UTC)
    Ship call sign
    Wind dir/speed (kt)
    Pressure (mb)
    05 / 0900 ELXL3 28.6 90.0 120 / 35 1015.0
    05 / 1000 WGXN 24.9 88.2 140 / 40 1012.0
    05 / 1100 V7HD3 28.2 88.5 150 / 35 1008.0
    05 / 1500 KRHX 26.8 89.5 170 / 37 1010.0
    05 / 1500 V7HC9 27.5 89.8 100 / 50 1010.2
    05 / 1700 V7HC9 27.6 89.8 130 / 60 1011.5
    05 / 1800 V2AW5 26.0 87.8 040 / 41 1017.0
    05 / 1900 V7HC9 27.5 89.8 150 / 46 1007.7
    06 / 0000 V7HD2 27.7 87.9 150 / 37 1016.0
    06 / 0100 KSYP 28.5 89.1 140 / 50 1006.5
    06 / 0600 V7HD2 27.7 87.9 170 / 37 1019.0

    The highest sustained wind reported by ship is 60kt, below hurricane strength. I also suspect that a reanalysis person seeing a 60kt with a 1011mb pressure reading (quite high for a storm) would assume that the reported 60kt wind measurement was wrong (instrument problem, reading too high) and ignore even the 60kt.

    The same principle applies to cyclones which briefly surge to tropical storm strength.

  289. David Smith
    Posted Jul 31, 2007 at 4:39 PM | Permalink

    Re #286 That is strange.

  290. David Smith
    Posted Jul 31, 2007 at 6:33 PM | Permalink

    Ken thanks for the link to the Holland Webster paper. I didn’t realize that it was available for free. I’ll take a look at it this weekend.

    Speaking of the paper, the Weather Channel is a popular US cable station whose management has long promoted the view of damaging/catastrophic global warming. It is home to Heidi Cullen, the punish-the-doubters lady. Yesterday one of the hosts asked their hurricane expert (Steve Lyons) what he thought about the Holland Webster paper.

    Lyons smiled, said something about changes in measurement techniques and the odd duration of the regimes in the paper and then he changed the subject. It was a ringing non-endorsement. I was surprised.

    Heidi Cullen will be knifing for his AMS Seal of Approval, or worse.

    (Lyons, by the way, is the best hurricane communicator and expert among the TV people.)

  291. Kenneth Fritsch
    Posted Jul 31, 2007 at 6:41 PM | Permalink

    I believe I saw elsewhere (maybe in another comment posted here) that Chris Landsea is the “unnamed” NHC representative who called it “sloppy science”.

    Landsea is a “named” NHC representative — unless the direct quotes I have heard from him are false.

  292. David Smith
    Posted Jul 31, 2007 at 9:02 PM | Permalink

    A few more words on storm observation (sorry) –

    I looked at the offical summaries for the 2005 Atlantic hurricane season, available here . The summaries report 12 tropical storms and 15 hurricanes for the season, a record.

    For 4 of the 12 tropical storms there is not a single surface (ship, land) report of tropical-storm force winds. Alternate modern methods (mainly aircraft and satellite) established the peak intensity and thus the tropical storm classification. Average life of these four storms was a brief 18 hours.

    For 9 of the 15 hurricanes there is not a single surface report of hurricane -force winds. Alternate methods (aircraft, satellite, radar) established the peak intensity and thus their hurricane classification.

    Of these 9 hurricanes which had no surface observation, 3 are special. For these 3, hurricane strength lasted just twelve hours for 1 and six hours for 2. I contend that these 3 would not have been detected as hurricanes prior to modern satellite, radar and aircraft.

    These are examples of how modern technology affects intensity measurement and thus the classification of marginal systems.

    Two impacts of improved technology on hurricane climatology are

    1. more named storms and
    2. more hurricanes (weak ones).

  293. David Smith
    Posted Jul 31, 2007 at 9:12 PM | Permalink

    The US National Hurricane Center has issued its final bulletin on Chantal. It was named at 6AM and is kaput at midnight, meaning the storm life was 18 hours. It joins Barry, which also lasted but 18 hours.

  294. John Lang
    Posted Aug 1, 2007 at 6:04 AM | Permalink

    On short-lived tropical storm Chantal, the seed for the storm developed in the Bahamas. It strenthened into an actual storm as it moved over the Gulf Stream south of Nova Scotia. I don’t think that makes a tropical storm just a large north Atlantic storm.

    You can see its development at the Goes East Full-Disk infra-red satellite which provides coverage for all of the Atlantic and provides a longer time-scale animation. For that reason, it is good for watching hurricane development.

    Click on: Quality – “100%”; image loop – “30”; Animation – “on”; “Animate Image Below”

    And you need to have Java capability.

  295. Bob Koss
    Posted Aug 1, 2007 at 8:20 AM | Permalink

    I was going to read the Holland & Webster paper, but gave up after the first paragraph of the introduction. I have inserted some bracketed numbers to comments below the quote.

    1. Introduction
    The North Atlantic Ocean (NATL) region experienced unprecedented tropical
    cyclone activity in 2005, with records including 28 named systems, 14 hurricanes
    (of which 7 were major hurricanes), 3 category 5 hurricanes and 2 category 4
    hurricanes in July[1]. The number of tropical cyclones and hurricanes was[2],
    respectively, 2.5 and 2.2 times the long-term seasonal means, and the previous
    record for overall tropical cyclone numbers was exceeded by 33%. While the
    entire season was active, August and September, each with five tropical cyclones,
    were below previous records[3]. The truly anomalous months were July and
    October, with five and six tropical cyclones, respectively. October experienced
    almost twice the mean of 3.1 hurricanes and equalled the previous record set in
    1950[4]. July tropical cyclones are relatively rare, typically occurring in less than
    one every two seasons, with an overall average of 1.2, providing less than 5% of
    Atlantic tropical cyclones[5]. July 2005 was a record with 20% of all 2005 tropical
    cyclones and an unprecedented two category 4 hurricanes[6]. Several cyclones also
    developed during June and November’€”December; however, for this study, we
    shall confine ourselves to the period from July to October, inclusive.

    At this point I got disgusted and moved on to check out the graphs. Upon viewing figure #1 I gave up. The storm count and temperature anomaly lines are indistinguishable from each other.

    [1] July had two TS, one cat 1, one cat 4, one cat 5. These guys seem to be numerically challenged.
    [2] Here they defines tropical cyclones as meaning tropical storms and hurricanes are different.
    [3] Now they defines tropical cyclones as meaning both tropical storms and hurricanes.
    [4] October 2005 had four hurricanes, two TS, and one sub-tropical. October 1950 had four hurricanes, and two TS.
    How does four October hurricanes equate to almost twice the mean of 3.1? And what basin and time frame did that 3.1 mean come from? Pick any fairly long time frame in the Atlantic and the yearly hurricane mean will be greater than five. If he is talking only the month of October then the hurricane mean is just under one. Still doesn’t fit.
    [5] How does an average of 1.2 equate to less than one every two seasons?
    [6] Yeah. So unprecedented he got all worked up and fumbled the count. There was only one category 4 in July.

    How can they put their names to such shoddy work? I can see why the reviewer wishes to remain anonymous.

  296. Bob Koss
    Posted Aug 1, 2007 at 8:44 AM | Permalink


    Thanks for the clarification.

  297. SteveSadlov
    Posted Aug 1, 2007 at 9:18 AM | Permalink

    RE: #295 – And this year, the Bahamas are dominated by the loopy / split polar jet and the semi persistent trough sitting over the Piedmont. Chantal, another FRAUDULENT named storm!

  298. steven mosher
    Posted Aug 1, 2007 at 9:36 AM | Permalink

    I hope they name 100 bogus storms this year. 100 bogus storms that never gain strength,
    never make landfall.

  299. Gerhard H.W.
    Posted Aug 1, 2007 at 9:42 AM | Permalink

    Re:#298 – But hey, bogocane Chantal was the strrrongesssst cane this season with unbelievable 994 mbar, 50 mph and 0,608 ACE! (To my opinion, everything under 24 h should not be named as a tropical storm. They should pick a number which turns into a name 24 h after defining a system a tropical storm.)

    Posted Aug 1, 2007 at 11:28 AM | Permalink

    H&W have apparently named the SS off the Azores,
    (between Stan and Tammy..).. the NHC did
    not name it!! Talking about this “Blow-up” business that was one of the few movies Antonioni made(1966) that Ingmar Bergman liked…They
    died the same day…Sergei Prokofiev and Josef
    Stalin (the latter not music but empire composer)
    another famous couple…Is there a plan??

  301. Bob Koss
    Posted Aug 1, 2007 at 11:48 AM | Permalink

    RE: #301
    I accounted for that sub-tropical in my comment [4]. They only claimed six during October. Four hurricanes, two tropical storms. That SS would have raised the number to seven for the month. I agree with them not including that one.

  302. Kenneth Fritsch
    Posted Aug 1, 2007 at 11:55 AM | Permalink

    I hope they name 100 bogus storms this year. 100 bogus storms that never gain strength,
    never make landfall.

    The naming and the very existence of so many of these tiny teeny storms helps make the point of
    those who (and I include myself here) are claiming a very significant and not small past history of undercounting.

  303. Bob Koss
    Posted Aug 1, 2007 at 12:07 PM | Permalink

    The reason for comment [4] was to point out that four October hurricanes is not almost twice a 3.1 hurricane average. It’s not even 150% of 3.1. Additionally, there is no time period of any reasonable length for which they justify making a claim that 3.1 hurricanes is some sort of long term average. The 3.1 figure is bogus.

  304. SteveSadlov
    Posted Aug 1, 2007 at 12:53 PM | Permalink

    RE: #301 – Three noir references – Antonioni’s “Blow Up,” Prokofiev’s tearful life, and Stalin’s empire – in one post! Impressive! 🙂

  305. steven mosher
    Posted Aug 1, 2007 at 1:04 PM | Permalink

    re 300.

    50MPH? So instead of hitting a 9 iron, I should hit a 4 iron.

    Been there done that. If you can play golf, it isnt a storm.

    Posted Aug 1, 2007 at 2:29 PM | Permalink

    #305 Steve Sadlov after all it came from Devilⳳ
    living room where so called TH “Vinceⳳ” eye
    in fact managed to watch a Bergman film before
    death…(see earlier post of mine in this thread!) Guess what? DRS2 has a program about
    Antonioni right now! If you wish to hear it in
    stereo use Windows Real is mono since some time!
    Now only 6 min left though…But perhaps good
    modern Swiss music…AFTER from 22.35 GMT+2

  307. David Smith
    Posted Aug 2, 2007 at 7:57 PM | Permalink

    The Eastern Pacific (EPAC) has been even tamer than the Atlantic so far this season.

    Normally by early August the EPAC has seen 7 total storms, including 3 hurricanes of which 1 is a major hurricane. In 2007 the actuals are 5 total storms and 1 very weak hurricane. (That 1 hurricane barely attained hurricane intensity for a total of……………. you guessed it, just 12 hours.)
    However, while storm count has been a bit below normal the intensity and duration have been well below normal. Here’s a look at the ACE index.

    The ACE index is a function of storm count, duration and intensity. The normal seasonal ACE for EPAC is 160 (the Atlantic = 100). The 2007 ACE to-date is a very meager 8.4 . The EPAC season is normally about 1/3 complete by this date, so ACE is far behind normal.

    It is true that the more-intense storms occur later, in September, and those will swell the ACE. However, to reach normal there will have to be a lot of strong activity soon. That looks unlikely.

    Apocalypse postponed.

    Posted Aug 2, 2007 at 9:39 PM | Permalink

    Gerhard, Steven Mosher …
    XTS “Chantal” is doing much better over 13-14 C
    waters…965 mb at midnight GMT ..same latitude
    as my livingroom …”Come here little girlie,
    and watch another Bergman”…Sorry the rabid rabbit
    must have overtaken my PC and he pressed “sub-
    mit comment”…LOL…

  309. steven mosher
    Posted Aug 2, 2007 at 10:00 PM | Permalink

    Chantal should be whirling around a pole

  310. steven mosher
    Posted Aug 2, 2007 at 10:05 PM | Permalink

    re 303.

    Well, I guess my point is this. If they want to inflate the Named storm number, let them.

    It will just screw the landfall proportion ( TM RPjr) into the ground. It will screw the

    proportion between NS and cat3-5 into the ground. Moving the goalposts makes a short headline,

    but without LANDFALL what is the point? If the wind blows in the ocean and no one is there

    to feel it, does it make a difference?

  311. Jonathan Schafer
    Posted Aug 3, 2007 at 8:54 AM | Permalink


    Or they will remove them at the end of the year after “re-analysis”. Depends on which targets they want to hit.

  312. MarkW
    Posted Aug 3, 2007 at 9:07 AM | Permalink

    maybe they can use teleconnections to prove that the storms actually did hit land?

  313. Kenneth Fritsch
    Posted Aug 3, 2007 at 9:32 AM | Permalink

    Re: #311

    Moving the goalposts makes a short headline, but without LANDFALL what is the point? If the wind blows in the ocean and no one is there to feel it, does it make a difference?

    That has been my point all along. Seems it should provoke a better explanation by those attributing more TSs and hurricanes to AGW, with both concern and alarm, and at the same time recognizing that landfall events have been constant. Lacking a good detailed explanation for a change in the ratio of landfall events to total events, what is a firm advocate of “higher SSTs cause more events” to do: invoke a friendly Storm God steering these storms into the sea?

    The recent Holland and Webster paper on the relationship of storm frequency to SST referenced some unpublished works that will purportly shed some light on the landfall ratio puzzle that will differ from Chris Landsea’s observations.It remains to be seen then whether it will a stronger case than the broad brush strokes ones made by Holland and Webster in their recent paper.

    Of course, as David Smith noted above the use of ACE helps keep these teeny but detectable storms in perspective. Curiously we do not hear much of this index of late (except from DS).

  314. beng
    Posted Aug 3, 2007 at 9:57 AM | Permalink

    David Smith, the sun’s at a minimum & the ITCZ has been meek in the W. hemisphere up to now. I wonder how much influence the solar cycle has on the ITCZ? Not sure how one would proxy ITCZ strength.

    I wonder about this because, from my limited meteorological understanding, the ITCZ’s spatial/temporal development is sensitive to the solar constant.

    Thinking alittle more, the SE Asian monsoon is a manifestation of the ITCZ in the E. hemisphere (the US SW “monsoon” would also be a much smaller, weaker example) so simple rainfall records might be decent proxies there.

  315. Posted Aug 3, 2007 at 10:50 AM | Permalink

    In the meantime evidence of the record breaking hurricane season is being constructed:
    “…From 1944-2005, the avearage date of the third named Atlantic storm is August 20, making Chantal’s appearance 3 weeks earlier than climatology.”

  316. David Smith
    Posted Aug 3, 2007 at 11:43 AM | Permalink

    Speaking of ACE (which incorporates storm strength and duration as well as storm count), here are the current global ACE values:

    South Pacific / Australia –
    Typical ACE is 60 (for a season)
    ACE for the 2006-2007 season, which is over, was 45

    Southern Indian Ocean –
    Typical ACE is 130
    ACE for the 2006-2007 season, which is over, was 125

    Western Pacific –
    Typical ACE is 305
    Year-to-date ACE is 60
    Season peak is September and lasts thru December

    Northern Indian Ocean –
    Typical ACE is 15
    Year-to-date ACE is 23
    Season has two peaks, the first is over and the second will be in a few months

    Eastern Pacific –
    Typical ACE is 130 (based on 1970-2000 avg, which is a different period than I posted above)
    Year-to-date ACE is 9
    Season peak is September and is usually over in late October

    Atlantic –
    Typical ACE is 100
    Year-to-date ACE is 2
    Season peak is September and is usually over in late October

    Globally it’s a normal-looking year so far, with the potential to be below-normal.

  317. David Smith
    Posted Aug 3, 2007 at 12:02 PM | Permalink

    Re# 315 beng, we looked for a good historical index/proxy of ITCZ strength and found none. I’ll take a look at the Asian monsoon, but I think it has drivers which are independent of the global ITCZ.

    Besides the seemingly unusual ITCZ this year I’ve noticed more low cloud cover in the southern Indian Ocean than what I recall in the past. It’s simply an observation and can be wrong or meaningless, but it just strikes me as odd.

  318. SteveSadlov
    Posted Aug 3, 2007 at 12:30 PM | Permalink

    RE: #316 – the photo really brings it home. The masses (and even supposed elites) are being reprogrammed by Big Brother to use newspeak, in this case, cold fronts are tropical storms. For the love of Big Brother …. I digress. This ongoing fraud is precisely the reason why there needs to be a highly aggressive campaign to both publicly and privately call out this false naming and call it what it is … padding the named storm count in order to “make the case” that “killer AGW is leading to longer, stronger and more dangerous tropical cyclone seasons, see, we told you so!” Call a spade a spade. Who cares if the perps slime you by calling you an “oil company shill” or “right winger.” The overwhelming majority of those who want to call this out are neither.

  319. David Smith
    Posted Aug 3, 2007 at 2:59 PM | Permalink

    The updated Gray/Klotzbach Atlantic forecast is here .

    The revised forecast calls for 15 named storms this season (excluding subtropical storms). We’ve had two storms to-date, be they arguable events each of which lasted only 18 hours.

    The 15-storm prediction is aggressive and depends on an active September plus an extended season, both thanks to ENSO-neutral / La Nina conditions in the Pacific.

    Their statistical model predicts 12 named storms, so they’ve tweaked their “official” forecast upwards.

  320. Bill F
    Posted Aug 3, 2007 at 3:54 PM | Permalink

    They must be applying the NHC observational bias adjustment to their model output. That would account for the extra 3 storms. I used a similar adjustment to my predictions 😉

  321. David Smith
    Posted Aug 5, 2007 at 11:17 AM | Permalink

    RE #296 Bob, I find similar problems when trying to read the Holland Webster paper .

    Besides the items you cited in their first paragraph they also got the total 2005 storm count wrong (in their opening sentence of all places): there were 27 named storms (27 tropical cyclones) in 2005, not 28. And, there were 15 hurricanes, not the 14 figure they cite. These are minor errors but when there are such errors in plain sight it makes one wonder whether they’ve used care in handling data in the more-substantive, away-from-plain-sight aspects of their paper.

    It didn’t take long for my concern to heighten. I kept reading and saw this on page 3:

    The exception to this conclusion is subtropical developments, which were not analysed prior to 1965 (eg, Spiegler 1971; Simpson & Pelissier 1971). These systems are therefore not included in this analysis.

    I take that to mean that they’ve excluded subtropical storms from this tropical cyclone study. I thought wow, they’re making progress.

    But then I look at their Figure 1, which is labeled “Tropical Cyclone Occurrence”. Lo and behold I see that they’ve included subtropical storms, despite their page 3 statement to the contrary. This is a more substantive error.

    (I spot-checked their Figure 1 dots for 2005, 2004 and the mid-1970s and it looks like they’re still counting subtropical storms. I’ll be glad to post their plot with my comments if anyone wishes to compare notes with mine.)

    Then I tried to check their Figure 6, as I have had difficulty in the past replicating the details of their SST plots. I thought that maybe I could replicate the details this time. But I struggled for part of my Saturday afternoon unsuccessfully trying to replicate their East Atlantic end points and finally gave up. Why bother?

    I have no problem with making mistakes – they are simply part of trying. But, it seems like something submitted for publication would be checked before submission.

  322. Posted Aug 5, 2007 at 11:40 AM | Permalink

    #322, ..and they use 9-year running mean in the Figure 1 and 5-year in the Figure 2. Then they make this regime change non-sense in Figure 4, similar as in Wilson 2007. Wilson cites Rodionov, and in that paper is is clearly stated that

    It should be emphasized that the method proposed here cannot tell whether the observed regimes are realizations of a Gaussian, red noise process or “true” regimes with different statistics. This is a general property of the currently available methods of regime shift detection.

    To me this is a big problem. See my nice regime shifts from AR1 process:

    ( URL )

    (Rodionov 2004: A sequential algorithm for testing climate regime shifts, GRL vol 31.)

  323. Kenneth Fritsch
    Posted Aug 5, 2007 at 11:51 AM | Permalink

    Re: #322

    Please be gentle, David Smith, as I have detected that is in your nature. Holland and Webster admit to using broad brush strokes in their paper and Chris Landsea appears to confirm this a bit more emphatically by rephrasing that to a sloppy approach. Could it be that Holland and Webster meant well by stating the need to omit the subtropical storms, but their graduate students doing the counting did not get the message? I think we need to consider your lack of confirmation of SSTs, merely a misinterpretation of the designation area from whence they came. After all should not we be able to (hunt and) find regions of SST somewhere in that great expanse of the NATL that will correlate with storm frequencies. Heaven forbid we attempt to look at SSTs in actual storm tracks as was attempted by that fossil fuel industry supported Pat Michaels.

  324. Bob Koss
    Posted Aug 5, 2007 at 12:37 PM | Permalink

    The numerous numerical errors suggests to me they didn’t think enough of their paper to bother checking it, or they have no respect for the reader, or both.

    The fact that someone reviewed it and it passed, speaks to how much confidence one should put in the review process.

  325. Bob Koss
    Posted Aug 5, 2007 at 12:46 PM | Permalink

    They evidently included the subtrop in October in order to reach 28 for the year. But then didn’t count it when counting October storms.

  326. Kenneth Fritsch
    Posted Aug 5, 2007 at 12:52 PM | Permalink

    It should be emphasized that the method proposed here cannot tell whether the observed regimes are realizations of a Gaussian, red noise process or “true” regimes with different statistics. This is a general property of the currently available methods of regime shift detection.

    Couldn’t we be a bit generous and consider under the definition of “broad brush” that a true regime shift has some “broad” chance of being correct? It is those damn media people who do not recognize deniability when it was there in plain sight all the time.

  327. Bob Koss
    Posted Aug 5, 2007 at 12:54 PM | Permalink

    I wonder if anyone will ever cite it for any purpose. lol.

  328. David Smith
    Posted Aug 5, 2007 at 1:06 PM | Permalink

    I’ve been playing with SST and various smoothing techniques and will post several plots later today.

    One thing about the 9-year simple smoothing which they use is that it eliminates those strange peaks and valleys in east Atlantic SST that come along on 11-year cycles.

  329. David Smith
    Posted Aug 5, 2007 at 1:35 PM | Permalink

    Ken the thing which surprises me the most, as I read climate science papers, is the presence of so many small errors which should have been easily caught by authors and their graduate students. The superficial ones have no bearing on the substance of the paper but they do make those of us from other disciplines wonder about what’s “under the hood” (OK, “under the bonnet” for those in the UK).

  330. Kenneth Fritsch
    Posted Aug 5, 2007 at 2:13 PM | Permalink

    Ken the thing which surprises me the most, as I read climate science papers, is the presence of so many small errors which should have been easily caught by authors and their graduate students.

    David, I have a theory that these miscues have something to do with the urgency that many of these authors (and publishers) feel they have in getting their messages out, i.e. it stems from the political and topical nature of GW and AGW. Looking forward to seeing your smoothing analysis this evening ‘€” and before my Cubbies take on the hated Mets.

  331. David Smith
    Posted Aug 5, 2007 at 8:55 PM | Permalink

    Here are some comments regarding the SST in the Holland Webster paper. I’ll try to keep this short.

    First I wanted to replicate their east Atlantic SST, which is the red line in Holland Webster’s Figure 6. The figure is here .

    One minor head-scratcher is that there are three points up at the maximum, which surprises me, as I expect just one. The final point is fine – it is composed of the years 1998 thru 2006. However, the two points before it involve dropping warm years (2006 and 2005) and adding cool years (1996 and 1997) so those two points should be lower than the end point.

    The bigger head-scratcher is that, between 1955 and 1975, the Figure 1 red line drops perhaps 0.2C, maybe a hair more, but, when I calculate it, I get something more like 0.4C drop. My time series is here . This difference (HW vs mine) is important because it affects the extent to which the East Atlantic is anomalous.

    Now, I used the SST in the NCEP plotter and checked it against the Kaplan SST plotter (NCEP and Kaplan correlated OK). The Holland Webster paper uses SST from a dataset at the NCDC. Perhaps the SST sets are materially different and perhaps the NCDC dataset is better for this purpose than NCDC – I don’t know. However, I’ll be surprised if SST datasets give materially different regional results in the modern period.

    If it turns out that the red line in Figure 1 is wrong by about 0.2C then a significant part of the Holland Webster paper becomes shaky. If anyone can get the NCDC plotter to work for them (I tried but was not successful) then please check if that 1955-1975 part of the HW red line can be replicated. Perhaps it can – I don’t know.

    The second topic has to do with smoothing. Nine-year simple smoothing is used on my East Atlantic chart , similar (I think) to the HW chart, and portrays a rather monotonic temperature rise. However, if one uses a 1-2-3-2-1 smoothing ( see here ) then richer detail arises. Specifically, that pesky 11-year (more or less) cycling becomes apparent (recall Holgate’s sea level rate of rise chart was also plagued by similar mysterious cycling). Also of note is that it becomes clearer that the SST in the 1990s is not much different from the SST in the 1950s and that the sharp rise is mainly in the last five or so years.

    Finally, a plot of East Atlantic, West Atlantic and Gulf of Mexico SSTs is here , just for fun. The odd cycling is apparent in all regions. The West Atlantic is only now reaching temperatures seen in the 1950s, and note that the 1950s were probably not the high temperature point of the last AMO cycle. The East Atlantic appears to be more amplified than the western regions – perhaps that is simply an AMO characteristic of the region.

    Well, this wasn’t short – my apology.

  332. Kenneth Fritsch
    Posted Aug 6, 2007 at 3:00 PM | Permalink

    Re: #332

    David, your smoothing analysis shows that you can get very different perspectives on the data by how you smooth. In the end, however, someone must explain what it means and why it looks the way it does. Holland and Webster talk of regime changes of sudden transitions from one regime to the other with rather flat storm counts within the regime. Their claim is that these regime changes fit the SST pattern or at least the one they used. They also use the sharp transitions to argue that the storm undercounts would be much less likely to follow such quickly changing patterns. Unfortunately when Holland and Webster claim a correlation between NATL SST and cyclone frequency we never get to see the regression scatter plot or whether they correlated annual or smoothed data. I would not think that a simple OLS approach would be in order unless the distribution was shown to be normal. Many here, and I include myself, claim it fits a Poisson distribution best.

    Anyway I found the following excerpt from the HW paper a bit perplexing wherein they state that the SST cyclone frequency correlation arises entirely from the regime changes and that within the regimes SST correlates weakly negatively with the frequency — and perplexing due to a lack of an explanation why this should be. We need to look at yearly data plots of SST and cyclone frequency and see what that looks like. I would be curious whether you feel the correct SSTs are being used by HW and you have others in mind that might be more appropriate or least could be used for a sensitivity analysis.

    Figures 1 and 2 clearly show that the transitions between the three main
    climatic regimes are closely aligned with changes in eastern NATL seasonal SST
    anomalies. The SST anomalies explain over 60% of the tropical cyclone variance
    since 1905 and tend to lead the changes in cyclone frequency. Similar results have
    been found by Mann & Emanuel (2006). This correlation arises entirely from the regime changes defined earlier. As shown in figure 2, the regimes are clustered within East Atlantic SST ranges of 25.8’€”26.28C (TC1, 1900’€”1930), 26.2’€”26.78C (TC2, 1940’€”1994) and 26.7’€”278C (TC3, 1995’€”2005). Within each regime, the variations in SST are weakly correlated negatively with tropical cyclone frequency. Hurricanes follow a very similar clustering, though with a slightly lower explained variance (56%) by eastern NATL SST anomalies.

    PS: My Cubbies defeat last evening put the final nail in my thinking that they do not have the player abilities to beat the better teams in the leagues.

  333. David Smith
    Posted Aug 6, 2007 at 4:52 PM | Permalink

    Re #333 Ken what I can say is that Holland Webster’s Figures 6 and 7, which are important to part of their argument, contain something I cannot replicate or explain. Their East Atlantic SST for 1950-2005 has a pattern that is at odds with the SST datasets I’ve checked, including Hadleyv2.

    Holland Webster show about a 0.4C rise (1950s vs 2000s) while what I can replicate is only about a 0.2C rise.

    Their Figures also contain West Atlantic SST which I can reasonably replicate. I just can’t replicate their East Atlantic plot, which is the important one. I’ll check the eastern Pacific SST to see if I can replicate it.

    The overall paper reminds me of a plaintiff’s filing, wherein a lawyer makes many claims in the hope that something sticks. The paper has a high adjectives-to-data ratio.

  334. Kenneth Fritsch
    Posted Aug 6, 2007 at 8:03 PM | Permalink

    David, are you sure that your comparison of SSTs with HW’s paper is covering the same months of the year. That has been a bone of contention before in analyzing these TC papers and even created some suspicions of data snooping with me. It would be great to get your data and HW’s side by side and then plot it against the counts for hurricanes, named storms and land falling events.

  335. David Smith
    Posted Aug 6, 2007 at 8:49 PM | Permalink

    Re #335 Ken I used the period July through September for the SST boxes defined on HW page 4, consistent with the page 4 description. I notice, though, that in the next paragraph they talk about averaging SST over “the full tropical cyclone season” which is truly different from just July thru September. Did they mean July thru September? I don’t know.

    Last year I tried to match the SST plots in Webster (2005) and Hoyos (2005). These were papers with a lot of author overlap. Some of the basin plots were the same in both papers, some were not. This is important because Hoyos uses SST as input to some kind of information-theory data processor.

    I’ll put together some data time series later this week.

  336. David Smith
    Posted Aug 7, 2007 at 8:54 PM | Permalink

    Re #335 Ken I made a couple of charts as you suggested. The results puzzle me.

    I compared sea surface temperature (SST) generated by the NCEP/NCAR Reanalysis plotter with the SST plot provided in the Holland Webster (HW) paper. The region covered is the East Atlantic, a region important to hurricanes and to an HW hypothesis.

    The NCEP/NCAR source for global surface temperature is listed as GISS, and I believe that GISS uses Hadley Reynolds v2 for SST. I did not find this clearly spelled out so I may be mistaken and I welcome any correction.

    The Holland Webster paper says their “…SST are derived from the extended reconstructed SST set archived at the NOAA Climate Data Center.” My assumption is that they took these NCDC values as-is but “derived” could also mean that they used the NCDC as a first step and applied Holland Webster adjustments. If HW did adjust the raw SST data then I missed their note and explanation in the paper.

    Here is an overlay showing the NCEP/NCAR derived SST anomaly (blue line) and the Holland Webster SST anomaly (red line) for 1950 to 2006. (The light blue trace is the raw data for NCEP/NCAR. Also, the absolute value of the anomaly has no meaning on this chart.)

    What is evident is a divergence beginning about 1960 and widening to about 0.25C. In the world of SST anomalies a divergence of 0.25C in modern times is substantial, I believe, especially in the period before satellite-derived SST estimates. It’s a head-scratcher.

    But I am even more intrigued by is this plot . This (blue line) is a plot of the difference between the two 9-year averages, with 1952-1962 arbitrarily set at zero by me.

    I expected to see a more or less gradual divergence between the two, as perhaps they used different methodologies which, over time, gradually diverged. But what I see are almost step changes (red lines), with a 10-year rise to a 0.15C divergence, where things held for 25 years, then another 0,1C increase over 5 years, then a return to 0.15 in the final several years. Remember that these are comparisons of 9-year averages, so such uniform changes are even more surprising to me.

    Perhaps the parties who created the two sets had good reasons, differences in professional judgment, which account for such well-defined changes in one part of the global SST. It would sure be educational to hear the explanation for this.

    Next, I’ll look at the adjacent region (WATL) and perhaps the eastern Pacific (EPAC). You’d think that similar steplike adjustments should also apply to them.

  337. David Smith
    Posted Aug 7, 2007 at 9:14 PM | Permalink

    Re #337 I’ll post the west Atlantic comparison tomorrow. Any guesses as to its appearance?

  338. John Lang
    Posted Aug 8, 2007 at 6:47 AM | Permalink

    Roger Peikle Jr. posted up a critical review of the Holland and Webster paper yesterday on Sr.’s weblog. Interesting read which shows their data contains lots of errors and doesn’t come from any published dataset.

  339. steven mosher
    Posted Aug 8, 2007 at 8:24 AM | Permalink

    331. Cubbies!

    Nothing better than blowing off Calculus, hopping the El south to west addison..
    Bleacher bumming it in wrigley. Best park ever!

  340. Kenneth Fritsch
    Posted Aug 8, 2007 at 11:55 AM | Permalink

    Re: #338

    David Smith, I will not venture any guesses but thanks for the SST info and plots. I would like to do some quick analyses of my own and your link to the SST data is just what the doctor ordered.

    Re: #340

    Steven Mosher, you and Ferris Bueller. My wife and I had a memory lane revisit to the bleachers a few years back. Friday afternoon, mostly young 20 and somethings in the bleacher crowd looking for dates, extra inning game and a husband/wife backache. It made us feel our age, but I would not trade that day and memory for anything. Oh yeah, Cubs lost. Also, I can thank you here for the OT links to the urban warming articles.

  341. Posted Aug 8, 2007 at 1:01 PM | Permalink

    Re 339: there is a mention in the linked item about landfalls becoming less common. Is this likely? Is it possible?

    I found a track for Hurricane Emily — does anyone know why she suddenly swerved not once but twice? Are there other hurricanes which show this sudden track change?


  342. Bob Koss
    Posted Aug 8, 2007 at 5:02 PM | Permalink

    What a coincidence. I just came here to post a graphic about landfall storms.

    Here is a chart of what should be Holland & Webeter’s storm data for 1905-2005. Blue and Red areas indicate storms reached landfall or came within 60 nautical miles of land somewhere in the Atlantic. Not necessarily US. All green storms stayed farther from land.

    Only 11% of the the 333 observed storms up to 1948 stayed farther than 60 miles from land. Whereas 25% of the 589 storms post 1948 stayed away from land. The detection at 35 Knot/60 mile blue portion doesn’t appear at all unusual at any point during period. The 34 knot/60 mile red portion looks to me as if they are noticing quite a few storms since 1949 that might have simply be considered squally weather previously. The green portion seems to indicate better observational ability away from land.

    I don’t see the regime changes H&W allude to, just observation ability changes. There is no doubt many storms away from land were simply missed the further you go back in time.

  343. Kenneth Fritsch
    Posted Aug 8, 2007 at 5:05 PM | Permalink

    David Smith, I want to confirm that I am inputing into the link you gave correctly. The inputs I used (to follow directly) would not produce the same anomalies, at first glance by ranking my actual temperature outputs, as you graphed under SST Reynolds.

    Period 1948 to 2007 NCEP Reanalysis Dataset:

    Variable = SST; Ananlysis Level = Surface; Latitude = 25 to 5; Longitude = 55 to 20; seasonal average; First Month = July; Second Month = Sep; Average Weights Grids = No; Output format = Raw Data Values

  344. Steve McIntyre
    Posted Aug 8, 2007 at 5:06 PM | Permalink

    Bob, what is your criterion for excluding subtropicals.

  345. Bob Koss
    Posted Aug 8, 2007 at 5:28 PM | Permalink

    The best track database trailer line which indicates HR, TS or SS. I excluded all SS storms. There are 22 of them in the database. All since 1969. One other might qualify in Sept. 1968, as all the tracks are subtropical or extratropical but Hurdat has it listed as a HR. A couple tracks are 70 knot. I have the database last revised in May 2007.

  346. Bob Koss
    Posted Aug 8, 2007 at 5:34 PM | Permalink

    My graphic is for the entire period. Actually H&W restricted their data to the months of July to October. Their storm counts seem to be way off anyway. I posted up the figures for their partial seasons over at Pielke Sr. site in the post Roger Jr. made. I can put them up here if you want.

  347. Bob Koss
    Posted Aug 8, 2007 at 5:42 PM | Permalink

    I’ll post up a graphic similar to the one above for only the period July-October. About an hour from now.

  348. Kenneth Fritsch
    Posted Aug 8, 2007 at 5:51 PM | Permalink

    My graphic is for the entire period. Actually H&W restricted their data to the months of July to October. Their storm counts seem to be way off anyway. I posted up the figures for their partial seasons over at Pielke Sr. site in the post Roger Jr. made. I can put them up here if you want.

    Bob Koss, I realize that your question was directed to Steve M, but I would like to see the HW data posted here. Your red, blue and green graph is an excellent one for making your point on changes in detecting TC and hurricanes, but I wonder if other readers here see it as clearly.

  349. David Smith
    Posted Aug 8, 2007 at 6:11 PM | Permalink

    Re #344 Ken try longitudes of -55 and -20 (west longitude is expressed as a negative).

  350. David Smith
    Posted Aug 8, 2007 at 6:35 PM | Permalink

    Re #342 Julian it’s not unusual for Atlantic tropical cyclones to make sudden changes of direction. I remember this one ( Elena 1985 ) which threatened Mississippi, swerved, stopped, and returned to Mississippi.

    This storm changed direction due to changes in steering currents (winds about three miles above the surface) and pressures, not an avoidance of land. Land masses can exert some slight influence on a tropical cyclone but not enough to change its destiny (landfall and dissipation).

  351. David Smith
    Posted Aug 8, 2007 at 6:45 PM | Permalink

    Re #343 Nice chart Bob.

    The other category of storms which were hard or impossible to detect in earlier years are those which had brief, weak existence (defined as TS strength for 24 hours or less) which I posted about ( link ) .

    I’m very interested in a plot of tropical cyclones which excludes (1) those which stayed at least 60 miles from land and (2) the weak, brief storms. I bet you’d find evidence of a 60-year oscillation with no upward trend. If you’re generating plots I sure hope you can do this one, too, if it’s little trouble. You’d have to exclude storms which had 35 (or higher) knot winds for less than 30 total hours (five observation periods).

  352. Bill F
    Posted Aug 8, 2007 at 6:47 PM | Permalink

    Just look at TS Allison once it hit Houston or Hurricane Wilma over the Yucatan in 2005. Both storms were motoring along, suddenly stopped and hung around for a day or two, then motored off rapidly in a completely different direction. The presence of land had nothing to do with why they changed direction.

  353. Bob Koss
    Posted Aug 8, 2007 at 7:05 PM | Permalink

    My bad. I described that first graphic as being their data set. Their set would be a sub-set of that data. Should be similar to one of these next two graphics.

    803 storms spawned Jul-Oct 1905-2005.
    Up to 1948 of the 288 observed storms 12% stayed farther than 60 miles from landfall.
    Post 1948 of the 515 observed storms 74% stayed farther than 60 miles from landfall.

    814 storms Jul-Oct 1905-2005 including June carry over.
    Up to 1948 of the 292 observed storms 12% stayed farther than 60 miles from landfall.
    Post 1948 of the 522 observed storms 74% stayed farther than 60 miles from landfall.

    All sub-tropicals storms removed.

  354. Bob Koss
    Posted Aug 8, 2007 at 7:19 PM | Permalink

    Error alert! error alert!

    This line is wrong in my previous post. I forgot to subtract from 100%.

    Post 1948 of the 522 observed storms 74% stayed farther than 60 miles from landfall.

    The 74% figure below both graphics should be 26%.

    That’s what I get for not taking my time.

  355. Bob Koss
    Posted Aug 8, 2007 at 7:23 PM | Permalink


    I’ll see if I can work something up for you tomorrow.

  356. TCO
    Posted Aug 8, 2007 at 8:06 PM | Permalink

    Mosh, I’ve done my el trips south. Also been drunk late at night at the Howard transfer station.

    (To keep this on topic, I’ve run the seawall during a hurricane.)

  357. Bob Koss
    Posted Aug 8, 2007 at 8:17 PM | Permalink

    Here’s the data set I generated for the storms spawned Jul-Oct.
    Third column includes second column storms. Both qualify the tracks for landfall, minimum speed, and time frame. If at least one track qualifies I count the storm. Fourth column only qualifies the time frame so it includes the storms in the other two columns. With the columns ordered properly and by using the area graph only column two and the differences from column two are visible.

    Hadn’t used the area graph previously. Has a nice look. Who says you can’t teach an old dog like me new tricks?
    35kts & up 34kts & less Jul-Oct Total
    1905 4 4 5
    1906 5 6 8
    1907 2 2 3
    1908 7 7 8
    1909 7 7 7
    1910 5 5 5
    1911 5 5 6
    1912 3 5 5
    1913 3 4 5
    1914 1 1 1
    1915 5 5 5
    1916 9 10 12
    1917 1 1 3
    1918 5 5 5
    1919 2 2 2
    1920 3 3 4
    1921 4 4 5
    1922 3 3 3
    1923 7 7 7
    1924 5 5 6
    1925 1 1 1
    1926 9 9 10
    1927 5 5 7
    1928 4 4 6
    1929 1 1 2
    1930 2 2 2
    1931 6 6 7
    1932 8 8 9
    1933 17 17 18
    1934 7 7 8
    1935 6 6 6
    1936 10 10 13
    1937 8 8 9
    1938 7 7 7
    1939 4 4 4
    1940 7 7 7
    1941 6 6 6
    1942 5 6 9
    1943 9 9 10
    1944 9 9 11
    1945 9 9 10
    1946 5 5 5
    1947 9 9 9
    1948 7 7 7
    1949 9 9 12
    1950 10 10 13
    1951 5 5 9
    1952 3 3 6
    1953 9 9 11
    1954 6 7 8
    1955 9 9 12
    1956 5 7 7
    1957 5 5 6
    1958 5 6 9
    1959 7 7 8
    1960 6 6 6
    1961 7 8 9
    1962 3 4 5
    1963 6 7 9
    1964 7 8 10
    1965 2 2 5
    1966 5 5 9
    1967 4 5 8
    1968 1 2 5
    1969 7 9 16
    1970 8 9 9
    1971 8 10 12
    1972 2 2 3
    1973 5 5 7
    1974 5 5 7
    1975 5 5 7
    1976 4 5 8
    1977 5 5 6
    1978 7 8 11
    1979 5 7 7
    1980 3 4 9
    1981 4 4 8
    1982 2 3 4
    1983 3 3 4
    1984 6 7 10
    1985 9 9 10
    1986 2 2 3
    1987 5 6 7
    1988 7 9 11
    1989 5 6 9
    1990 8 8 14
    1991 3 6 7
    1992 4 5 6
    1993 5 5 7
    1994 2 2 4
    1995 11 12 18
    1996 9 9 11
    1997 3 3 6
    1998 8 9 13
    1999 8 8 10
    2000 10 10 14
    2001 8 9 12
    2002 9 9 12
    2003 9 10 12
    2004 10 11 13
    2005 16 17 21

  358. Kenneth Fritsch
    Posted Aug 8, 2007 at 8:21 PM | Permalink

    David Smith, thanks for the help. I can now duplicate your seasonal SST graph. I down loaded all the months just so I can look at some sensitivity issues for season. I can, at the same site you linked to, find the data set that HW used, but it cannot be down loaded in the convenient form that the one we are using is in — or at least as well as I can currently tell. It is a .nc file and I do not know how to read those. Steve M does it with R and I could draw a lecture from him for not using it. Maybe now’s the time.

  359. David Smith
    Posted Aug 8, 2007 at 8:57 PM | Permalink

    I looked at Holland Websters’s west Atlantic sea surface temperature plot, compared to the NCEP-generated plot. The difference is here . Just like the east Atlantic comparison (post #337) the west Atlantic shows those remarkable steplike changes, but the east and west go in different directions (the east is warmer in HW while their west is mostly cooler, relative to NCEP). Go figure!

    What is the significance of all of this? Let me see if I can get there. HW’s Figure 7 shows the three tropical Atlantic basins minus the east Pacific (I believe the assumption is that the east Pacific is a proxy for the global warming trend). By backing out this global trend, one can see how the Atlantic basins have behaved. If the east Atlantic has warmed more than the others, then perhaps the east Atlantic is generating more storms now than in the past, suggesting that the increase in at-sea storm count over recent decades is real and not a detection issue. Also, if for some reason the east Atlantic SST is galloping away, temperature-wise, from the global warming trend, then we are witnessing something ominous.

    Here is HW Figure 7, using the HW SST: Link
    The red line is the eastern Atlantic while the western Atlantic is black and the Gulf of Mexico is blue. I added the green line to denote 1950, which is when my charts begin.

    As one can see, using the remarkable Holland Webster SST, the red line (eastern SST) has indeed become a rogue, heating faster than the others and faster than the global trend. This suggests more east Atlantic storms relative to the other two basins and suggests greater tropical east Atlantic warming than the global trend – scary.

    But, if one uses the NCEP plot generator (which I take to be based on Hadley Reynoldsv2) one gets this plot: Link . (Note that I started the lines at the same point.) What I see is that, since 1950, the three basins have weaved and bobbed a bit but generally behave in concert and have not outrun the global trend. (The 1970s peak may be ENSO-related, but that’s a different story.)

    At a minimum, if one accepts the SSTs at face value, then one is left wondering why the HW dataset shows a regional pattern, with ominous implications for North America, while the other shows no such thing.

  360. Posted Aug 9, 2007 at 5:34 AM | Permalink

    Re replies to 342:

    Thank you. I was reminded of the tracks of certain cunims which have a mind of their own and wander off regardless of the rest of the weather. With a big one reaching up to 45k ft they don’t seem to care about winds at 15k.

    I can never look at a hurricane picture without wondering how to gut its heat engine.


  361. David Smith
    Posted Aug 10, 2007 at 6:20 PM | Permalink

    The US National Hurricane Center (NHC) released its updated forecast for the 2007 season ( link ). Their forecast is little changed from earlier, with 14 or 15 tropical cyclones as their forecast midpoint.

    I continue to think that the NHC, and Gray/Klotzbach, are right about this becoming a busy season. The thinking is that it will be stacked towards the back end, with a lot of activity in October and November. It is looking like I, Mike Mann, Gray/Klotzbach and the NHC are the most aggressive in forecasts. We’re quite the combo – if we’re right maybe we can do a group hug.

    Meanwhile, the first unarguable tropical storm(s) of 2007 are about to form, with two active areas in the Caribbean and near Africa.

    I’ll post the special August update contest this weekend. There’s still time if anyone wants to offer a revised forecast for this second contest.

  362. Kenneth Fritsch
    Posted Aug 15, 2007 at 8:41 AM | Permalink

    David Smith, I was able to find the HW SST data set that they used in their recent paper. I confirmed it by comparing 9 year MAs. You can obtain it in easily down loadable form by going here:

    where you will receive the instructionsfor using this link for downloading various time series.:

    For the HW data set you will need to go to the row with “Smith Reynolds Extended Reconstruction SST’s” and click on plot. The Reconstructed data set is the only one I could find that has data back into the mid 1800’s which confirms that it is then one HW used. I was curious why they needed data that early when they have themselves noted that the hurricane counts are uncertain, but have since concluded that there correlation look better with this SST data set and by going back further in time.

    I used hurricane/TC official data from NOAA and plotted frequencies versus the SST from your Reanalysis data set and the HW data set for the period 1948 – 2006. The SST used in the plots was always for the months July-September. I included a plot of the ACE index for each SST set. I calculated R^2 to explain the portion of frequencies and index was explained by changes in SST and also the trend of number of storms and units of ACE index per degree C of SST. Although I would caution against using such plots without determining how well the data fits a normal distribution, I do it in attempts to repeat the HW results.

    For the Reanalysis data set I obtained:

    Named Storms: R^2 = 0.15 Trend = 4.9 storms
    Hurricanes: R^2 = 0.20 Trend = 3.8 storms
    Major Hurricanes: R^2 = 0.17 Trend = 2.6 storms
    ACE: R^2 = 0.21 Trend = 91 units

    For the HW (Extended Reconstruction) data set I obtained:

    Named Storms: R^2 = 0.30 Trend = 8.6 storms
    Hurricanes: R^2 = 0.20 Trend = 4.7 storms
    Major Hurricanes: R^2 = 0.14 Trend = 2.9 storms
    ACE: R^2 = 0.20 Trend = 110 units

    Finally I used what I suspect HW used in their paper and that is a plot of 9 year MA for both the SST data set that they used and for the storm frequencies. Remember that HW used the time period 1905 to current while I am using only 1948 to current. Why they used back to 1905, I am not certain, because they claim the data in the first half of the 20th century is questionable. A real correlation should not change over time unless one has a reasonable explanation for it. Doing that I obtained:

    Named Storms: R^2 = 0.54 Trend = 7.5 storms

    When I substituted the Reanalysis data set of SSTs for the HW set and did the same 9 year MA versus 9 year MA plot I obtained:

    Named Storms: R^2 = 0.23 Trend = 4.9 storms

    The correlations to which HW refer in their paper are described in the excerpt below:

    (ii) Relationship between SSTs and tropical cyclones

    Figures 1 and 2 clearly show that the transitions between the three main
    climatic regimes are closely aligned with changes in eastern NATL seasonal SST
    anomalies. The SST anomalies explain over 60% of the tropical cyclone variance
    since 1905 and tend to lead the changes in cyclone frequency. Similar results have
    been found by Mann & Emanuel (2006). This correlation arises entirely from the
    regime changes defined earlier. As shown in figure 2, the regimes are clustered
    within East Atlantic SST ranges of 25.8–26.28C (TC1, 1900–1930), 26.2–26.78C
    (TC2, 1940–1994) and 26.7–278C (TC3, 1995–2005). Within each regime, the
    variations in SST are weakly correlated negatively with tropical cyclone
    frequency. Hurricanes follow a very similar clustering, though with a slightly
    lower explained variance (56%) by eastern NATL SST anomalies.

    From my analysis I see several unanswered questions coming out of the HW results:

    1. Why do we see differences in correlations using a different SST data set and what are the differences in set constructions? It bothers me that data sets with these differences exist and that their numbers could potentially influence data snooping the results using the set that “works”.

    2. Why should one use a nine year MA in looking at and correlating these data and how are hurricanes and TCs sensitive to conditions averaged over 9 years. Also what would this mean in terms of the Pat Michaels paper where he looks at hurricanes and TCs as function of SSTs within days of the events?

    3. Why would not one want to look at these same relationships with the idea of eliminating some of the detection variables over time, since we know the detection has improves at the same time SSTs have been generally increasing?

    Based on question number 3, I want to use the Bob Koss’ off-shore frequencies in similar plot trends and other indicators that Bob or David or other posters here might suggest. I suspect that the R^2s will decrease from even the smallish ones that I found when plotting year by year data (as opposed to 9 year MAs).

    Since the down period of Climate Audit allowed more time for analysis, I went back to the HW data set of SSTs for 1905-2006 and found the following (presented in the same form as was used above and for year by year plots of storm frequency/index versus July/August/September SST for 5-25N, 55-20W):

    Named Storms: R^2 = 0.35 Trend = 8.2 storms
    Hurricanes: R^2 = 0.25 Trend = 4.3 storms
    Major Hurricanes: R^2 = 0.14 Trend = 2.3 storms
    ACE: R^2 = 0.17 Trend = 77 units

    For the same 1905-2006 period for the HW SST data set and using 9 Year MA of storm frequency versus 9 year MA for SST instead of year by year plots I obtained:

    Named Storms: R^2 = 0.74 Trend = 8.2 storms

    Adding these additional results to those from above would certainly seem to present evidence that HW where looking for the most favorable SST data set, time period, moving average and storm category to demonstrates a correlation between frequencies and SST.

    As I waited longer than anticipated for CA to come online, I decided to look at the months July/Aug/Sept that HW apparently used and determine how well they stood up to any a prior reasoning and .

    I looked at the hurricane/tropical cyclone data for the NATL for the period 1905-2002 and determined the following distribution by month for named storms:

    May = 1.3%; June = 5.5%; July = 7.1%; Aug = 21.6%; Sept = 37.5%; Oct = 20.5%; Nov = 5.4%

    Based on these data, the three months that would appear more reasonable to use for SST are Aug/Sept/Oct and thus my choice to test the robustness of the above analyzed relationships was to use these 3 months in place of July/Aug/Sept. The results are given below with a description of the data set used and the time period. All SST results were as above for the area 5-25N, 55-20W and all plots/correlations were on a year by year basis (no MA).

    Reanalysis data set for 1948-2006:

    For the Reanalysis data set I obtained:

    Named Storms: R^2 = 0.15 Trend = 4.9 storms
    Hurricanes: R^2 = 0.16 Trend = 3.3 storms
    Major Hurricanes: R^2 = 0.12 Trend = 2.2 storms
    ACE: R^2 = 0.17 Trend = 80 units

    HW data set for 1905-2006:

    For the HW (Extended Reconstruction) data set I obtained:

    Named Storms: R^2 = 0.20 Trend = 6.2 storms
    Hurricanes: R^2 = 0.11 Trend = 2.9 storms
    Major Hurricanes: R^2 = 0.05 Trend = 1.3 storms
    ACE: R^2 = 0.08 Trend = 51 units

    Now, since these three months appear to be a more rational choice for attempting to construct a correlation of SST with seasonal storms, but have smaller R^2 values one must then either come up with a better and reasonable rationale for using July/Aug/Sept or assume the correlation is spurious.
    Continuing with the analysis as CA remains down, I looked at using a combined average SST from the HW Extended Reconstruction data set for the period from 1905-2006 for the two NATL regions where tropical storms develop, i.e. the Eastern NATL (5-25N,55-20W) used previously and the Western NATL (10-25N,90-55W). These plots are all year by year (without MAs). I looked at the 3 months July/Aug/Sept and Aug/sept/Oct.

    For July/Aug/Sept I obtained:

    Named Storms: R^2 = 0.33 Trend = 8.8 storms
    Hurricanes: R^2 = 0.23 Trend = 4.7 storms
    Major Hurricanes: R^2 = 0.13 Trend = 2.4 storms
    ACE: R^2 = 0.15 Trend = 80 units

    For Aug/Sept/Oct I obtained:

    Named Storms: R^2 = 0.23 Trend = 7.5 storms
    Hurricanes: R^2 = 0.14 Trend = 3.7 storms
    Major Hurricanes: R^2 = 0.06 Trend = 1.8 storms
    ACE: R^2 = 0.08 Trend = 62 units

    It would appear from this exercise that one obtains nearly the same correlations using either Eastern NATL SSTs alone or in combination with the Western NATL SSTs and the correlation with the combined like NATL Eastern alone is reduced when using the Aug/Sept/Oct months.

  363. Kenneth Fritsch
    Posted Aug 15, 2007 at 9:38 AM | Permalink

    Plotting Bob Koss’ data from his third column for tropical storm frequencies within 60 miles of landfall for the period 1905-2005 against the HW (extended reconstructed data set) for year by year comparison I obtained the following:

    Using the months July/Augt/Sept for the SST:

    R^2 = 0.03, Trend = 1.6 stroms

    Using the months Aug/Sept/Oct for SST:

    R^2 = 0.00, Trend = -0.3 storms

    Using 9 year MA for storms and SSTs for the months J/A/S I obtained:

    R^2 = 0.31, Trend = 2.2 storms

    Using 9 year MA for storms and SSTs for the months A/S/O I obtained:

    R^2 = 0.22, Trend = 1.9 storms

    From this one can see that using the more readily detectable storms causes the correlations to be reduced to values much lower than those obtained when using all the storm count data.

  364. SteveSadlov
    Posted Aug 15, 2007 at 10:44 AM | Permalink

    By my count, which excises the two midlatitude cyclones and the one subtropical jet wave which were counted, we now have 1 named storm with the possibility of a second, later today.

  365. John G. Bell
    Posted Aug 15, 2007 at 12:11 PM | Permalink

    Re #365 That is how I see it also. But at the rate they are naming them we will be in the low 20s by the time the season is over.

  366. Jonathan Schafer
    Posted Aug 15, 2007 at 12:45 PM | Permalink


    Second storm has now been named.

  367. SteveSadlov
    Posted Aug 15, 2007 at 1:06 PM | Permalink

    Indeed. Bogus count is now 5. Real count is now 2.

  368. Mark T
    Posted Aug 15, 2007 at 2:04 PM | Permalink

    How many “storm days” are there? As I recall, at least one or two lasted less than a couple days.


  369. fFreddy
    Posted Aug 15, 2007 at 2:43 PM | Permalink

    Re #368, SteveSadlov

    Bogus count is now 5. Real count is now 2.

    Steve, you can imagine that later this year, the Warmers and their PR firms are going to be making a big noise about these inflated numbers.

    Have you considered opening one of John’s audit blogs to make a record of these storms and bogus storms through the year ? Have one post per “official storm” , with detailed reasons why you don’t agree with it.

    It could be a really useful resource for any remaining honest journalists.

  370. Bob Koss
    Posted Aug 15, 2007 at 3:48 PM | Permalink

    Kenneth Fritsch,

    Seems like you keep yourself pretty busy while CA was down.

    I’m not surprised at the low correlations you found. Basically I’m of the opinion that when they use storm counts in whatever relationship they are trying to establish, they’re starting with a very crude metric. The variation in strength and duration is huge.

    Seems to me analysis should be based on as discrete a time interval as possible, the track. At least that way consistently sized data points are available for comparison. Storm size has too large a variation.

  371. Bob Koss
    Posted Aug 15, 2007 at 3:53 PM | Permalink

    Dave Smith,

    Graphed up the storms starting 1881 after removing the storms of 2.5 days or less that stayed more than 30 miles away from land. Only minor differences since I only found about 13 of the short ones. No apparent cycle to be seen. Haven’t been able to get to the site to check exactly what you wanted. Hope I covered it.

    There does seem to be somewhat of a cycle in this data.

    I plotted up Accumulated Cyclone Energy by year and divided that by the yearly track count to get the average track ACE. I plotted the yearly track count also, but that stays pretty much in step with the yearly ACE for the whole series. Made the graph too busy so I took it out. The graph seems to show an association with the AMO cycle. Seems longer though.

    Looking at the divergence between series around 1922. With the yearly ACE down and track ACE up, it appears the tracks were more intense or many weak tracks were missed. The last few years seems to show the tracks being average, but with many more occurring to raise the yearly ACE.

  372. David Smith
    Posted Aug 15, 2007 at 8:08 PM | Permalink

    It’s good to see CA back and good to see such interesting, detailed work by Kenneth and Bob. Ken, I especially appreciate the link to the SST data, which I’ll plot and post.

    I have some thoughts on #364, 365 and 372, as well as the topic of those ever-present faux storms, but this evening I have some obligations related to Tropical Storm Erin as it nears Texas which will keep me occupied until it officially comes ashore, so posts will have to wait. I do have time to visit the tip jar though.

    Hello to all.

  373. Jonathan Schafer
    Posted Aug 17, 2007 at 3:49 PM | Permalink


    Well, Erin is ashore, and if you live in S TX or San Antonio/Austin regiion, you are being inundated with rain. Up here in teh D/FW area, we had a brief shower with lots of lightning last night, but nothing today.

    Keeping a very close eye on Dean though. If it does hit TX and tracks anywhere close to what Erin did, the flooding and damage potential will be enormous. Current models mostly show TX/MX border or south, with GFDL the outlier. Will have to see what happens with the TUTT and whether Dean slows down, speeds up, or stays the same. Just too far out at this point to know if it will be an issue.

  374. steven mosher
    Posted Aug 17, 2007 at 5:15 PM | Permalink

    RE 366 and 368.

    Let them boost the bogus storm Count. As I’ve said before it just deflates the
    Landfall proportion ( Tm Roger Peilke Jr.)

    So anybody taking bets that Dean threads the Cuba/Yucatan needle and slices into


  375. SteveSadlov
    Posted Aug 17, 2007 at 5:28 PM | Permalink

    So we’ve got two proper, named storms that came directly from Easterly waves. It will be interesting to see how the following plays out. Autumn has come early to the northern plains, north of the US – Canadian border. Some of the long term forecasts depict a series of Alberta Clipper type storm tracks over the coming 10 day period. Depending on if this indeed occurs and how far the cold fronts / polar jet dips south, it might or might not constitute a notable interuption to the “prime time” portion of the current TC season. Stay tuned ….. I’m still stubornly sticking with 9. We can debate about what to do with storms A – C at the end of the season.

  376. David Smith
    Posted Aug 17, 2007 at 6:35 PM | Permalink

    Re #374 Tropical Storm Erin dropped 7 inches of rain in my backyard gauge in slightly under two hours yesterday afternoon. I doubt that I’ll have to water the flowers for the rest of the summer.

  377. John Lang
    Posted Aug 17, 2007 at 6:47 PM | Permalink

    Actually, tropical storm Erin probably shouldn’t be called a tropical storm since its rotation from the beginning was backwards (clockwise) and this continued even into today.

    It is more a collection of thunderstorms pushed around by the weather systems around it.

    Hurricane Dean, however, is starting to look nastier and nastier.

  378. David Smith
    Posted Aug 17, 2007 at 7:58 PM | Permalink

    Erin is the third example this season of a named tropical storm which would be hard to detect by the pre-satellite/radar methods of ships, sea states and barometer readings. (The apparent clockwise rotation may be the anticyclone (outflow) at the top of the thunderstorms.)

    Dean is cat 4 and headed to cat 5 this weekend. Anxiety is rising along the northwestern Gulf Coast, with local and state emergency action plans already activated. My weekend plans have been scratched and replaced by emergency response planning meetings. I hope it proves to be just a practice run.

  379. David Smith
    Posted Aug 17, 2007 at 8:40 PM | Permalink

    Kenneth your analysis is impressive and indeed raises some tough questions. If HW is robust then the one-month shift in SST should not create the large change in correlation you found. If HW is robust then correlations should be robust using unsmoothed data and not have to rely on 9-year smoothing to get a decent r value.

    I am able to replicate the East Atlantic SST used in HW, thanks to the link you provided. At the same time I’m perplexed by the fact that the various SST constructions vary in the ways they do and that a person can select a SST construction to best suit a hypothesis (ie, cherry-picking). I’d be delighted to find an explanation by the SST constructionists for the specific case of EATL, where they explain why the constructions go through periods where they match then methodically diverge then match again. My suspicion is that there are many assumptions, adjustments and grafts built into their black boxes.

  380. John G. Bell
    Posted Aug 17, 2007 at 9:06 PM | Permalink

    The north shore of Jamaica may be in for a very ugly time. I wish them all the luck in the world. Plenty of yacht owners are depending on these forcasts. Probably heading to Costa Rica from Jamaica. I hope they are well on their way.

  381. David Smith
    Posted Aug 17, 2007 at 9:22 PM | Permalink

    Here is a time series of Atlantic ACE divided by the number of storms. This measure (ACE per storm) is a good indicator of the overall severity of a storm or season.

    The ACE per storm (seasonal average) has generally declined since the 1950s. The specific pattern shows a large decline in the late 1960s followed by flat-to-slowly increasing ACE per storm since about 1970.

    I suspect that the large drop in the late 1960s was, to a large extent, technology-related, as satellite coverage improved and more weak storms were detected, diluting the averages. The rise since 1995 reflects greater activity in the MDR, but even that has not made enough of a difference to return the index to 1950s levels.

  382. Kenneth Fritsch
    Posted Aug 18, 2007 at 11:16 AM | Permalink

    Re: #380

    David, thanks for the comments on my correlation comparisons. I was just now going to pose the same questions to you and any other interested posters here. The correlation, i.e. lack of correlation, with Bob Koss’ 60 mile near landfall storms, was the clincher for me. I also have a growing opinion that many of these climate scientists, like H&W, do not have an appreciation for what it means statistically to search out these “best” fits without providing some detailed and a prior reasons for doing it. If one were more cynical one might think that they search the fit that fits a conclusion and then assume that the peer reviewers will see the conclusion and not look as closely at how it was arrived at – but that might be giving them too much credit for statistical awareness. I could use some help myself in understanding better how a year by year plot with no trend or correlation can show a trend and correlation when plotted as 9 year moving averages. I need to do some serial correlation calculations on the raw data.

    More importantly I was thinking about hurricane Dean and trying to put myself in the position of the many people that are potentially in the track of this monster and growing storm. In the Midwest of the US, where I have spent most of my life, we get tornados and bad thunderstorms, but these events develop quickly and with little anticipation so I do not have a feel for what it must be like to be on the projected storm track and waiting for something to happen or not over a period of days. I would guess that that situation would be very stressful. Anyway here’s hoping that it avoids your area and, for that matter, other populated areas as well. With Dean becoming more intense and traveling over waters described on the Weather Channel as at temperatures not unlike bath water, I was wondering what other intense storms, like Dean and in that position, have done on their trips into the Gulf of Mexico.

    On a lighter note, these 5 storms to this date, as my portions of storms per month of the season would indicate, put you, Gray/Klotzbach, Mann and me on track to be very close on the seasonal totals. Your suggested victory group hug, real or virtual, works better for soccer players and assorted other athletes. I would much prefer a stiff and formal scientific, virtual or real, handshake.

  383. steven mosher
    Posted Aug 18, 2007 at 12:51 PM | Permalink

    RE 357. TCO

    Howard Street! I spent four years staggering down howard street

    At the Tally Ho… and what was the other place…

    And then Back to campus in the bitter bitter cold.

    Dang, maybe we know each other

  384. Judith Curry
    Posted Aug 18, 2007 at 1:13 PM | Permalink

    Re Hurricane Dean, the closest analogue is Hurricane Gilbert (1988), which was very strong and followed the same path projected for Dean (Wikipedia gives a good overview on the storm)

  385. steven mosher
    Posted Aug 18, 2007 at 1:37 PM | Permalink

    DR Curry. Your comments on RC about the need for openess are appreciated.
    I said as much there. I’ll say the same here. You and Ruedy stand out. Nuff said.

  386. David Smith
    Posted Aug 18, 2007 at 2:06 PM | Permalink

    Re #383 Kenneth I keep going back to this time series which is similar to Bob’s work. Storms detected from land (including islands) show cyclical behavior but little increase. Storms detectable only at-sea began to increase when detection (aircraft, satellites) improved.

    There are other considerations (poor Caribbean records prior to 1886 and the recent weak-storm phenomena discussed here) that improve the near-land cyclicity even more.

    Judith Curry is correct about Hurricane Gilbert but with an important addition: people in areas of risk should never assume that a current storm will follow the same path or intensity as a historical storm. I’m sure she agrees with that.

  387. David Smith
    Posted Aug 18, 2007 at 2:13 PM | Permalink

    Re #387 By the way, I’ll update the time series to 2006. The impact is modest – it will add a net 5 storms to the final decade, about 4% of the decadal total.

  388. Kenneth Fritsch
    Posted Aug 18, 2007 at 2:33 PM | Permalink

    Our local meteorologist here in Chicago reminded us this morning that we could feel the effects of a storm like Dean if it hits the Texas coast – it has happened before.

    Oh, and since we are talking about Chicago here’s something on Howard Street. Steven, did you attend Northwestern University?

    If you’ve written off Howard Street as a “won’t go there, no thank you” location, it may be time to change your tune. The occasionally unpleasant divider between Chicago and Evanston isn’t the hotspot it once was, but it’s making a comeback.

    During the 1940s and 1950s, Howard Street was a bustling entertainment district whose latitude marked the northern-most entrance or “gateway” into Chicago’s city limits. Most suburbs were dry at the time, making the decidedly un-dry Howard Street a vibrant entertainment hub for North Shore residents.

  389. steven mosher
    Posted Aug 18, 2007 at 3:22 PM | Permalink



    Evanston was dry. My dorm overlooked the WCTU.
    We had to walk to the friggin EL take it down to howard street
    And walk to either the “tally HO” or the “PM club” in rogers park.

    Since the PM clubs jukebox played frank sinatra 24/7 I walked the extra few blocks.

    The alternative was Rush street…

  390. David Smith
    Posted Aug 18, 2007 at 3:23 PM | Permalink

    The Galveston (TX) Hurricane of 1900 killed an estimated 6,000 people in Galveston. The remnants traveled across the US Midwest and passed near Chicago with sustained winds estimated at 60 mph. I believe it sank a few ships on Lake Michigan. It wasn’t tropical at that point but it was still loaded with energy.

  391. Jonathan Schafer
    Posted Aug 18, 2007 at 4:47 PM | Permalink


    No, but it’s looking more and more likely that it will follow a similar path to Gilbert. The TUTT moviing through the GOM is supposedly weakening, and thus will not pull Dean on a more N track. I’d say by tomorrow night, we’ll know reasonably well if it will be TX or MX. Current bet is on MX.

  392. Posted Aug 18, 2007 at 4:54 PM | Permalink

    #387 I’ll take that bet, I have 75 Linden Dollars on Texas.

  393. Kenneth Fritsch
    Posted Aug 19, 2007 at 4:41 PM | Permalink

    It is once again time for a sanity check.

    In my analysis of the recent HW paper, I determined that HW used 9 year moving averages (MA) to obtain and reference some correlations. They mentioned 0.60 plus R^2 values for explaining the variance of NATL named storm frequencies by way of SST changes for the months July/Aug/Sept and for the period 1905-2006. When one uses a year by year regression the R^2 values are dramatically reduced. They are further reduced by using the months Aug/Sept/Oct and still further (to R^2 = 0) when using the Bob Koss storm frequencies that occurred within 60 miles of the shoreline.

    Now I have learned enough about time series analysis over the past few years to know that I know only a small portion of what would be required of the practicing statistician in this area. I do, however, believe that one would not use a MAs for regression since that is what the theoreticians warn against in terms of not accounting for AR and ARMA and providing methods as those from Box-Jenkins to avoid what HW apparently created.

    My simplistic view sees MAs as a convenient means of visualizing trends in noisy time series and indeed we know there are trends in the SST data and trends in the named storm data. The point with the storm data is how much of the trend is caused by actually occurring increases in frequencies and how much by the increases in capabilities of finding these storms. One indication of the latter is to compare easily detectable with more difficultly detectable storm categories.

    In my judgment, HW simply confirmed with this exercise with MAs and regressions on them, that trends exist in both storm frequencies and in SST, but did not shed any light on the correlation of storm frequencies with SST and the potentially overwhelming effects of detection changes over time.

    May I impose on any of the well versed and experienced posters here in time series to give me my sanity check?

  394. David Smith
    Posted Aug 19, 2007 at 6:11 PM | Permalink

    Kenneth, your concerns are well-posed and shared. I look forward to reading what people outside the topic, statistically skilled and/or one of the HW proponents have to say.

    On Dean, the prayers of Jamaicans seem to have nudged the eye of the storm to pass just south of the island. However, the northern eyewall is slamming the southern half of the island and damage will be bad. Let’s hope all are safe.

    Dean has some unusual aspects, particularly the lack of very cold cloud tops. Usually a category 4, 920mb storm extends very high in the atmosphere, with cloud top temperatures occasionally around -80C. Dean’s coldest run around -60C. I have no guess as to why.

    On the Atlantic season, the mid-range models don’t show a particularly active Atlantic for the next two weeks. Steve Sadlov, John A and others at the low end of the predictions may yet prevail.

  395. steven mosher
    Posted Aug 19, 2007 at 6:43 PM | Permalink

    After a Hurricane passes over a patch of ocean ( say a 5*5 grid) How much
    heat does it suck out of the ocean, if any..

  396. Kenneth Fritsch
    Posted Aug 19, 2007 at 7:37 PM | Permalink

    Re: #396

    After a Hurricane passes over a patch of ocean ( say a 5*5 grid) How much
    heat does it suck out of the ocean, if any..

    I believe I have heard that it is not an insignificant amount but there are others here in a better position to answer that. I also recall Kerry Emanuel discussing a theory where with sufficient global warming hurricanes and tropical cyclones will become a significant conduit of heat out of the tropics (towards the polar regions I assume). The good news with that is that it provides a negative feedback on warming; the bad news is that it generates more storms and more intense storms (hypercanes, if you will).

    And while we are on the subject of Kerry Emanuel, I noticed a recent paper that Emanuel coauthored where the authors were proposing a more serious search for hurricane and TCs going back thousands of years. I got the idea that the recent Kossin reanalysis poured cold water on Emanuel’s ability to experimentally confirm his theories on the increases in intensities of TCs and hurricanes with SST increases and he is now on a path to test his theories using storms proxies in times when the climate was as warm and warmer than currently. I think he may be looking for past evidence of hypercanes. Sounds like a noble effort to me. Except it could be fraught with the same, if not worse problems, of confounding and data snooping as is evidenced with dendrochronology today.

  397. David Smith
    Posted Aug 19, 2007 at 8:01 PM | Permalink

    Re #396 Steven a rough look at a typical hurricane’s energy release is here . The ocean is the major source of the sensible and latent (water vapor) heat used by the storm so it indeed loses heat to the hurricane.

    The churning also mixes warm water downward and cool water upwards so it’s tricky to state a net temperature drop on the surface. I think I’ve read that globally tropical cyclones cool the upper tropical oceans on the order of 0.5C, so they are a significant part of the global heat transfer.

  398. steven mosher
    Posted Aug 19, 2007 at 8:30 PM | Permalink


    Negative feedback was where I was going with this line of thought. The kinetic energy
    of the wind isnt a free lunch..

  399. Jonathan Schafer
    Posted Aug 19, 2007 at 10:30 PM | Permalink


    It could be any number of things. Hurricanes tend to pulse up/down during the day, with the cloud tops warming/cooling accordingly. I wouldn’t chalk it up to anything other than the standard pulsing they go through. Other thoughts could be related to how much the storm is travelling over land, and I believe an eye wall replacement cycle started yesterday. I think those take 24 hours to complete, so it could be related to that as well. In any case, I fully expect Dean to briefly drop to Cat 3, then jump back to Cat 4, possibly Cat 5 before dropping to Cat3/4 as it hits the Yucatan. I wouldn’t want to be in Cancun/Cozumel. All latest model runs except CLIP5 still show a MX hit. TX is likely safe at this point, but should know with much more assurance by tomorrow night.

  400. Posted Aug 21, 2007 at 7:43 PM | Permalink

    I think all people should be aware about the hurricanes because it is a disaster that will destroy our houses and other establishments.

  401. SteveSadlov
    Posted Aug 22, 2007 at 11:55 AM | Permalink

    Things are going quiet ….. nada …… only a few showers in two disorganized blobs …. a bet within a bet …. no more named storms for August ….

  402. Jonathan Schafer
    Posted Aug 22, 2007 at 1:19 PM | Permalink


    Yeah, Invest 92 completely dissipated. Ingested too much dry air from the N. I wouldn’t be surprised if there were no additional storms in August, since there are no waves of any significance around. Unless there is a tail end of a trough somewhere, it may be quiet for a couple of weeks.

  403. steven mosher
    Posted Aug 22, 2007 at 1:23 PM | Permalink

    RE 402.

    I want to see the heat content before and after. Since the first CAT out of the
    hat was a 5, how long does it take SST to recover?

    SST can’t recover instantly.

    I now add additional observational hypothesis.

    1. GW will lead to MORE named storms OR
    2. GW will lead to EARLIER storms OR
    3. GW will lead to stronger storms OR
    4. GW will lead to EARLIER STRONGER storms OR
    5. GW will lead to hypercanes OR
    6. GW wil lead to greater landfalls OR
    7. GW will lead to more destructive landfalls OR
    8. GW will lead to ealier land falls OR

    you get the idea.

  404. SteveSadlov
    Posted Aug 23, 2007 at 11:46 AM | Permalink

    RE: #402 – I’ll critique my own bet within a bet. Between 28 and 30 Aug, there will be a minor window of opportunity owing to the Bermuda High’s expected position. After that, it appears that an early Alberta Clipper will close the window, for how long, who knows.

  405. Jonathan Schafer
    Posted Aug 23, 2007 at 5:36 PM | Permalink


    I’m not sure how long it takes, but since hurricanes rarely take the exact same path, it’s not likely to be a big issue, IMO.


    Dr. Master’s blog says none of the reliable models show any development over the next 7 days. I just don’t see anything out there that would cause me to disagree.

  406. David Smith
    Posted Aug 23, 2007 at 6:21 PM | Permalink

    We’ve had four named tropical storms so far in 2007 and a to-date ACE of about 35. The GFS model shows few seedlings and barely favorable upper winds for the next two weeks which, if true, means only a modest chance of tropical development.

    If the GFS is correct then we’ll find ourselves at the midpoint of the season with only normal levels of activity. (“Normal” total-season activity for the last 60 years is a total storm count of about 10 storms and an ACE of 100.)

    2007 continues to be an odd weather year worldwide.

  407. steven mosher
    Posted Aug 23, 2007 at 7:54 PM | Permalink

    RE 406…

    Yes, they wont take the same path ( for a bunch of reasons I suppose)

    Historically, how many times has the first cat out the bag been a 5?

  408. steven mosher
    Posted Aug 23, 2007 at 8:05 PM | Permalink

    Found it.

    Pre Dean:

    Post Dean.

  409. David Smith
    Posted Aug 23, 2007 at 8:27 PM | Permalink

    Re #408 There was a catagory 5 storm named Andrew in 1992 that pounded Miami. I experienced Andrew’s second landfall and vividly remember watching a chimney collapse from the wind followed by the owner climbing onto the roof to try to cover the hole with plastic (he almost para-sailed).

  410. David Smith
    Posted Aug 23, 2007 at 9:09 PM | Permalink

    Re #410 It’s amazing how visible typos become just one nanosecond after pressing the “Submit Comment” button.

    World Climate Report has two interesting commentaries this week on hurricanes and one on the thermohaline circulation. They look intriguing. It’ll be a few days before I can get to the originals as I’m headed out this weekend to try to survey ten USHCN stations for Anthony’s website.

  411. Jonathan Schafer
    Posted Aug 27, 2007 at 10:54 AM | Permalink


    According to Dr Masters, it will take several weeks for the temps to recover, except over the warm gulf loop.

  412. MarkW
    Posted Aug 27, 2007 at 11:47 AM | Permalink


    Of course there’s Andrew.

  413. Judith Curry
    Posted Aug 28, 2007 at 8:23 AM | Permalink

    The EUROSIP paper on the seasonal forecasts by the 3 European centers has now been published
    unfortunately there isn’t a free online version available.

  414. steven mosher
    Posted Aug 28, 2007 at 9:46 AM | Permalink

    RE 412

    Thanks Jonathan.

  415. Kenneth Fritsch
    Posted Aug 28, 2007 at 9:50 AM | Permalink

    Chris Mooney in “Storm World” professes that the tropical storm climate scientists need to show a united front in presenting their results and theories but early in the book he chooses to divide them with a passing quote from Ernest Rutherford’s reference to “All science is physics or stamp collecting.” He then proceeds to anoint Kerry Emanuel as the physicist and labels William Gray the stamp collector. Later he adds to the list of physicists, who are primarily in the camp claiming significant effects of AGW on hurricanes and tropical storms, and to the list of stamp collectors, who primarily fall into the camp of those less certain about the effects of increasing SST on hurricanes and tropical storms.

    At the end of his book Mooney points to the somewhat theory deflating results of reanalysis that was done by Kossin and Vimont, who would almost surely be considered in this endeavor to be the stamp collectors, on tropical cyclone frequencies and intensities back to the 1980s. Now, if one was aware of the results of the stamp collectors in their attempts to demonstrate that the storm data cannot necessarily be used in the fashion of the physicists in their stamp collecting endeavors, one might want to show that since the data cannot show significant effects of SST on storm frequencies and/or intensities that all we have is the physicists’ theories. It would create a nice packaged version of heads I win tails you lose that could be weaved through the story by the author.

    Anyway I became curious as to the background of Mooney’s physicists (The four that appear below are those that Mooney labels physicists in his slide presentations that he delivers periodically at conferences. He would also consider those scientists that model hurricanes and tropical storms as physicists.) and the non-physicists in the storm discussions and looked online for some of their backgrounds. Ages and up-to-date peer-reviewed publication output were difficult for me to find for most of the participants and thus I give approximate numbers in those categories. The pictures of the participants made them look even younger than my calculated ages and therefore my approximations for age could be due to: child prodigies or dated pictures or everyone looking younger to someone my age or climate science providing such a good and easy life keeps the participants young.

    My only distinctions between the two groups, for what they are worth, is that the physicists tended to be older and have published significantly more peer-reviewed (and non peer-reviewed publications, for that matter) papers than the non-physicists. I thought I might list what I found for those that might be curious and to perhaps encourage some posters to add more exact information to my profiles.

    The Physicists:

    1. Peter Webster. Peer-reviewed and published papers – Over 110. Approximate age – 52. BS degree –The Royal Melbourne Institute of Technology, 1972. PhD – MIT, Atmospheric Science, 1972.
    2. Greg Holland. Peer-reviewed and published papers – Approx. 100. Age – 59. BS degree – U of New South Wales, 1972. MS – Colorado State U., Atmospheric Science, 1981. PhD – Colorado State U, Atmospheric Science, 1983.
    3. Kerry Emanuel. Peer-reviewed and published papers – Over 100. Approximate age – 55. BS degree – Earth and Planetary Science. PhD – Meteorology, MIT.
    4. Judy Curry. Peer-reviewed and published papers – over 140. Approximate age – 54. BS degree – Northern IL University, Geography, 1974. PhD – University of Chicago, Geophysical Sciences, 1982.

    The non-Physicists:

    5. Chris Landsea. Peer-reviewed and published papers – Approximately 50. Approximate age – 42. BS degree -UCLA, Atmospheric Science, 1987. PhD – Atmospheric Science, Colorado State University,1994 .
    6. Phil Klotzbach. Peer-reviewed and published papers – Approximately 7-10. Approximate age – 25. BS degree – Geography Bridgewater State College, 1999. MS degree – Atmospheric Science Colorado State University 2002.
    7. Jim Kossin. Peer-reviewed and published papers – 20+. Approximate age – 44. BS degree – Clarkson University, 1985. MS degree – Clarkson University, Mathematics, 1987. PhD – Colorado State University, Atmospheric Science, 2000.
    8. Dan Vimont. Peer-reviewed and published papers – 10+. Approximate age – 30. BS degree –Gonzaga University,. PhD – Colorado State University, Atmospheric Science, 1994.
    9. Paul Knappenberger. This scientist, who has teamed many times with Pat Michaels (and other authors), has a significant number of peer-reviewed papers published. I suspect that he is relatively young, but I was not able to find a complete biography on this apparent mystery man. He did graduate work at the University of Virginia.

    10. Pat Michaels. Approximate age – 57. Peer-reviewed and published papers – Unable to determine even approximately. He has had published 200 plus peer-reviewed and popular papers. AB and SM degrees – University of Chicago, Biological Sciences and Plant Ecology. PhD – University of Wisconsin, Ecological Climatology, 1979.

    By the way, on searching Pat Michaels’ profile online I found that he is considered by many as the “bad boy” of climate science and approaching “evil incarnate” with others.

  416. SteveSadlov
    Posted Aug 28, 2007 at 10:08 AM | Permalink

    RE: #416 – the Boomers and the X-ers. LOL!

  417. SteveSadlov
    Posted Aug 28, 2007 at 10:10 AM | Permalink

    And now, for something completely different …. (that is, in the context of 416 abd 417:

    Borrrrrrrr-ring …….. looks like we will exit August stuck at the letter “E” …. when will “F” occur?

  418. steven mosher
    Posted Aug 28, 2007 at 10:11 AM | Permalink

    Dr. Curry is a Univ Chicago Phd? She better be a cubbies fan.

  419. Kenneth Fritsch
    Posted Aug 28, 2007 at 10:32 AM | Permalink

    Re: #414

    The EUROSIP paper on the seasonal forecasts by the 3 European centers has now been published
    unfortunately there isn’t a free online version available.

    I attempted to purchase this article (for $9.00 US) and after doing the paperwork I got a notice that this article was not for sale without a subscription. I have emailed for details. One half of my storm estimate in the David Smith contest was tied to Meteo/France’s predictions so I have my interests to protect here.

  420. steven mosher
    Posted Aug 28, 2007 at 11:40 AM | Permalink

    I fully expect you guys to back me in my shameless promotion
    of moshers blowhard ratio.

    Knots of wind at landfall/dollars (pesos) of damage

    And if webster steals it, that’s ok by me.

  421. Kenneth Fritsch
    Posted Aug 28, 2007 at 1:11 PM | Permalink

    Re: #421

    I fully expect you guys to back me in my shameless promotion
    of moshers blowhard ratio.

    Knots of wind at landfall/dollars (pesos) of damage

    Steven, I think that we should call that ratio either Moshers ratio or blowhard ratio — no need to use both modifiers. The monetary value will need to be in a chosen currency and convertible to others with PPI and of course in constant dollars or other currency.

    Seriously, I saw an article linked by David Smith above in #411 that compared the damage in Mexico of Dean (minimal) to a comparable hurricane in the 1950s (much more damage and loss of life) and attributed the difference to better economics that allow for more adaptability to prevent storm damage. A measure of that adaptability would of course be covered partly with the Mosher ratio. You need to somehow include a measure of loss of life in your ratio or index.

  422. Posted Aug 28, 2007 at 2:01 PM | Permalink

    re: #421

    I suggest two changes to the ratio. First use a measure of the power in the storm at landfall in place of the knots of wind. Second turn the ratio over so that given the power the potential damage can be estimated.

    I suspect Pielke Jr. has lots of info over on his blog. This entry has a link to this paper. I’m sure there are many other discussions on the subject over there.

    I do not see a causal relationship between landfall and damage. The damage is a strong function of exactly where the landfall is located plus the nature of the structures of value under the path of the storm, not the mere act of landfall.

  423. Kenneth Fritsch
    Posted Aug 28, 2007 at 2:09 PM | Permalink

    The below quote was excerpted from a Judith Curry post here on Dec 9, 2006, showing the developing line of demarcation between Chris Mooney’s stamp collectors and physicists. Obviously from this, Judith has a stake in the Meteo France results also. Am I going to need to join AGU to get the Meteo predictions? I hedged my prediction square on the line between the stamp collectors and the physicists.

    3. Re prediction. Three European groups (METEO FRANCE, ECMWF, and UKMO) have commenced seasonal forecasting using coupled atmosphere ocean models. Frederic Vitart has been leading this effort. The past two years, experimental forecasts have been made; they will go operational next season. They have also done hindcasts back to 1993. The results have been astonishing; all three groups predicted > 20 NATL tropical cyclones for 2005 in the june 1 forecasts (Meteofrance picked up the large activity as early as their April forecast). This year all three June forecasts were lower than the statistical forecasts, with ECMWF being the lowest because it successfully caught the the incipient El Nino. THis success is being achieved with fairly coarse resolution model (about 200 km); their operational forecasts next season will be around 125 km resolution. At this resolution, they can capture number of storms; sensitivity studies indicate that a resolution of 40 km is needed to successfully simulate intensity and tracks. The U.S. NOAA is lightyears behind in this kind of seasonal numerical forecasting. I predict that after next years debut of the European operational seasonal forecasts for global tropical cyclone activity, that the empirical statistical forecasters will be out of business.

  424. Kenneth Fritsch
    Posted Aug 28, 2007 at 2:34 PM | Permalink

    Re: #423

    I do not see a causal relationship between landfall and damage. The damage is a strong function of exactly where the landfall is located plus the nature of the structures of value under the path of the storm, not the mere act of landfall.

    The insurance bet would go with your view of landfall damage for a particular location. The Mosher index, on the other hand, gives a global average effect of hurricanes and dollar damages which could be used in an historical perspective to measure how well civilization is adapting on the positive side and how much government assistance is negatively affecting civilization’s move from harm’s way. The Mosher ratio also goes with the storms that we are really interested in measuring and the easiest historically to detect, i. e. landfall. In the example given in the David Smith link of 2 comparable hurricanes, the Mosher ratio would provide a good historical comparison. A more complete analysis would, of course, be needed to sort out negative and positive effects.

    In the end it is the Mosher ratio that we need to change, regardless of the effects of SST on storm activity and it is something that changing would have very immediate benefits.

  425. Kenneth Fritsch
    Posted Aug 28, 2007 at 3:14 PM | Permalink

    Re: #420

    I have completed reviewing the paper referenced by Judith Curry and in my post #420 and it does not give the Meteo France or other European predictions for 2007 — and only discusses in detail those for 2005 and 2006. I know that recently David Smith had not found Meteo France predictions for 2007. Why would not an organization such as this one provide their predictions in a timely fashion. Publishing prediction results after the fact is a little silly.

  426. John Baltutis
    Posted Aug 28, 2007 at 3:19 PM | Permalink

    Re: #419

    Why? U of C is only a couple of miles south of the White Sox playground.

  427. steven mosher
    Posted Aug 28, 2007 at 3:25 PM | Permalink


    I know that dillweed.

  428. tetris
    Posted Aug 28, 2007 at 3:59 PM | Permalink

    Re: 426
    Getting one series/year [2005] more or less right is one thing. Is there a Meteo France forecast for 2006, verifiable against observational data?

    The absence of a timely Meteo France forecast for 2007 is interesting and unfortunate, but maybe not too hard to explain. Elaborating a complex, non-linear, multi-variate predictive model that must include reliable feedbacks in order to calibrate, is not only a very tall order in and of itself, but clearly requires several years of work.
    For that reason alone, I would not put any money on Curry’s forecast that the empirical “stamp collectors” are soon going to be out of business.
    In intelligence circles there is a saying that one observation is an occurance, two highly similar ones probably coincidence, and that only after three such observations is it probably time to start thinking pattern.

    Judy Curry’s views appears to be firmly on one side of the now all too familiar demarkation line that separates trumping unproven and unreliable model outcomes over understanding what’s actually happening out on the field.

  429. steven mosher
    Posted Aug 28, 2007 at 4:34 PM | Permalink

    re 423.

    Yes, I’m fiddling with the quantities and final form.

    I considered the following to make it resonate emotionally.

    ( damage in 2004 dollars/Katrina Damage in dollars)/

    (wind speed in Knots at landfall/Katrina knots at landfall)

    In words the numeratr is normalized to a Katrina dollar, where
    more damage > 1 and less damage is 1 and and a lessor storm is

  430. steven mosher
    Posted Aug 28, 2007 at 4:36 PM | Permalink

    re 427?

    Why.. well the white soxs suck.

  431. David Smith
    Posted Aug 28, 2007 at 4:39 PM | Permalink

    A while back I commented on some of the Meteo France global results, which were not very impressive. The thread was “The 2006 Hurricane Season”, the post was #80, which I think is linked here .

  432. Bill F
    Posted Aug 28, 2007 at 5:07 PM | Permalink

    I am going to propose another damage index for use. I call it the “Dumbenoughtobuildthere” index or the New Orleans Index. The index is calculated by dividing the value of what was damaged (Vd) by the total value of the properties within the landfall footprint (Vt) to get the damage ratio (D). Then you take the maximum windspeed at landfall (Wmax) and divide it by the size of the hurricane force windfield (A) to calculate the storm’s strength (S).

    To calculcate the New Orleans Index, you just multiply the S by the D. A really strong storm that does lots of damage to a sparsely populated area will be ranked higher than a weak storm that does relatively minor, but expensive damage to a place like South Beach, because both the D and P will be higher in the former. Obviously there will have to be adjustments made for situations like New Orleans, where engineers, residents, governments, and others conspired to unwittingly build a city in the worst possible location and then surround it with levees that only made the problems worse. I propose assigning areas of coastline a value between 1 and 5 before the season begins, based on how well the area has prepared and attempted to prevent the damage that could be caused by a hurricane. The factor would be used as divisor to the NOI to take into account how predisposed the area was to catastrophic damage. NO would obviously get a 5, while places that restrict building directly on the beachfront and actively enforce laws designed to make buildings storm resistant would receive a 1 or a 2.

    Any thoughts?

  433. steven mosher
    Posted Aug 28, 2007 at 6:41 PM | Permalink

    RE 430.

    Hey WTF ??????? my genius calculation got snipped. Evil Mohel.

    Anyways the function looks like this..

    Numerator = Damage normalized to Katrina damage dollars ($100B)… SO, Dean is 1Billion/100B or .01

    Denominator = Energy at landfall normalized to Katrina (125knts) so dean is 150/125 or 1.2

    Final units are “katrinas” so Dean is (.01/1.2) Katrinas

    SO the final blowhard ratio mosher index for dean is .008 Katrinas.


  434. David Smith
    Posted Aug 28, 2007 at 6:54 PM | Permalink

    I bought the EUROSIP article but it will be a few days before I read it, as I need to get some USHCN surveys posted on Anthony’s site and take a look at mid-tropospheric satellite-derived temperatures.

    On the USHCN sites I found one MMTS that had apparently been “egged” ( a US teenager tradition of throwing uncooked eggs at things) and one next to a big propane storage tank. If that propane ever lights off I guess we’ll have a new US high temperature record.

  435. Kenneth Fritsch
    Posted Aug 28, 2007 at 6:59 PM | Permalink

    Re: #433

    Any thoughts?

    I do not want to get politics involved here so I will say simply that at least one prominent US presidential candidate has made a pitch to rebuild New Orleans. I doubt that you will hear an up roar from the AGW inclined climate scientists about that even though when they talk about protecting future generations they can get shrill. Without a doubt building or rebuilding in harms way will do more and more immediate damage than some not so certain distance climate change will. While I have no problem with private citizens making a calculation on potential damages that they will face from rebuilding without others help, this is not what politicians have in mind.

    With politicians willing to buy immediate votes for such irresponsible actions how in the world do we expect them to do the right things in mitigating any of the potential effects of AGW. We have to some how put together a political storm index that would probably start at dumb and proceed to dumber and dumbest.

  436. David Smith
    Posted Aug 28, 2007 at 7:23 PM | Permalink

    This is the wrong thread for a MMTS photo but since I mentioned it, here is a shot of a rural USHCN station (Columbia, MS) with the fuel tank in the background. The fuel tank is about 12 feet from the MMTS. (I said propane earlier but it may be some other fuel.)

  437. steven mosher
    Posted Aug 28, 2007 at 7:26 PM | Permalink

    RE 433.

    Its a sad thing.

    When I first moved to LA and drove up PCH I saw Ocean to the left of me and Mountains to the right.
    Million dollar homes on stilts nearly below high tide. and million dollar homes on stilts carved
    into hillsides.

    Midwestern boy who lived through tornados shook his head. Fires came. Storm surges came.
    And they rebuilt on quicksand.

    I used to get SCIamerican. Long before Katrina they had published an artcile on the threat
    to NOLA. I loved NOLA. ( somewhere here find a story about my trip to Natchez MISS.. which ended in NOLA)

    Anyway, when you travelled there you just shook your head.

    2-3 days out from landfall I pulled out that old scientific american. I shook my head.

    I hoped they wouldnt rebuild.

    I think If you take AGW seriously you have to impose costs ( taxes) where the ROI is greatest.
    Essentailly, dont control carbon. Control develpment in areas likely to see prblems

    (Drought areas, flood areas, coastal) Tax MALIBU.

  438. Bob Koss
    Posted Aug 28, 2007 at 8:18 PM | Permalink

    If the fuel tank is 12 feet then the concrete block wall must be within 10 feet.

  439. David Smith
    Posted Aug 28, 2007 at 9:56 PM | Permalink

    Re #439 Indeed it is.

    The dark wooden pallets and poor drainage are also of interest, because they affect nearby temperatures to a small extent and they can change over time (pallets get moved or rot, drainage patterns change), thus creating spurious trends.

    Also, there is a large concrete parking area about 30 feet to the south and a large asphalt parking lot 40 feet to the north.

    This site is not what I consider “rural” but it is listed as such.

  440. fFreddy
    Posted Aug 29, 2007 at 12:43 AM | Permalink

    Re 430, 434, steven mosher
    The blog software thinks that a less than sign is the start of an HTML tag. IF it can’t resolve the subsequent text to a tag, it goes into a sulk and throws it away.
    If you want a <, use the following four characters & l t ; without the spaces. The semi-colon is not needed in the preview window, but is needed once posted.

    We really need this as a perma-note by the preview window …

  441. MarkW
    Posted Aug 29, 2007 at 5:51 AM | Permalink

    Propane tanks have rounded ends, the end of this tank looks flat.
    Additionally that blue switch mounted on the wall looks like a fuel pump control.

    I would guess the tank holds an unpressurized liquid fuel, probably diesel.

  442. SteveSadlov
    Posted Aug 31, 2007 at 10:17 AM | Permalink

    No additional named storms for August. Possibility of one for early Sept if the wave at the junction of the Windwards and South America does not dissipate against the SA land mass. Of course, this very scenario attests to a key factor preventing named, truly tropical storms this year, the ITCZ being stuck so close to the equator. The weather is not able to make it around the NE corner of SA and as a result, few TCs:

  443. David Smith
    Posted Aug 31, 2007 at 10:46 AM | Permalink

    Re #443 It probably depends on how name-happy the NHC is this afternoon. The system near the Antilles has a strong chance of becoming a tropical cyclone though it may never amount to much due to proximity to land. There is also a swirl near New England which would have been ignored in years past but which now might get named.

    Maybe we’ll see Felix and Gabrielle by sunset, to help pad what has been a mediocre season so far 🙂

    The remarkable thing about 2007 in the Atlantic is how stable the air has been. Seedlings have withered.

    Worldwide 2007 has been, at most, an average year in storm count. I’m suspecting it will be markedly below-average on ACE.

  444. Kenneth Fritsch
    Posted Aug 31, 2007 at 6:49 PM | Permalink

    Re: #414

    The EUROSIP paper on the seasonal forecasts by the 3 European centers has now been published

    unfortunately there isn’t a free online version available.

    I spent my $9 to buy the article to which Judith Curry linked and did not find any predictions from Meteo France for 2007. The article discusses the work of Meteo France which combined with two other organizations (Met Office and ECMWF) gives a forecast known as EUROSIP. I googled EUROSIP and found that it does not have a published prediction either. Can anyone please tell me if EUROSIP indeed publishes their predictions before the fact and if not why not and why should we take their predictions serious without reporting before the fact?

    Judith Curry seems to think that the modelers who supposedly use the laws of physics to predict cyclone activity in the NATL will beat the pants off the empirical users of past data. Now I have no dog in this … sorry Michael V. and US football fans… no horse in this race. As a matter of fact I have my doubts about the capabilities of empirical models (that keep changing year over year and are tweaked by outside human intervention). A model truly encoded with the laws of physics or “dynamical model” as they are called would not be subject to the over fitting that empirical models are and one could confidently use past data to confirm the models capability to predict as in “re-forecasting”. In- and out-of sample testing would become less important with the dynamical models.

    Now as I see it a model based purely on the laws of physics could yet remain susceptible to over fitting if one could pick and choose the laws they used to construct the model and then peek at the fit before going back and adjusting the model to physical laws that gave a better fit. The paper linked here does indeed talk about model “drift’ away from “observed” climate and using past years’ data for an adjustment. It also talks about using the media and mean storm frequencies for this adjustment and changing from the mean to the media for a “better fit”

    When one looks at EUROSIP results it must be noted that 1993-2004 was from reforecast while only 2005 and 2006 were in real time.

    When I look at empirical or dynamical forecasts I see a large part of process involved with forecasting all the variables in let us say June that are known to effect tropical storms for the entire storm season. I would prefer to see the process broken down by how well the model predicts given all the effective variables let us say a week ahead. From results I have seen the closer the predictions are made to the events occurring the more accurate they become. In effect what I see these models doing in a critical fashion is making relatively long term weather forecasts.

    The paper linkeed above statistically compares the EUROSIP dynamical process with the empirical processes of CSU (Gray and Klotzbach) and TSR (Tropical Storm Risk) and shows that over the period 1993-2006 EUROSIP performed significantly better. The dynamical systems do not, however, attempt to forecast storm intensities or landfalls as the empirical ones due, so despite, Judith Curry’s encouraging words they appear to have a ways to go before being practically depended upon. I suspect that dynamical models have a greater dependency on SST predictions and SST is a more important variable in that system than it is in the empirical models and that might explain who is in the respective cheering sections.

    The linked paper states that EUROSIP made real time forecasts each month for the last 2 years. In that case we should have a June, July and August forecast for 2007 to compare with the actual numbers. Does anyone know where I might find these numbers? If the forecast is monthly, then one would want month by month comparison to actual to go along with the seasonal averages. As a mattter of fact the seasonal average under those predicting conditions becomes very much less important than the monthlies.

  445. John Lang
    Posted Aug 31, 2007 at 7:16 PM | Permalink

    The Eurosip paper does not contain an actual forecast for 2007 ???

    I mean, who can take them seriously at this point. Sorry Ms. Curry, it is ridiculous to hype some forecasting model which is always carried out after the fact.

    My model of the S&P index is 95% accurate (before the fact) and 99% accurate (after the fact.) That just means I am losing money and do not perform as good as a random model.

    I think the real problem (in all of this) is that global warming researchers do not understand basic number theory (see any paper by Mann or Curry).

  446. David Smith
    Posted Aug 31, 2007 at 8:25 PM | Permalink

    Ken I’ve made my first read of the EUROSIP paper and have sat it aside for a day before making my second read. I saw forecasts for 2005 and 2006 only, and neither was impressive (for 2005 EUROSIP says they forecast 16 storms while 27 actually occurred, and for 2006 they say they forecast 12 while 10 actually occurred, one of which was not detected until several months after the season ended). Anyway, I’ll comment after the second read.

  447. SteveSadlov
    Posted Aug 31, 2007 at 8:40 PM | Permalink

    I just looked at the imagery for so called TD6. What a joke. Three blobs of moisture / convection, a garden variety, run of the mill T-storm clustes. Seen scads of them near the Philippines, almost a daily event this time of year. I note that the NHC has not issued an advisory update for 5 hours. Hmmmmmm….. We’ll it’s nearly midnight ADT, we’ll see if they try to pull a fast one and slip it in as a “borderine TS” a few minutes from now. Failing that or other chicanery, no more named storms for August.

  448. tetris
    Posted Aug 31, 2007 at 9:33 PM | Permalink

    Am I missing something? The 2005 and 2006 forecasts in the EUROSIP paper are “postcasts” unless they were actually produced before those dates.

  449. David Smith
    Posted Aug 31, 2007 at 9:56 PM | Permalink

    Re #449 tetris, my understanding is that the 2005 and 2006 forecasts were provided to paying customers and not to the general public until after the season was over. I take them at their word though of course it would be nice to see their full range of forecasts (all basins) prior to the start of the seasons.

  450. windansea
    Posted Aug 31, 2007 at 10:26 PM | Permalink

    Not much left to say about this. Instead of being unusually active, it looks like the current hurricane regime is simply a return to more normal conditions, following an unprecedented tranquil couple of decades.

    Nyberg, J., B.A. Malmgren, A.Winter, M.R. Jury, K.H. Kilbourne, and T.M. Quinn. 2007. Low Atlantic hurricane activity in the 1970s and 1980s compared to the past 270 years. Nature, 447, 698-702.

  451. Kenneth Fritsch
    Posted Sep 1, 2007 at 11:26 AM | Permalink

    Re #449 tetris, my understanding is that the 2005 and 2006 forecasts were provided to paying customers and not to the general public until after the season was over. I take them at their word though of course it would be nice to see their full range of forecasts (all basins) prior to the start of the seasons.

    If they are providing their information to paying customers then, yes, I can understand why they do not provide the predictions upfront. The problem with that situation is that the model they use will be proprietary and not available for independent verification.

    If indeed they use an unchanging model not based on peeking at past results and data then, of course, their re-forecasts are as valid as their real time forecasts. This in turn brings up 2 other important items to consider. If they do indeed do monthly forecasts as indicated in the paper then I would hope that they would give the results on a monthly basis. Finally, if I have a model based on the laws of physics and not peeking at recent past data and results, I should be able go back in time much further than they have (1993) and look at model performance. I do not know what the computer time would cost but surely going back further in time to give more data points of validation should be a major marketing point for paying customers – as would, of course, using monthly results.

    By the way, if the most critical part of the tropical storm forecasting process is long term weather forecasting I have no doubt in my mind that a computer program can best the best efforts of humans attempting to use regression calculations with past data. My calculations of CSU results said that their earliest forecast were not statistically better than random selections. I think that instead of comparing their results to empirical methods, the dynamical methods will have to demonstrate they have practical value to the paying customers. I have great faith in the market place (unfettered) to judge these issues.

  452. tetris
    Posted Sep 1, 2007 at 5:19 PM | Permalink

    Re: 452
    I agree with your faith in the [unfettered] market place. Only those models that either allow for the verifiable avoidance of costs or the verifiable increase in revenues will survive. If it’s indeed proprietary, I doubt we will see monthly results.

  453. David Smith
    Posted Sep 1, 2007 at 8:46 PM | Permalink

    The Vitart et al paper discusses their method (EUROSIP) for seasonal forecasting of Atlantic tropical cyclones.

    ** EUROSIP, like Gray/Klotzbach, NHC and others, attempt a valuable service: seasonal forecasting of tropical storms. That is laudable. It is also very difficult, as storm count is a function of broad weather patterns which have to interact with each other in just the right ways and locations, as well as being a function of mesoscale factors and changes in detection technology.

    ** My view is that current knowledge allows a reasonable guess as to whether a season will be average, below-average or above-average, but that’s about it.

    ** Like Kenneth Fritsch, I have no dog in this fight other than a suspicion of anything Judith recommends 🙂

    ** Vitart’s storm count and SST charts can be replicated and they calculate correlations only with unsmoothed data! That’s an improvement over some other hurricane papers.

    ** Vitart uses three computer models, similar I think to the ones used today for weather forecasting. Initial conditions are fed into the models and then they generate runs covering a number of months.

    ** However, these models generate with Vitart calls “…climates that are somewhat different from the observed condition.” That’s a polite way of saying that the models drift away from reality (similar to the weather models). Errors compound.

    ** So, Vitart applies a “calibration” factor (known in the US as a “fudge” factor). The fudge factors are derived by seeing how each model behaved in prior years (1993-2004) and then calculating the fudge factor which creates the closest match to the actual storm counts.

    ** Then the three models are combined (median or mean, the paper looks at both) to produce the EUROSIP forecast. The final product is basically an ensemble forecast.

    ** One bone to pick: the paper compares how the current EUROSIP would have performed against historical analog/statistical forecasts (Gray, etc) for 1993-2004. This is a rather disingenuous comparison unless the analog/statistical forecasters are allowed to use their current knowledge to re-forecast 1993-2004. Comparisons need to be apples to apples.

    ** The EUROSIP approach reportedly does a good job of forecasting SST (the Pacific El Nino region and the Atlantic), perhaps better than the forecasts used by the analog/statistical folks. This is quite important, especially the Pacific forecast. The analog folks failed to foresee the 2006 El Nino, which apparently brought the season to an early end, while EUROSIP picked up on it.

    ** The only true forecasts are the ones reported for 2005 and 2006. That’s an extremely short record to reach an opinion, let alone a conclusion.

    ** For 2005, EUROSIP forecast 16 storms while Gray forecast 15 and TSR forecast 14. The actual number was 27. All were poor.

    ** For 2006, EUROSIP forecast 12 storms while Gray forecast 17 and TSR forecast 14. The actual number was 10 (arguably it was only 9). EUROSIP was closest. It detected the pending El Nino while the analog/statistical forecasters did not.

    ** Since ENSO plays an important part in Atlantic activity, a method which does a superior job of detecting ENSO changes would have a forecasting edge. If EUROSIP does that then that is a good thing. But if the EUROSIP ENSO method is superior then the results will tend to become known and incorporated into ENSO forecasting, producing ENSO forecasts available to all, including the analog/statistical people. The edge will diminish.

    ** Conclusion: the EUROSIP results (two years) are unimpressive but their 2006 was not as unimpressive as the analog/statistical methods. If the reputation for best forecaster hangs on 2006 then the prize goes to Steve McIntyre, who came closest 🙂

  454. David Smith
    Posted Sep 2, 2007 at 6:20 PM | Permalink

    Global notes, as we enter the peak of the Northern Hemisphere’s season:

    Hurricane Felix has achieved category 5 status in the Caribbean this Sunday evening. It is experiencing zero shear and deep warm water and has another couple of days over open water, so it may end up in the 900mb pressure range, which is extremely strong (but not a record). The Yucatan is in its path.

    Turbulence inside Felix is so strong that the recon aircraft had to abort. (On a side note, the NHC used a term (“groupel”) which is not in the English language so far as I can tell. It does seem to have a specific meaning to them. Perhaps this is how words are born.)

    Near Japan is a typhoon whose track may take it quite close to Tokyo. It will likely be losing strength at that time but could still bring strong winds to a major metropolitan area.

  455. steven mosher
    Posted Sep 2, 2007 at 6:55 PM | Permalink

    re 455.

    did they mean

  456. David Smith
    Posted Sep 2, 2007 at 7:04 PM | Permalink

    steven, I think you’re right! They misspelled it twice but graupel seems close enough.

    Here’s an excerpt:


    At first glance it seems unlikely that 2 to 5mm snow balls would cause a problem but then again they may be quite bad at 400mph and especially for instruments. Maybe they’ll explain later.

  457. Kenneth Fritsch
    Posted Sep 2, 2007 at 8:24 PM | Permalink

    Re: #454

    Good review, David. Just curious but did you pay $9 for the article also.

    I look at these forecasts with the same eye (a skeptical one from experience) that I do investment strategies. If, in this case of EUROSIP, they can show that it indeed is constructed from first principals, the last 2 years are not differentiated from the previous years for purposes of out-of-sample and in-sample designations and we can use as many past years as practical in determining the skill of the forecast. I have some doubts because they talked about the adjustment(s?) for drift and peeking to find that using median frequency for the adjustment may be better than the mean frequency.

    I really think that to legitimize their claims they need to use the forecast going back in time to look at a complete cycle of minimum and maximum storm activity. This exercise is comparable to determining whether an investment strategy is only good for example when value equities are outperforming the growth category.

    If they are making monthly forecasts, which actually make more sense for practical use, then they should give their monthly forecast versus actual performance. If they do not provide extended data in this form or for that matter the cheering sections do not demand it then I will have some major doubts about the skill of their forecasts. Like I said before truly paying customers are the best sources of these demands. On the other hand, if EUROSIP is a government run or funded organization and it is being paid by these same governments for their proprietary information, I would have to revise my customer expectations.

    Do we know how EUROSIP, and/or the 3 organizations that contribute to it, is organized and who their customers are?

  458. steven mosher
    Posted Sep 2, 2007 at 8:54 PM | Permalink

    will felix out blow fifi

  459. David Smith
    Posted Sep 2, 2007 at 9:42 PM | Permalink

    Re #458 Yes, Kenneth, I paid the $9. Maybe Exxon will leave three gallons of gasoline under our doormat to cover our out-of-pocket costs. 🙂

    The EUROSIP models generate unrealistic climates and storm counts which have to be adjusted by an empirically-derived fudge factors. For example, model A is found to generate roughly twice as many storms as actually formed in 1993-2004, so in the future model A’s output is multiplied by 0.5 . The same approach is taken for the other models. The final result is part physics and part empirical.

    Ultimately EUROSIP may fall into the trap of having been tuned (fudge factors determined) in a particular type of climate (the early years of an active phase of AMO), then, when the climate shifts, the fudge factors no longer work well. Sounds almost like a stock market.

  460. fFreddy
    Posted Sep 3, 2007 at 1:52 AM | Permalink

    Re #460, David Smith

    For example, model A is found to generate roughly twice as many storms as actually formed in 1993-2004, so in the future model A’s output is multiplied by 0.5

    And somebody pays for such a forecast ? Wow …

  461. David Smith
    Posted Sep 3, 2007 at 7:16 AM | Permalink

    Re #461 Individual weather models have their strengths and weaknesses and “moodiness”, almost like personalities.

    The CMC model creates far too many Atlantic hurricanes (maybe 50 to 100 a year) while others, like ECMWF and UKM, are more stoic and much slower to create storms. The GFS is like the girl in the nursery rhyme, “When she’s good, she’s very very good, but when she’s bad she’s horrid”.

    The GFS, for example, forecast Dean far in advance, even before the seedling was detectable. On Felix the GFS is at the other extreme, failing to even show Felix’ existence even though Felix was a major hurricane.

    But, if you combine these into an ensemble, and if the models are predicting about the same thing, then you can get a reasonably good prediction of events. However, the farther out in time you go, the worse the prediction.

    If I was an emergency response planner would I pay modest money for these seasonal forecasts? Yes, but only because the other costs and impacts dwarf the cost of a seasonal outlook.

    Perhaps the thing to do is to get Las Vegas interested in seasonal hurricane forecasts, and let them develop odds. They have no motive other than to be right and have to stake their positions unambiguously and in advance of the season.

    Regarding ensembles, it’ll be interesting to see if the ensemble, or subgroups, of our contest forecasters come closest ot the actual 2007 seasonal results.

  462. fFreddy
    Posted Sep 3, 2007 at 9:24 AM | Permalink

    Re #462, David Smith

    The GFS, for example, forecast Dean far in advance, even before the seedling was detectable. On Felix the GFS is at the other extreme, failing to even show Felix’ existence even though Felix was a major hurricane.

    David, I’m sorry, but I think that means that it is guessing.

    If I was an emergency response planner would I pay modest money for these seasonal forecasts? Yes, …

    Well, would you ? Surely the main decision such a planner has to make is whether to issue a “batten down the hatches” warning to his population. It would be easy to do this whenever there is the slightest chance of a storm, but the population would rapidly tire of the inevitable false alarms. It is easy to do this too late, when the roof has been blown off the mayor’s office, but this is not much use to the population.
    I would have thought that the only thing that is really useful to the planner is improved certainty on the scale of 12-48 (?72 ?96) hours. Anything beyond that is just telling him that he lives in a hurricane zone.

  463. Judith Curry
    Posted Sep 3, 2007 at 9:42 AM | Permalink

    The EU funds ECMWF, and French govt funds Meteo France and UK govt funds UK Met office. Meteo France has gone public with their seasonal forecast this year, the other two have not. Since hurricanes don’t directly impact Europe (but only their islands in the caribbean and possible shipping operations), hurricane forecasts are a low priority for the european weather offices. Nevertheless, their global coupled atmosphere/ocean seasonal forecast models (unlike NOAA, which is atm only) show great promise for seasonal hurricane forecasts since these models have pretty good credibility in forecasting el nino/la nina, and hence they are investing a small amount of effort in seasonal hurricane forecasts.

    David, in your review of the Vitart et al paper you missed the essential contribution of the EUROSIP ensemble forecasts, which is the potential for probalistic forecast and capturing extremes. In 2005, while the average of 16.4 is not too impressive, the ensembles predicted 37% probability above 20, and 13% probability above 27. The other key issue is model resolution. From a ppt presentation given by vitart (not published or publicly available: At T95 resolution (about 200 km), which was the resolution reported in the vitart GRL paper, you can pick up a majority of the storms, but intensity and tracks are not captured. At T159 (about 125 km), which is the resolution ECMWF is using in the 2007 seasonal forecasts, you pick up much more credible tracks and can start to distinguish intensity. At T511 50 km (which is what ECMWF uses in its 15 day ensemble prediction system), you can capture both intensity and tracks, in fact a hindcast test simulation of the 2005 seasonal forecast showed quite credible tracks and intensities.

    The ECMWF ensemble prediction system (15 day forecasts, 51 ensembles, about 55 km resolution) can do very well with medium range TC prediction. THis is not a high priority at ECMWF, although they are apparently running TC tracking scheme through their simulations. In fact, i enquired at ECMWF as to whether their TC forecasts could be purchased, and the said no they are experimental. I think this is also why ECMWF has not gone public with its seasonal forecasts; this has not been a priority for them, they view them as experimental. Several private sector forecasting groups have purchased the ECMWF real time ensembles (15 day forecasts) and are apparently doing such forecasts for private clients. I have seen some of this, and ECMWF ensemble prediction system has done an excellent job so far this season in terms of picking up the tropical cyclones.

    How to do the tracking is a subject of active research. further, the whole issue of how to put together a multi-model ensemble forecast is a topic of active research, see the attached link for further info on this.

    Click to access 3a.3Stockdale.pdf

    The bottom line is that as the coupled atm/ocean seasonal forecast models go to higher resolution (say 60 km or so), the ensemble seasonal forecasts of tropical cyclone activity will become far better that the regression/analogue statistical forecast schems such as used by Gray. Even at coarse resolution, they are already better. further, sectors such as reinsurance, energy trading, etc who make use of the seasonal and medium range forecasts for financial decisions all hire a plethora of private sector forecasting groups who presumably are making useful probabilistic forecasts to support their decision making. This is a pretty competitive business and some of these advances are not made public, since the “payoff” here is making money in the financial markets and in the weather risk management industry (rather than scientific publication, scientist prestige, etc). The 3-5 day forecasts made by the National Hurricane Center are what is needed for emergency management, but these longer range forecasts are of apparent great utility and consequence in financial markets.

  464. Demesure
    Posted Sep 3, 2007 at 9:59 AM | Permalink

    Meteo France uses 6 models in addition to their Climat-Arpege model to make ensemble forecast.
    Unfortunately, their seasonal bulletins for precipitation are absolutely useless; they keep saying systematically since the beginning “for precipitation, there is no privileged scenario” (“pour les précipitations, aucun scénario n’est privilégié)”.

    The bulletin of May 30 2007 for JJA said:
    Models are unanimous : temperatures will be above seasonal mean values” (just a reminder: France hasn’t had such cold and rainy summer for years).

  465. tetris
    Posted Sep 3, 2007 at 11:32 AM | Permalink

    Re: 464
    The industry sectors you mention more often than not make their money by forecasting high [re-insurance and oil futures in particular] since the play the fear factor so I have a problem with those as some sort of a justification for anything. More to the point, what is the ECMWF forecast for named NA storms/actual hurricanes for the 2007 season [we’ve only got some 6 weeks to go]?

  466. Judith Curry
    Posted Sep 3, 2007 at 12:14 PM | Permalink

    Actually, the private forecasters make money by providing useful forecasts that verify with the actual observations, there is little long term pay off for systematically biasing the forecasts. The only thing I have heard from ECWMF for this season (and last i heard was the result of their july forecast) was a very informal email that said that they were not predicting an overly active season, looked pretty average (the average they are working off of is 12). I assume that they still view this as a developmental product and are not going public with it (and are not selling the forecasts). Hopefully their forecasts will be available for the 2008 season, but again hurricane forecasting is a low priority for them. The private sector guys of course are not making public what they do with the ECMWF simulations.

  467. steven mosher
    Posted Sep 3, 2007 at 12:58 PM | Permalink

    My Hurricane prediction model has been 100% accurate for the past 20 years.

    My 2008 prediction is available today at the reasonable sum of 1Million dollars .

    This is a money back offer. If our prediction is off by 7 or more storms, your
    entire deposit will be reimbursed.!!

  468. steven mosher
    Posted Sep 3, 2007 at 12:58 PM | Permalink

    My Hurricane prediction model has been 100% accurate for the past 20 years.

    My 2008 prediction is available today at the reasonable sum of 1Million dollars .

    This is a money back offer. If our prediction is off by 7 or more storms, your
    entire deposit will be reimbursed.!!

  469. Kenneth Fritsch
    Posted Sep 3, 2007 at 1:16 PM | Permalink

    Re: #464

    Judith Curry, thanks for confirmation of my research of the 3 organizations that contribute to EUROSIP. Evidently these are organizations that are directly funded by government and do much work for governments, but in at least one of them and probably all, I noticed that they are expected to operate in the private sphere where they would have paying customers. It is private paying customers that would be the best judges of the skill of these projections. I personally would have little faith in the marketing decisions of something transacted from government to government. Can these private customers be identified?

    David, in your review of the Vitart et al paper you missed the essential contribution of the EUROSIP ensemble forecasts, which is the potential for probalistic forecast and capturing extremes. In 2005, while the average of 16.4 is not too impressive, the ensembles predicted 37% probability above 20, and 13% probability above 27. The other key issue is model resolution.

    Just to put the computer models probabilistic forecast into perspective based on a Poisson distribution model with an average storm frequency of 16.4, one would obtain a probability of 15.6% for 20 or more storms and 0.6% for 27 or more. For an average of 12 storms the Poisson model would yield probabilities of 1.6% for 20 or more and 0.01% for 27 or more storms.

    The important point I learned for my $9 is that the resolution must be increased dramatically before the models can forecast quantities such as would be required for determining ACE.

    Judith, you seem to have insider information that you are coyly referencing about the model forecasts. You sound as if you may be under a nondisclosure agreement with the forecasting organizations. Can you reveal the periods used for the model forecasts, i.e. are they monthly? And if monthly why would not the modelers reveal monthly forecasts versus actual results? Why would they not use the model in re-forecast back in time to encompass an entire max and min frequency cycle? Unless more can be revealed publicly about these models, I would doubt that scientific statements or analysis of skill could be objectively made.

  470. David Smith
    Posted Sep 3, 2007 at 7:42 PM | Permalink

    Judith thanks for your contribution. I hope you had a good (US) holiday and I imagine at least some of your students are quite happy following the trouncing of ND.

    In #464 you note:

    David, in your review of the Vitart et al paper you missed the essential contribution of the EUROSIP ensemble forecasts, which is the potential for probalistic forecast and capturing extremes. In 2005, while the average of 16.4 is not too impressive, the ensembles predicted 37% probability above 20, and 13% probability above 27.

    Actually I did see the statement, which is one sentence in a five-page paper otherwise devoted to storm count. Storm count was the yardstick noted in the abstract and throughout the body of the paper and is what Vitart et al chose as the measure of EUROSIP’s performance versus Gray, TSR, etc.

    To stay apples-to-apples in their comparison Vitart et al should have provided any appropriate cautions or qualifiers which Gray, TSR, NHC, etc had (or lacked) in their 2005 forecasts, or omitted the sentence.

    Despite that, the idea of a seasonal probability curve is a good one and would be a considerable improvement over the present count-range approach. Storm count alone, as I think people know, is an inadequate measure of a season (or a forecast). Perhaps Vitart will offer a paper on the topic one day.

    On the rest of it, there’s simply no EUROSIP data for us on the outside to form a meaningful opinion about EUROSIP’s performance. As Jerry Maguire might say, “Show me the data”!

    On models, I took a look at the European (ECMWF) model forecast made seven days ago (27 August) for today (3 September). The reality is that, at 12Z today (September 3), there was a category 5 hurricane in the Caribbean (Hurricane Felix). For anyone who is curious what the ECMWF computer model forecast just seven days ago, check here . (For those unfamiliar with these maps, a hurricane would be a small but distinct collection of circles with some yellow and green – you’d recognize it if it was there on the ECMWF forecast map.)

    I simply have little confidence in computer models beyond a week or so, especially on small-scale features like hurricanes. This current ECMWF performance does nothing to grow my confidence.

  471. steven mosher
    Posted Sep 3, 2007 at 7:57 PM | Permalink

    It’s Poisson.

    But some can predict it better than others

    V2 rockets. Gravity’s rainbow. I think I will be the Tyrone Slothrop of Hurricanes.

  472. David Smith
    Posted Sep 3, 2007 at 8:17 PM | Permalink

    For anyone interested in what a hurricane looks like on the European (ECMWF) map, note the small circles and color south of Baja California on this older map . Compare that to the map for 3 September ( here ) and see if you can spot Hurricane Felix in the Caribbean.

  473. steven mosher
    Posted Sep 3, 2007 at 8:31 PM | Permalink


    What the best link to histrical records. When was the last time batter #1 and #2 were

  474. David Smith
    Posted Sep 3, 2007 at 8:47 PM | Permalink

    And, for anyone who likes animations, here’s a list of some of the key computer models used for hurricane forecasting. Click on a “submit” button, then on the next screen click on “FWD” to start the animation. Hurricanes are concentrated circles in the warm regions. The colors represent air pressure. You can watch the computers generate and move storms. Neat.

    When the models agree on storm formation then the odds of a storm forming are strong. For instance, they currently (September 3) mostly agree that Gabrielle will form north of The Bahamas later this week. I think that’s a pretty good prediction. We’ll see!

  475. David Smith
    Posted Sep 3, 2007 at 8:58 PM | Permalink

    Re #474 steven the easiest source is Unisys . Also, for interesting tidbits and “color commentary”, whoever does the Atlantic hurricane web pages for Wikipedia does a very good job.

  476. tetris
    Posted Sep 3, 2007 at 11:59 PM | Permalink

    Re: 476
    On the international “feed” {Reuters, AFP, etc.] we have in the past few hours gone from a Felix “catastrophic” cat 5 to “still catastrophic” cat 4, with all sorts of allusions to hurricanes past, while leaving wide open the posibility that the outcome may be quite less dramatic.
    In doing so, Reuters earlier this morning explicitely argued that Felix was yet more “proof” of an increase in NA hurricane activity and growing strength due to AGW.

    AGW and largely all of its purported tangible consequences, e.g. from the melting of Antartica to the increase in hurricane frequency and strength, remain unproven.
    So, with a smile, I do have have a dog in this fight. Given that we’re half way through the hurrricane season [with Felix as the “cat in the bag’, as it were..] many of us would no doubt appreciate your comments as to why what’s happening out on the water doesn’t appear to be living up to the billing.

  477. SteveSadlov
    Posted Sep 4, 2007 at 10:06 AM | Permalink

    Well Felix happened against all odds (that’s right, it was not off to a very robust start, bouncing along the Venezuelan coast – had it gone just a click more on a direct westerly course, it never would have been more than a TD. What this points out in dramatic fashion is just how much the equatorward squashing factor is in force. The ITCZ is squashed in toward the equator. The formative regions are therefore very constrained versus normal. Something else to keep an eye on. It looks like the jet stream wants to adopt a Siberia Express orientation over the course of the next week. A block may get set up out between the 120s W longitude and the international date line. The jet may start to loop through the Yukon, down through the Columbia Plateau, over California heading SSW then dipping into northern Mexico, then back up into Texas and the mid South. It will be insteresting to see how, if this happens, it impacts TC development in the Atlantic basin.

    Oh, and here are the bogus and real counts (bogus counts Gulf of Alaska origninated Andrea and Chantal, Real excises them … I am no longer quibbling about Barry and Erin, my resolution for September is to be generous!):

    Bogus: 6
    Real: 4

  478. SteveSadlov
    Posted Sep 4, 2007 at 2:12 PM | Permalink

    So, it probably comes as no surprise that, although TD6 did not become TS Felix until after midnight local time, the NHC counted it as an August named storm. I am really trying hard to avoid seeing conspiracies and book cooking …. but …..

  479. Judith Curry
    Posted Sep 4, 2007 at 3:16 PM | Permalink

    Well in terms of cats and dogs, i would say that two cat 5 hurricanes already in the North Atlantic this season is pretty impressive. In terms of the predictions of 17 by Bill Gray et al., I have said all along the statistical forecasting methods have little to no skill.

  480. Kenneth Fritsch
    Posted Sep 4, 2007 at 3:37 PM | Permalink

    Re: #464

    Meteo France has gone public with their seasonal forecast this year, the other two have not.

    Judith, could you (or any other poster) please give us a link to the Meteo-France seasonal forecast or state what that organization forecasted for 2007. I was able to find the link below to the Met Office (UK) forecast for 2007 in USA Today of all places. I find nothing on Meteo-France.

    By Jessica Gresko, Associated Press Writer

    MIAMI — British government forecasters are predicting that the Atlantic hurricane season may not be as busy as their American counterparts expect.

    It is most likely that 10 more tropical storms will form from July to November, the British forecasters said Tuesday. An expected cooling trend in Atlantic Ocean surface waters favors fewer tropical storms than in recent years, the British meteorologists said in their first-ever hurricane season forecast.


    FIND MORE STORIES IN: Hurricane | British | Storm | Atlantic | National Oceanic and Atmospheric Administration | British government | Met Office
    Colorado State University researcher William Gray’s team also predicted double-digit tropical storm numbers. Gray predicted 17 named storms and nine hurricanes, five of them intense.
    Matt Huddleston at the U.K.’s Met Office, a weather tracking agency within the British Ministry of Defense, said its numbers are based on a “brand new forecasting system” using a global climate model.

    Huddleston said his office’s prediction is based on a so-called “dynamical” model, more like what is used for daily weather forecasts. American forecasters do not currently use that model for seasonal predictions.

    “Our method is very different in the sense it uses models of the climate that include all the laws of physics and supercomputers together to predict what’s going to be happening over the coming season,” Huddleston said.

    The Met Office ran but did not make public their model in the 2005 and 2006 seasons. They correctly predicted the change from the active 2005 season to the below-normal 2006 season, the Met Office said.

    The Atlantic season has already had two named storms, Andrea and Barry.
    NOAA and Colorado State will update their seasonal predictions in August.

    The British scientists did not predict a number of hurricanes that would form or how many would become strong, as American forecasters do. There is a 70% chance that the number of storms will be in the range of seven to 13, according to the British.

    In May, U.S. government forecasters predicted 13 to 17 tropical storms in the season that runs from June 1 to Nov. 30. The National Oceanic and Atmospheric Administration scientists said they expect seven to 10 tropical storms to become hurricanes and three to five of them in the strong category.

    I was able to find a link with a good description of the ECMWF organization given below. It would appear that this organization is less public than Meteo-France or Met Office and may explain why it is less public in announcing its forecasts. It would appear that ECMWF does 1 month and 7 month forecasts. The link goes into some detail about the models that are used in forecasting.

    I need to look further at the private customers of these organizations and also determine if these three organizations break down their separate forecasts or simply averaged them to give the EUROSIP forecast performance? Once one has the extra selections of which model results and how many of them to average together to obtain a result it is imperative to see how the individual components performed. If the selection choice occurred after the fact we have to be cautious against data snooping.

  481. steven mosher
    Posted Sep 4, 2007 at 3:45 PM | Permalink


    Two big cats in the hat. I think that is a first time event. It might be instructive to
    calculate the prob of that.. since the distinction between categories is anthropomorphic
    I’m not sure what this would mean ( hey why not cat 3.5, and cat 4.87564532)

    Anyway Assume that Hurricanes are Poisson. Assume that Categories (1-5) are ??? distributed.
    Whats the probability ( in the last XX years of good record keeping)
    that a year leads off with 2 cat 5s?

    That should be directly calculable

    Quite Honestly ( see the thread) I was toying with an assumption that the first big
    Cat would pull enough heat from the Ocean to diminish the probability of another
    big cat running down a similiar path… within a given time horizen ( following a gradient)
    Dean cooled things down, but the rebound was astounding.

    Anyway.. two big cats that do little damage on landfall. GW is leading to stronger storms that
    kill no one and do no damage. ( joking ok)

  482. SteveSadlov
    Posted Sep 4, 2007 at 4:53 PM | Permalink

    Stephen – Felix was significantly to the south of Dean. It came no where near Dean’s deplected zone. Now, between the two of them, both the southern Caribbean and the Gulf have had lots of heat content sent to outer space. Let’s see what the rest of the season is like. What is truly impressive is just how far south the seed areas are this year. Felix almost didn’t make it around the corner, one more degree West of North and it would have fizzled into Venezuela at the get go. If the ITCZ moves any further south, that’s all she wrote.

  483. SteveSadlov
    Posted Sep 4, 2007 at 4:58 PM | Permalink

    RE: #475 – To quote Gavin, *sigh* ….. if they name that occulded front off of Florida, it will be shameful.

  484. Kenneth Fritsch
    Posted Sep 4, 2007 at 8:49 PM | Permalink

    Re: #482

    Steven Mosher, you make a good point about man made and rather arbitrary categories and divisions and how one should fit them into a distribution. One could and should also look at the global tropical cyclone records. Category 5 hurricanes (or least recorded ones) are sufficently rare in the NATL that fitting a distribution to them would be a very uncertain endeavor to my untrained mind. We should perphaps look at global category 5 hurricanes or cyclones since as I recall that category in less rare in other tropical storm basins.

    Anyway here are the years with the most category 5 hurricanes in the NATL back to 1854:

    2005 = 4; 1960 = 2; 1961 = 2; 2007 (inc) = 2 and 20 total years back to 1928 with 1 category 5 hurricane. No category 5 hurricanes were reported from 1854 to 1928. By the way, I suppose, if one is looking for strange combinations then 1960 and 1961 with 2 category 5 hurricanes in the NATL 2 years in row would be considered a very rare combination of events.

    If we chose category 4 and 5 hurricanes in the NATL we have:

    2005 = 5; 1999 = 5; 1926 = 4; 1961 = 4; 1964 = 4; 2004 = 4; 1932 = 3; 1950 =3; 1958 = 3; 1964 = 3; 1988 = 3; 1995 = 3; 2007 (inc)= 2 and tied with 17 other years with 2 category 4 and 5 hurricanes.

  485. David Smith
    Posted Sep 4, 2007 at 9:02 PM | Permalink

    I read the Dean and Felix hurricane advisories (the “discussion” sections) and noted that their maximum windspeeds were determined using four methods:

    1. Dropsondes employing GPS technology
    2. Satellite imagery capable of Dvorak estimates
    3. Flight-level (700mb, about 2 miles up) windspeed extrapolated at a 0.9 factor
    4. SFMR (Stepped-frequency microwave radiometer)
    5. Central pressure determined by GPS-type dropsondes (note: data was gathered but I saw no indication that they directly used this data).

    I have a question/exercise for anyone wanting to explore the hurricane measurement quandry is: How many of these measurement techniques were employed during the 1930s thru the 1960s (the last active phase of the Atlantic)?

    Answers tomorrow.

  486. Kenneth Fritsch
    Posted Sep 5, 2007 at 9:09 AM | Permalink

    Re: #486

    David, that’s a nice touch in making your ongoing point about the historical detection capabilities for tropical cyclones. It works better than the TCO (and too frequently Ken Fritsch) method of repetitious phraseology in attempts to make points.

    I am going with method 3, because I do not think the other technologies were up to snuff in the 1960s and, of course, 1 out of 5 would best make your point.

  487. MarkW
    Posted Sep 5, 2007 at 10:51 AM | Permalink

    My guess would be none of those methods were used from the 1930’s to 1960’s.

    I believe the only method was ground or ship based wind speed gauges.

    I don’t believe they started flying into hurricanes until the 60’s. And I’m pretty sure that none of the planes that existed in the 30’s would have been able to stand up to a hurricane. (The British were still using cloth covered bi-planes in their military in the late 1930’s.)

  488. Kenneth Fritsch
    Posted Sep 5, 2007 at 2:24 PM | Permalink

    Re: #488

    Since the first flights into hurricanes and Pacific typhoons in 1944, military and civilian crews have shown that it’s safer than you might think. Only three Air Force airplanes have gone down in Pacific Ocean typhoons, and one Navy plane in an Atlantic hurricane. All 36 men aboard the four airplanes were lost…

    ..It all began in 1944 after an Army Air Forces pilot, Col. Joseph Duckworth, became the first pilot to intentionally fly into a hurricane’s eye. He made two flights in a two-seat, AT-6 trainer from Bryan Field, Texas, into a hurricane over land near Houston on July 27, 1943.

  489. mccall
    Posted Sep 5, 2007 at 2:40 PM | Permalink

    With Hurricane Dean & Felix early track predictions not very good (Dean was horrible with most models), it will be interesting to see whether the upcoming(??) TS Gabrielle fulfills it’s model predictions.

    Here’s an early one (don’t forget, these are overwritten) so save locally to keep history.

  490. SteveSadlov
    Posted Sep 5, 2007 at 2:43 PM | Permalink

    RB-50 program also started in the late 1940s or early 50s, not long after the Texan flying the Texan.

  491. SteveSadlov
    Posted Sep 5, 2007 at 2:45 PM | Permalink

    RE: #491 – You meant “Cut Off Low Gabrielle” didn’t you?

  492. mccall
    Posted Sep 5, 2007 at 2:51 PM | Permalink

    BTW, my unofficial recollection is BAMM & XTRP were pretty good at predicting Dean (I didn’t save interims) & Felix (did save) Central Am landfalls — early CLP5 had a W-LA/E-TX track? As of today, BAMM shows a Gabrielle track reversal to approach east coast, while CLP5 and XTRP have no reverse, continuing out into the Atlantic.

    If this discussion should stay on Unthreaded 19, pls move and advise.

  493. mccall
    Posted Sep 5, 2007 at 3:07 PM | Permalink

    Assuming 492 refers to 490: “upcoming(??)” is a functional equivalent to #492 “cut off low” … I agree that models predicting TS and some even Hurricane Cat 2 are of course suspect at this time.

  494. SteveSadlov
    Posted Sep 5, 2007 at 3:18 PM | Permalink

    What I mean is, if they name this one, it will be yet another bogus naming. Out West, we get cut off lows with great frequency. They retrograde, amble about, and do all sorts of things. During times of El Nino, they may even get so called tropical characteristics. But they are never named. I submit that this year, there is an overt effort to name with abandon.

  495. mccall
    Posted Sep 5, 2007 at 3:43 PM | Permalink

    Also convenient grist and speculation from The Weather Channel:
    a “could turn North” Dean, that might hit the gulf states (especially LA) again?
    rampant Erin speculation — appeared to me, some were rooting for hurricane?
    a Cat 4 Felix, that balloons to Cat 5 just before landfall?
    a stable 75mph (barely) Cat 1 Henriette?
    and now a “cut off low” TBD TS Gabrielle, so the east coast has something to worry about.

    Hmm??? Oh well, we have been lucky with the quiet season… so far!

  496. Bill F
    Posted Sep 5, 2007 at 5:11 PM | Permalink

    I would also point out in regards to the distinction between Cat 4 and 5, that the threshold between the two seems to be right at the point where the eye of a hurricane becomes unstable and subject to more frequent eyewall replacement cycles. As a result, it is rare for a hurricane to maintain Cat 5 strength for longer than about 24-36 hours. As a result, and mindful of David’s point about changes in technology, I am dubious of our ability to distinguish between cat 4 and cat 5 storms prior to the last 35 years or so. The combined counts of cat 4 and cat 5 storms in previous years is pretty informative in pointing out the importance of our ability to determine that threshold when evaluating how unprecedented any recent activity has been.

  497. David Smith
    Posted Sep 5, 2007 at 6:49 PM | Permalink

    Re #486

    Here’s some background:

    1. Dropwindsondes (GPS-equipped dropsondes capable of excellent windspeed measurement) have been in use since 1998.
    2. Satellite imagery useable for Dvorak-type hurricane intensity estimates began in the 1970s.
    4. SFMR began operational deployment a couple of years ago (2006 I believe) and may not be fully deployed yet.
    5. GPS dropsondes, capable of reporting whether they landed in the hurricane eye or not, have been around since the 1990s.

    All of these were obviously not around in the 1930s thru 1960s. But, what about aircraft?

    In the mid-1940s aircraft began to (intentionally) fly into hurricanes and to report maximum “flight-level wind”. There have been several changes in this –

    a. The pilot has to find the location of the maximum wind, which is considerably easier today with modern aircraft radar, satellite support and storm knowledge. Otherwise he misses the peak.
    b. “Flight-level” is typically 10,000 feet above the surface, a height which experiences different windspeeds than exist at the surface. So, a correction factor has to be applied. The NHC has used various correction factors over the years and tends to use higher factors (= higher windspeed estimates) in recent years.

    Even though aircraft were used in 1945-1959, the routine use of 0.9 flight-level correction factors, an important part of wind estimates, didn’t begin until the 1980s to 1990s.

    So, the answer to the question posed in #486 is, “None of the above”.

    The way we measure storms today is much different from the way we measured storms during the last Atlantic peak. Today we measure more-frequently and more-thoroughly. The more frequently we sample and the more-thoroughly we sample the greater our chances of catching the peak intensity.

  498. David Smith
    Posted Sep 5, 2007 at 7:01 PM | Permalink

    RE #498 I should have summarized the 1930 thru 1960s as being heavily-dependent on the skill of the aircraft crew to find the maximum wind and to apply appropriate correction factors. Sometimes they did, sometimes they did not. They did their best but their technology was quite limited compared to today.

  499. David Smith
    Posted Sep 5, 2007 at 7:59 PM | Permalink

    Re #499 That should be 1945 thru the 1960s for aircraft, not 1930 thru 1960s. Before 1945 there was heavy dependence on chance encounters with land or unfortunate ships.

    Regarding cat 4 and 5 storms, there’s an arguable case that the two categories should be lumped together anyway. A few years ago, someone (Klotzbach?) found that the transformation from the common weak hurricanes into severe (cat 4 and 5) storms tends to be sudden rather than gradual. The central pressure suddenly “bombs” (rapid deepening). See Dean and Felix of 2007 for examples.

    Cat 4 and 5 storms may be somewhat structurally different versus their weaker cousins, maybe some upper-air feature. My conjecture.

  500. SteveSadlov
    Posted Sep 5, 2007 at 8:25 PM | Permalink

    Member of the “AGW begets more storms” faction at NHC: “Please, pretty please …. may you rise to the level of being countable!” Puh-leeeeeeeeeaaaase!”


  501. David Smith
    Posted Sep 5, 2007 at 8:37 PM | Permalink

    I’ve updated the weak storms plot to include 2007 to-date ( link ).

    Short-lived tropical storms have a duration of 24 hours or less and are quite weak.

    They are hard/impossible to detect without modern technology (like modern satellites and Doppler radar) and were generally missed or ignored in earlier times. As the plot shows their numbers continue to increase, inflating the modern storm count.

  502. Kenneth Fritsch
    Posted Sep 5, 2007 at 9:05 PM | Permalink

    Re: #500

    Regarding cat 4 and 5 storms, there’s an arguable case that the two categories should be lumped together anyway.

    As I recall the categorization of storms was to provide a measure of potential damage they could do to building structures (for the time the categories were initialized). In fact that is why a category 5 exists and not anything higher because the damage to structures at the time of the introduction to categories a 5 would do in any structure. Certainly what we want here is something that can be related to the storm itself like an ACE score with maybe a maximum intensity kicker. As long as we are all aware of the arbitrariness of categorization and concentrating on the NATL basin, I think we will be able to rationally discuss the issues.

  503. Bill F
    Posted Sep 6, 2007 at 12:08 AM | Permalink

    Actually, the very best way to measure storm strength would be to use something similar to the Richter Scale for earthquakes that would measure the total energy of the storm instead of the instantaneous strength at its maximum. That would account not only for instantaneous measures such as max windspeed and min central pressure, but also diameter of the hurricane force winds, storm surge, total rainfall, duration at a given intensity, etc. The measurement of the total storm energy would be a much better way to evaluate if AGW is making storms larger/stronger, but would be nearly impossible to accurately estimate for older storms.

  504. Judith Curry
    Posted Sep 6, 2007 at 5:23 AM | Permalink

    There are new and better ideas on the table for hurricane classification, see esp the following

    Click to access 2006EO010003.pdf

    (note this is agu pub that normally costs $9, but i managed to find an open copy)

    Note, the National Hurricane Center seems to be rejecting any changes at all based on the principle that we shouldn’t change at all.

  505. Kenneth Fritsch
    Posted Sep 6, 2007 at 10:01 AM | Permalink

    RE: #503

    As long as we are all aware of the arbitrariness of categorization and concentrating on the NATL basin, I think we will be able to rationally discuss the issues.

    I should have added, of course, and particularly so after another David Smith object lesson on the subject, a rational discussion must include an acute awareness of the changing capabilities of detection of tropical cyclones over time.

%d bloggers like this: