## Hurricanes 2008

Discussion of 2008 Hurricane seasonal forecasts, old and new research papers, and interpretation of trends over the entire globe.

1. Jonathan Schafer
Posted Apr 8, 2008 at 11:21 PM | Permalink

Let me be the first…#420 from Hurricanes 2007

TSR has issued their April forecast for the Atlantic: 15 storms and an ACE of 136. That ACE value qualifies as an “above-average” season by the guidelines of post #409.

The dynamical models should be able to see into the heart of the hurricane season (August-September) by now. Perhaps they could give us a peek at the future and let us know how bad it’ll be.

Hmmm….it would be interesting to see a breakdown by month. La Nina is forecast to be with us through early summer. I’m wondering if that is playing a part of an above normal forecast, since La Nina generally indicates weaker shear over the NATL.

It will be interesting if there is ARGO information on sea temps at depth for the NATL as well. There has been some indication that it is not SST that is important, but the depth of the temps. Shallow depth is not good for hurricane development, as the energy will quickly be used. It requires heat at sufficient depth. Now, from what I’ve read previously, ARGO was not finding increased temps at depth, but I think that was in the Pacific at the time I read it.

Does anyone know where I can find good information on the precip in the Sahel this winter/spring?

I’ll release my early season predictions now, some of which will be totally revised before the season officially starts. I believe it’s going to be an average to slightly below average ACE year despite La Nina, with an average to slightly above average named storm count. I predict Sadlov will go ballistic at every sub tropical storm/cold core low/occluded front that gets a name. I predict the media will continue to hype the storm count, as the average person would not likely understance ACE. I predict The Weather Channel will continue to have people stand outside in hurricane force winds, because people will continue to watch and be fascinated (whether at the hurricane or the stupidity of standing in hurricane force winds, I’m not sure). I predict there will be further discord in this thread between Dr. Curry and others over various papers/reports, but everyone will still be glad that she shows up, even if they don’t agree with her.

Posted Apr 9, 2008 at 2:39 AM | Permalink

Where can I find the latest & updated hurricane statistics (especially for US landfalls) for 2007? (And also previous years.)

3. Anders Valland
Posted Apr 9, 2008 at 5:31 AM | Permalink

This is probably OT but a recent article in a Norwegian newspaper claimed the past 3 months has seen
2 1/2 times normal tornado activity in the US. Is this correct?

4. HFL
Posted Apr 9, 2008 at 6:30 AM | Permalink

Re #1

Values for the Sahel Rainfall Index through December 2007 are available at http://jisao.washington.edu/data/sahel/

5. MattN
Posted Apr 9, 2008 at 7:29 AM | Permalink

#3

I wouldn’t doubt that claim much. Winter tornadoes are very common in La Nina years. In fact, I believe I read on ICECAP that every time the US has had significant winter tornadic activity, there has been a moderate-strong La Nina.

BTW, if you’re in the mid-west right now, hang on to your hats. Big outbreak to rival the 1974 super-outbreak is forecast for Thursday.

I’d call winter tornadoes “normal”…for La Nina.

6. Jonathan Schafer
Posted Apr 9, 2008 at 8:41 AM | Permalink

#4,

7. Hoi Polloi
Posted Apr 9, 2008 at 9:22 AM | Permalink

From World Climate Report:

8. David Smith
Posted Apr 9, 2008 at 9:26 AM | Permalink

Klotzbach/Gray have released their April forecast ( here ). It calls for 15 storms and an ACE of 150, which puts the forecast into the “much above average” category.

Kenneth, check out Figure 1 for their hindcast comparisons.

This year they use two inouts: (1) SST contrast between two parts of the Atlantic, and (2) sea-level pressure near Africa.

9. Posted Apr 9, 2008 at 9:26 AM | Permalink

Bill Gray’s research team predicted on Wednesday that 15 tropical storms would form in the coming hurricane season, of which 8 would strengthen into hurricanes.

This is an increase of 2 Tropical Storms and of those, 1 more hurricane since the December forecast. The PDF link for Gray’s forecast is published on the CSU website here: Forecast PDF

This is a “well above average” forecast. ACE is expected at 150. In 2007, the ACE was about 67.

From the PDF:

Current conditions in the Atlantic basin are quite favorable for an active hurricane
season. Both of our early April predictors call for a very active hurricane season in 2008. The current sea surface temperature pattern in the Atlantic is a pattern typically observed before very active seasons. Waters off the coast of Iberia as well as the eastern tropical Atlantic are very warm right now (Figure 6). The Azores High has also been quite weak during the month of March. Typically, a weakened Azores High leads to weaker trade winds that enhance warm SST anomalies due to reduced levels of evaporation, mixing and upwelling in the eastern tropical Atlantic.
Our final April statistical model calls for a hyper-active season with an NTC of 190 (Table 10). Due to the uncertainty with current ENSO conditions, we do not feel confident enough to raise our forecast that high at this point, however, if current trends in the Atlantic persist, there is a possibility that the forecast could be increased more in early June.

10. steven mosher
Posted Apr 9, 2008 at 10:49 AM | Permalink

this year I thought we were going to bet ACE? so that sadlov wont whine about tiny tims
being counted.

Posted Apr 9, 2008 at 11:08 AM | Permalink

I give here my NEPAC 2008 forecast. Well, it’s actually a forecast update, since, in the Brave New World, the season starts on Jan 01, and ends Dec 31.
Named NEPAC storms: 41 *
Hurricanes: 25 *

* Based on categorization by the NHC, NEPAC Division:

http://nationalhysteriacenter.blogspot.com/

12. Dave Blair
Posted Apr 9, 2008 at 11:20 AM | Permalink

The political candidates should disclose their hurricane predictions.

13. Posted Apr 9, 2008 at 11:25 AM | Permalink

Gray and Klotzbach are kind enough to provide verification of their forecasts: for the past 6 years. I screen captured the page 32 in their April 2008 forecast update and have dumped it here ; 2002-2007 Gray Forecast Verification
As we saw from last year, the expected ACE of 140 in December 2006 was increased to 185 during the April 2007 update (verification of 68). Thus, the April update went in the wrong direction. Yet, the number of named storms was accurately predicted last year — which was due to Tiny Tim’s and other forgetful whirls. Thus, for the seasonal forecasting competition, I suggest creating trying to beat Gray and make the same predictions for:

Hurricanes ____ Named Storms _____ Hurricane Days _____ Named Storm Days _____
Intense Hurricanes _____ Intense Hurricane Days ______ ACE ______ and Net Tropical Cyclone Activity _______

Note the above metrics of seasonal activity are NOT independent, you must do very well in the Intense Hurricane Days department to have a prayer with the ACE and NTC. And for fun, I think we could have a couple more categories:

Number of sub-40 knot storms that last less than 24 hours _____ Number of non-tropical named storms _____

14. crosspatch
Posted Apr 9, 2008 at 12:32 PM | Permalink

It seems to me that one can do a more accurate job of forecasting hurricanes by simply looking at 1: SST and 2: position and strength of the Bermuda High. Hurricane development is going to depend on ocean temperature and winds. A strong Bermuda High results in strong Trade Winds resulting in cooler SSTs and more difficult conditions for development.

The position of the weather feature tells you where the hurricanes are going to go if they do form. Positioned more to the Southwest, they go into the Gulf. More to the Northwest and they threaten the Atlantic coast. My guess is that a highschool student can do a pretty good job of forecasting hurricanes knowing the starting SSTs and following development of the Bermuda High. The only thing these guys are basically doing is forecasting the strength and position of the Bermuda High before it develops.

15. Posted Apr 9, 2008 at 12:39 PM | Permalink

that’s awesome that Gray and Klotzbach do forecast verification.
thanks for the link ryan maue.

It’s a shame we can’t get the IPCC/Mann/Hansen/et al to do the same.

Posted Apr 9, 2008 at 1:15 PM | Permalink

The political candidates should disclose their hurricane predictions.

They will take the time worn political position of taking both sides–more hurricanes will prove global warming and less hurricanes will prove global warming.

Hey, that sounds like a warmer press release!

17. bill-tb
Posted Apr 9, 2008 at 1:21 PM | Permalink

You didn’t know the weather channel has hurricane proof reporters? They got them from Fox News.

It’s hard to play in the game of named storms, naming isn’t what it used to be. It now seems naming is now on a quota sport.

I have my own unscientific forecasts, we are still getting cold fronts in south Florida, hurricanes won’t be very numerous, same as last year or less. Last year the first cold front that made it to our area in south Florida came in September. That ended hurricane season, from there they just kept naming swirls until the quota was met.

18. steven mosher
Posted Apr 9, 2008 at 2:34 PM | Permalink

re 13. with that many predictions this is starting to take on the flavor of GCMS or nostradamous.

But I like the idea. maybe we can assign points and weights and fudge factors.

19. George M
Posted Apr 9, 2008 at 5:09 PM | Permalink

I will be issuing my forecast, which I guarantee will be absolutely accurate in every detail, at 2359 on Dec. 31, 2008. Why waste time revising every couple of months?

Can someone please tell me what is the point of these forecasts anyway? Unless the exact landfall can also be predicted far enough in advance to permit orderly evacuation (and I point to Rita and Katrina), they are just smoke and mirrors. I would love to hear from someone who admits moving away from the coast permanently based on one of these forecasts issued months in advance. Which is really the only reason to forecast hurricanes far in advance anyway, as far as I am concerned. But as long as taxpayers are allowed to build and live in dangerous places, the results will be the same, predictions or not.
Now here is what I challenge one of you mathematicians to do: Calculate (or retrieve) the long term averages in each category. Construct a chart of yearly deviation of actual from these averages. Then do the same for the assorted forecasts vs. actual. Is it any better overall than using the simple numerical average as the forecast? A question I have previously asked, and never had answered.

20. Kristen Byrnes
Posted Apr 9, 2008 at 5:15 PM | Permalink

18 Mosher,

How about a non-GCM KBSF prediction? 18 storms, 10 hurricanes.

21. steven mosher
Posted Apr 9, 2008 at 6:14 PM | Permalink

re 20. Kristin, the trick is to guess at 50 different metrics and then argue like heck
that the one you hit spot on is the BEST METRIC.

Lets face it. When it comes to hurricane metrics there is one metric that matters.
D&D.

Death and destruction. It’s gruesome to put a number on the former, so focus on the latter.

DOLLARS of destruction. I hate to be mercenary about this but if a wind blows over the ocean
and nobody is there to have their hair blown back, then it just doesnt matter.

22. Kristen Byrnes
Posted Apr 9, 2008 at 6:39 PM | Permalink

21 Mosher,

In that case, invest in insurance companies and hope that high pressure persists over Georgia again. And the only metric that matters is wind shear. 18 and 10!

23. JP
Posted Apr 9, 2008 at 6:54 PM | Permalink

Perhaps someone should see what the latest trends of the Bristle Cone Pine tree ring growth. They could teleconnect to NATL Hurricanes

24. Posted Apr 9, 2008 at 7:26 PM | Permalink

2008 Contest

I think we should have two levels. Contestants can enter one, or both, depending on your level of climatological prowess and/or blood alcohol. This contest is for fun and to illustrate how Nature so often humbles us all.

Entries can be made from April 9 thru June 5, 2008. You can change your entry at any time up to June 5.

“Named storms” applies to tropical systems only.
________________________________________________________________________

“Wizard” Level (two pieces of information should be submitted)

1. Forecast the ACE category for the season ____
– much below normal (0 to 40)
– below normal (40 to 80)
– near normal (80 to 100)
– above normal (100 to 150)
– much above normal (150+)

2. Forecast the number of named tropical storms ____
(This is a tiebreaker)

_________________________________________________________________________

“‘You Da Man” or “You Da Wo-man” Level (nine pieces of information should be submitted)

1. Forecast the number of named tropical storms ____
2. Forecast the number of hurricanes ____
3. Forecast the number of intense storm-days ____
4. Forecast the number of hurricane-days ____
5. Forecast the number of storm-days ____
6. Forecast the seasonal ACE ____
7. Forecast the number of Wimps (less than 40 kts or 24 hrs or less) ____
8. Forecast the number of subtropical storms ____
9. Name your forecast reasoning (SST, Sahel rainfall, AMM strength, ENSO state, SWAG, Madden-Julian Oscillation, ideological bias, my psychic, my psychosis, 1995-2007 climatology, 150m ocean temperature, Brasilian liqueur, other)

____________________________________________________________________________

Entries are posted at CA. I’ll compile the records and publish all the entries shortly after June 5.

25. Kenneth Fritsch
Posted Apr 9, 2008 at 8:52 PM | Permalink

Re: #21

DOLLARS of destruction. I hate to be mercenary about this but if a wind blows over the ocean
and nobody is there to have their hair blown back, then it just doesnt matter.

Listen up, Mosher. This contest is a serious scientific exploit that should not get sidetracked into the economics of estimating dollar costs of destruction or be subject to the uncertainty of a landfall storm hitting a populated area (that’s too much like a guessing game). Besides there are oil rigs out in those open seas and estimating the costs of a hit or even a suspected hit can get very complicated.

Finally, if you say it doesn’t matter 10 times real fast and real loud – well then it does matter.

Personally I think I am going to base my prediction(s) on the two European dynamical models that have not yet reported their 2007 results — not that I am convinced of their models’ accuracies just that I think a prediction after the fact has a few advantages.

26. Kenneth Fritsch
Posted Apr 9, 2008 at 9:02 PM | Permalink

Re: #19

Now here is what I challenge one of you mathematicians to do: Calculate (or retrieve) the long term averages in each category. Construct a chart of yearly deviation of actual from these averages. Then do the same for the assorted forecasts vs. actual. Is it any better overall than using the simple numerical average as the forecast? A question I have previously asked, and never had answered.

http://www.wunderground.com/education/seasonal2007.asp

George M, you just need to try harder.

27. John A
Posted Apr 10, 2008 at 1:24 AM | Permalink

The official JohnA forecast: $6 \pm 3$ hurricanes
Remember: You read it here first.

28. ma in va
Posted Apr 10, 2008 at 7:39 AM | Permalink

I’ll go on the high side – 10 hurricanes with 4 making landfall in the US. Why a high number? Well if I forecast that (with absolutely no basis – scientific or otherwise)for enough years I’ll eventually be right. And, since I have nothing at stake (I won’t bet money and my reputation isn’t worth squat) what the heck!

I’ll also predict that Pittsburgh or Chicago Cubs will win the World Series – same bet as before. Maybe in a hundred years someone will look back and declare me a prophet.

29. Paul Linsay
Posted Apr 10, 2008 at 7:44 AM | Permalink

#27, John A, you stole my thunder. I’m out.

30. George M
Posted Apr 10, 2008 at 7:50 AM | Permalink

Kenneth:

Thank you for that link. It will provide entertainment and amusement for the rest of the day. The TSR article Dr. Masters links to can be summarized as: “Well, we can predict with great accuracy unless the hurricanes refuse to follow our predictions, like they did in 2006″.

31. David Smith
Posted Apr 10, 2008 at 9:02 AM | Permalink

When I look at the Gray/Klotzbach forecast methodology I wonder how their indicators address the issue of atmospheric dryness and stability. That seems to have been a fly in the ointment in recent years.

I admire their courage in publicly tackling a difficult question but my sense is that, in the end, science will be able to accurately forecast “average”, “above-average” or “below-average” seasons and that’s about it. But, that may be all that’s needed anyway.

32. Posted Apr 10, 2008 at 9:57 AM | Permalink

In my judgement,2008 will continue as a solar minimum type of a year. Solar minimum in my opinion will be in the first half of 2009. Hence the hurricane season is governed by a minimum of solar wind ram pressure activity [less heat] and less solar electricity and hence a cool year[ very cold next winter with lots of snow] and very similar to the final years of other past long solar cycles of 12 years and up. These years have fewer high level solar wind dynamic pressure spikes[ 10 nPa and more] and most spikes are very minor . Hence little joule heating of the planet’s atmosphere as we have already seen this year. My best estimate is 15 named storms , 6 hurricanes and an ACE OF 70-80, a year similar to 2007. I disagree with the yet high high forecasts of both TSR and CSU both of whom have been off the last 4 years.WRC from TEXAS has also a forecast that is much more resonable.This will be a below normal hurricane season in my opinion

33. Posted Apr 10, 2008 at 10:56 AM | Permalink

I used the term below normal in a more generic sense in that it would be a more quiet hurricane season in my opnion as opposed to the predicted very active season.. Since I predicted an ACE of 70-80 , the formal guidelines define my forecast to be a normal season . I would have to predict an ACE OF 66 or less in order it be a below normal season.

Here are some long solar cycles
#4 1784-1798 13.6 years
#6 1810-1823 12.7 years
#9 1843-1856 12.5 years
#5 1798-1810 12.3 years
#13 1889-1901 12.1 years
#14 1901-1913 11.9 years
#20 1964-1976 11.6 years

I am guided by the more recent and yet long cycles #20 and #14. Our records get less reliable as we go further back . For example the years 1974, 1975,1976 could correspond with 2006, 2007 and 2008 respectively .

34. Posted Apr 10, 2008 at 11:22 AM | Permalink

#24 David Smith:

Are we posting our contest entries here or on the bulletin board? I’m game!

35. David Smith
Posted Apr 10, 2008 at 4:10 PM | Permalink

Re #34 terry I suggest that we use one thread (this one) to record entries. In the future maybe the contest moves to the message board, but not yet.

36. Judith Curry
Posted Apr 10, 2008 at 7:12 PM | Permalink

Here is TSR’s forecast (dated April 7), pretty much the same as Gray/Klotzbach (but with decimal points, 14.8, love the precision)

http://tsr.mssl.ucl.ac.uk/

37. steven mosher
Posted Apr 10, 2008 at 7:39 PM | Permalink

dr. curry and JEG both need to play! it’s all in good fun. come on in. the waters fine.

38. Posted Apr 10, 2008 at 7:55 PM | Permalink

I want a forecast for Sarasota. That’s where Dad lives. I need to know whether I need to plan to swoop in and rescue him!

39. Posted Apr 10, 2008 at 7:57 PM | Permalink

32 (Matt Vooro):

In my judgement,2008 will continue as a solar minimum type of a year. Solar minimum in my opinion will be in the first half of 2009. Hence the hurricane season is governed by a minimum of solar wind ram pressure activity [less heat] and less solar electricity and hence a cool year

Apart from your physics being wrong, the solar wind ram pressure does not vary much with the solar cycle and is, in fact, usually highest near the minimum of the cycle. The electric field in the solar wind does not vary much either, so the cooling is not explicable from correct physics and the variables you were using.

40. Posted Apr 11, 2008 at 7:40 AM | Permalink

I like contests.

“Wizard” Level (two pieces of information should be submitted)

1. Forecast the ACE category for the season: 103
– much below normal (0 to 40)
– below normal (40 to 80)
– near normal (80 to 100)
– above normal (100 to 150)
– much above normal (150+)

2. Forecast the number of named tropical storms: 16
(This is a tiebreaker)

_________________________________________________________________________

“‘You Da Man” or “You Da Wo-man” Level (nine pieces of information should be submitted)

1. Forecast the number of named tropical storms: 16
2. Forecast the number of hurricanes: 7
3. Forecast the number of intense storm-days: 10
4. Forecast the number of hurricane-days: 20
5. Forecast the number of storm-days: 54
6. Forecast the seasonal ACE: 103
7. Forecast the number of Wimps (less than 40 kts or 24 hrs or less): 9
8. Forecast the number of subtropical storms: 3
9. Name your forecast reasoning: AMO, active cycle, and my raging Soju hangover.

41. Judith Curry
Posted Apr 11, 2008 at 7:45 AM | Permalink

Lucia, as far as i can tell, when we are in warm AMO and LaNina/cool PDO, it is very bad news for Florida if we also have positive NAO. Looks like we are on with the AMO and LaNina, too soon to tell what the NAO will look like. I will be down in Miami i think on May 9, so i will need to firm up my florida analysis before then.

For seasonal forecasts, what i look at this early in the season are this years AMO index and climate model projections of ENSO for aug/sept/oct. These predictors pretty much agree with Gray and TSR, although they use very different predictors. Predictions this early arent worth much, pretty much need to wait until early june.

So i would say we need to evaluate the forecasts made in apr/may separately from those made in jun/jul and in aug.

42. Mike B
Posted Apr 11, 2008 at 8:07 AM | Permalink

Predictions this early arent worth much, pretty much need to wait until early june.

When I lived in New Orleans, I always felt much more comfortable with my September forecast than my June forecast. My October forecasts also tended to be better than my September forecasts. By November, I usually had it pretty much nailed.

Here is my intrepid prediction: given that it is an election year, they Hysteria center WILL find a way to make it to VICTOR by election day, and to use the whole alphabet for the season (do drive the point home, they’ll make sure to re-use “A”). So I predict 27 named storms. 14 hurricanes and 5 reach cat 4 or higher.

And yes, I believe science has been corrupted by politics.

43. Posted Apr 11, 2008 at 8:17 AM | Permalink

LEIF
My own analysis of comparing individual months of the year 2005, a very active solar ramp down year which was very active in terms of hurricanes and record temperature anomalies
with the first three months of 2008 a solar minimum period which is showing the opposite trend revealed the following interesting comparison.

2005[a solar ramp down year]
Number of very high solar wind ram pressure spikes of over 10 nPa was 60 or an average of 5 per month.
The peak pressure spike of each month averaged about 25[9 were over 15 nPa]

2008 JANUARY –MARCH [three months] a solar minimum period
Number of very high solar wind pressure spikes of over 10 nPa was zero. There were 4 spikes for an average of 1.3 per month of only 5- 9.5 nPa
-As you are aware the temperature anomalies were way down for these three months

I also found that the year 2007 which is another near solar minimum year had an average of only 3 major solar wind ram pressure spikes of 10nPa or more and again the spikes were of lower average value. The land /global temperature anomaly was also lower than 2005.

What I have generally found that high global land temperature anomaly months [global/ land] have a higher number of high[10 nPA and more] solar wind ram pressure spikes and low temperature anomaly months have the opposite.
Also the average number of hurricanes/year in all solar minimum years is about 6.1 while the average of all hurricanes since the last solar minimum is closer to about 8.[1995-2007]

44. Posted Apr 11, 2008 at 8:38 AM | Permalink

Weather’s not my game, but may I play anyway?

“Wizard” Level (two pieces of information should be submitted)

My predictions for the two-pieces-of-information level are the average values obtained for the reference period. I don’t have a clue what the numerical values are, but I assume they are readily available.

1. Forecast the ACE category for the season:
– much below normal (0 to 40)
– below normal (40 to 80)
– near normal (80 to 100)
– above normal (100 to 150)
– much above normal (150+)

2. Forecast the number of named tropical storms:
(This is a tiebreaker)
_________________________________________________________________________

“You Da Man” or “You Da Wo-man” Level (nine pieces of information should be submitted)

My Predictions for these is the average value obtained for the reference period. I don’t have a clue what the numerical values are, but I assume they are readily available.

1. Forecast the number of named tropical storms:
2. Forecast the number of hurricanes:
3. Forecast the number of intense storm-days:
4. Forecast the number of hurricane-days:
5. Forecast the number of storm-days:
6. Forecast the seasonal ACE:
7. Forecast the number of Wimps (less than 40 kts or 24 hrs or less):
8. Forecast the number of subtropical storms:
9. Name your forecast reasoning: I’m an average kind of guy.

45. Hoi Polloi
Posted Apr 11, 2008 at 9:15 AM | Permalink

The best way to forecast the future is to create it.

46. Posted Apr 11, 2008 at 9:22 AM | Permalink

Lief

http://web.mit.edu/space/www/

ftp://space.mit.edu/pub/plasma/publications/jdr_global/jdr_global.withthumbs.pdf

The graphs of these two refernces show considerable change in solar wind velocity, density and ram pressure during typical solar cycles as measured by them. Do these not apply to earth as well?

47. Posted Apr 11, 2008 at 9:29 AM | Permalink

43 (Matt): Here is the solar wind pressure [each day since the start of cycle 22]:

1990 and 1991 were very very active solar maximum year, yet very few hurricanes, 1995 was coming into solar minimum, yet many hurricanes. It is not a good practice to forecast based on single examples.

48. steven mosher
Posted Apr 11, 2008 at 9:29 AM | Permalink

ok. I’ll stick my neck out early

“Wizard” Level (two pieces of information should be submitted)

1. Forecast the ACE category for the season 127

2. Forecast the number of named tropical storms 15

_________________________________________________________________________

“‘You Da Man” or “You Da Wo-man” Level (nine pieces of information should be submitted)

1. Forecast the number of named tropical storms 15
2. Forecast the number of hurricanes 8
3. Forecast the number of intense storm-days 8 ( you mean Intense hurricane days?)
4. Forecast the number of hurricane-days 33
5. Forecast the number of storm-days 66
6. Forecast the seasonal ACE 127
7. Forecast the number of Wimps (less than 40 kts or 24 hrs or less) 2
8. Forecast the number of subtropical storms 3
9. Name your forecast reasoning: handicapping gray

49. John A
Posted Apr 11, 2008 at 9:47 AM | Permalink

Paul Linsay:

You might call it stealing, I call it “background research”

50. John A
Posted Apr 11, 2008 at 9:53 AM | Permalink

Leif Svalgaard: You obviously haven’t heard of Paul Linsay’s blockbuster posting on this subject. Then you might have cause to wonder about any cause….

51. Posted Apr 11, 2008 at 11:06 AM | Permalink

http://lepmfi.gsfc.nasa.gov/mfi/plots.html

Leif
I did my wind analysis on a daily count basis which is in more detail and more accurate and reliable.[see the reference above]

I note that of the 14 last solar minimum years only 4 had more than 7 hurricanes . This not a single example but 152 years of history

52. Sam Urbinto
Posted Apr 11, 2008 at 11:48 AM | Permalink

ZERO!

53. Judith Curry
Posted Apr 11, 2008 at 11:51 AM | Permalink

There’s a new paper by Kerry Emanuel out
ftp://texmex.mit.edu/pub/emanuel/PAPERS/Emanuel_etal_2008.pdf
It is discussed by SciGuy at http://www.chron.com/disp/story.mpl/front/5693436.html

54. Posted Apr 11, 2008 at 12:26 PM | Permalink

Judy #52, thank you for the link — I was unsure when the embargo was going to be lifted. I will create a thread shortly for discussion.

55. Michael Jankowski
Posted Apr 11, 2008 at 12:29 PM | Permalink

The generally emerging view, she said, seems to be that global warming may cause some increase in intensity, that this increase will develop slowly over time, and that it likely will lead to a few more Category 4 and Category 5 storms. How many? When? No one yet knows.

So IF we see an impact on canes from global warming, it hasn’t happened yet, and it could be a long time before it is detectable.

According to the Gospel According to Gore in 2005, “…the science is extremely clear now, that warmer oceans make the average hurricane stronger, not only makes the winds stronger, but dramatically increases the moisture from the oceans evaporating into the storm – thus magnifying its destructive power – makes the duration, as well as the intensity of the hurricane, stronger…”

http://www.commondreams.org/views05/0912-32.htm

56. Posted Apr 11, 2008 at 12:34 PM | Permalink

Kudos to Judith, Peter Webster, Kerry Emanuel and their co-authors for raising an important topic and speeding the advancement of tropical cyclone science. The attention they’ve focused on the topic has, in my opinion, saved lives. The next step is for all the science parties to join together to urge smarter development in the storm-prone areas.

57. Posted Apr 11, 2008 at 1:03 PM | Permalink

Leif

Does your solar wind ram pressure source show 1986. I notice that both 1986 and 1996 which are solar minimum years both had low global land temperature anomalies of 0.119 and 0.205[per CRUTEM3]. You will note on the graph that you showed earlier, the 1996 SOLAR RAM WIND PRESSURE is also again quite low as I said it would [ mostly below 5 nPa which is about the peak we got in March 2008 ]. 1987 is also similar but a pick up in temperature anomaly [o.285] and some modest solar wind ram pressure increase.Your graph seems to confirm what I said earlier.

58. Sam Urbinto
Posted Apr 11, 2008 at 1:19 PM | Permalink

Thanks Judith. Interesting that Eric Berger quotes Vecchi, Pielke Jr and you in the article. Cool.

59. steven mosher
Posted Apr 11, 2008 at 1:28 PM | Permalink

I love the sound of ‘cyclone science’

cyclone science seems sound.

Sorry I needed a distraction from taxation without synapasation

60. Posted Apr 11, 2008 at 1:30 PM | Permalink

Judy

Lucia, as far as i can tell, when we are in warm AMO and LaNina/cool PDO, it is very bad news for Florida if we also have positive NAO.

Thanks.

Looks like we better get the upstairs remodeling done so Dad can have a room and comfortable bed when he evacuates.

61. Richard Sharpe
Posted Apr 11, 2008 at 1:33 PM | Permalink

If this is an accurate quote of Emanuel:

The new work suggests that, even in a dramatically warming world, hurricane frequency and intensity may not substantially rise during the next two centuries.

then I smell sandbagging.

That is, of hurricane frequency and intensity does not increase over the next few years, it does not necessarily mean that global warming is not happening or that cooling has set in.

Perhaps I am being too Machiavellian … but what impact should global warming have?

62. Barney Frank
Posted Apr 11, 2008 at 2:08 PM | Permalink

then I smell sandbagging

Jeez, how can the guy win?
If he says GW increases hurricanes he’s an alarmist,
If he says it doesn’t he’s an alarmist enabler. :)
Maybe he’s just muddling his way through a difficult subject as best he can.
At least he was willing to admit his previous work may have been off base. That alone puts him miles ahead of the Team.

63. Sam Urbinto
Posted Apr 11, 2008 at 2:30 PM | Permalink

Kerry and Gabriel could be hedging their bets, sure. But if you had seven model simulations, two suggested an intensity decline and five showed a moderate increase, what would you say? At least he’s publically saying theres a lot of uncertainty, and power will be a mixed bag. Isn’t that the case?

Regardless of the facts of the data guiding the statements, I detect the aura of sincerity in his words and deeds.

And what Roger said was true too, that his work was seized upon in the debate; just because you publish findings that others take and run with because it supports their agenda doesn’t mean you have one yourself. So.

And I also agree with Judith “The issue probably will not be resolved until better computer models are developed”

But then again, look at what Kerry ” adding that global warming may still play a role in raising the intensity of hurricanes” and Gabriel “possibly in addition to global warming” say in the article. So don’t get your head all wrapped up in the towel rack.

Still, admiting questionable results are indeed questionable results is more than I’ve come to expect, so, there is that.

64. steven mosher
Posted Apr 11, 2008 at 2:44 PM | Permalink

re 59. It might be bad in Florida. right before free elections.

65. Posted Apr 11, 2008 at 2:44 PM | Permalink

leif

Many thanks for your solar wind ram presssure source and comments. Idid find the source on my own.

66. steven mosher
Posted Apr 11, 2008 at 2:56 PM | Permalink

we need sockpuppet prophilatics. I dont want sadlov beating me
by stuffin the ballot box with bogus junk.

JUDY?? we need a number . I’d would bet on record year. I think it would be a good bet at this stage.

10000 quatloos ( orginal moshpit noncirc quatloos) That 2008 will
set a record… of some sort.

I pick landfall.

67. Sam Urbinto
Posted Apr 11, 2008 at 3:13 PM | Permalink

I pick an ice age starting in 10 minutes. Hurry up and turn on the burners on your oven, there may be hope yet.

Then again, I’m smart enough not to live below sea level or in a desert or over a fault.

68. Judith Curry
Posted Apr 11, 2008 at 4:00 PM | Permalink

I’m waiting til june to make my official forecast

69. Richard Sharpe
Posted Apr 11, 2008 at 4:31 PM | Permalink

I withdraw my earlier remark as I had not looked at the paper itself, and drew that interpretation from the review article.

70. Sam Urbinto
Posted Apr 11, 2008 at 6:13 PM | Permalink

Dude, like always read the paper.

71. Judith Curry
Posted Apr 11, 2008 at 7:40 PM | Permalink

I just received this reference via email, sort of interesting

http://blog.wired.com/wiredscience/2008/04/underwater-micr.html

72. Posted Apr 11, 2008 at 9:44 PM | Permalink

46,50,56 (Matt): Here is a plot of ALL available data since spacecraft measurements began [note the 1-day resolution]:

Note that during 1972-1997 the pressure seems to be a bit higher [look at the bottom of the black curves. I believe this is an artifact due to ‘uneven’ [read: wrong] inter-calibration of the spacecraft data by NASA.

73. Dave Dardinger
Posted Apr 12, 2008 at 4:39 AM | Permalink

re: #72 Leif,

note the 1-day resolution

I don’t care what you say or it says, there are not 365 points in each year. It doesn’t even look like a weekly average. Of course maybe the original had that high a resolution, but no computer screen will print about 4000 dots per inch…. Right clicking and looking at properties I see your picture is a total of 640 pixels wide for about 64 years if it was all data, so you’ve got about monthly data showing.

74. Dave Dardinger
Posted Apr 12, 2008 at 4:46 AM | Permalink

Ok, looking more closely I see what was meant. These are monthly highs and lows based on a daily average. But it should have been stated that way.

75. Posted Apr 12, 2008 at 6:31 AM | Permalink

74 (Dave):

I don’t care what you say

But you should care about what people say [especially me :-) ]. These are, indeed, daily values [based on hourly values, themselves based on much higher resolution]. Agreed, that you on a small computer screen and 640 pixels can’t see every day. But, if you had a very high pressure on a given day, it would stick out and rise about the blur. Now, when dealing with hurricane activity, one-day time resolution is probably silly anyway. Monthly [or better solar rotation , 27 days] averages would be better, so here is such a plot:

There are a few spikes [e.g. in 1968] that seem out of place. These are mostly caused by the data being incomplete, and the 27-day average ending up being based on only a few real data points.

76. Dave Dardinger
Posted Apr 12, 2008 at 6:53 AM | Permalink

re: 75 Lief,

Well, the thing is that this second graph shows a completely different thing. Here the monthly “bars” are just grainy connections of the monthly average data while the first graph shows how much the daily averages vary each month. Since the average for a given month must by definition be within the bar containing the range of daily averages, it would be easy to create a graph which contains both graphs with a white dot within each bar representing the monthly average. This would have the advantage of making the average values clearer. Also it’d let you know if a particular month had many high values or just one or two.

77. Posted Apr 12, 2008 at 7:30 AM | Permalink

76 (Dave): it would [and I have such graphs], but since the flow pressure has very little [if anything] to do with hurricanes, I’ll let [enthusiastic] others do that. The NASA website has lots of ways of plotting many solar wind parameters.

78. Posted Apr 12, 2008 at 7:37 AM | Permalink

Here are the early entries. As noted, these can be changed at any time up to June 5 as contestants study their crystal balls.

Let me know if I’ve misinterpreted some of the non-ACE entries.

79. Posted Apr 12, 2008 at 12:38 PM | Permalink

Leif
Your ram pressure source came in real handy.
I apologize for the detail but the answer lies in the detail only. I mentioned that I thought that we were near the end of long solar cycle #23 which is 12 months to date and could be longer still in my judgment. A similar period was at the end solar cycle #20 mostly 1974-1976. I have shown a year before and after 1973 and 1977 to show the change.

1973
Global land temp anomaly 0.151[Crutem3]
# of hurricanes 4
Named storms 8
ACE 43
# of solar wind pressure spikes 5nPa and over, 20
—————————————-6nPa and over, 12
—————————————7nPa and over, 5[one 9]

1974
Global land temp anomaly -0.295
# of hurricanes 4
Named storms 11
ACE 61
# of solar wind pressure spikes 5nPa and over, 18
————————————–6nPaand over, 12
————————————– 8 nPa and over, 2[one 9]

1975
Global land temp anomaly -0.111
# of hurricanes 6
Named storms 9
ACE 73
# number of solar wind pressure spikes 5 nPa and over, 19
———————————————–6nPa and over, 5[one 11]

1976 SOLAR MINIMUM
Global land temp anomaly -0.368
# of hurricanes 6
Named storms 10
ACE 81
# of solar wind pressure spikes 5nPa and over, 5
—————————————6 nPa and over, 7[0ne 15]

1977
Global land temp anomaly 0.081
# of hurricanes 5
Named storms 6
ACE 25
# of solar wind pressure spikes 5nPa and over, 8
————————————–6nPa and over, 5[one 10]

OBSERAVATION

As the number of higher level solar wind ram pressure spikes dropped, [from 18-20 spikes of level 5 nPa and > to 5-8 spikes of level 5nPa and >, so did the global temperature anomaly [drop of 0.519 from +0.151 to -0.368]. The number of hurricanes, number of named storms and the Ace remained quite low. This pattern is similar to the pattern of 2006-2008 to date at least. Hence for my below normal hurricane forecast for 2008.

80. Posted Apr 12, 2008 at 12:47 PM | Permalink

79 (Matt): do the same for the other minima, 1986 and 1996, as well

81. Posted Apr 14, 2008 at 1:13 PM | Permalink

Tropical Storm 2 in the Western Pacific is slowly gathering strength. It will likely reach typhoon status in a couple days, starting the tally on Northern Hemisphere ACE for 2008. Based upon the last 30 years or so, and last year’s inactivity, it is likely that the NH ACE will be around 400 or so, well below normal. Due to La Nina (weakening) conditions, the WPAC should be about 80% of normal, the EPAC may not have a hurricane (based upon its collapse), and the Indian Ocean will be its usual 2-5% of hemispheric output. So, that leaves the North Atlantic as the wild card. However, if one pursues a persistence forecast, the NATL will repeat last year’s performance.

82. Posted Apr 14, 2008 at 6:12 PM | Permalink

Leif

Here are the comparison for 1986 and 1996, both of which are also solar minimum years

1986 SOLAR MINIMUM [end of 10.3 year solar cycle]
Global land temp anomaly 0.119
Global sea surface temp anomaly -0.020
# of hurricanes 4
# of named storms 6
ACE 36
El Nino Sept -Dec
# of solar wind pressure spikes 5 Npa and over, 6
————————————–6 nPa and over , 2
Solar wind spikes pattern seemed to be decreasing during latter part of year, the hurricane season

1996 SOLAR MINIMUM [ end of 9.7 year solar cycle]
Global land temp anomaly 0.205
Global sea surface temp anomaly 0.113
# of hurricanes 9
# of named storms 13
ACE 166
La Nina first three months
# of solar wind pressure spikes 5nPa and over 3
—————————————6 nPa and over 1
Solar wind spikes pattern seemed to be increasing during latter part of year, the hurricane season.

OBSERVATION

The 1996 more active hurricane season than 1976 and 1986 might be due to the shorter solar cycle prior to 1996[ less earlier cooling?], warmer average sea surface temperatures and land temperatures and an increasing pattern of solar wind ram pressures during the latter part of the year .

83. Posted Apr 15, 2008 at 4:45 PM | Permalink

Ryan re #82

I agree with you that 2008 Hurricane season will most likely repeat the 2007 season. For a while I thought I was the only one in the left field from the other forecasts.

84. Posted Apr 15, 2008 at 11:24 PM | Permalink

Thanks Matt, 2008 ACE is 2 so far this year for the Northern Hemisphere, but nothing really gets going until July and August. WPAC TC activity is very predictable with a successful ENSO forecast through the April barrier. Since WPAC ACE is a nearly constant 55% of NH total activity since the 1970s, a 2008 forecast ACE of 190-230 would yield an NH ACE of about 400. So that leaves the EPAC and the NATL to fight over about 200 units of ACE. If this does not happen, then a fundamental change in ENSO would have occurred or some sort of other unknown climate fluctuation (apparently like 2007). Need June SSTs to know for sure.

85. Posted May 2, 2008 at 7:43 PM | Permalink

The PDO has flipped possibly? http://www.jpl.nasa.gov/news/news.cfm?release=2008-066 NASA Press Release The last flip was in 1977 to the warm phase, also a year when TC activity (as in 1976) really waned in the NH. I presented the results of my work at the Tropical Conference in Orlando this last Monday, and received boisterous praise from Dr. Bill Gray in the audience. I am not sure how to take that endorsement. Anyways, I posted a link to the powerpoint on my tropical web page. Ryan’s Tropical Page . I am finishing a write up this weekend.

86. bender
Posted May 2, 2008 at 8:42 PM | Permalink

#86
How do you diagnose a “flip” vs. a brief anomaly? Don’t you have to wait 2+ decades and then look back at the trend? i.e. Isn’t the question unanswerable?

87. Posted May 2, 2008 at 9:23 PM | Permalink

bender, 87, without any supporting evidence or knowledge about the past, we would be in for a long wait. analogous to the change in phase of the AMO — which supposedly coincided with Atlantic hurricane activity…i think we should know after the first El Nino comes along, in 2-3 years

88. bender
Posted May 2, 2008 at 10:07 PM | Permalink

How do you distinguish a flip in one circulatory mode from a redefintion among circulatory modes? Can redefinition not happen? Is our knowledge of these so-called “modes” absolute? Or do our definitions improve over time as the circulation plays itself out?

89. David Smith
Posted May 3, 2008 at 8:58 AM | Permalink

Jeff Master’s Weather Underground has started a hurricane forecast contest in which people predict the number of 2008 storms. Our CA contest, which predates Weather Underground’s, won’t be able to match their number of contestants so we’ll just have to outdo them on forecast quality.

The CA contest entrants as of April 12 are given here:

We’re now exiting the “April barrier” so that reasonable ENSO forecasts should soon be available, plus Atlantic weather patterns are coming into view. And, the US Kentucky Derby is about over so US readers need something else on which to focus our forecasting wizardry.

You can enter by posting on this thread your forecast of the number of storms and hurricanes or, preferrably, choose one of these ACE categories:

Much Below Normal
Below Normal
Normal
Above Normal
Much Above Normal

I’ll convert those with storm numbers into equivalent ACE numbers, using historical values.

We need to beat Weather Underground in accuracy so sharpen your crystal ball and enter here!

90. bender
Posted May 3, 2008 at 9:26 AM | Permalink

Statistics says 2008 is too early for a rebound back to normal. Should be “below normal”. 2009 normal. 2010 above normal. How’s that for long lead time?

91. bender
Posted May 3, 2008 at 9:28 AM | Permalink

Reasoning: 5- and 10-year periodicity in Atlantic eigenthingies have a yet-to-be-identified physical cause.

92. David Smith
Posted May 3, 2008 at 10:12 AM | Permalink

Yes, the eigenthingies indicate nothing spectacular in 2008.

That reminds me – seems like someone (James Elsner??) made multi-year predictions a few years ago. I’ll post them if I can find them.

93. Posted May 3, 2008 at 11:58 AM | Permalink

Elsner presented some rather marginal solar cycle – Atlantic hurricane linkages at the Tropical Conference. I am still mystified at the connections, but he said “it is obvious” and “clear to anyone with intuition” that there was indeed a connection. Something to do with raising the MPI (Emanuel) due to warming of the lower stratosphere (more uv; ozone) is causing stronger hurricanes to get stronger. My prediction for Atlantic ACE is the same as last year, 68.

94. David Smith
Posted May 4, 2008 at 10:30 AM | Permalink

I often map and plot things simply to see what they look like. Most end up buried and forgotten in the electronic basement of my computer. Here’s one that’s headed for that burial ground but I’ll post it since it did surprise me somewhat.

This shows where each US landfalling hurricane originated (the point where it first reached tropical storm strength). The red dots designate major (cat 3,4 and 5) storms while the black dots cover the weaker hurricanes.

In the region east of the pinkish line 42% of the hurricanes were major hurricanes. In the region west of the line 40% were major hurricanes.

What I expected to see was the eastern area with a major-hurricane percent that was notably higher than the western area’s. This would be due to the eastern storms being longer-lived. In general, the longer a storm exists, the greater its opportunity for strengthening and maintaining that strength.

What’s the significance? Not much, though if the eastern Atlantic heats up and becomes more active due to AGW then there’s a piece of evidence here that the eastern Atlantic storms may not be worse (intensity-wise) than had they formed in the western Atlantic. I do think they’d have greater aerial extent, though, with destructive winds covering a larger area, but maybe I’m wrong there too. I may explore that, if the data allows.

On a somewhat related note, I posted a plot a few weeks ago which showed that the farther east a storm formed (reached tropical storm strength in 1979-2007) the less its chance of reaching the Caribbean (and other land by extension). Stated differently, the farther east a storm formed, the greater the chances that it remained entirely-at-sea. This is due to a formed storm being steered by somewhat different winds than those that steer a seedling.

I calculate that, for the eastern Atlantic storms of the last 30 years which entered the Caribbean, if their formation had shifted eastward by an average of 350 miles due to warmer water, then about 30% of them would have curved out to sea rather than have entered the Caribbean.

95. David Smith
Posted May 4, 2008 at 10:37 AM | Permalink

Re #95 The map covers storms of 1950-2007.

96. Bob Koss
Posted May 4, 2008 at 6:59 PM | Permalink

David,
It appears to me you used the value for the safir scale at US landfall instead of the storm peak wind which usually occurs away from land. I count a total of 107 landfall US hurricanes. There are 19 hurricanes that had a safir scale of zero at US landfall which don’t appear to be in your graph. About 75% of those were spawned in the west half of the basin. That would alter the distribution considerably.

When done by peak wind here is what I get. This includes US landfall hurricanes at safir zero. I used first track location when dividing the storms by basin area, rather than using TS strength. This results in more storms in the east basin since I’m picking most of them up as subtropical stage.

East basin
cat 1-2, 4
cat 3-5, 34
34 / (34 + 4) = 89% majors

West basin
cat 1-2, 41
cat 3-5, 28
28 / (41 + 28) = 40% majors

So your intuition was correct about the higher percentage of majors in the east basin.

97. David Smith
Posted May 4, 2008 at 7:33 PM | Permalink

Re #97 Thanks for the check, Bob.

You are right – I used the storm strength at US landfall, not the maximum strength during the life of the storm, as my interest was in the storms as they reached the US.

What your analysis may tell us is that

1. eastern-basin storms, probably thanks to their long duration, have additional opportunity to reach high intensity at some point in their journey

2. storms wax and wane as they travel and by the time the eastern-basin storms make it to the western extremes of the Atlantic they mainly reflect the western-basin conditions. They don’t carry “extra uumph” simply because of their birthplace (other than perhaps the size of their windfield).

98. Posted May 5, 2008 at 6:22 AM | Permalink

Ryan RE #94
You and i both expect the 2008 Hurricane season to be somewhat similar to 2007.
I also agree with Elsner that solar cycles seem to have an impact on the number and level of hurricanes in the Atlantic Basin. I am perhaps coming from a slightly different point of view. The number and magnitude of high level [over 10Npa and more] solar wind ram pressure spikes during the peak hurricane season seem to affect the number and size of the subsequent hurricanes. The year 2005 best illustrates this. For example :

The four LEVEL 5 hurricanes during 2005 were all preceded by or took place at the start of the hurricane by high level solar ram pressure spikes.

EMILY preceded by two 10-12 nPa level solar wind ram pressure spikes
KATRINA fed by a 30 nPa spike at the start of the hurricane
RITA fed by a 20 nPa and three 10 Npa spikes
WILMA preceded by a 15 Npa spike and fed by two 10 nPa solar wind spikes

The nPa data comes from OMNIWEB[ hourly intervals] plots
All other smaller hurricane of 2005 have similar patterns with varying and often lower level of solar wind ram pressure spikes associated with each.

I may be a minority blogger in believing that high level solar wind ram pressure spikes project more solar wind induced electricity into our Troposphere and atmosphere which causes joule heating of both, but that the extra electricity also provides the extra energy to power the hurricanes.
The hurricanes originating in the eastern Atlantic accumulate this extra electricity as they pass over the water.

99. Posted May 6, 2008 at 2:14 PM | Permalink

Cyclone Nargis made landfall in Myanmar this weekend and caused extensive damage and loss of life on the scale of the 2004 Tsunami. However, before aid is even being allowed in by the military government, Al Gore declared that the cyclone was a consequence of global warming. Al Gore and Fresh Air Interview on NPR

“And as we’re talking today, Terry, the death count in Myanmar from the cyclone that hit there yesterday has been rising from 15,000 to way on up there to much higher numbers now being speculated,” Gore said. “And last year a catastrophic storm from last fall hit Bangladesh. The year before, the strongest cyclone in more than 50 years hit China – and we’re seeing consequences that scientists have long predicted might be associated with continued global warming.”

There is nothing new to the argument made by Gore, except he added in some more anecdotal evidence from the prior couple of years. There are anywhere between 20-30 major tropical cyclones (Cat 345) globally each year, largely depending upon the phase of ENSO. Naming two of them maybe makes Gore sound smart — something the liberal media will surely lap up, just like NBC’s footage of penguins floating on ice near the North Pole…

100. Severian
Posted May 6, 2008 at 2:17 PM | Permalink

It figures that it wouldn’t take long for Ghoulish Gore to jump on this tragedy to try and sell sell sell that AGW mantra. Rising sea surface temps eh Al? Seems to me the sea temps are stable or declining, so how’s that figure into it? Rising temps mean more storms, stable or falling temps mean more storms?

101. aurbo
Posted May 6, 2008 at 10:19 PM | Permalink

Re the Myanmar cyclone and the large fatalities total. Deadly cyclones in the Head Bay (the northern portion of the Bay of Bengal) are not unusual. Most deaths are from the flooding caused a 12-foot storm surge as the multitude of islands in the Head Bay are mostly between 1-5 feet ASL. Here is a list of particularly deadly cyclones in and near Bangladesh since 1960. Deaths are for Bangladesh unless otherwise indicated:

Date Deaths

1963, May 28-29 22,000
1965, May 11-12 17,000
1965, June 1-2 30,000
1965, Dec 15 10,000
1970, Nov 13 300,000*
1985, May 25 10,000
1991, April 30 130,000
1999, Oct 29 9,372**
2008, May 4 22,000+

*Some reports put the total as high as 500,000 for this storm.

** In East India

+ In Myanmar

1970 was during the mid 20th Century cool period. So what does AGW have to do with the Myanmar storm, or anything else for that matter?

N.B. One wonders if the sesonal count of Atlantic hurricanes isn’t based on the same methodolgy that we’re seeing applied to counting votes in Gary IN this evening…and for similar reasons.

102. aurbo
Posted May 7, 2008 at 12:02 AM | Permalink

As they say in media-land; This just in:

The UAH April 2008 Global temperature Anomaly was +0.015°C. This compares with the March 2008 UAH reading of +0.089°C and the April 2008 RSS reading of +).080°C.

103. Severian
Posted May 7, 2008 at 6:00 AM | Permalink

Re: 104

Those are staggering casualty numbers. And they are a prime example of the reason that being poor, underdeveloped, and having no access to affordable energy or development are not the ways to live in response to “climate change” or natural disasters.

104. Michael Jankowski
Posted May 7, 2008 at 6:08 AM | Permalink

There is nothing new to the argument made by Gore

105. Michael Jankowski
Posted May 7, 2008 at 6:11 AM | Permalink

There is nothing new to the argument made by Gore

Actually, it almost seemed as if he’s taken a step back in your quote. He didn’t say that the consensus of scientists is that global warming is making hurricanes stronger, he said, “we’re seeing consequences that scientists have long predicted might be associated with continued global warming.”

106. MarkW
Posted May 7, 2008 at 6:40 AM | Permalink

[snip- I ask that people be polite to one another and snip offending posts when I notice them.]

107. Posted May 7, 2008 at 6:52 AM | Permalink

Re 106
MARK [w]
I have not stated that solar wind ram pressure spikes only affect the Atlantic Basin. To-date I have only studied this area in more detail. I think the entire planet is affcted by the solar wind spikes . I will check out this latest hurricane

108. Posted May 7, 2008 at 7:05 AM | Permalink

114 (MarkW): As I have said before, it is hard to know what is sarcasm and what is not. I personally think that vooro’s ideas are ‘not even wrong’, and if you really thought that there was something to his ideas, then I was contemplating a comment on the ‘solar wind pressure influence on hurricanes’. If vooro is alone on this, I wouldn’t. That’s all.

109. Steve McIntyre
Posted May 7, 2008 at 7:18 AM | Permalink

MAtt Vooroo, I really don’t like this blog to become a forum for people’s pet ideas. I wish as much as possible to limit discussion to conventional and influential literature. We’ve been through this before.

110. Posted May 7, 2008 at 8:41 AM | Permalink

Steve
The science of measuring solar wind and its impact on global weather has been around since the early 1960’s. Since I seem to be the only one bringing this science into our discussions in this blog, I may be giving you the wrong impression that this is my personal pet study only. All that one needs to do is to read the multitude of web pages and technical papers on this topic to see that the scientific community is deeply into this topic .Conventional science is far from agreement with many ‘influential’ aspects of current science as it relates to our weather as clearly shown by the recent debate about what causes global warming .Hurricane forecasters have been struggling with conventional science and have been coming short the last three years. Just like you are trying to get the credibility to the numbers , I was trying to bring fresh science into the dialogue to try to explain some of my numbers.

However this is your web page and you set the rules. I will not bring this topic up again on this web page.

111. Posted May 7, 2008 at 9:04 AM | Permalink

#106 I listened to the NPR interview with Gore. He obviously does radio so we can’t see him furiously reading prepared remarks by his own “hockey team” of so-called climate specialists. NPR is a safe outlet for him to propagandize; they will not challenge a word he says, ditto for PBS. Gore did mention:

…an emerging consensus among climate scientists that hurricanes are becoming more intense…

Also, Gore dropped in a mention of “more Category 5’s” as a consequence of global warming, which reflects back to Greg Holland’s sawtooth time series/trends of Atlantic Cat 5’s at the New Orleans annual meeting. There were “5” category 5’s last year globally, hardly an astounding number.

What in blazes is an “emerging consensus”? Does anyone have an analogous field or example of what that could be in the past?

112. MarkW
Posted May 7, 2008 at 9:16 AM | Permalink

It’s your site Steve, it’s just that to me it seems that your standards of what constitutes “polite” vary depending on who’s on the receiving end. This is especially true for comments with a political bent.

113. MarkW
Posted May 7, 2008 at 9:18 AM | Permalink

matt,

You said you would check out this relationship for hurricanes outside the NA months ago.

114. Posted May 7, 2008 at 9:39 AM | Permalink

#113, Climate science is so intertwined into the political (atmo)sphere that we have Gore saying, “this is not a political issue, but a moral issue”. There are other variations or talking points that have been foisted upon society over the past several years. Yet, this is part of polite discourse and the exchange of ideas. So, I would argue that the context and motives are critical components to the “climate auditing” process. The best way to combat charges of bias and agenda-driven science is to provide incontrovertible facts.

115. Posted May 7, 2008 at 9:46 AM | Permalink

113,115 (MarkW,Ryan): true, but still Mark’s ‘political bent’ shouldn’t be an issue as he suggests. And, so I agree that the debate should provide and discuss facts and not sarcasm.

116. MarkW
Posted May 7, 2008 at 9:52 AM | Permalink

I’m wondering why Leif is so dead set against “sarcasm”. Some of the greatest scientists in history are famous for their use of sarcasm. Leif himself uses it from time to time. This thread even.

117. Posted May 7, 2008 at 10:05 AM | Permalink

117 (MarkW): Sarcasm at times can be used to counter sarcasm, but is barren. Does not further the discussion, brings nothing to the table. When people are joking they often attach a smiley. One could do the same for sarcasm, clearly mark it as such, e.g. with :-[ . But, while I would say that levity is at times useful, sarcasm is not, as it is a negative that is intended to reflect badly on some individual which I find repulsive.

118. Posted May 7, 2008 at 10:06 AM | Permalink

As long as nothing offensive is said in terms of lousy language, gross distortions of others’ posts, or conspiracy theory level discourse, censorship should not apply. I started this thread to discuss the Tropical Cyclones of 2008 as well as “interpretation” of such trends, events, etc. Many posts on this board receive high visibility on Fox News, in the blogosphere, and in many email inboxes. I am not going to pretend that bias exists in the coverage of hurricane science, and prefer not to justify my remarks or suspend debate because something political may be contextually important. It is why climate skeptics are also referred to as climate realists.

119. Posted May 7, 2008 at 10:14 AM | Permalink

117 (MarkW):

Some of the greatest scientists in history are famous for their use of sarcasm.

Isaac Newton famously said: “If I have seen further, it is by standing on the shoulders of giants.” . While sounding good, it probably was sarcasm in its day. There are suggestions by some scholars that he was actually using the phrase “on the shoulders of giants” as a veiled insult of Robert Hooke, who was a rather short man. Newton had a reputation as somewhat of a petty and vindictive man whose ego clashed with those of his rivals in the scientific and mathematical communities. One of these rivals was Robert Hooke, who had been involved in a long-running feud with Newton over which one had discovered the inverse square law. Although Newton’s letter to Hooke appeared courteous on the surface, some historians have concluded that he cleverly employed the phrase “on the shoulders of giants” to ridicule Hooke’s lack of physical stature and imply that he lacked intellectual stature as well.
I don’t see how this kind of behavior is useful.

120. Posted May 7, 2008 at 10:23 AM | Permalink

119 (Ryan):

As long as nothing offensive is said in terms of lousy language, gross distortions of others’ posts, or conspiracy theory level discourse, censorship should not apply

I tend to agree [except that 'offensive' is somewhat in the eye of the beholder]. What I’m urging is self discipline and restraint. Ask: “what is the benefit of what I’m going to post?” “does this further the discussion?”

RyanM: we just went in a big circle and ended up right where we began…back to discussion on Hurricanes — like the developing storm in the Western Pacific, that is destined to affect lots of fish.

121. Sam Urbinto
Posted May 7, 2008 at 11:13 AM | Permalink

“If I have seen further, it is by standing on the shoulders of giants.”

I always thought he was downplaying his fame, that the only reason he has done better is because of the support of others. Although I suppose it could have been a snipe at Hooke (I’d call it an insult, but hey).

Anyway, here’s a 1991 event that’s interesting, more stuff at the home page. ( NASA’s Convection and Moisture EXperiment, part of Marshall)

122. Kenneth Fritsch
Posted May 7, 2008 at 2:28 PM | Permalink

Re: #120

While sounding good, it probably was sarcasm in its day. There are suggestions by some scholars that he was actually using the phrase “on the shoulders of giants” as a veiled insult of Robert Hooke, who was a rather short man.

While sarcasm can be a waste of time it can be used to make a point and in a clever if not subdued manner. What I prefer for my time spent here at CA is that when a point is made in a manner thought impolite or ill advised by some that comment be made and then we all carry on with the topic at hand. What I find a big waste time is a thread getting side tracked by the protocol police and the endless discussion from there. St. Steve then fells obligated to get involved when we all know that he should be using that time to write and publish papers.

As a kid I always had an image of Einstein as a kind and gentle old man until I read some of his conversations with other (and younger) physicists and in particular when discussing quantum mechanics and theory. It did not change my views on relativity — or quantum mechanics.

123. Posted May 15, 2008 at 2:15 PM | Permalink

Joe Bastardi has weighed in with his seasonal forecast of Atlantic Hurricane activity Accuweather Link

“Although we are forecasting a total of 12 named storms in 2008, much more important than the forecast storm number are the facts that a relatively high percentage of tropical storms are expected to make landfall and that the major threat area is farther north than normal.,” said Bastardi. “We believe at least 40% of named storms will cause tropical storm or hurricane conditions on the US coastline, which is about 1.6 times the norm.”

Also, in case you missed it, the WSI outfit has issued an above normal forecast as well: WSI Tropical Outlook

The 2008 forecast now calls for 14 named storms, 8 hurricanes, and 4 intense hurricanes (category 3 or greater).

Posted May 15, 2008 at 4:27 PM | Permalink

By this time last year, NATL storms had already been named and claimed.

Interesting – NAO seems to be going negative.

125. bender
Posted May 15, 2008 at 4:28 PM | Permalink

#125 NAO:

126. Posted May 15, 2008 at 8:14 PM | Permalink

Do you mean the PDO? The SST anomalies in the Gulf, near the Bahamas, and in the Caribbean are not very large at all. However, there are areas of large negative values in the East Pacific, off the coast of California, and in the Gulf of Alaska. SST Anomaly

127. David Smith
Posted May 15, 2008 at 8:35 PM | Permalink

Here are the entries as of May 15. There are less than three weeks left to make your entry and possibly enter the CA Hall of Hurricane Fame:

Much Below Normal

Sam Urbinto

Below Normal

matt vooro
Ryan Maue

Normal

Dan Hughes
1950-2007 climatology
John A
Paul Linsay

Above Normal

Terry
Steven Mosher
TSR
ma in va
1995-2007 climatology
Gray/Klotzbach

Much Above Normal

Mike B

Leif, we need you entry. Judith, if you’re around, we need yours and your Georgia Tech students. Ken, Bob, Steve, bender, willis, Lucia, everyone, use your short-term climate models to let us know what nature has in store for us.

128. bender
Posted May 15, 2008 at 8:39 PM | Permalink

David Smith: I already said ‘below normal’, in #91.

129. David Smith
Posted May 15, 2008 at 9:00 PM | Permalink

Re 3129 Very good, bender. I missed that.

130. paminator
Posted May 16, 2008 at 6:36 AM | Permalink

Put me down for below normal.

131. David Smith
Posted May 16, 2008 at 6:56 AM | Permalink

Updated:

Much Below Normal

Sam Urbinto

Below Normal

matt vooro
Ryan Maue
bender
paminator

Normal
Dan Hughes
1950-2007 climatology
John A
Paul Linsay

Above Normal

Terry
Steven Mosher
TSR
ma in va
1995-2007 climatology
Gray/Klotzbach

Much Above Normal
Mike B

132. Kenneth Fritsch
Posted May 16, 2008 at 10:18 AM | Permalink

David Smith, I do not believe I have seen your predictions and whether you are going to use a model for predicting that you were working on.

I planned on using a Poisson distribution, dependent on the AMM phase positive or negative, as related to Easy to Detect TCs and ACE and then converting/relating Easy to Detect TCs and ACE back to actual counts and ACE measures. Do you have any predictions of the AMM phase for this storm season? Obviously the toughest part of predicating TC activity is predicting the seasonal climate conditions.

Any word on the Meteo-France and ECMWF NATL TC counts for 2007? Does not this delay in announcing their 2007 results make one suspect that they must not have done well in forecasting? Maybe they want to average the 2008 and 2007 seasons together in order to improve the appearance of skill.

133. Posted May 16, 2008 at 11:45 AM | Permalink

Here is a prime example of how difficult it is for bloggers/journalists to accurately report the findings of new research:

Warning: this is from the Daily Green: New Study from Knutson at GFDL

A long-term forecast for Atlantic hurricane activity later this century predicts fewer but stronger storms that will pack stronger winds and more rain…Scientists don’t see this as the last word on the subject, but it supports the general, if still controversial, view that hurricanes will become stronger because of global warming. The results apply to the Atlantic basin only, which has the most impact on the U.S., but may not apply to other regions of the world, like the Bay of Bengal, which has recently produced two massively deadly storms, Cyclone Sidr in Bangladesh in November and Cyclone Nargis this month in Myanmar.

This is a tested and true tactic of journalists: take a study and then apply current analogue anecdotes to add weight to the study. Cyclone Sidr and Nargis have nothing to do with the new Knutson study, as the author blithely states, yet he throws it in there anyways.

The study contradicts a recent high-profile study by hurricane expert Kerry Emanuel which found that hurricanes are unlikely to become more frequent as the world warms.

Again, this does not contradict anything Emanuel said, and must be a typo (to be generous). The author misinterpreted the findings of Emanuel. But why even add this line? Now, for the icing on the cake:

2007 saw fewer hurricanes than expected, though there were several rapidly intensifying cyclones in the Atlantic basin. Three killer storms – Dean, Felix and Noel – were just retired by the World Meteorological Organization, and their names will not be used to describe tropical cyclones in the future.

The early predictions for 2008 are for an above-average year filled with frequent storms and several intense hurricanes. The storm season officially begins June 1.

Now the circle is complete. Take a study which models future hurricane activity — based upon past history (to prove accuracy of the future model, this is a dicey way to do business), add some doom and gloom anecdotes from the previous year (killer storms), and then extrapolate to the upcoming season.

This template has been repeated and will be repeated in the future as journalists scramble science in order to be climate change witnesses. When the new Nature Geosciences paper hits the web, a thread will be dedicated.

134. David Smith
Posted May 16, 2008 at 12:07 PM | Permalink

Kenneth, the AMM index can be found here . The SST forecasts I’ve seen, like here , don’t show any extraordinary SST contrast (=favorable AMM) in the peak of the hurricane season, so my guess is that the AMM will be only mildly positive.

My 2008 forecast has been calculated and will be released shortly. First, though, I figured I need to go ahead and publically release my 2005 forecast (25-28 named storms) and 2006 forecast (10 storms). I would have released these forecasts earlier but I’ve been busy. Later today I’ll check the actual records and see how my forecasts performed.

135. Bob Koss
Posted May 16, 2008 at 12:59 PM | Permalink

David,

Put me down for an ACE of 73.

136. bender
Posted May 16, 2008 at 1:09 PM | Permalink

Make my ACE 78.

137. Kenneth Fritsch
Posted May 16, 2008 at 2:44 PM | Permalink

Re: #135

My 2008 forecast has been calculated and will be released shortly. First, though, I figured I need to go ahead and publically release my 2005 forecast (25-28 named storms) and 2006 forecast (10 storms). I would have released these forecasts earlier but I’ve been busy. Later today I’ll check the actual records and see how my forecasts performed.

Well, David, let me be the first to inform you that you did just great in your forecasts. How did you do that?

Anyway since you have given me confidence in your predictions I am going with your prediction of a positive AMM to make my ACE and count forecasts. When I did my analyses of a Poisson distribution with a positive and negative AMM, I found that the value (how positive or negative) of the AMM did not affect the counts, i.e. the correlation between counts and AMM was determined by a cluster around a positive AMM and a negative and if one plotted the counts versus a positive or negative AMM value, the correlations were much lower. In other words, I only need to known whether AMM is positive or negative and not its magnitude.

I should also make clear that AMM positive and negative Poisson distributions were used for the TC counts and for ACE two separate normal distributions where used depending on a negative or positive AMM.

I will be spending some time putting together my forecast, and as as any forecaster worth their salt knows, will spend a goodly portion of that time determining good explanations for my forecasts going into the tank. First on my list will be David Smith’s forecast for the AMM.

138. David Smith
Posted May 16, 2008 at 2:58 PM | Permalink

Re #138 Thanks for checking the record for me – looks like my abilities are beyond belief.

Stay tuned because, later today, I’ll be issuing my predictions for the 2003 and 2004 Super Bowls.

139. David Smith
Posted May 17, 2008 at 1:06 PM | Permalink

Updated:

Much Below Normal

Sam Urbinto

Below Normal

matt vooro
Ryan Maue (72)
bender (78)
paminator
Bob Koss (73)

Normal

Dan Hughes
1950-2007 climatology
John A
Paul Linsay

Above Normal

Terry
Steven Mosher
TSR
ma in va
1995-2007 average
Gray/Klotzbach
raleighwx
David Smith

Much Above Normal
Mike B

I’ve chosen to go with the 1995-2007 average. I see no sign of a hyperactive season nor do I see an indication of season-shinking unfavorable ENSO conditions.

140. bender
Posted May 17, 2008 at 4:32 PM | Permalink

I’ve chosen to go with the 1995-2007 average.

Why 1995?

141. David Smith
Posted May 17, 2008 at 7:05 PM | Permalink

Re #141: An apparent shift in Atlantic activity, and the Atlantic Meridional Mode (AMM), occurred about 1995

142. frost
Posted May 17, 2008 at 8:15 PM | Permalink

I’m predicting … Below Normal.

My method is I flipped a coin 4 times. Heads was for high activity and T was for low activity. I got 3 Ts and 1 H, ergo below normal.

I hope the payoff to the winners is in quatloos. I sure could use some of those.

143. bender
Posted May 17, 2008 at 9:06 PM | Permalink

#144
Does a cross-spectral analysis of AMM and ACE yield coherence at around 5y and 10y? If so, then you have a paper, we have a model, and the correct answer may be the average of our guesses: borderline normal-below normal.

What’s Judith’s prediction?

144. David Smith
Posted May 17, 2008 at 10:24 PM | Permalink

Re #144 I don’t know the answer to either question, bender.

AMM and ACE values are available here . Kenneth Fritsch has done a lot of work on AMM and hurricane season values, including ACE, and may be able to shed some light on how they relate. A smoothed plot of June-November values is below, which visually suggests that they do relate with AMM and ACE possibly having similar eigenthingies:

If there’s something to the 5 and 10 year frequency idea then indeed 2008 may be a ho-hum season, ACE-wise. Interesting.

Regarding Judith’s forecast, maybe she’ll post. I hope Leif participates, too.

145. Posted May 17, 2008 at 11:11 PM | Permalink

Re #145, you don’t even have to do any smoothing, the AMM and ACE are well correlated whether looking at low or high frequency. This is not true with SST, which has zilch relationship with ACE at high frequencies.

April 2008’s PDO index value was the lowest since Jan 2000, when many thought the PDO phase had changed back to cold. This is not too surprising, since the PDO has El Nino spatial-like characteristics, which is still a topic of research, with some disagreement upon the temporal mechanisms involved in the relationship. So, the record strength La Nina for Feb 2008 is simply being reflected in the Pacific SST’s …

The recent series of Pacific storms, including Nargis in the Bay of Bengal, are very typical of La Nina springs, with development in the South China Sea or near the Philippines.

Everyone must also factor in global warming into their forecasts. Since no one has mentioned the actual ACE or number of storms that may entail, perhaps David could simply adjust everyone’s predictions upward by 0.

146. steven mosher
Posted May 18, 2008 at 6:51 AM | Permalink

do an ensemble forecast of the forecasters. hey, its climate science.

dont include bender or me. we are biased.

147. David Smith
Posted May 18, 2008 at 8:54 AM | Permalink

Re #147 moshe, believe it or not, the ensemble (average) forecast of CA readers last year did well (don’t ask about our spread, however).

Re #146 per Ryan’s note, AMM correlates much better with ACE than does SST on a seasonal basis:

(Footnote: the SST and AMM are for August-October and the ACE is for the entire season, which is a reasonable (but not perfect) combination as most of the ACE is from August-October activity. I would have done ACE for the August-October period but I don’t have that data handy.)

148. bender
Posted May 18, 2008 at 9:59 AM | Permalink

Well, well. Surely this relationship is known in the primary literature?
Where the heck is Judith?

149. Posted May 18, 2008 at 2:05 PM | Permalink

#149, the relationship between AMM and ACE is well established in a series of very well written papers by Dan Vimont and Jim Kossin from Madison UofW (BAMS, GRL).

The Tom Knutson paper was released today: Lots of press coverage. Judy is busy talking to journalists :-) Eric Berger Houston Chronicle

My question about all of these studies which attempt to balance the competing effects of vertical wind shear with SST: does the current relationship between vertical wind shear and storm development hold in the future? More plainly, does shear matter more or less over a warmer sea surface?

RealClimate doesn’t much like this paper…

150. Bill P
Posted May 18, 2008 at 3:32 PM | Permalink

151. Posted May 18, 2008 at 4:15 PM | Permalink

David Smith Re 140

My ACE forecast remains at 70-80 as before . If you need a single number for me ,use 75.

152. Posted May 18, 2008 at 4:46 PM | Permalink

david smith

Another organiztion that does hurricane forecasts is WEATHER RESEARCH CENTER. I attach their web page address They seem to be predicting a near normal forecast although they do not give an ACE figure. They are predicting 11 named storms and 6 hurricanes for 2008

http://www.wxresearch.org/press/2008huroutlook.pdf

153. Kenneth Fritsch
Posted May 18, 2008 at 5:58 PM | Permalink

I present in this post my long awaited May NATL TC activity forecast for the 2008 season. As explained previously, it is based on named TCs and hurricanes belonging to 2 Poisson distributions depending on AMM being negative or positive. The ACE index is similarly based on 2 normal distributions depending on a negative or positive AMM. The TC and hurricane counts and ACE indexes were adjusted for changing detection capabilities based on their Easy to Detect counterparts.

I used AMM positive for my forecasts based the Klotzbach/Gray 2008 forecast notes a favorable (positive) AMM for the upcoming NATL TC season – kind of – and found in this link:

http://typhoon.atmos.colostate.edu/Forecasts/2008/april2008/apr2008.pdf

Now for my forecast:

Named Storm Counts:

Mean = 14.2. Plus/Minus 1 St Dev Range = 10.4 to 18.0.

Hurricane Counts:

Mean = 8.1. Plus/Minus 1 St Dev Range = 5.3 to 10.9.

ACE Index:

Mean = 145. Plus/Minus 1 St Dev Range = 89 to 201.

154. Kenneth Fritsch
Posted May 18, 2008 at 6:11 PM | Permalink

Re: #148

David, as I recall, when I looked at the correlations of ACE to AMM from 1948 to 2007 the R^2 was about 1/2 the value you obtained from 1979-2007 and the correlation (R^2) for ACE to SST for the same period 1948-2007 it was close to zero. I can dig out the information to confirm my recollections.

Also I found a good and consistent correlation of ACE with wind vectors for the NATL over all time periods, whereas the SST versus ACE broke down to nearly zero in certain time periods. I concluded (conjectured) that these correlations were indicative of a spurious SST to ACE index relationship.

155. Kenneth Fritsch
Posted May 18, 2008 at 6:28 PM | Permalink

Re: #150

A strength of the new study by Knutson is that it uses an average of all the computer models used by the IPCC, a technique that generally leads to better forecasts.

Oh no, we must look for that one model result that gives the correct answer and in this case it may not have been run yet, since mumble, mumble these damn models just do not have the resolution (fill in any other suspected model shortcomings here) to produce that correct result — even though, and this is important to remember when doing these types of excercises, together they show that we have a climate problem.

156. David Smith
Posted May 18, 2008 at 8:13 PM | Permalink

Ken, I’ll update the list with your submission and matt’s WSR nomination in a moment. Matt, I note that the WSR technique involves solar cycles? It’ll be our first solar-powered submission.

Ken I chose 1979 as my start point due to the questionable quality of SST in the remote tropical Atlantic prior to 1979. That marks the start (more or less) of modern satellite-derived SST estimates. Both SST and AMM are dependent on good SST estimates, of course.

When I stretch the data back to 1948 I get r-squares of 0.41 for AMM/ACE and 0.13 for SST/ACE.

Thanks for the forecast and reasoning – I just skipped the reasoning part with mine and cut to the chase :)

On the point raised earlier by bender, his work a year or so back showed some kind of periodic behavior in the hurricane data. It looks like that behavior may also be present in the AMM numbers. Below is AMM (1953-2002) created by removing the 9-yr average then, for visual presentation, I did a 3-yr smoothing of those remainders:

I ponder (and I think this is also true for Bender) the appearance of periodic behavior – is it real or just noise, and if it’s real then can be be explained?

(Regarding the 1950s portion of the plot, I have little confidence in that SST data – I just dunno about it.)

157. David Smith
Posted May 18, 2008 at 8:22 PM | Permalink

Updated:

Much Below Normal
Sam Urbinto

Below Normal
matt vooro (75)
Ryan Maue (72)
bender (78)
paminator
Bob Koss (73)

Normal
Dan Hughes
1950-2007 climatology
John A
Paul Linsay
WSR

Above Normal
Terry
Steven Mosher
TSR
ma in va
1995-2007 average
Gray/Klotzbach
raleighwx
David Smith
Kenneth Fritsch (89-201, give or take)

Much Above Normal

Mike B

158. Kenneth Fritsch
Posted May 19, 2008 at 10:23 AM | Permalink

Re: #157

Ken I chose 1979 as my start point due to the questionable quality of SST in the remote tropical Atlantic prior to 1979. That marks the start (more or less) of modern satellite-derived SST estimates. Both SST and AMM are dependent on good SST estimates, of course.

David, I think Judith Curry and other climate scientists have frequently noted that their studies start in the 1970s because of the satellite data being more reliable starting then. That is all fine and good as long as one also notes that with those steep valleys and peaks in TC activity (and AMM for that matter) starting at or near the bottom of a valley and ending at or near the peak can seriously confound any relationships one is attempting to establish.

I think attempts to get data from before the 1970s is important to overcome the confounding that the cyclical nature of TC activity can cause. Using Easy to Detect Storms is a way of doing this for TC activity. I do not have a good feel for the validity of SST data in these localized areas, but if one needed one could go to less localized temperature data and long term trends.

If the long term trends in TC activity are relatively flat after compensating for changes in detection capabilities and we assume that SST has been increasing then we have another way of showing little or no correlation between the two.

159. Kenneth Fritsch
Posted May 19, 2008 at 10:35 AM | Permalink

The Tom Knutson paper was released today: Lots of press coverage. Judy is busy talking to journalists Eric Berger Houston Chronicle

Knutson was talking to Fox this AM.

My very amateurish analysis would indicate that wind shear can be better and more consistently correlated with TC activity than SST. Knutson’s paper is the second recently published paper to evidently point to the previously under estimated wind shear to explain potential future TC activity. More importantly I think it shows that the SST trends are not likely to account for much of the past TC activity increase in the NATL and that puts changing detection capabilities well into play for explaining much of the increase.

160. David Smith
Posted May 19, 2008 at 12:05 PM | Permalink

Re #159. You’re right, Kenneth.

Regarding the early SST data, part of my concern about pre-79 SST comes from the comparisons I did with various SST sets last year. I forget the details (on the road at the moment) but it seems like I compared three tropical Atlantic SST reconstructions and found odd differences in the pre-satellite era data. Sometimes the sets matched for a period but then one would diverge by set amounts (say, 0.15C) then converge back into agreement. That left me with the impression that the reconstructionists were making assumptions, and adjustments, to the raw data from these remote regions. I’ll dig out those plots this evening.

161. Posted May 21, 2008 at 12:06 PM | Permalink

I’m guessing NOAA will release a seasonal forecast tomorrow similar to that of Gray and Klotzbach. More details at my blog at the Sun-Sentinel, Ken Kaye’s Storm Center,weblogs.sun-sentinel.com/news/weather/hurricane/blog/

162. Posted May 21, 2008 at 1:32 PM | Permalink

Nice blog Ken. I think an oil company should sponsor the upcoming hurricane forecast, like a corporation that puts their name on a stadium. This way, when the forecast is horribly off, we can blame the Exxon hurricane forecast. In all seriousness, I would be very interested in knowing about the internal discussions that go on in the energy industry in terms of their forecasting guidance for Gulf of Mexico rigs.

The futures market already has the “hurricane insurance” premium built in (tough to see it now with oil at 133.50/barrel), which was in evidence when just after the 2006 NOAA seasonal forecast came out. Unfortunately, we do not see a refund or recompense for NOAA’s continued historically bad forecasts (2004, 2005, 2006, 2007).

163. Kenneth Fritsch
Posted May 21, 2008 at 1:52 PM | Permalink

Re: #161

I forget the details (on the road at the moment) but it seems like I compared three tropical Atlantic SST reconstructions and found odd differences in the pre-satellite era data.

Yes, I recall looking at two different SST reconstructions that seemed to differ and differ more as one went back in time. I remember I wanted to determine the sensitivity (to the use of a particular reconstruction SST data set) of some claims made by climate scientists in a paper they had authored. Sure enough they used the data set that best made their case. I was surprised that they did not give reasons for chosing the data set that they used in their analysis.

A similar case was encountered in the Douglas et al. use of a radiosonde temperature series for comparing model and observed surface to troposphere temperature trend ratios. The choice was evidently affected by the availablity of the latest reconstruction when the paper was submitted, but at least Douglas indicated (later) that the latest reconstruction provided data that were inconsistent with the physical science.

164. Judith Curry
Posted May 21, 2008 at 5:12 PM | Permalink

I see lots of activity on thread, hope to catch up over the next few days. FYI, here is my review on Knutson et al. paper (that I gave to Eric Berger, etc). Lots more to discuss in terms of concerns about their methodology, if anyone is interested

Knutsen et al. use a regional modelling approach, which has a resolution where
individual storms can be identified, although their model does not do a good job
at simulating the intensity (systematic underestimation). I note that his
resolution of 18 km is only slightly smaller than the 20 km resolution used in a
global climate model simulation by Oouchi et al. using the Japanese Earth
Simulator Supercomputer (which found an increase in the number of North Atlantic
tropical cyclones). So this regional modelling approach does have some
advantages over the coarse resolution climate model simulations (but not the
highest resolution climate models), but there are some disadvantages to this
regional methodology also. The use of a regional model with forcing from
climate models does not allow for any feedback between the hurricanes and the
climate itself (or even the sea surface temperatures), i.e. it assumes that
hurricanes are passive players in the global climate system (which is almost
certainly not true).

Apart from the overall methodological issues raised above, the main first order
reason that i am not convinced by Knutson’s paper in terms of tropical cyclone
counts is that they limit their future simulations to the period August-October.
This limitation precludes any change in seasonal distribution or season length
that would influence number.

There is strong agreement among scientists that intensity will increase in a
warmer climate, although there are arguments about the magnitude of the increase
(knutson’s paper generally falls within the range of other estimates here).
With regards to number of tropical cyclones, there is much less convincing
evidence one way or another. The highest resolution climate model simulations
tend to show a slight decrease in numbers globally, although a few have shown no
change or even an increase in North Atlantic numbers. In Knutsen’s paper, this
is the first time i have seen anything as extreme as a 27% reduction in the
frequency of north atlantic tropical cyclones. The IPCC states: “There is less
confidence in projections of a global decrease in the number of tropical
cyclones”. I don’t see that any of the recent studies have increased our
confidence very much in this regard owing to methodological problems in the way
this has been done, although most studies are showing a slight decrease in
overall tropical cyclone numbers.

The situation in the North Atlantic remains hotly debated with regards to
hurricane number. While observations of global numbers of tropical cyclones since 1970 show
essentially no change, numbers in the North Atlantic have increased beyond
anything seen in the historical record, and this increase is unlikely to be
explained away (in its entirety anyways) by undercounting in the earlier part of
the historical record. The argument for at least a near term continued increase
in number of North Atlantic tropical cyclones is that the North Atlantic has the
coolest SSTs of any of the ocean basins with tropical cyclones along with plent
of “seeds” in the form of african easterly waves, and hence there is unrealized
potential for further increases of activity in the North Atlantic. The Knutsen
et al. paper is notable in this debate in that it is the first paper to suggest
such a large decrease 27% for the North Atlantic. So until we understand what
is going on physically that controls tropical cyclone number, the various data
analyses, high resolution model simulations, and various proxy/downscaling
strategies will not be very convincing owing to their individual deficiencies,
and scientists will continue to disagree and assessments (like IPCC) will
continue to assign a high level of uncertainty to this.

Knutson is arguing that any signal of increased intensity would be too small to
have noticed yet, if you believe his numbers in terms of expected intensity
increase. There is another problem with looking at average intensity
(knutson’s 3%). What webster et al. and (indirectly emanuel) found is that the
increase in average intensity isn’t so great (6%), but that there is a shift in
the intensity distribution to more cat 4 and 5 (we found a doubling since 1970).
If you look at Knutsons figure 1c, you see his simulated intensity distributions
compared to the observed one. the main thing you see is that there is little
relations between the observed and simulated intensity distributions (the
simulations substantially underestimate intensity), and there are no cat 4 or 5
storms in the simulations. So they are completely missing the main signal that
you would expect in the intensity increase, which is the shift to more category
4 + 5s.

But we do see an increase in global intensity since 1970. given the uncertainty
in the data, the increase may be smaller than what Webster et al. originally
found, but it is still a discernible signal with any credible estimate of
uncertainty in the intensity data. However, others argue that the uncertainty
is too great to identify a credible signal. So there is still disagreement.
The only thing that people agree upon is that we have seen an intensity increase
since 1970 in the North Atlantic (this is in the IPCC), but there is
disagreement in terms of whether this is associated with global warming or
natural variability (it is almost certainly a combination of both). Frankly, i
think Knutson’s paper adds little to this particular debate since his
simulations systematically underestimate intensity.

165. Kenneth Fritsch
Posted May 21, 2008 at 6:32 PM | Permalink

Re: #165

Judith Curry, no disrespect, but I think I need to await a detailed reading of this paper and perhaps a detailed analysis on this thread as I unfortunately judge this to be one of your least informative reviews.

There is strong agreement among scientists that intensity will increase in a warmer climate, although there are arguments about the magnitude of the increase (knutson’s paper generally falls within the range of other estimates here). With regards to number of tropical cyclones, there is much less convincing evidence one way or another. The highest resolution climate model simulations tend to show a slight decrease in numbers globally, although a few have shown no change or even an increase in North Atlantic numbers. In Knutsen’s paper, this is the first time i have seen anything as extreme as a 27% reduction in the frequency of north atlantic tropical cyclones. The IPCC states: “There is less confidence in projections of a global decrease in the number of tropical cyclones”. I don’t see that any of the recent studies have increased our confidence very much in this regard owing to ethodological problems in the way this has been done, although most studies are showing a slight decrease in overall tropical cyclone numbers.

Perhaps Knutson’s paper, like Emanuel’s papers preceding it, should be considered a break through. The IPCC comment certainly preceded the publication of Knutson’s paper as would the consensus view that you seem intent on pitting against the Knutson paper. Science marches on – doesn’t it?

Apart from the overall methodological issues raised above, the main first order reason that i am not convinced by Knutson’s paper in terms of tropical cyclone
counts is that they limit their future simulations to the period August-October.
This limitation precludes any change in seasonal distribution or season length
that would influence number.

Looking at the TCs of named storm magnitude from 1905-2002, 79% occur in the months of August, September and October. In the period from 1970-2002, 78% of these TCs occur in the months of August, September and October. This would indicate that the ratio of these TCs within these months and without has been fairly constant. If Knutson is comparing the same months for current and future I do not see a major disconnect here.

While observations of global numbers of tropical cyclones since 1970 show essentially no change, numbers in the North Atlantic have increased beyond anything seen in the historical record, and this increase is unlikely to be explained away (in its entirety anyways) by undercounting in the earlier part of the historical record.

If one corrects (or attempts to correct) for the detection capability changes the ACE index and named storm counts for the NATL TC activity are almost flat when one goes back in time prior to 1970. Obviously when there is a cyclic character to the storm activity and one starts at or near a valley and ends at or near a peak one can claim to detect an increasing trend.

Furthermore correlating annual SST with annual TC activity does not show the consistent correlation that wind shear does and particularly so when going back in time before 1970.

Knutson is arguing that any signal of increased intensity would be too small to have noticed yet, if you believe his numbers in terms of expected intensity
increase. There is another problem with looking at average intensity (knutson’s 3%). What webster et al. and (indirectly emanuel) found is that the increase in average intensity isn’t so great (6%), but that there is a shift in the intensity distribution to more cat 4 and 5 (we found a doubling since 1970).

Which could be strong evidence that the WHCC estimates were wrong and that one must go back prior to the 1970s and consider detection capability changes and the cyclical nature of the TC activity in the NATL.

166. Posted May 21, 2008 at 6:58 PM | Permalink

Quoting Dr. Curry,

The use of a regional model with forcing from climate models does not allow for any feedback between the hurricanes and the climate itself (or even the sea surface temperatures), i.e. it assumes that hurricanes are passive players in the global climate system (which is almost certainly not true).

A critical limitation indeed..

Frankly, i think Knutson’s paper adds little to this particular debate since his simulations systematically underestimate intensity.

Even if the simulations did accurately estimate intensity, the aforementioned issues are not addressed.

167. Posted May 21, 2008 at 7:13 PM | Permalink

A broader observation: TC intensity change is poorly predicted on a 3-5 day time scale using the best computer models and technology we have available. Using an ensemble of these forecasts really doesn’t really help either. So, why are we devoting any more valuable resources to prediction in the years 2080-2100 especially when additional studies are simply inconclusive academic exercises?

None of these papers are addressing Judy’s main concern at the beginning of her post: how do TCs interact with the current climate?

168. Kenneth Fritsch
Posted May 21, 2008 at 7:17 PM | Permalink

Re: #167

I am not clear exactly what is meant by feedback between hurricanes and climate, but would not that feedback most likely be negative and if ignored overestimate TC activity. Did not Emanuel hypothesize the potential for hypercanes in a future climate with significantly higher SSTs and the negative feedback on climate temperatures that would have. Has anyone shown or quantified a feedback to this point.

169. Kenneth Fritsch
Posted May 21, 2008 at 7:31 PM | Permalink

Re: #168

None of these papers are addressing Judy’s main concern at the beginning of her post: how do TCs interact with the current climate?

Well if climate scientists do not have any clues in this regard then how can Judith state the we all agree that increasing SSTs will lead to more intense TCs and that Knutson’s analysis does not show sufficiently intense historical TCs . It seems a convenient partitioning of theory/observations to me.

170. Posted May 21, 2008 at 7:56 PM | Permalink

#170, It is a convenient partition, especially when theory and historical observations don’t agree.

171. Kenneth Fritsch
Posted May 22, 2008 at 8:10 AM | Permalink

Re: #168

A broader observation: TC intensity change is poorly predicted on a 3-5 day time scale using the best computer models and technology we have available. Using an ensemble of these forecasts really doesn’t really help either. So, why are we devoting any more valuable resources to prediction in the years 2080-2100 especially when additional studies are simply inconclusive academic exercises?

Ryan, is that a fair comparison? I would guess that the models looking years out are merely looking at TC activity related variables such as the prevailing/changing wind shear conditions and SST. The models than take into account the estimated effects from observational and/or physical science of these variables on TC activity. I say this without knowing the uncertainty of these model predictions/projections. I would guess they are large. Yet I would think that at least empirically it is less difficult to use trends to project over the longer term than it is to know what a TC will do in the next 4 to 5 hours.

The recent Knutson and Emanuel papers have brought wind shear effects strongly into play with regards to NATL TC activity. Perhaps on further analysis we will find that the wind shear effect on NATL TC activity has been a major confounding variable when attempting to relate SST to TC activity over the past thirty years or so.

172. Posted May 22, 2008 at 9:31 AM | Permalink

Yawn…

Federal forecasters Thursday predicted a busier-than-average hurricane season in 2008, with as many as 12-16 named storms expected to form, of which six to nine should be hurricanes. Two to five of those hurricanes will be classified as major, with wind speeds of 111 mph or higher.

The term “average” is terribly misused with regards to hurricane activity in the Atlantic.

173. Posted May 22, 2008 at 9:38 AM | Permalink

#172

I do not know which is more difficult — seasonal forecasting which has shown little accuracy or 3-5 day intensity forecasts or 7-10 day genesis forecasts.

When running a climate model of TC changes over the next 100 years, the papers all have a common theme — what is going to happen in the years 2080-2100. However, little is said about the cyclical changes in between — perhaps 2080-2100 will be a very active time period or a lull — akin to the previous cycles we have seen. The changes in TC intensity with global warming will doubtfully be linear, but interact with the AMO, AMM, PDO, etc. The simple fact that our seasonal forecasts are so far off and often leads me to believe that serious work needs to be done on the current climate — not 80-100 years in the future.

We should then look at the results of Knutson et al. and Emanuel et al. from 2010-2030 and make policy decisions based upon that!!!

174. JCH
Posted May 22, 2008 at 9:42 AM | Permalink

I made my hurricane bet in February – 40,000 shares. It’s up 40% in the hurricane “preseason”. Maybe I should sell.

175. Larry T
Posted May 22, 2008 at 10:33 AM | Permalink

Some of the best climate models are the hurricane tracking models. If you look at the ensemble of early tracks of a hurricane they may predict it to hit from South America to England. And these are the most accurate of any climate models.

176. Kenneth Fritsch
Posted May 22, 2008 at 10:41 AM | Permalink

I finally downloaded (free for non-subscribers) and had a quick read of the Nature Knutson et al. paper. Here’s a listing of results/conclusions from the paper as noted for changes for the 21st century:

1. No indication from models that the TC formation region expands with SST increases.
2. Threshold TC formation temperature increases with SST.
3. Model projected decreases in TC activity is more pronounced in western half of NATL (due primarily to increased wind shear in that region) and thus landfall hurricanes decrease more than hurricanes in general (about 30% versus 18%).
4. Model projects TC storm counts decrease by about 27%.
5. The models fail to simulate the most intense hurricanes in the observed period as model cuts-off at around 55 m/sec wind speed. This item points to the major reservation/deficiency of the model outputs.
6. Models project that wind speeds for hurricanes increase 1.7%.
7. Near hurricane rainfall rates are projected to increase by 37%, 23% and 10% at 50km, 100km and 400 km, respectively.
8. The model ensemble hind casts the hurricane counts for the Aug-Oct time period with an annual correlation of 0.84.
9. The authors point to changes in the region of reduced vertical wind shear being displaced by the gradient of SSTs in the NATL and that that could contribute to the 1970s to present change in TC activity in the NATL.

Posted May 22, 2008 at 10:56 AM | Permalink

Memorial Day Weekend … and still nothing in NATL.

178. Posted May 22, 2008 at 11:51 AM | Permalink

One of Ken’s points:

5. The models fail to simulate the most intense hurricanes in the observed period as model cuts-off at around 55 m/sec wind speed. This item points to the major reservation/deficiency of the model outputs.

Real-time TC prediction models (HWRF, GFDL, global models) do not have sufficient skill for intensity in 3-5 day time periods for the most intense storms either. So, if the climate models could produce category 5 TC’s, probably with 1-4 km resolution, what hurdle does that overcome? This is a red herring in my opinion.

179. Posted May 22, 2008 at 11:57 AM | Permalink

Journalistic malpractice

TAMPA, Fla. (AP) – The 2008 Atlantic hurricane season should be about as bad as normal or slightly busier, with a good chance of six to nine hurricanes forming, federal forecasters said Thursday in a new way of making predictions.

…as bad as normal or slightly busier…

The headline Up to 9 Atlantic hurricanes is naturally a good place to start…

180. JCH
Posted May 22, 2008 at 12:12 PM | Permalink

9 would be good! I just need up to 1 normally bad or even normally average hurricane to end up in the GOM. After that it can slightly peter out in a new way for all I care. Slightly knocking over one oil rig would be good, but I don’t want to ask Santa Claus for slightly too much.

Didn’t Knutson predict more Cat 4-5 hurricanes in a prior paper?

181. Judith Curry
Posted May 22, 2008 at 12:37 PM | Permalink

Ryan, I agree, what is the point of using a fairly high resolution limited area model if you still can’t simulate a rapid intensification or even a cat 3 hurricane? this paper makes me wonder about the GFDL mesoscale model. other models seem to be doing better even at lower resolution (e.g. 20-55 km) in terms of intensity simulations, including WRF and even the ECMWF global weather forecast model (at 55 km resolution). The big tropical cyclones (in horizontal extent), like the ones you get in the west pacific, indian oceans and sometimes NATL, are fairly well simulated in terms of intensity by the ECMWF 55 km operational ensemble weather forecast system, where at least some of the ensemble members show up with major hurricane wind speeds. Lacking resolution, what kerry emanuel tried in his most recent paper is a fairly sophisticated downscaling technique to get at intensity. In principle, this could be a good way to go, but in practice some of Emanuel’s assumptions in his downscaling method are questionable.

182. Posted May 22, 2008 at 2:47 PM | Permalink

The GFDL hurricane model’s bogus is the only way to analyze a TC anywhere near its intensity for initialization purposes. The HWRF’s variational procedure is a work-in-progress, to put it politely. And neither have shown much intensity skill over the long haul. I am not shocked by this since these models rely on the global model analysis for boundary conditions. The climate modeling efforts suffer from the same problem.

Judy, I think it is sufficient to say that the modeling efforts are still inadequate, inconclusive, and essentially in their infancy. I don’t know if we agree on my previous statement that policy decisions should not be made with these predictions, scenarios, or whatever you call them…

183. Judith Curry
Posted May 22, 2008 at 7:15 PM | Permalink

Ryan, totally agree that policy decisions should not be made based upon the future scenario model simulations of hurricanes The hurricane/global warming thing is sufficiently uncertain that it shouldn’t serve as any kind of a driver for overall AGW issue, but we have sufficient certainty that the next 20 years or so are going to bad in the atlantic whether you think it is only natural variability, only global warming, or combination of both. We need to be improve our land use policies, coastal engineering, etc. to prevent continuing catastrophic losses. We know the hurricanes are going to continue to come and likely get worse; we were lulled into complacency during the 70’s and 80’s when things were very quiet in terms of U.S. landfalls.

184. Posted May 22, 2008 at 8:39 PM | Permalink

#184 — It does not seem difficult to find agreement on these broader issues about preparations regardless of global warming.

However, I have noticed a trend lately in reactions to various studies. Kerry Emanuel and Kevin Trenberth were quoted with respect to the Knutson study as it being “poor” and “demonstrably wrong”. Conversely, Chris Landsea’s comments were cast in a tone of long-coming vindication. So, journalists know which doors to knock on to get their appropriate soundbites. It is becoming increasingly clear that objectivity is decreasing in terms of assessing a peer study, especially in hurricane climate science. Judy’s review above is an example of what should be done: explain the advantages and disadvantages and then the implications of each. Obviously writing up these comments takes time and effort, but is a lot more useful than pithy soundbites like “this study sucks because mine was right.”

185. Kenneth Fritsch
Posted May 22, 2008 at 11:30 PM | Permalink

Re: #174

When running a climate model of TC changes over the next 100 years, the papers all have a common theme — what is going to happen in the years 2080-2100. However, little is said about the cyclical changes in between — perhaps 2080-2100 will be a very active time period or a lull — akin to the previous cycles we have seen.

Ryan, you make a good point here and especially to those of us who have been critical of those climate scientists who have made some strong statements about increasing NATL TC activity and by confining their view to the 1970s to present. In the Emanuel et al. (2008) comparing the TC activity changes for 20 plus years points to the problem you mention:

This suggests that comparisons between 20-yr periods in the twentieth and twenty-second centuries for at least some models may be seriously affected by multidecadal variability, a point we shall return to presently.

In Knutson et al. (2008) we see a similar reference that indicates at least some acknowledgment of the problem you describe.

There is little question that Atlantic hurricane counts and the power dissipation index (PDI, a measure of the destructive potential of storms) have increased markedly since 1980, along with a recent increase in Atlantic SSTs (refs 4,5). Looking further back in time, observational analyses have produced intriguing but often conflicting results, depending in part on details of the adjustments for storm undercounts or intensity biases in the pre-satellite era4,11–15.

Yet I find it ironic that those who would extrapolate the current 1970s to present trends into the future seem to think that the increasing SST trend is the important factor and thus the cyclical nature of TC activity becomes secondary.

I am curious as to how one can run these climate models for 100 years and than only compare 20 or 20 plus year periods at the start and end. Are the models run for climate change throughout the 100 years and the TC activity results extracted for only the first and last 20 years? Is it a matter of availability of scarce computer time? Why have nearly the same 20 year periods been selected for comparison in so many different analyses/papers.

186. Kenneth Fritsch
Posted May 22, 2008 at 11:46 PM | Permalink

Ryan, totally agree that policy decisions should not be made based upon the future scenario model simulations of hurricanes The hurricane/global warming thing is sufficiently uncertain that it shouldn’t serve as any kind of a driver for overall AGW issue, but we have sufficient certainty that the next 20 years or so are going to bad in the atlantic whether you think it is only natural variability, only global warming, or combination of both. We need to be improve our land use policies, coastal engineering, etc. to prevent continuing catastrophic losses. We know the hurricanes are going to continue to come and likely get worse; we were lulled into complacency during the 70’s and 80’s when things were very quiet in terms of U.S. landfalls.

I assume the 20 year period to which you refer has to do with the past cyclical nature of TC activity in the NATL. You presented testimony to the US Congress a year or so ago in which you combined the effects of the cyclical TC nature and SST to present the threatening picture that you are painting in the above posted comment. What has the US Congress initiated in the meantime with regards to potential mitigating policies? It would appear to me that they are continue to be more than willing to bail people out of the consequences of bad or risky decisions relative to TC activity (and home financing, for that matter).

187. Judith Curry
Posted May 23, 2008 at 9:55 AM | Permalink

kenneth, i and others have spoken out publicly against policies that encourage risky behaviour, see

http://wind.mit.edu/~emanuel/Hurricane_threat.htm

188. Posted May 23, 2008 at 10:02 AM | Permalink

Global climate change has hit Jupiter as well and is increasing hurricane activity there as well (or red spot making).
Red Spots on Jupiter

The PDI is a crappy metric to use especially when your climate model cannot resolve major hurricanes. Based upon observations, storm size and intensity are not well correlated, so a model could easily resolve storms unrealistically. There are ways to get around this intensity underestimation bias in current operational models where you have some idea of truth, but not in the climate community.

The planet’s TC activity is intimately tied to ENSO which is related to the PDO. Observations since 1970 show this quite well. I would be curious to know if a climate model reproduces this relationship either in hindcasts or future runs.

189. Kenneth Fritsch
Posted May 23, 2008 at 1:35 PM | Permalink

Re: #188

The statement that you linked to above was one that I linked to CA sometime ago and pointed to its reasonableness. My query was whether you have heard of any policy initiatives/changes with regards to this matter. I have not.

190. Posted May 23, 2008 at 2:34 PM | Permalink

Bloomberg Commodities Report on Hurricane Season

Thank you to all of the brilliant seasonal hurricane forecasters out there. I am still waiting for my refund on coffee beans, insurance rates, gasoline, etc. from the last few blown forecasts. I foreshadowed this in comment #163.

Crude oil rose, headed for a third weekly gain, after a report forecast that the 2008 hurricane season may be more active than usual, threatening oil platforms and refineries in the Gulf of Mexico.

The journalists at Bloomberg are sharper than the average Associated Press or Reuters liberal arts oriented writer. They even manage to drop in a reference to the worthless forecasts that NOAA puts out:

The agency predicted 10 hurricanes last year, and only six formed. In 2006, NOAA forecast as many as nine hurricanes and only five were recorded. In 2005, the government forecaster underestimated storm activity.

I have a new slogan for NOAA’s forecasts: our predictions are like a stopped clock, after enough time, we will be right…

191. David Smith
Posted May 23, 2008 at 9:24 PM | Permalink

Update:

Much Below Normal

Sam Urbinto

Below Normal

matt vooro (75)
Ryan Maue (72)
bender (78)
paminator
Bob Koss (73)

Normal

Dan Hughes
1950-2007 climatology
John A
Paul Linsay
WSR

Above Normal

Terry
Steven Mosher
TSR
ma in va
1995-2007 average
Gray/Klotzbach
raleighwx
David Smith
Kenneth Fritsch (89-201, give or take)
US NOAA
Accuweather
Weather456

Much Above Normal

Mike B

By the way, an informative blog on current Atlantic tropical weather is here . It’s weather, not climate but the writer illustrates climate-related phenomena like wind shear, stable air, threshold SST and so forth using real-time situations.

192. Judith Curry
Posted May 24, 2008 at 9:55 AM | Permalink

Kenneth, there have been a few hints of rationality in dealing with this issue. North Carolina has bought up most of the coastal land in the cape hatteras region, no building there. people on the virginia coast are also moving inland. after the great galveston hurricane in 1900, pretty much everything moved to houston, which then developed into a major city. places like New Orleans (with alot of history, culture) and Florida (economy tied to coastal tourism) are much harder to convince to move inland.

The situation with federal disaster relief continues to get worse and worse in terms of encouraging the wrong behavior. Insurance companies are trying hard to discourage property on the coasts by raising insurance rates and dropping insurance, but there is much backlash on this. The situation is especially tense in florida.

Overall, we are not really moving in the right direction. The Army Corps of Engineers works very slowly and uses 50 year climatology data, is not factoring in longer term climate variability or change. its not just the ocean coasts that need work, but also rivers and lakes close to the coast (okeechobee, ponchartrain) that need work.

Lets say that we can expect avg of $10B damage per year in the U.S. for the next 20 years, totalling$200B (could be much worse). And something like 3/4 of that damage could be preventable with engineering, building codes, land use. hard to imagine that spending $100B over the next decade on mitigation/adaptation responses to hurricanes wouldn’t be a good investment. 193. JCH Posted May 24, 2008 at 2:21 PM | Permalink Judith, you’re confusing Texans with rational thinkers. After the Galveston flood, the city’s survivors built an extremely impressive seawall, the building of which by chance had been highly recommended to them by storm experts whose advice went unheeded before the flood. Guilt-driven-catch-up mitigation effort? Houston built a ship channel. It was started shortly after the Galveston flood. Opportunistic? Naw, just coincidence. but Galveston’s wharf did not get dwarfed by the ship channel without a fight. 194. Posted May 24, 2008 at 5:05 PM | Permalink #193. Judy, the government has “given” taxpayer money to coastal states or cities to buttress levies, build sea walls, or other adaptation boondoggles — and they decided to spend it on other things. Case in point is the Minnesota highway bridge collapse. This money is allocated as earmarks with little oversight. We are naive to expect the government to come to the rescue and continue to bail out risky property ventures. I do favor your other suggestion of better building codes, land use, etc. which would bring down insurance premiums for everyone. Here in Florida, on the radio there are public service advertisements which tell people: “when in a hurricane, you must evacuate if ordered to do so — because rescuers are not going to risk their lives to save yours — until well after the storm has passed. The New Orleans folks sitting on their roof waiting for rescue illustrate what Florida wishes to avoid. So, a combination of governmental oversight of adaptation and mitigation at the locale level coupled with something called “personal responsibility ” would go a long way to preventing needless loss of life. 195. Posted May 24, 2008 at 6:20 PM | Permalink For those that made predictions for Atlantic hurricane activity, one should not forget that the Eastern Pacific season started May 15, without much fanfare. Upon examination of my crystal ball (NCEP GFS), a major hurricane is forecast to develop and intensify in the 5-7 day time frame. Near surface wind forecast The GFS global model has been able to produce Category 3 winds (>100 knots) at 925 hPa quite readily over the past couple seasons with nominal grid spacing at 0.5 degrees. The Eastern Pacific activity was lackluster in 2007, with an ACE of 52, with is less than many historical Atlantic storms. NOAA made a rather helpful figure: 196. Kenneth Fritsch Posted May 29, 2008 at 10:46 AM | Permalink Discussion of the Knutson paper brought forward the issue of the authors confining their climate model study to the months Aug/Sep/Oct (ASO) and how that could affect the estimation of future TC frequencies for the entire storm season. I had previously put the TC storm counts by month into an Excel SS that needed updating. I updated the SS and then thought about making some statistical tests on the how increasing ASO MDR SST and total storm counts might affect the tendency of percent of seasonal storms occurring in ASO. In order to obtain a reasonable sampling I decided to use decade long counts, percentages and ASO MDR temperature anomalies. I also made the comparison using Jul/Aug/Sep/Oct/Nov (JASON) to seasonal total percentages. The results are in the table below. I used a non-parametric Spearman rank correlation to determine whether the correlations were statistically significant. The results listed below show that no correlations exist for any of the relationships tested. All rankings were performed by giving the lowest number ranking to the highest value of the item being ranked. Spearman rank correlation TC Count versus ASO %: r = -0.18 R^2 = 0.03 Stdev = 0.32 Z = -0.57 p = 0.57 Spearman rank correlation TC Count versus JASON %: r = -0.25 R^2 = 0.06 Stdev = 0.32 Z = -0.77 p = 0.44 Spearman rank correlation Temperature Anomaly versus ASO %: r = -0.15 R^2 = 0.02 Stdev = 0.32 Z = -0.46 p = 0.65 Spearman rank correlation Temperature Anomaly versus JASON %: r = 0.09 R^2 = 0.01 Stdev = 0.32 Z = 0.29 p = 0.78 197. Judith Curry Posted May 29, 2008 at 4:11 PM | Permalink Kenneth, nice work, using your analysis can you test the hypothesis: trends determined from ASO are different from trends determined from JJASON 198. Kenneth Fritsch Posted May 29, 2008 at 5:12 PM | Permalink Re: #198 I compared the decadal NATL TC counts for ASO and JJASON and used the Spearman rank correlation as before to calculate r, R^2, z and p. There is a very strong and statistically significant correlation between the ASO and JJASON TC counts in the NATL for the period analyzed. The results are presented below. Spearman rank correlation r = 0.936364 R^2 = 0.876777 Stdev = 0.316228 z = 2.961042 p = 0.003 199. Judith Curry Posted May 29, 2008 at 6:25 PM | Permalink having trouble posting this, will try again **** interesting new paper discussed on pat michaels blog, world climate report. pat doesn’t really get the point, this paper actually provides substantial support for Webster et al. (2005) Science paper that found global increase in the number of category 4 and 5 hurricanes. an online version is not yet available, but here is website and abstract http://www.agu.org/pubs/crossref/2008/2007GL032983.shtml GEOPHYSICAL RESEARCH LETTERS, VOL. 35, L14S08, doi:10.1029/2007GL032983, 2008 On tropical cyclone activity in the Southern Hemisphere: Trends and the ENSO connection Y. Kuleshov National Climate Centre, Bureau of Meteorology, Melbourne, Australia L. Qi National Climate Centre, Bureau of Meteorology, Melbourne, Australia R. Fawcett National Climate Centre, Bureau of Meteorology, Melbourne, Australia D. Jones National Climate Centre, Bureau of Meteorology, Melbourne, Australia Abstract A collective list of historical El Niño and La Niña events has been developed, based on an examination of different indices describing the El Niño-Southern Oscillation (ENSO) phenomenon. Based on this list, tropical cyclone (TC) data from a newly created TC archive for the Southern Hemisphere (SH) have been stratified accordingly and significant changes in TC occurrences depending on warm or cold phases of ENSO have been identified. TC trends in the SH (area south of the equator, 30°E to 120°W) have been examined. For the 1981/82 to 2005/06 TC seasons, there are no apparent trends in the total numbers and cyclone days of TCs, nor in numbers and cyclone days of severe TCs with minimum central pressure of 970 hPa or lower. However, significant positive trends in occurrences and cyclone days of severe TCs with minimum central pressure of 945 hPa or lower have been identified. ****** 200. John A Posted May 29, 2008 at 7:35 PM | Permalink I’m impressed by all of this statistical analysis of a stochastic process which shows a clear Poisson distribution. 201. Posted May 29, 2008 at 8:14 PM | Permalink #200, Judy, you have to be joking — substantial support? Talk about cherrypicking a result from a marginally important study. Pat Michaels’ summary is accurate in its simplicity. Kuleshov et al. 2008 is a weak paper regardless of its conclusions because it says nothing about data reliability — where do these pressure readings come from, and what significance is 945 mb? What version of the Dvorak technique, was there a reanalysis? Kossin et al. (2007) has different results. Referencing Knutson and Tuleya (2004) or Webster et al. (2005) does not independently confirm squat. We already know that ENSO changes the distribution of cyclone locations, where is the mystery in this? I downloaded the Australian’s fancy new dataset from here: ftp://ftp.bom.gov.au/anon2/home/ncc/cyclone/cyclones_newformat.zip No clue what is in it… 202. John Norris Posted May 29, 2008 at 9:08 PM | Permalink 1. Forecast the ACE category for the season 90 2. Forecast the number of named tropical storms 10 203. Judith Curry Posted May 30, 2008 at 6:07 AM | Permalink Ryan, the interesting part of the paper is that they find a TRIPLING of the most intense TCs (pressure less than 945 mb) since 1980. That is the main finding of the paper of relevance to the climate debate, i.e. they are saying that Webster et al. (2005) is on the low side! Michaels totally misses this, the main point, and talks about no increase in TC frequency (which webster and everyone else has been saying all along). We are currently in the process of revisiting the southern hemisphere and north indian ocean data sets, i will try to get more info from kuleshev on their data set. Their analysis obviously hinges on the credibility of their data. But this is the first paper i’ve seen to suggest Webster et al. trend is too low! surely this is interesting. Also, it is getting increasingly difficult to get things published in GRL these days. I’m surprised this slipped through without more documentation of the data. 204. Judith Curry Posted May 30, 2008 at 6:16 AM | Permalink The significance of 945 mb is that this is the demarcation between cat 3 and cat 4 on the saffir simpson scale, so below 945 mb is cat 4+5. Kossin’s analysis says nothing about the extremes (cat 4+5), his analysis scheme is geared towards getting the average intensity correct. note this was discussed on the listserv several months ago and clarified by Kossin. Upon reading the fine print, the new dataset has focused on creating continuity in individual tracks across the different jurisdictions of australian, new zealand and french meteorological centers. so at the borders, each storm has apparently received some sort of scrutiny and where discrepancy exists, the intensity was adjusted based on something (not clear what). apparently the discrepancies were so large prior to 1980 that they didn’t attempt any adjustments 205. Judith Curry Posted May 30, 2008 at 6:27 AM | Permalink One other point. pat michaels doesn’t dismiss the paper, he thinks it is important and will be overlooked since it doesn’t agree with the “party line” (e.g. Webster et al.). The exact opposite is true. Since the authors didn’t issue a press release (after all, they are australians and probably don’t promote science that way on a regular basis), the title is el nino, and the trend stuff is buried deep in the paper, someone would have to dig deep to smoke this out. i hadn’t notice the paper until i checked in at michaels blog. Interesting example of the need to market your paper to get it noticed. 206. Jonathan Schafer Posted May 30, 2008 at 6:48 AM | Permalink After intense study (also known as waiting til the last minute and not putting any prep work into it), my entries are as follows: 13 Named Storms 6 Hurricanes 3 Intense Hurricanes ACE Below normal (72) 207. kim Posted May 30, 2008 at 7:48 AM | Permalink 207 (JS) Still premature; surely the bucket adjustments will affect everyone’s forecasts. =============================== 208. Posted May 30, 2008 at 7:57 AM | Permalink Ryan, the BoM database you downloaded is quite useful regarding the quality of it’s info, especially in the last few decades where quality control measures have been applied. In at least the last decade, the aussies outperformed JTWC in operational forecasting most of the time, and their intensity forecasts were far more frequently validated by ground truth observations when TCs came near instruments. While there were TCs where JTWC definitely did do it better, overall JTWC forecasts often had errors of intensity estimation (at times quite serious underestimates, especially with the aussie ‘midget’ cyclones), probably because of a strict application of Dvorak that does not account for regional quirks that are well known to the BoM through their extensive TC research – and the BoM forecasters have a free hand to adapt the results of Dvorak analysis to local conditions (plus access to lots of live instrumental data in many areas) wheras JTWC forecasters have their hands tied and must stick to the rules in the handbook. However, when it comes to climatological studies, you should note that the BoM uses 10-min-av wind speeds while US based data uses 1-min-av, the aussie category scale is not based on Saffir-Simpson and is not compatible with it, and the BoM definition of what constitutes a TC is different to the US definition, meaning that you are trying to compare apples with oranges when it comes to more global studies. 209. Posted May 30, 2008 at 8:30 AM | Permalink Carl Smith, I appreciate the update on the BoM cyclone database. and the BoM definition of what constitutes a TC is different to the US definition, meaning that you are trying to compare apples with oranges when it comes to more global studies 210. Posted May 30, 2008 at 9:28 AM | Permalink Okay, this is from the new Kuleshov et al. (2008) GRL paper on Southern Hemisphere storms. I went ahead and did some painstakingly hard research and created a time series of 81 (4), 82 (1), 83 (5), 84 (4), 85 (4), 86 (1), 87 (2), 88 (6), 89 (3), 90 (3), 91 (8), 92 (2), 93 (9) 94 (8), 95 (7), 96 (5), 97 (2), 98 (7), 99 (8), 00 (5), 01 (3), 02 (10), 03 (9), 04 (7), 05 (6) That is a total of 129 cyclones, average of 5.2 and standard deviation of 2.7 — the time series is highly variable. 129 cyclones from 1981-2005 Hmmm…Webster et al. (2005) from 1975-2004 found a total of 105 as gleaned from the helpful table. Something isn’t registering here. When I take the BoM dataset and do a few simple commands on this trimmed dataset: http://www.coaps.fsu.edu/~maue/blog/sh_bom_tc.data cat sh_bom_tc.data | awk ‘($5 list_of_cat45

After trimming out a couple blank lines, I find a total of 70 cyclones.

So Judy, #205-207, I have no clue what is going on with their dataset. Please David Smith or someone verify my results from scratch.

211. Kenneth Fritsch
Posted May 30, 2008 at 9:45 AM | Permalink

Re #199

Using the number of TCs that occur in the ASO months versus those that occur in the other months for the entire year in decade increments and assuming a binomial distribution for the TC falling into ASO or the other months of the season, it can be shown that decadal ratios for ASO versus season total for the period analyzed in #199 are all within +/- 1 stdev of the average.

Re: #201

I’m impressed by all of this statistical analysis of a stochastic process which shows a clear Poisson distribution.

John A, I have noticed that you on occasion do a drive by Poisson comment on these threads. I think you need to be bit more explicit in what you mean.

I agree that the TC (and hurricane) counts in the NATL fit a Poisson distribution (so do Mann and Sabbatelli) if one accounts for the AMM being positive/negative and the counts are corrected for changing detection capabilities by using something like an Easy to Detect index. These fits are much improved over looking at the recorded TC counts long term without any adjustments. The ACE index is similarly better fit to a normal distribution by correcting for AMM phase and changing detection capabilities.

Re: #205

Kossin’s analysis says nothing about the extremes (cat 4+5), his analysis scheme is geared towards getting the average intensity correct. note this was discussed on the listserv several months ago and clarified by Kossin.

If Kossin’s analyses show the global storm regions, excluding the NATL, not showing an increasing trend in frequency or in average intensity (PDI or ACE?) how does one get more Cat 4 and 5 hurricanes out of this? Would not the Cat 4 and 5 hurricanes have lots of leverage on the average intensity? It would seem to require that the gain in intensities from the Cat 4 and 5 storms would have to be “compensated” by a substantial decrease in intensities of hurricanes and TCs below those levels. How would one hypothesize such a situation?

212. Kenneth Fritsch
Posted May 30, 2008 at 10:12 AM | Permalink

Re: #211

Ryan, I would like to analyze these data you present here a bit more in detail. Notice that Judith’s claim on the face of it and without the benefit of looking at the actual time series is impressive for the increase of intense hurricanes.

The graph shows that the intense storms appear to stay at an average plateau for long periods of time and then jump to a higher plateau. As I recall Webster and Holland referred to these breaks as regime changes. I do not recall whether the breakpoints were handled by eyeballing or by statistical methods.

If one is attempting to connect or correlate SST changes with more intense hurricanes on an annual basis these plateaus tend to get in the way of that explanation — unless the SSTs plateau coincidentally with the change in intense storms.

I think one must also consider any cyclical components in the time series when looking at these relatively short time periods. There is a real danger of selecting peaks and valleys in the series and confounding that change with a spurious connection to SST.

213. Kenneth Fritsch
Posted May 30, 2008 at 11:46 AM | Permalink

Ryan, I took your SH intense hurricane counts for the period 1981-2005 and did a change point analysis per the link given below. An eyeball change would have given me a break at 1990-1991, but the statistical method that I used gave it at 1992-1993. I show the trend lines for (1) the entire period, (2) 1981-1992 and (3) 1993-2005.

For this change point analysis I used the following parameters:

Probability = 0.05
CutOff length = 10
Huber Parameter = 1
AC correction method = IP4

Notice the two plateaus with little or no trend and the jump in the 1991-1993 period. Unless the relevant SSTs had this same “regime change” I think an argument for a correlation, on an annual basis, of SST with more intense storms would be difficult to make. Of course, an eyeballed or statistically derived break point is of less value without an explanation.

http://www.beringclimate.noaa.gov/regimes/rodionov_overview.pdf

214. Posted May 30, 2008 at 12:14 PM | Permalink

Ken, did you subset by pressure, which is the 5th column of that dataset?

I sent an email to the author of the GRL paper, waiting for a response.

From Klotzbach:

If the increases in TC activity found by Emanuel (2005) over the past 30 years (based on data from 1975-2004) and Webster et al. (2005) over the past 35 years (based on data from 1970-2004) are robust, one would expect to see similar trends over the shorter time span evaluated in this paper (1986-2005), especially since SST increases have accelerated in the past twenty years.

Judy’s response, which I agree with:

This is flawed logic, fallacy of distribution of the divisional type, whereby you cannot assume that what is true of the class is true of its members. You cannot dice up the 35 year period and expect the same statistical relationships to be present in each segment. 35 years is marginally short to identify a statistically significant trend (people who criticized our study because the length of the data record is too short raised a legitimate point). 20 years is definitely too short…

So how can a new study of 25 years be any better? Just trying to be consistent here and say that this new study is not an independent confirmation of previous studies. And it is folly to say their results are consistent with Knutson and Tuleya (2004) — unbelievable.

215. Kenneth Fritsch
Posted May 30, 2008 at 1:10 PM | Permalink

Re: #215

Ken, did you subset by pressure, which is the 5th column of that dataset?

I used your SH numbers listed and graphed in #211. I need to look at the Webster tabled results, but they appear on first glance to be in line with what you graphed from Kuleshov.

If the increases in TC activity found by Emanuel (2005) over the past 30 years (based on data from 1975-2004) and Webster et al. (2005) over the past 35 years (based on data from 1970-2004) are robust, one would expect to see similar trends over the shorter time span evaluated in this paper (1986-2005), especially since SST increases have accelerated in the past twenty years.

I tend to agree with more Klotzbach and disagree with Curry here — given that we are looking at TC activity versus SST. The point taken away from Klotzbach is that if someone is making a claim for a correlation/influence of SST with TC activity then one should be able to show that on an annual basis the relationship is reasonably consistent over time. The point I take away from Curry’s comment is that, ignoring the claimed SST versus TC activity connection, that the time period is short in determining whether the TC activity has changed. In fact knowing the cyclical nature of TC activity one must go back further than Curry suggests to detect the period of the cycle and whether one is selecting a near valley to near peak when claiming trends.

216. Mark H.
Posted May 30, 2008 at 5:48 PM | Permalink

I didn’t notice a direct link to the Knutson paper, so here it is: http://www.gfdl.noaa.gov/~gav/REPRINTS/KSGVH_08_HURR.pdf

217. Posted May 30, 2008 at 7:12 PM | Permalink

Ryan, I am not sure exactly what you are doing, but you need to be aware that the BoM dataset includes only cyclones that for at least part their journey effected the BoM AOR, which extends from SH 90E to 160E.

SIO cyclones that remained west of 90E are not included, so you will need data from MFR (Reunion Is.) to complete the SIO totals.

SWP cyclones that remained east of 160E are not included, so you will need data from both FMS (Nadi Fiji) and NZMS (Wellington NZ) to complete the SWP totals.

I know it’s a pain, but such is the reality of trying to do TC research when responsibility lies with different agencies for different regions!

218. Posted May 30, 2008 at 8:47 PM | Permalink

Carl, I am trying to get the data that they used in the paper and compare it to JTWC and previous studies.

[7] A new TC archive for the SH (the SHTC) has been
prepared at the Australian Bureau of Meteorology’s National
Climate Centre. The data for the Australian region (90E
to 160E) has been provided by the Australian Bureau of
Meteorology, for the area from 30E to 90E by Me´te´o-
France (La Re´union) and for the area east of 160E by the
Meteorological Services of Fiji and New Zealand. TC tracks
from these three archives have been merged in one homogeneous
archive, ensuring consistency of trajectories and
intensities when TCs cross regional borders. Record lengths
of the TC archives and their quality vary. Australian region
TC records are relatively complete only after meteorological
satellites came in operational use in the late 1960s [Holland,
1981]. Solow and Nicholls [1990] further discussed changes
in observations of TC activity in the Australian region over
time arising from changes in observational capabilities.
Examining the SHTC data, we confirm the findings of these
previous studies and conclude that TC records in the SIO
and SPO can be considered homogeneous from the 1981/82
TC season. The SHTC archive (available at http://www.bom.
gov.au/climate/change/) now consists of TC best track data
for TC seasons from 1969/70 to 2005/06. The data from the
SHTC archive were used in this study to calculate trends
and analyse the influence of ENSO on TC activity.

What archive are they referring to? So I need to track down 59 sub-945 hPa cyclones from the other agencies — wonderful.

219. Bob Koss
Posted May 30, 2008 at 10:07 PM | Permalink

The BoM database has a mishmash of wind speeds in both m.sec and knots. The classifications are not consistent. Look at Harry 1988. Has Class H3-H5 tracks and highest wind speed is 51 m.sec. There are others similar. Don’t think I’d put any confidence in the data.

220. Bob Koss
Posted May 31, 2008 at 2:42 AM | Permalink

Ryan,

You might find some of this hurricane data useful. They are .csv formated.
I added ACE, SST, Landfall, Travel, and Heading information to the Best Tracks files. Unfortunately, three of the basins only go up 2003.

221. Posted May 31, 2008 at 3:18 AM | Permalink

Bob Koss:

All wind speeds in the current version of the BoM database file ‘cyclones_newformat.zip’ are m/sec 10-min-av.

If I remember correctly, when parsing the data through a simple program to generate some subsets a while back, I found one cyclone that was still in knots, however character number 10 was clearly marked ‘N’ instead of ‘S’ in that case – however the ‘cyclones_oldformat.zip’ file was a real mish-mash and should be avoided.

I would not trust the intensity codes (H1 etc), as they are inconsistent and appear not to have been updated when quality control was last applied.

For climatological studies, I would suggest selecting data using CP in hPa, as Ryan appears to be doing.

Ryan:

You are right that you will have to chase the rest of the cyclone data from elsewhere, unless you can figure out how to scrape the data (since 1969-70) from behind this archive access page that only draws track graphs:

http://www.bom.gov.au/cgi-bin/silo/cyclones_sh.cgi

Update: Just looking into it, I found that on my Mac using Safari, as long as one only chooses a single season, the “View source” option opens a page that includes the 6 hourly Lat/Lon and CP in hPa for all systems in that season – if that is all the data you need then it should not take you that long to grab it all and figure out a way to extract it from the HTML junk that accompanies it.

222. Bob Koss
Posted May 31, 2008 at 5:55 AM | Permalink

The file I was talking about is the file Ryan linked in comment #202. The name says it is the new format file. It is still has many storms handled in knots. I’m not intending to do anything with it anyway.

223. Kenneth Fritsch
Posted May 31, 2008 at 10:14 AM | Permalink

Ryan, I downloaded the data you linked and did my Pivot Table magic from Excel on it and came up with 69 SH storms that meet the 946 or less pressure criteria. A total of 289 storms were found in this data set and thus the intense storms were 24% of the total. The results are plotted in the graph below and show no trend over the period 1980-2005. We will have to ask Judith whether we should invoke the “length of time period rule” as it sometimes seems a bit subjective.

As Bob Koss noted the wind speeds are not consistent or even always recorded. I think that can be corrected and will look at it next. Anyway the take away is that in this SH region for the period 1980-2005 we see no trend in intense storm counts – so these data must be flawed.

224. David Smith
Posted May 31, 2008 at 11:15 AM | Permalink

Well, tropical storm Arthur has formed over Belize this morning. The 2008 contest is open until June 5, so there’s still an opportunity to participate. I’ll post the revised board later today.

225. David Smith
Posted May 31, 2008 at 11:29 AM | Permalink

Arthur is actually a regenerated Pacific Tropical Storm Alma, which moved inland a day or two ago. The mid-level center of Alma reached the Caribbean and reformed the surface circulation there.

Thus, we get to count the system as two tropical cyclones, even though it’s really but one system.

Arthur has moved inland with a total ACE of 0.12 (normal ACE is about 9 or 10 per storm). That’s probably it, unless it regemerges over the Bay of Campeche with a new surface circulation and gets named “Bertha”. Conceivably this could become a three-fer system, getting counted three times.

226. Posted May 31, 2008 at 11:53 AM | Permalink

226: (David Smith)

If it were to re-emerge over the BOC, it would still be Arthur. If you’ll recall, Hurricane Ivan entered the USA in Louisiana and died the usual death of a TC over land, meandered for several days up through the TN Valley, emerged off the East Coast, raced south off the East Coast, crossed Florida, and BAM. Ivan was reborne.

The only unusual thing about Arthur is that it started in the Pacific. Most double-named storms start in the Atlantic, and migrate to the Pacific.

Tom

227. David Smith
Posted May 31, 2008 at 1:04 PM | Permalink

Re #226 Tom, yes, that’s been the historical practice, and I was actually trying to dryly tweak the NHC’s nose over storm count inflation. Failed humor!

I noticed that the NHC has changed its two-name policy and will now carry a storm’s name into its new basin. There is one exception – if the surface circulation dissipates (mid-level remains) then reforms in the new basin, the reformed center will get a new name, like Alma-Arthur.

228. Posted May 31, 2008 at 6:27 PM | Permalink

228: (David)

Sorry I did not get the humor the first time. I agree completely with you about storm counts. I often find myself wondering whether Arthur would have been named 20, 40, or 60 years ago. I believe I know the answer.

In the Svalgaard discussions, people question sunspot counts over time due to our ability to see the sun better now than in the past. The same goes for tropical systems. How does one reconcile changes in technology?

tom

229. John A
Posted May 31, 2008 at 6:54 PM | Permalink

Re: #212 Kenneth Fritsch:

Yes, I’m sorry it’s a little drive-by in my approach but I’m busy with the rest of my life and can’t post like I used to.

My point is that Paul Linsay established that the distribution of hurricanes in any year is indistinguishable from a stochastic process which has a Poisson distribution (incidentally that’s another result that should have been written up for publication).

So for the rest, trying to derive trends in a time series which has such a distribution seems to me to be a classic example of mining that series for spurious trends.

If Paul Linsay is correct, there is no trend. Nothing. Rien. Nada. Zilch.

To me, it’s like mining the time series dataset of results from a roulette wheel. People can see trends in anything when they don’t exist, and climate modellers are not immune.

230. Posted May 31, 2008 at 9:53 PM | Permalink

WHAT IS NORMAL?

The notion of “normal” when it comes to the Atlantic hurricane activity is becoming a pet peeve of mine. If we are indeed in an active period due to some sort of nebulous multidecadal cycle, then normal changes/has changed. It seems clownish to keep referring to the average frequency (number o’storms) of the period 1950-2007 as “normal”. From 1970-2007, the average ACE is 92 with a standard deviation of 60. So the years that were closest to normal over the past 20 years are

2006 (ACE=76), 2001 (ACE=108), 1992 (ACE=75), 1990 (ACE=97), 1988 (ACE=103).

Okay, these years’ ACE values seem reasonable, so about 75-108 is normal, over the past 20-years.

However, the NOAA forecasts use the median of the last 57 years in this image:

So, by eyeball, the above/below normal years are not changed (I used 1970-2007. I assume that NOAA uses 1950-2007, but who knows). No surprises here. But, on their image, when they use the term Seasonal Averages: ACE = 167% of the median, is this an average of the departures from the median, or the average ACE as compared to the median?

Anyways, in a long winded way, I don’t know what number of ACE that NOAA is expecting, 100-210% of the median. Can someone provide real values or opinions on just what “normal” is?

231. David Smith
Posted Jun 1, 2008 at 6:57 AM | Permalink

Re #214

Notice the two plateaus with little or no trend and the jump in the 1991-1993 period. Unless the relevant SSTs had this same “regime change” I think an argument for a correlation, on an annual basis, of SST with more intense storms would be difficult to make.

Kenneth, here’s the Smith-Reynolds SST anomaly for the Southern Hemisphic region in question:

Hard to see any indication of regime change in the 1990s.

232. David Smith
Posted Jun 1, 2008 at 7:10 AM | Permalink

Here are the updated contest entries. The 2008 contest remains open until June 5 so there’s still time to enter:

Much Below Normal
Sam Urbinto

Below Normal

matt vooro (ACE of 75)
Ryan Maue (72)
bender (78)
paminator
Bob Koss (73)
Jonathan Schafer (72)

Normal
Dan Hughes
1950-2007 climatology
John A
Paul Linsay
WSR
John Norris (90)

Above Normal
Terry
Steven Mosher
TSR
ma in va
1995-2007 average
Gray/Klotzbach
raleighwx
David Smith (ACE of 147.119352614, more or less)
Kenneth Fritsch (ACE of 89-201)
US NOAA
Accuweather
Weather456

Much Above Normal
Mike B

233. David Smith
Posted Jun 1, 2008 at 10:03 AM | Permalink

Here’s Jeff Masters’ comment on Arthur:

Would Arthur have been named 30 years ago?
Arthur is one of those weak, short-lived tropical storms that may not have been recognized as a named storm thirty or more years ago. Arthur was named primarily based on measurements from a buoy that didn’t exist 30 years ago, and from measurements from the QuikSCAT satellite, which didn’t exist until 1999.

234. Bob Koss
Posted Jun 1, 2008 at 10:23 AM | Permalink

RE 231

LOL no normal range to be found.

Typical way for them to do things. Use three unequal size buckets of data with the most recent period being the shortest. You’d think they’d catch on that they look foolish doing that sort of thing. OCD maybe? ;)

The full data set has a median of 88, a mean of 101, and stdev of 61. I would call all those periods normal. Most recent period has a mean of 144 after neatly skipping over four consecutive years of below median ACE.

The oldest time period has a median of 113 and a mean of 115. Calling that period above normal range doesn’t work since that range starts at 117.

235. (terry)
Posted Jun 1, 2008 at 10:34 AM | Permalink

David: since there’s another Terry here, put parenthesis around my name (I’m first on above normal)

236. paminator
Posted Jun 1, 2008 at 10:50 AM | Permalink

David Smith- Here is an update for my forecast-
ACE below normal, ACE=79.

237. Kenneth Fritsch
Posted Jun 1, 2008 at 11:06 AM | Permalink

Re: #231

WHAT IS NORMAL?

Well, I know for a fact that my reactions are normal, even though I sometimes doubt that for the rest of you.

Seriously Ryan, I think that normal and levels of normal are just convenient ways of classifying events. Like you say they can be misleading if one applies these labels out of context and that context here would depend heavily on the long term TC activity. If a reader is unaware of these long term trends and what could potentially affect them, then I think no matter how the events are labeled there will be confusion. The mainstream media is generally so acknowledged in looking for opportunities to label events as abnormal that I think their pronouncements are not taken all that seriously.

238. Kenneth Fritsch
Posted Jun 1, 2008 at 11:58 AM | Permalink

Re: #211

Hmmm…Webster et al. (2005) from 1975-2004 found a total of 105 as gleaned from the helpful table.

Ryan, I took the tabled data from this post and looked at the TCs falling into a Cat 4 or 5 classification and lesser classification as events guided by a binomial probability function. (I attribute my tendency not to be able to refrain from taking these data and statistically test it to (1) being retired and (2) to not being a statistician and thus not having a reputation to uphold in doing the correct statistical tests.)

I calculated the individual basin mean Cat 4 and 5 storms to total storms from the entire time period 1975-2004 and then determined the binomial means and standard deviations for each basin for the 1990-2004 period. I then calculated the difference between the mean number of Cat 4 and 5 storms for the period 1990-2004 and the entire 1975-2004 period assuming the total storms were the same for each time period. This difference was divided by the standard deviation to determine how many standard deviations the 1990-2004 mean fell from the 1975-2004 mean. The results were as follows:

East Pacific Ocean: 1.31 Stdev.

West Pacific Ocean: 3.14 Stdev.

North Atlantic Ocean: 0.53 Stdev.

Southwestern Pacific: 1.83 Stdev.

North Indian Ocean: 0.70 Stdev.

South Indian Ocean: 2.04 Stdev.

From this one could conclude that the West Pacific and South Indian basins had Cat 4 and 5 storm percentages for the 1990-2004 period that were greater than the period 1975-2004 by a statistically significant amount. Of course if one combined all the basin data, the statistical significance for a 1990-2004 increase in percentage of Cat 4 and 5 storms would be on the order of 4 standard deviations.

Notwithstanding a potential scolding from Dr. Curry (after delivering the same to PW) for using 15 year regime changes, I think that something could be causing a regime change of the percentage of Cat 4 and 5 storms in some storm basins. I just do not think it can be shown to be caused by changing SSTs.

239. Reference
Posted Jun 1, 2008 at 12:20 PM | Permalink

My projection:

28 named storms, including 15 hurricanes with 6 of Category 4 or higher.

ACE: 202

Projection basis is protected under FOI any attempt to release these details will be resisted to the full extent possible. So as not to appear totally uncooperative about this, note that these data are designed to ensure a top rating and maximum visibility in the media.

240. David Smith
Posted Jun 1, 2008 at 12:38 PM | Permalink

Update:

Much Below Normal

Sam Urbinto

Below Normal

matt vooro (ACE of 75)
Ryan Maue (72)
bender (78)
paminator (79)
Bob Koss (73)
Jonathan Schafer (72)

Normal

Dan Hughes
1950-2007 climatology
John A
Paul Linsay
WSR
John Norris (90)

Above Normal

(terry)
Steven Mosher
TSR
ma in va
1995-2007 average
Gray/Klotzbach
raleighwx
David Smith (147.319353)
Kenneth Fritsch (89-201)
US NOAA
Accuweather
Weather456

Much Above Normal

Mike B
Reference (202)

241. Posted Jun 1, 2008 at 2:45 PM | Permalink

I agree that we need a simpler and better system to define the hurricane seasons and what constitutes, above normal, normal, and below normal that everyone uses

For example:
The 2008 hurricane forecasters call the season as follows

We are calling for a very active hurricane season [CSU]

TSR slightly reduces its forecast but continues to predict an active Atlantic hurricane season in 2008

The Climate Prediction Center outlook calls for considerable activity with 65 % probability of an above normal season and a 25 % probability of a near-normal season [.NOAA]

WRC did not state a specific type of season but called for 11 named storms and 6 hurricanes which would be about a near normal season according to the definitions

The public receives this information from the media in different headlines as follows;

Active 2008 hurricane season predictions reinforce the need to prepare [ Red Cross]

2008 Hurricane season will be ‘Well above Average “[national Geographic]

Super-active 2008 Atlantic hurricane season predicted [Environmental News service]

The market place then digests this and we all end up paying

Oil Rises on forecast of Worse- than- average hurricane season [First Word/Bloomberg .com]

All of these forecasts could be high like they were in 2006-2007 when active sesaons were also forecasted but we had near normal seasons instead, but meanwhile we are all paying higher oil prices until November for three years in a row.

242. David Smith
Posted Jun 1, 2008 at 4:06 PM | Permalink

Re #214, #232

Interestingly, the satellites for the operational phase of Meteosat (Euro weather satellites) were launched in 1989, 1991 and 1993 (Meteosats 4,5 and 6). These are the ones which provide coverage of the southern Indian Ocean. To my knowledge the US doesn’t have weather satellite coverage in that region.

Any exploration of the apparent increase in SIO intense hurricanes would also have to look at changes in detection capabilities.

243. Posted Jun 1, 2008 at 6:34 PM | Permalink

I just received a response to my email from Yuriy Kuleshov concerning the Southern Hemisphere dataset. He confirmed that the link at the URL in the GRL article is only for Australian area storms, which was confirmed by Carl.

SH TC data are not available yet.

I may suggest you to look for SH TC data in

https://metocph.nmci.navy.mil/jtwc/best_tracks/shindex.html

Gee wiz, thanks for that help! This response is unacceptable. An important paper concerning trends in intense cyclones uses a new dataset of storms from various agencies and IT IS NOT AVAILABLE YET?

244. Posted Jun 2, 2008 at 7:18 AM | Permalink

One of the other things that I also observe is that the hurricane forecasters do not consistently use the terms above normal, near normal and below normal. Instead they use terms like, active, very active, etc.

“We are calling for a very active hurricane season” . The media quickly translate this into” well above average” or ‘super- active’ and a false panic quickly sets in across the nation.
Even FEMA picked this up and in their GET READY97 news bulletin prior to the 1997 season and said,

‘The 2007 hurricane season starts in a couple of weeks ,and the forecasters say this hurricane season could be nearly as destructive as 2005, the worst on record.’

In 2007, the use of terms like ‘” a very high 75% chance of an above normal season, a 20% chance of near normal and only 5% chance of a below normal season” were not helpful when the actual 2007 season was near normal and only a few points away from below normal . How are these high confidence levels arrived at?
I did notice that the 2008 confidence level was lowered to 65% for an above normal season but the media did not appreciate this and continued to define this season as very active or super active as in the past.

245. Posted Jun 2, 2008 at 10:05 AM | Permalink

#222, Carl Smith

Update: Just looking into it, I found that on my Mac using Safari, as long as one only chooses a single season, the “View source” option opens a page that includes the 6 hourly Lat/Lon and CP in hPa for all systems in that season – if that is all the data you need then it should not take you that long to grab it all and figure out a way to extract it from the HTML junk that accompanies it.

I hoped to avoid this issue, assuming the authors would point me in the right direction, but a cshell script and about 5 minutes of sorting has given me a Southern Hemisphere dataset. I do not know if this is the exact data used for the study, but the numbers of sub-945 mb storms looks about the same. Here is the data SH Web cgi page dataset

246. Kenneth Fritsch
Posted Jun 2, 2008 at 2:57 PM | Permalink

snip…

247. Kenneth Fritsch
Posted Jun 2, 2008 at 6:42 PM | Permalink

That less than symbol bit me again. Here’s the complete post.

Ryan, I finally went back to the original (partial) data set from Kuleshov (see Post #202 above) for SH storms and looked at that data going back to 1906. An immediate question coming to the fore is the capturing of storm pressure data all the way back in the time series but wind speeds listed starting in the middle of 1984?

I did a correlation and graph of maximum wind speed versus minimum pressure after converting all wind speeds in knots to m/sec (see below). The R^2 is not as high as I would have expected from doing the same calculations with NATL storm data where R^ ~ 0.90. There may have been some knots data that were not properly coded.

I also took the opportunity to look at the intense storm counts (hPa less than 946) going back further in time than I did in a previous post. Those results are graphed below. Notice that the 1980-2005 time period shows little or no trend but the entire 1906-2005 period shows a healthy trend that not only was obviously distorted by the lack of observed intense storms in the early part of the period, but extrapolates the trend back to a negative storm count by 1900.

248. David Smith
Posted Jun 2, 2008 at 7:59 PM | Permalink

Re #248

An immediate question coming to the fore is the capturing of storm pressure data all the way back in the time series (1906)

200 quatloos to the person who can explain how they did that.

Nice graphs. I agree that some of the outlier windspeeds are likely actually knots but are mislabeled as m/s.

249. Posted Jun 2, 2008 at 10:14 PM | Permalink

If any of you are wanting to compare US derived wind speeds with BoM derived wind speeds, you need to be aware that the BoM data is in WMO standard 10-min-av while US based data is 1-min-av, so BoM derived wind speeds will usually appear a bit less than US derived wind speeds.

The operational BoM Dvorak analysis conversion factor from 1-min-av to 10-min-av is usually either 0.88 or 0.9, depending on who is doing it – if I recall correctly, Perth and Darwin use 0.88 while Brisbane (and Fiji) use 0.9.

250. Bill F
Posted Jun 3, 2008 at 9:32 AM | Permalink

David,

Put me down for an ACE of about 112. I ran it through a proprietary model that is powered by Jack Daniels and that is what came out. If conditions change later in the year, I will be sure to let you know that my model predicted EXACTLY those kinds of changes to occur.

251. Kenneth Fritsch
Posted Jun 3, 2008 at 9:46 AM | Permalink

Re: Linked data @ Post #246

Ryan, I downloaded your SH complete (more complete?) tropical storm data and condensed the data into the table and chart listed below.

Interesting notes on the data are that the Total, and Non-Intense storm counts trend downward while the Intense Storms (pressure less than 946hPa) trends less steeply upward. I also included the NAs for pressure data that accompanied a surprising number of storms in the 1980s and into the 1990s.

The coefficient for the regression of Intense Storms versus time for the period 1980-2007 showed on analysis to be significant with a p = 0.02 and a mean rate of increasing Intense Storm counts of 1.35 counts per decade with a 2 SE range of 0.25 to 2.45 counts per decade increase.

I did not see a statistical significant change point for the Intense Storm time series, but one can eyeball a rather characteristic two plateau time periods from 1980-1993 and 1994-2007. While these plateaus may not have any statistical significance they usually have indicated in similar past analyzed cases that the underlying storm counts will have poor correlation to a corresponding increasing trend like SST on an annual basis. It would be of interest, I judge, to correlate Intense Storm counts in the individual SH basins to the storm season SST.

252. Posted Jun 3, 2008 at 10:47 AM | Permalink

#252, Ken this is a great analysis, I loved condensed things, especially laundry soap. Remembering that the Southern Hemisphere season straddles the New Year, when comparing your helpful table to the GRL article, I find 129 sub-945 mb cyclones. So, I would be confident in saying that this dataset, which required 5 minutes of scripting to extract from the cgi database, is indeed the one that the authors of the GRL article used.

I will take the time stamps in that database and cross reference with JTWC and add a column for its winds, which are reported as 1 min winds in knots. Then we will have a cross-validation between pressure in the Aussie SH dataset and the winds in the JTWC dataset.

253. D. Patterson
Posted Jun 3, 2008 at 4:28 PM | Permalink

There is a new 3 June 2008 Reuters article titled “Forecaster expects busy Atlantic hurricane season”, which reports Klotzbach and the Colorado State University (CSU) forecasting an increased number of hurricanes for the 2008 season.

254. David Smith
Posted Jun 3, 2008 at 5:21 PM | Permalink

Update:

Much Below Average

Sam Urbinto

Below Average

matt vooro (ACE of 75)
Ryan Maue (72)
bender (78)
paminator (79)
Bob Koss (73)
Jonathan Schafer (72)

Average

Dan Hughes
1950-2007 climatology
John A
Paul Linsay
WSR
John Norris (90)

Above Average

(terry)
Steven Mosher
TSR
ma in va
1995-2007 average
Gray/Klotzbach
raleighwx
David Smith
Kenneth Fritsch (89-201)
US NOAA
Accuweather
Weather456
Bill F (112)

Much Above Average

Mike B
Reference (202)

The contest will close at 12Z 6 June. There’s still time to join us and become a hurricanewizardologist. No experience required.

255. David Smith
Posted Jun 3, 2008 at 6:20 PM | Permalink

Kenneth, below is a plot of Southern Hemisphere Intense Storms (Cat 4 and 5) versus seasonal SST. The SST is a weighted average of the two basins and seasons as defined in Webster 2005. The SST source is NCEP reanalysis data.

Not much of a correlation.

Ken, if you’ll break out the list of intense cyclones by basin I’ll plot them against each basin’s SST and use Smith-Reynolds data for the SST. Those will be more appropriate than the generalized exercise but I doubt that the outcome will change.

256. Kenneth Fritsch
Posted Jun 3, 2008 at 6:33 PM | Permalink

Re: #256

Ken, if you’ll break out the list of intense cyclones by basin I’ll plot them against each basin’s SST and use Smith-Reynolds data for the SST. Those will be more appropriate than the generalized exercise but I doubt that the outcome will change.

I obtained my data from Ryan and the basin was not listed in that data set. Perhaps we can impose on Ryan to add that information.

I do have the date of and name for the tropical storms, so I would suppose with some effort on my part I could back the basin of occurrence out with that information. SWMBO has me on a current project of editing family videos and thus she is very much aware on the time I spend at my computer not doing the editing.

257. Posted Jun 3, 2008 at 8:16 PM | Permalink

I obtained my data from Ryan and the basin was not listed in that data set. Perhaps we can impose on Ryan to add that information.

Yes, I got caught up in the Red Wings game last night.
Cyclone Lat Lon Positions First is Lat (S) and then Lon (degrees E)
Since there is little metadata I could find with some sifting through Fiji and Reunion, I have no clue what Dvorak scheme was used to come up with the pressures. The updated Dvorak (1984) technique may not have been implemented there, and who knows what satellite data was available to those agencies. It is a shame that the lead-author of the GRL paper does not wish to share the dataset.

258. David Smith
Posted Jun 4, 2008 at 4:43 AM | Permalink

I’ll do the plots this evening. I’ll be surprised if the results are significantly different from # 256.

I’ll also do some comparison of the SH data vs North Atlantic data to see if the characteristics are similar.

259. Grant Anthony
Posted Jun 4, 2008 at 8:25 AM | Permalink

The following statements are most likely true about the new BoM 90e to 160e TC dataset:

1. It attempts to remove duplicate entries, erroneous and out of bounds values, incorrect units and obviously missing TC’s.
2. With the possible exception of missed systems, the key underlying location and intensity data does not change. That is, this is not a re-analyse dataset.
3. It is therefore “probably” valid, under those circumstances, to begin to use the database prior to a full disclosure of its new nature. I believe rather mundane and droll reasons are behind the fact it has not yet been (completely) peer reviewed and published externally.
4. Perth, Darwin and Brisbane TCWC’s have, in the past, used very different pressure/wind relationships in their respective basins.
5. Perth, Darwin and Brisbane TCWC definitions of what is/is not a TC, and the handling of hybrids, has also varied through time.
6. Therefore any attempt to use pressure values directly out of the database is fraught with difficulty. Serious consideration has to be given to each individual storm, the basin it was located in, and the method used to proscribe its intensity.
7. Having best tracked many many TCs past and present, I continue to marvel and how much new information and science has been brought to bear on the subject in recent times – and how much better I am at Dvorak analysis. The further one goes back in time, the less certain one should be about categorical intensities.

After looking at the data sets in detail I can only say that I no opinion on whether there is a trend in intense TCs over recent decades.

260. Posted Jun 4, 2008 at 8:48 AM | Permalink

#260, Grant, thank you for your opinions on the BoM dataset. I am highly concerned about a couple of your points, but I do not discount the impacts of the other points:

2. With the possible exception of missed systems, the key underlying location and intensity data does not change. That is, this is not a re-analyse dataset.

The Kuleshov et al. (2008) paper explicitly states that a “new” TC archive for the Southern Hemisphere was developed. 90e to 160e is the BoM dataset with 30e to 90e provided by La Reunion (Meteo-France) and 160e eastward by Fiji + New Zealand. “TC tracks from these three archives have been merged in one homogeneous archive, ensuring consistency of trajectories and intensities when TCs cross regional borders.”

How was this accomplished? How do you homogenize differing intensities without a “truth” or reanalysis?

3. It is therefore “probably” valid, under those circumstances, to begin to use the database prior to a full disclosure of its new nature. I believe rather mundane and droll reasons are behind the fact it has not yet been (completely) peer reviewed and published externally.

Not only is the BoM database used in this paper, it was also used by Ramsay et al. (2008) J Climate Interannual Variability of TCs in the Australian Region: Role of Large-Scale Environment.

If anyone has the data loaded up and ready to go, please give a check at those boundaries — and see what happens when/if a storm crosses 90e or 160e.

261. Kenneth Fritsch
Posted Jun 4, 2008 at 12:29 PM | Permalink

Re: Ryan at Post #258

Ryan, I am not sure how to apply the long/lat data, that you provided in this post, to the previous data set. Is it sequential with the previous storms? David, if you need the condensed storm data with minimum pressures by individually named storm I can email you that information. I can break it out as intense storms also.

262. Posted Jun 4, 2008 at 2:13 PM | Permalink

Yes, those lat/lons are simply like additional columns. Should be 25,683 or so points. I have yet to put together a cross-referenced JTWC dataset.

263. Kenneth Fritsch
Posted Jun 4, 2008 at 5:30 PM | Permalink

David Smith, I emailed you my Excel SS with the latitude and longitude of the individual storms with name and year and the minimum pressure.

264. David Smith
Posted Jun 4, 2008 at 9:04 PM | Permalink

Kenneth, thank you. I’ll pull data from it in the next several days.

Besides looking at basin storm characteristics and SST vs storm count, I want to compare this database with the Unisys one, which I think comes from the Joint Typhoon folks.

265. Posted Jun 4, 2008 at 10:21 PM | Permalink

There is a history file for the BoM 90E to 160E TC database that lists the corrections here:

ftp://ftp.bom.gov.au/anon2/home/ncc/cyclone/amends.txt

However the line breaks seem to be missing so it all runs together as a single stream of text on my Mac making it impossible to read, so I spent some time adding (Unix) line breaks to make it readable, and uploaded it here:

http://plasmaresources.com/ozwx/cyclones/bom_tcdb/amends.txt

266. Orson
Posted Jun 5, 2008 at 7:55 PM | Permalink

For hurricane watchers in Colorado, Phil Klotzbach will be appearing in Fort Collins to talk about his hurricane predictions. Free and Open to the public, it meets Wednesday June 11th at 6PM, Lucky Joe’s, 25 Old Town Square and runs until 7. Arrive earlier (or stay late) to socialize.

267. STAFFAN LINDSTROEM
Posted Jun 5, 2008 at 11:34 PM | Permalink

David Smith … I assumed NATL HURRICANE SEASON
PREDICTION/GUESSING/WHATEVER would be in by
Texas Time June 5 2008 … that is in 7 minutes
or so so … Mr Fogg can wait ….
Here is the Swedish Jury: NATL ACE 2008: …NOTB
BUT IF EXCEEDING 53 BY SEPT 27, DOUBLE…IN
NUMBERS 66.6 OR 133.2 CLEAR?? ALT 1 9/3/1
alt 2 15/6/4 (Number of storms/number of hurricanes/
number of major hurricanes)…submitted

268. STAFFAN LINDSTROEM
Posted Jun 5, 2008 at 11:39 PM | Permalink

#268 Ooops! Correcting myself …NOTB/10…of
course if not our dear Devil always is split in 10…

269. Blair Trewin
Posted Jun 6, 2008 at 12:17 AM | Permalink

I’ve dropped in to clarify a few things about the data sets we are talking about. There are two ‘new’ data sets which have recently been produced by the Australian Bureau of Meteorology – one covering the Australian region only (90-160 E) from 1906 to the present, and one covering the Southern Hemisphere from 1969 to the present. As these data sets have been developed in parallel they should match where they overlap in time and space. The Southern Hemisphere set is the one that is used in the Kuleshov et al. paper.

One important point: the set which is currently (6 June) on the web at ftp://ftp.bom.gov.au/anon2/home/ncc/cyclone/cyclones_newformat.zip is NOT the ‘new’ set. The ‘new’ set was only formally adopted as the official Australian best-track set a couple of weeks ago and we haven’t yet uploaded it to that part of the web, but expect to do so within the next week in place of what is there now (when it does go up it will be a zipped Excel file rather than a zipped text file, so you’ll be able to tell when the change has happened).

Grant Anthony (#259) is broadly correct as to the nature of the new data sets (although there have also been some changes to tracks). A reanalysis of intensities wasn’t, in general, part of the scope of this part of the project – that will come later. However, we have done a fair bit of analysis about what issues exist with intensities in general, and have reached the conclusion that there are no significant systematic biases in intensity after 1980 in the Australian region (I can’t speak for the broader SH), although the error bars on the intensity of individual cyclones are still substantial. The major intensity biases are pre-1970 but the analysis under discussion doesn’t extend to that period.

The bulk of the amendments to the data set that is on the web now will be in the pre-1970 period, especially between 1955 and 1970.

The database work hasn’t got as far as a peer-reviewed publication yet but hopefully that isn’t too far away. In the meantime the best available documentation is an extended abstract I presented to this year’s AMS Annual Meeting (available via the AMS website).

270. Geoff Sherrington
Posted Jun 6, 2008 at 4:32 AM | Permalink

Re # 270 Blair Trewin,

Since you are a climatologist with the Australian Bureau of Meteorology, might you please be so kind as to give the official definition adopted for “surface sea temperature” for the the purposes of comparison with storm intensity? Thanks Geoff.

271. David Smith
Posted Jun 6, 2008 at 7:04 AM | Permalink

The contest will close at 12Z 6 June.

It’s now 12 Zulu time (which is the same as 12 GMT or 12 UTC). The contest is closed and the race for prognostication immortality is on. Staffan, I’ve noted your ACE entry as 133.

Much Below Average

Sam Urbinto

Below Average

matt vooro (ACE of 75)
Ryan Maue (72)
bender (78)
paminator (79)
Bob Koss (73)
Jonathan Schafer (72)

Average

Dan Hughes
1950-2007 climatology
John A
Paul Linsay
WSR
John Norris (90)

Above Average

(terry)
Steven Mosher
TSR
ma in va
1995-2007 average
Gray/Klotzbach
raleighwx
David Smith
Kenneth Fritsch (89-201)
US NOAA
Accuweather
Weather456
Bill F (112)
Staffan/Swedish Hurricane Centre (133)

Much Above Average

Mike B
Reference (202)

272. David Smith
Posted Jun 6, 2008 at 7:42 AM | Permalink

Below is a plot of seasonal SST (Nov-Apr) vs count of intense cyclones (below 945 hPa) for 40E-120E (Indian Ocean from Africa to Australia) (thanks to Kenneth’s spreadsheet):

This uses NCEP reanalysis SST, which I’ll check against other SST constructions, but I doubt that there’s much difference.

There doesn’t seem to be a relationship.

273. Posted Jun 6, 2008 at 8:48 AM | Permalink

#270, Thanks Blair, we figured out most of what you said already. I asked the lead author of Kuleshov et al. (2008) for help with the data, and got a worthless response. It seems only readers of this blog who also work at BoM seem to have a handle on this new dataset.

274. David Smith
Posted Jun 6, 2008 at 8:21 PM | Permalink

Here are the x-y plots for SH intense cyclones near Australia and further east:

Not much of a SST/cyclone relationship in those splits. If I combine the two then the r-squared rises to 0.27, but I need to learn a bit more about the region to see if it makes sense to do that combo.

275. Kenneth Fritsch
Posted Jun 7, 2008 at 11:08 AM | Permalink

David, your graphs made it very easy for me to pick off the annual counts of intense storms (pressure less than 946) for each of the 3 storm basins you graphed. With that data, I determined how well each basin’s count frequencies fit a Poisson distribution. I was surprised that with relatively small total counts that the chi square goodness of fit was reasonably good for a Poisson distribution in each basin. The results are listed below.

Southern Indian Ocean:

Mean = 3.73 Counts per Year
Probability (p) from chi square test = 0.33

Near Australia:

Mean = 0.92 Counts per Year
Probability (p) from chi square test = 0.92

SW Pacific:

Mean = 0.78 Counts per Year
Probability (p) from chi square test = 0.34

276. David Smith
Posted Jun 7, 2008 at 12:52 PM | Permalink

Ken, that raises a question which I hope I can articulate.

Suppose that higher SST indeed causes a greater frequency of intense cyclones. How can that be statistically detected? My simple approach above is to check for linear correlation between SST and storm count. My assumption is that, if higher SST leads to higher intense storm count, then the population mean should shift and that shift should be evident in a linear correlation.

But, my sense is that in a Poisson situation I may need a large number of data points to truly explore for any SST/cyclone relationship.

My thinking is fuzzy and anything you or anyone else can say to de-fuzz me will be much appreciated.

277. David Smith
Posted Jun 7, 2008 at 1:28 PM | Permalink

When I re-read #277 I have trouble making sense of what I wrote, which is a bad sign. Perhaps the way to ask the question is this: is the scatterplot approach (SST vs cyclone count) a proper tool for exploring whether higher SST leads to greater cyclone frequency?