David Smith, a regular commenter on hurricanes, writes.
Tiny Tim is a Charles Dickens character. Tiny was a young lad, small, very weak, in a struggle to survive and of little notice in the hustle-bustle streets of London. Later, of course, his fortunes improved and he and Scrooge became “part of the record” of Victorian England.
In a similar vein (OK, it’s a stretch) there is a type of Atlantic tropical cyclone that is like Tiny Tim: generally of short duration, weak winds, small aerial extent and often in a remote part of the ocean. Its impact on its environment is tiny (a very small “footprint” in the Atlantic).
My operational definition of “Tiny Tim storms” are those that were so minimal that the NHC end-of-season reports do not report a single ship or single shore report of storm-force winds. This is not a matter of report oversights – storm analysts consider surface verification of wind estimates to be an important matter and list shore weather reports and ship reports in their reports.
And, the lack of ship or shore reports is quite significant if someone is looking at storm climatology. Storms lacking ship or shore reports of storm-force winds would, prior to 1945 (the start of recon), not have been classified as a tropical storm/hurricane. Why? Because, prior to 1945, all the meteorologists had were ship and shore reports. No aircraft recon, no satellites, no buoys and no Doppler radar – just ship and shore reports.
So, in this era of strong and many ships, rapid reporting and (US) shores lined with windspeed devices like onshore CMAN stations, a seeming plethora of data, are there still Tiny Tim storms, ones that modern technology sees but which lack storm-strength impact on ships and shores and which would have been ignored in the past?
The US National Hurricane Center (NHC) makes its end-of-season archives available at its website . An example of a storm report is here, the infamous Hurricane Katrina which shows a wealth of information including about 70 selected ship observations (hmmm, must have been some old dumb ships out there left over from the 1930s) of storm-forced winds and about 50 selected onshore observations.
Another example is Tropical Storm Ernesto , a report with with much less content because, well, it wasn’t much. There were no ship reports of storm-force winds and, as the NHC acknowledges, even in reanalysis it is of questionable strength or organization. Yet it carries as much weight in storm count trends as does Katrina.
(A word about reviewing the NHC archives – the ones for recent years are well-written and organized but that quality (for climatological purposes) diminishes as one goes back in time. There is a lot of verbage and data to review. I say this because I reviewed about 250 reports for this post and I may have missed some detail, one way or the other. I doubt that it’s material but I want to mention it anyway and welcome anyone who might audit my list.)
I reviewed the last 20 years of records as I figured that covered the modern increased-activity era (and the 1980s record quality became a bit more challenging).
So, the question of the hour is: how many Tiny-Tim storms, ones with nary a ship or shore report of storm winds, occurred in the last 20 years? The answer is here .
Frankly I was surprised. There are 52 storms on the list.That’s 52 out of the 252 storms in the official record, or 20% of the total. That’s 20% of the modern storms which lack a single classical (ship or shore) report of storm winds. Wow.
The obvious question is: how can one compare these satellite- and aircraft-based storms, which left no ship or shore evidence, with pre-1945 records which were based solely on ship and shore observations?
A little data.
First, here’s a look at a couple of characteristics of the Tiny-Tim group. Here is a bar plot of the duration of the Tiny Tims, grouped by days of existence (6 to 24 hours = 0 to 1 day, 30 to 48 hours = 1 to 2 days, and so forth).
The median duration is 1.7 days (42 hours). A few storms lasted beyond four days, ones that tended to be in remote open waters. For perspective, something that moves at, say, 10mph and lasts 2 days doesn’t cover a huge amount of real estate.
How about winds? Here’s a bar plot of windspeed distribution for the Tims. It shows that the group combined had 182 six-hour periods (45.5 days) of winds in the 35 to 39 knot range, as estimated by aircraft, satellite or buoy. The distribution has a mean windspeed of 43 knots (“strong gale” on the Beaufort scale) with 85% of the time spent below 50 knots.
There is an important graphic which I wish I could present but cannot because, to my knowledge, the data does not exist. The graphic would convey information on the geographical extent of storm-force winds. This is important because Tims likely have peak winds in only a small area on the eastern side of the center, perhaps 30 to 50 miles across typically. Tropical storms often lack symmetry and have their strongest winds in a relatively small area of thunderstorms.
As an exercise for perspective, figure that the hurricane-prone portion of the Atlantic covers 8 million square miles and that a Tiny-Tim has storm-force winds 100 miles across and moves at 10 mph for 2 days before weakening. That equates to the Tim covering 0.6% of the tropical Atlantic, which is not much.
Another useful graphic, which I have not done, would be a map of the storm locations. I think we’d see Tiny-Tims in the Gulf, along the eastern US seaboard (frontal-zone Tims), in the remote open Atlantic and scattered elsewhere.
OK, that’s a view of the group. Now for the main question: how have these storms affected the all-important trend in Atlantic storm count? What does the long-term time series of Atlantic tropical cyclones look like if the recent Tims are omitted?
Since my data covers only the most recent 20 years the plot is rather odd but does offer some information. The blue line is the official record (Tims included) while the red line is what the 5-year average would look like without the recent Tims. The comparisons should be (1) between the red and blue lines for 1988-2007 and (2) the red line (1988-2007) versus the blue line before 1945 (pre-aircraft). The plot shows notably fewer recent storms and shows recent activity more in line with historical (pre-1945) activity.
A closeup, with a few comments, is here . The impact of the Tims on the recent record is clear. I added several comments on the pre-1945 period, lest the question of a peak(1930s)-to-peak(2000s) comparison arises. The 1930s through the mid-40s was a period of limited global commercial activity, due to the Great Depression followed by World War 2 (this was shown on an earlier graph on CA a few months ago). Fewer ships in the 30s and early 40s meant less chance of an encounter with weather of any sort, including tropical cyclones. I suspect that this affected storm reports.
To me, this is further evidence of the problems with long-term comparisons of Atlantic storm counts and reinforces my view that improvements in storm detection are the main drivers, and perhaps the sole drivers, of the increase in reported Atlantic storms.
Climate Audit 
134 Comments
In Canada, maybe they’d be called “Timbits”.
I’m convinced David is right when he discusses weak storms being missed in the past. Decide for yourself.
Point of interest. Look at 1950 on chart. That year had only 13 recorded storms. A record eight major hurricanes, three others, and two tropical cyclones. No weak sisters. Compare with 2005 which had 28 storms and a weak sisters family reunion. The 2005 ACE was only 5 points higher. Which year do you think was actually worse?
David, I’ll email you this data.
David Smith, your Tiny Tim storm analysis is well researched and presented and, obviously, required some major efforts on your part. This was just a quick post before retiring for the evening to let you know that your efforts are appreciated.
Just curious, but Tiny Tim would put us of a mind to discuss The Ghost of Tropical Storms Past and The Ghost of Tropical Storms Present in this thread and, of course, for your storm predicting contest we would have had the Ghost Of Tropical Storms Yet to Come.
While it may be true that ship traffic was down in the ’30s due to economic factors, that effect must have been in large part canceled out after 1939 by military traffic (air and sea) across areas that were seldom visited by commercial ships in peacetime.
There was the air link shipping planes on the way from Brazil to Takoradi, and a lot of antisubmarine and (earlier in the war) hunting for surface raiders in out of the way corners of the southern north Atlantic and northern south Atlantic.
And going back to the whaling days the words of a sea shanty went;
On the 23rd of March, me boys, we hoisted our top sail,
Crying, heaven above protect us, with a sweet and pleasant gale.
The Tiny Tims were simply good sailing breezes, not storms!
sorry Steve, but i am unconvinced by this sort of analysis. you are using the improved technology that is being used for storm detection to discredit the current storm count.
on the other hand you are taking pretty unreliable data from the past at face value.
are you 100% sure, that not a single of the pre-1900 storms was a “tiny Tim” only?
our last discussion was on error bars. adding funnel-formed error bars to this graph, would change the outcome significantly and would cast massive doubts on your conclusion.
http://www.climateaudit.org/?p=2405#comment-164632
lol, i just noticed that by your definitions you actually are 100% sure 🙂
so let me rephrase:
are you 100% sure, that modern equipment is not downgrading some storms?
are you 100% sure, that ship sightings might not have upgraded some storms in the past?
Let’s say you’re the captain of ship in the late 1800s to early 1900s. If you did happen across some heavy winds would you continue into the storm to observe the peak winds or would you try to steer away?
re 1
I believe that the American term for what Canadians call Timbits is “doughnut hole.” So the metaphor is even more apt.
It would seem that in the official counts, at least, the ratio of minor storms to major ones is increasing. Leaving aside the strong possibility that this is an artifact, doesn’t this fly in the face of global warming theory that suggests that storm intensity and duration, rather than storm number, should be increasing?
In wartime a primary reason that ships would convoy was to limit the chances that a submarine would encounter them. Either they would all encounter the submarine or they would not. Individual submarines scattered randomly across an ocean would have fewer ship encounters. They would either encounter no ships or a great many at one time. A single submarine could not do much about a protected convoy. Wolf packs were a response to this but these were dealt with in other ways
A similar principle could be operational for tropical storms. The number of ships sailing might have remained constant or even grown but the chances a storm would encounter a ship would have decreased. ( The chance that a ship would encounter a torm would remain constant but this is not material). So there would be more reports of an encounterd storm but fewer encountered storms in total.
I don’t know if the analysis above rises above the level of arm waving but it might have an effect.
sod
are you shure, that Tiny Tims were recorded in earlier times?
Reread the post and you will see, that nothing ist mentioned about 100%! You just don’t get it once more.
sod,
You need to go back to the Hurricanes 2007 thread and review the material presented in 244, 248, 250, 251, and 252. If you have spent any time at sea, you would know of Bowditch and would know that the dumb ship theory in truly stupid. Weather at sea, is not a matter of convenience, it is a matter of life or death. In the 1800’s most dumb ships, did not report hurricanes, because they died. Ships that lived could not report hurricane force winds, because they had enough sense not to go there.
As to fish tales, when it comes to reporting weather by a ship’s master in the 1800’s and early 1900’s (or even today), a master that wanted to keep his command, would damn well tell it straight or his company and/or Lloyds of London would put him ashore. On the other hand, the deck hand’s bar tale would not be in the record.
Cliff
Probably an example of random recurrence in time series I have noted several times before, but something odd happened in 1888 …
Sod,
You might want to consider what charts in this post indicate. http://www.climateaudit.org/?p=2114#comment-144535
i noticed my error and corrected it. will you as well?
i am really glad that you old salt and sea dogs are explaining these things to me poor landlubber.
so please excuse me, if i will give you a “dry” land based example:
did traffic jams really increase? or are we just measuring them more accurately?
so we would compare trucker reports and modern traffic surveillance.
should we factor in that perhaps one of the drivers, now and then, fabricated some late delivery onto a small or non-existing traffic problem? and wouldn t they stop doing this, as soon as their claims could be verified?
please let me get this straight: i am not 100% sure, what effect this will have on storm counts. but i am sure, that your approach is one-sided!
“a person responsible for naming storms would damn well tell it straight, or the government would put him into a different position.”
see?
sod
I can’t follow your comparison with traffic, something really different from storm counts. Try it with common sense.
sod:
Your “dry land example” shows that you did not bother to read the material I cited and that you have no experience at sea. Given that, why do you think anybody should value your opinion on this matter?
Cliff
sod,
Ok, one more try at understanding.
Storm count before 1945, storm count after 1945 – that is what we are talking about. Storms counted before 1945 depend on landfall or dumb ships. After 1945, storm count is assisted by aircraft. After 1979, storm count is assisted by satellites. If you are going to compare historical storm counts, you need to have a common basis. The thesis of this thread is that the only common basis is to count (for comparison only) only those storms after 1945 that could have been reasonably observed before 1945.
The dumb ship theory says that most of the storms before 1945 were observed by dumb ships and that the count before and after 1945 is comparing apples to apples. This is a stupid position because it did not happen that way, because most pre-1945 ships would not survive steering through a hurricane (and very few modern ships would do it by choice). Very soon after Bowditch was published in 1802, ship masters knew and applied Buy Ballots’ Law, which kept their ships out of harms way (major storms). A ship’s course is not defined by a highway or rails, but rather by what gets the ship to its destination fastest and at the lowest cost. Driving through storms is rarely considered (or in practice) low cost.
The key to this issue is not questioning modern storm detection, but rather putting both modern and past detection into context so that modern and past storm counts can be fairly compared.
Cliff
As Bill Gray points out, the presumption of increased H counts in recent decades ought to be reconcilled against all US hurricane landfall records – which is flat. This constitutes the most complete and hardest single long-term count.
When one goes reaching beyond it (as Judith Curry, Greg Holland, Peter Webster, Kerry Emanuel and others do), one goes into waters only recently completely observed. (See for example, here.) Thus, arises the question of how to account for the undercount? Tiny Tim (or temporary) storms is one dimension that creates undercounts of the earlier record.
Since ship captains normally avoid large seasonal storms like hurricanes, another dimension is storms missed or under-rated in size, in the early records. A third dimension is the fact that most US trade is with Europe, especially Britain and Northern Europe. Especially, shipping routes via the industrial and transportation centers of the Northeast US.
Thus, one needed interdisciplinary study needed to account for undercount combines history, economic geography, and climatology to map these less travelled 20th century shipping lanes (further South), and their frequency of travel. The next step is to estimate the captain’s route-finding habits in the face of threatening and extreme weather. How often and how far did they vary their routes? Similar analyses, estimates, and adjustments throughout the Atlantic and Carribean basins need to be conducted in order to come to any honest picture of the hurricane counts before the frequent satellelite coverage era.
Chris Landseas may well be onto this path. A year ago, in an interview with Gary Taubes, he touched on this “normalization” or “bias correction” issue in regard to another aspect: was the 1990s exceptionally severe in inflicting hurrican property damage? He found that it was not:
After you take into account [various] factors, you find that the 1990s were not an outlier. Actually the amount of hurricane damage in the 1990s was about average, compared with the rest of the 20th century record, if you normalize for societal changes. The 1970s and 1980s were actually the quietest era for damage. The big conclusions in that paper were that the 1920s, 1940s, and 1960s had more damage than the 1990s.
Perhaps parallel to Steve’s dendro time-series data quality cautions, Landsea says:
And so it goes again.
sod, thanks for asking an interesting question about the historical data.
I’ll offer some evidence about pre-1945 practices. This ( link ) is a page from the October, 1936 Monthly Weather Review, which is the magazine-of-record in that era. Each issue contained a list of ship reports of gale-force winds in the Atlantic.
The type of information recorded, and my read of the storm summaries, indicates to me that the meteorologists look hard for corroborating evidence of a storm and discounted simple wind reports. They looked for key indicators like barometric pressure, wind direction, rapid windshifts at the time of lowest pressure and so forth (see green circles). They were looking for characteristics of tropical cyclones.
As examples of discounted wind reports I’ve marked two ship reports (red). Both reported tropical storm force winds in the subtropics (around 30N) with one reporting a suspiciously low barometer (29.33 Hg), 4-day gale and an unusual wind shift. Yet neither ship report resulted in a tropical cyclone being put into the official database.
It’s an educated-guessing game, but I think they erred on the side of underreporting.
Cliff:
I do not think that ‘the sod’ has ever lived near the coast; one can observe the approach of one of these larger scale weather phenomonons (TS or hurricanes) as a result of the circular cloud pattern extending out several days ahead of it (one need not feel the gale force winds to recognize the approach of an organized storm!) as I expect all mariners to this day would recognize.
It is likely, however, that the sod, a product of the contemorary i-sod -er- i-pod/vid-game/internet generation has little experience with out-of-doors, personal, real-world observations; vid game experience is no substitute for reality.
sod,
Can you explain why the storm count at sea is trending up and the landfalling count is flat?
Hmmm…quick word association:
Timbits come from Tim Horton’s donuts….Tim Horton was an NHL Hall-of-Fame defensemen…a user of hockey sticks…
sod,
A better landlubber’s example:
You are driving on the freeway and notice that traffic ahead is stopped. Rather than being stuck in traffic, for who knows how long, you take the next exit and a few miles of back roads to another freeway that will take you to where you want to go. When you get to your destination, can you fairly report the magnitude of the backed-up traffic on the first freeway? Was it a temporary stoppage or a major problem that lasted for hours?
The same applies for ships that avoid storms. Not knowing, they can’t fairly report on the storm. A ship’s master might log a report that guesses that a major storm was avoided, but the report would likely be discounted as an excuse for the trip taking longer than expected, however the excuse would be accepted as a ship’s master has the final word on getting his ship safely to port. The excuse is accepted, but the storm is not, because there is no data. This resulted in storms being under-reported pre-1945. Past 1945, the storm might be reported by an aircraft, past 1979 it would be reported by a satellite, but before 1945 it would not be recorded as a major storm.
Cliff
It seems like it would be interesting to plot shipping traffic on the graph as well. I would guess that there might have been an increase in shipping traffic over the years, (maybe not). It would also be interesting to see if the “shipping lanes” are the same now as in the past.
Patrick
Thank you David for this illuminating and well researched post.
A timely article which concurs with this storm accounting problem. But hey, logic and facts are ennemies of climate science.
And another thing, if a ship has radar on board, does it count as a ship report of storm force winds if they see it on radar as opposed to actually experiencing the winds?
Some of the lack of ship reports may have been due to better onboard weather info, and thus better avoidance. As I recall, WEFAX, though available since the 70’s really didn’t start taking off until the 80’s when low cost and compact WEFAX/HFFax (HF radio) units started to appear. Also about 1987, PC based WEFAX/HFFax systems started showing up. I remember the now defunct ALDEN corp pushing an inexpensive compact all in one unit about that time which could work either via satellite (suitable for shipping HQ/dispatch) or onboard via radio. I was doing some work with the Sunnyvale based company Oceanroutes (now WeatherNews) about that time and there was a big push for onboard weather info delivery.
Affordable GPS also started to emerge then.
Having a chart showing a TS/Hurr position onboard, plus accurate lat/lon makes avoidance easier. Better avoidance = less ship reports
Here is some WEFAX (and its replacement, LRIT) info http://noaasis.noaa.gov/LRIT/
Of course all this is just conjecture on my part. But I tend to see much of our NOAA data in terms of the technology coming available at the time. MMTS also came out in the mid 80’s and look what it did for surface temperature. 😉
David Smith, #21: “sod, thanks for asking an interesting question about the historical data.”
Ditto that. He has elicited several fascinating responses, the bogus analogy notwithstanding.
Bill Read, deputy director of the National Hurricane Center, disputed this in published reports saying, For at least the last two decades, I am certain most, if not all, the storms named this year would have also been named.
And the source of his certinty is?
What would be the reaction of the scientific community if I wrote a paper showing the astounding increase in tornadoes in Central North America over the last 500 years, concluding that the number of tornadoes is the result of the increasing caucasian population of the region? Of course, I would have to ignore the fact that no one was recording tornadoes 500 years ago, and the number of potential observers has been steadily increasing since then. I would also have to ignore the advent of Doppler radar, NWS spotter programs and so on. In other words, I would have to be fairly obtuse about scientific reasoning.
I am very sure that such a paper would be summarily dismissed as being silly. Just submitting such a paper should destroy any hope I would ever have of being a respected atmospheric research scientist!
Yet the same rational is being applied with the hurricane studies attempting to show a connection between hurricanes and AGW. Chris Landsea is being too kind when he refers to these researchers as being “a bit naive”. Frankly, I can not understand how any of these studies were ever taken seriously.
While the underreporting of the NUMBER of tropical cyclones may have ended with global satellite coverage, we are still improving our ability to recognize intensity. Just in the last 3 years, two tropical storms were upgraded to hurricanes well after most people forgot the storms even occurred. This was due to the reanalysis of data from radar technology that didn’t even exist 4 years ago.
Advances in our detection ability have been continuing nonstop. Ignoring this, or downplaying it as unimportant, is inexcusable.
Tiny Tim 1967 – The Other Side:
I wonder if Hansen and Gore and a few others around the place got bent on acid and listened to this? 🙂
Sod, I think you are defending the Judith Curry comment on the potential for past tropical storms missed being countered closely by the number that were incorrectly measured/observed as tropical storms. That comment would be considered a good naive starting point for a hypothesis if one lacked further analyses, information and insights.
In the meantime, David Smith, Bob Koss, Steve M and others here have presented no small amounts of data and analysis (evidence, if you will) that point to a significant number of tropical storms counted in the modern times that would have probably been missed in the past times.
If I were of your mind, I would look for the evidence supporting the Curry comment and the Dumb Ships theory and bring it back here for discussion. You, I would suspect, might be motivated to find something that I have not been able to uncover.
Re 32
Huh, actually there have been statements that there will be an increase in tornadoes and of course blaming it on global warming.
One example
However the data shows the number of F3-F5 tornadoes to be decreasing of course as you pointed out there are better detection capabilities increasing the tornado count overall.
Seems similar to the situation with hurricanes.
Jim C, speaking of upgrades in reanalysis, I notice that Tropical Storm Karen (September 2007) has been upgraded to Hurricane status ( link ). The upgrade was not based on an observation of hurricane winds but rather on cloud appearance.
I also noticed this, which was new info for me:
So Karen was a Tim.
One thing I forgot to mention is that I’m personally fine with the NHC’s practice of naming Tims and other weak systems, if that’s what their technology reports. Your never know whether a Tim will grow into a Katrina or an Ivan so let people know that a Tiny Tim is out there. My problem is with the apples-to-oranges comparisons of modern storm count with that of decades ago.
thanks for this explanation.
i do understand what is being done, and it is a reasonable approach.
but i do think as well, that you are ignoring the other part of the data. when looking at a dataline, that is made up by “hear-say” on one end and by accurate measurement on the other, most people would focus on the first part for problems.
the idea that modern technology is overcounting is fine. the remaining pen question is, whether the method before might not have been overcounting as well.
thanks for this information and for the polite answer.
the journal page is very impressive and data like this basically at least allows some investigation into pre 1945 reporting. i think writing at least a short paragraph about this would improve your analysis!
looking at landfall storms only is very much limiting the database.
cliff, you seem to imply that you have some oceangoing experience, but the idea that today’s ships, being smarter, avoid storms is not correct.
Some storms are too big to avoid, particularly when passing through constricted seas. I don’t mean through narrow straits. I was recently talking to a master who had made 140 passages through the north Pacific. He told me the biggest storm he ever encountered was off Kamchatka. There was no place else to go for him then.
A storm, even a big hurricane, is not usually fatal to ships in the open ocean. Being driven ashore is what killed ships in the 19 c. Even quite small ships — like the destroyer my father served on during WWII — weathered truly enormous storms.
Here’s an alternate way ( link ) of removing weak storms, courtesy of Bob Koss’ ACE data in #2 .
In this graph the storms with ACE below 2.0 have been removed (recall that a typical storm has an ACE of about 9 or 10, far above this 2.0 threshold). This effectively purges the marginal systems.
The AMO cycle is visible but I see little trend, with the current cyclic peak about matching that of the 30s-50s.
“Some storms are too big to avoid”
#38 The fact that ALL boats try to avoid storms remains true. Even biggest container cargos don’t go straight through storms even if the avoidance means very costly delays (an operating day of a mid size cargo of about 5,000 TEU costs 60,000$).
The difference with the past is now, captains have much sooner and better observations and forecasts on storms and on how to avoid them.
Perhaps the science of storm statistics is another real life example of Heisenberg’s uncertainty principle (modified a la Jurassic Park), where the act of observing is changing the phenomenon.
Re: 16 and 27
sod
Are we to assume that what is highlighted in post #27 is a proper example of your contention that: ” a person responsbile for naming storms would damn well tell it straight, or the government would put him into a different position”?
i was aware of that article, before i posted that example.
i was trying to point out, that “adjusting” storm intensity upwards might not be an invention of the late 20th century.
ps: as a poor fellow, grounded deep inland, another two questions for you seafaring folks: (i m just boiling a soup for dinner…)
1. what effect would a rise in water temperature have on storms in your oppinion?
2. do you belive that water temperature has increased lately?
“opinions” and “belief” are irrlevant, sod. what do the data say? look at the data.
Unfortunely, civilian ship-board navaigation RADAR is optimized more for navigation at much shorter ranges than that of RADAR used for the observation of weather (precip) artifacts; narrower pulse-widths required to view such details as channel width, navigation bouys and the like. Suffice it to say if you’ve got good storm-detail being painted on the face of your long-persistance Phosphor PPI CRT you will also be experinecing those gale-force winds …
temperature of the oceans is increasing.
i noticed some pretty strong wind over my boiling soup.
And during your “noticing” what sort of correlation did you get? Numbers, please.
this is going offtopic, so i will end it with this response.
my soup. slow but massive increase in heat. massive increase in air movement. obvious correlation.
there is a theory behind the increase in storms. if you assume a flat line, you will have to find more explanations.
Let’s not short change the other radio-based maritime global navigation systems in use prior to adoption of the NAVSTAR-GPS system, notably OMEGA which started in the 1950’s … a short summary of the available maritime rnav systems and time scale can be found here:
http://www.climateaudit.org/?p=1272#comment-96213
It is not off topic. You are simply evading the question because you are trapped. Shall I continue? Or will you concede that you were trying to advance an untenable argument in #43. Choose.
Re #46 – **temperature of the oceans is increasing.
http://www.cpluhna.nau.edu/images/climate1999.gif**
That is only 8 years old. Is the temperature still increasing?
Re: 43
Sod,
An increase in water temperature has little effect on storm intensity unless it results in an increase in delta T between the atmosphere and the ocean’s surface or between different parts of the ocean’s surface. Temperature differential, not termperature, is a driver of storm formation and intensity. If SST increase without a corresponding increase in atmospheric temperatures, then the delta T would be larger and there would be an effect on storms. If SST increase non-uniformly, then the delta T between different parts of the ocean’s surface would increase and this would effect storms. If both the SST and atmospheric temperatures increase, or the SST increases uniformly; then the delta T would be more or less unchanged and the effect of temperature on storms would be minimal.
This is basic thermodynamics.
There is data indicates that SST has been increaseing.
thanks for the answers Brooks.
And landfalling hurricane counts are dead flat, possibly DECLINING.
Sod,
You wrote:
I aqm not suggesting that modern technology is ‘overcounting’ the storms. Also, refering to pre-1945 data as ‘hear-say’ is a bit unfair to those who recorded the measurements that justified the classifications. The definition of what makes a tropical storm and what makes a hurricane has not changed in the historical record. The only thing that is changing over time is our ability to detect the physical charachteristics of a cyclone that define its classification. Then as now, storms were not upgraded until there was sufficient physical evidence to support the upgrade.
When we look at proxies for temperatures and say that there is uncertainty in the data, we really do not know which way the uncertainty goes. The proxy may be indicating temperatures that were warmer or colder than the actual temperatures. This is not true in hurricane data. Hurricane data is not proxy data, but an actual measurement of what occurred, based on our ability to record it at the time. There is no question that historical storms have been undercounted because we simply did not have the ability to count them all in the past, and even less ability to detect their peak intensity!
Your whole argument is based on speculation without evidence. It is a faith-based argument that may be more appropriate on a religion or philosophy blog, but lacks credibility here. I am not suggesting that you go elsewhere, only that you present physical evidence to support your argument if you wish to make a scientific point.
Re: 49
Sod:
It is not the temperature of the soup, but rather the difference in temperature between the soup and the air around. Increase the temperature of the air so that is is equal to the temperature of the soup and see what happens.
The theory that increased water temperature leads to more storms is based on an underlying assumption that only water temperature increases. If this is the case, then the delta T increases with increases in water termperature. This is the only way that increasing water temperature will increase the driving force for storm creation. If water temperatures increase without a corresponding increase in air tempertures, then one would have to accept that air temperatures are not rising.
If both air and sea temperatures are increasing then the delta T, and thus the driving force for storm creation and intensity, is not increasing.
Sod,
You are welcome.
Re: 47
bender
At least the soup data is current. sod’s reference in #46 is from 1999.
Re: 52
Brooks Hurd
It would be of interest to see the most recent SST data. My impression was that they have been declining, in particular in the SH. What about the tropical Pacific which is now La Nina bound? Do we have any recent updates on the AMO and PDO and a possible 30 year switch from positive to negative?
Re: 59
Tetris,
I will have to admit that I know a lot more about thermodynamics than about the recent trends in SST. I will gladly defer to other posters who have studied this topic.
By the way, my opinion on SST trends is not germane to the argument about the effect of temperature on storm formation and intensity.
I posted the link to global mean SST yesterday. There is a clear upward trend from the 1970s to 2000. Since then no trend or even downward trend. 2007 looks to confirm the downward (cooling) trend. And note the divergence between sea and land in 2007.
#61
Of course it isn’t. sod just hasn’t figured that out yet. He thinks he’s been excused from class.
The problem with dodge-and-weave, sod, is that the next time you return I will provide a link to your unfinished business here. Pay me now or pay me later. See you then.
RE 63. Buy some herbicide.
To detect trends in total storm activity perhaps the barometric records should be examined. It seems to me that storm intensity and counts would both correlate with the distribution of pressure readings.
Obviously one station would mean very little but the aggregate of many stations on the Atlantic coast and islands near NA and SA should be useful. Statistics like total days by month and year with a pressure recorded below 900 mb, 910 mb, 920 mb, etc.
Any comments about flaws in this approach? It wouldn’t bear upon the SST argument but it also wouldn’t require that storms be named by honest but subjective methods.
I had been looking at similar data.
I think my favorite storm was Tropical Storm Chris in 2000.
“until a burst of convection occurred and satellite estimates indicated that the depression reached tropical storm status at 1200 UTC 18 August. Soon thereafter, convection became disorganized and by the time the reconnaissance plane reached the area, the system had already weakened. In fact, data from the plane suggested that there was no longer a well-defined closed circulation.” Seemes it’s designation is entirely because of satellite estimate.
I looked at the last 10 years. Excluding 2007, and noted the average maximum winds for the weakest system for the period was 38 kts. Barely a tropical storm.
#64 Only took me a week and two tries to get that one 🙂
Barometric pressure readings won’t help. Storms prior to satellite era had few if any pressure readings. If you searched land-based records attempting to reconstruct them, you end up still only knowing about the close to land storms which already show no detectable change.
i think that there is a significant difference between a storm being observed from a far distance by one ship and continuos observation of a storm by satellite, planes and other technological means.
i surely don t know a lot about this field.
but water temperature is a real limiting factor . no tropical cyclones above cold water. so at best it is extending reach of the storms.
wouldn t warmer water allow for greater differences in temperature as well?
looking at the ups and downs in those graphs, i would not interpret too much into a two years trend.
a rather bizarre threat, coming from a person who wrote his last post to me a while ago.
i don t know what you mean with “dodge ” anyway. i told you that i will not comment further, because i think it is only peripherical related to the topic of tiny tim storms.
talking about data, there is ample evidence of a warming sea temperature.
there are a couple of articles about warm water and cyclones as well.
http://tinyurl.com/2ftfny
http://tinyurl.com/2fx9rx
http://tinyurl.com/yu3egl
what more do you want?
mosher suggests glyphosate.
sod, what is wanting in your argument is a real-world data link between temperature and a dangerously steep rise in hurricane frequency. i.e. That which does not exist. Forget your pots of boiling water. We have real data. There ain’t the trend you wish for.
It is what you are doing here right now – suggesting that you do not need to defend your argument here on this thread just because I’ve let you off the hook on another question on another thread.
sod, you are making assertions that are provably false. If you persist in doing this, dodging the necessity for defending these assertions, ignoring flawless counter-arguments, then this is called being a “troll”. Try not to go there. It is bad.
You must realize: it is not me that demands these proofs. It is your community and your conscience.
i did NOT claim any “dangerous steep rise”.
here are my questions again:
if you do not think that there is a connection between water temperature and storms, just say so. or give a reasonable answer, as Brooks did above. or ignore it.
but please do not put words into my mouth.
btw, this comment on another topic was not challenged:
http://www.climateaudit.org/?p=2471#comment-170160
sod,
The problem with limiting the data to landfalling hurricanes is that it cannot then be fudged to justify the cover art of AIT.
sod,
The discussion is about the actual numbers, not theoretical links to SST. How can we even talk about that until it is established whether the number is going up or down? Just like your traffic jam example, you are assuming the answer that you want, and arguing from there.
Sod,
You wrote:
Of course there is, but how does ‘one ship’ OVERCOUNT the number of storms or overestimate the intensity of an existing storm? The less ability one has to make detailed observations of the tropics, the more likely it becomes that details will be missed, resulting in an undercount of storms and intensities.
You have implied that there could have been nefarious reasons for ships at sea to ‘make up’ storms and intensity observations, but have not supplied any evidence that this occurred once, much less on a regular basis.
You are implying a vast conspiracy among mariners that could be easily discovered in a number of different ways. For example, imaginary storms would never make landfall. If sailors were adding imaginary storms to the records, landfalling hurricanes would make up a lower percentage of the total number of storms before WWII. Of course, the exact opposite is true, indicating more storms were missed in the past, not falsely generated. More directly, a ‘made-up’ storm could be easily falsified if other ships were in the area reporting clear skies and light winds at the time, making it awfully risky to generate false reports.
There is simply no logical argument supporting the notion that historical tropical cyclone activity was overcounted, while all evidence, logic and reason indicates that storms and storm intensities were undersampled the further back we go.
sod, you are a nice guy. patient, too. But listen. There IS a link between water temperature and storms.
The problem for your argument is that it, so far, appears to be a very weak link. Follow? There are factors other than global mean temperature driving storm dynamics, and these factors must be fairly important because the statistical association between temperature and storm frequency is weak to non-existent. This is why 2007 was a bust. There are processes we don’t understand fully that are influencing things. Argue from data, not theory. Theories are conjecture. Facts are facts. The AGW-hurricane link is so weak as to not be newsworthy. But look at the press it gets.
Bender asks:
And Sod replies #37:
sod,
You missed Bender’s point. If the storms at sea are trending upward and the landfall count is not then increasing ability to spot storms is far more a probable cause than an increasing number of storms that don’t happen to touch land. Unless, of course, you can name a mechanism that would affect the landfall rate.
Re 52 Brooks Hurd says (and echoed by others):
That’s not a very good argument. Water has a higher specific heat than air (nearly 4 times as high) which makes the ocean a low pass filter of air temperature (assuming SST is driven mostly by air temperature) meaning it is proportional to average air temperature. Presumably, the water is also absorbing solar radiation as well. Yes, the delta-T necessary but remember that means instantaneous delta-T. Air temperature has a diurnal cycle of higher amplitude than water’s cycle. That means there are times during the day when the SST may exceed the instantaneous ambient air temperature even if there is no influx of cooler air due to induced circulation. The air must also have adiabatic properties (lifted index) similar to (or maybe the same as) those required for thunderstorm creation.
So, sod is correct in saying SST drives storm creation but there are other variables that must be considered as well.
DAV makes good points.
sod, I think that the Landseas and Michaels and Klotzbachs and the people like me agree that increases in SST may well increase storm intensity, in line with Emanuel’s hupothesis, but the increase will be small (on the order of 5% higher windspeed per degree C of SST rise). That is too small for us to detect in the historical record.
At the same time, there are also plausible arguments that the tropical atmosphere would be more stable in a warmer world and (in key regions) more prone to windshear, which would tend to result in fewer and possibly weaker storms.
thanks a lot for all the friendly and interesting replies. i am learning a lot from the expertise of you guys.
i think that it would need a rather massive change, to show up in landfall numbers.
has anyone got a (landfall) trend line with error bars?
there is nothing wrong with having theory before looking at data. manipulating data obviously is wrong, but there is no connection between this two things.
(i think that it is a strong point of the global warming science, that they combine theory and lots of different data. )
if i understand the “tiny tim” methodology, then it is an attempt to limit this undercount. another limit attempt would be a closer look at ship numbers and routes. (it is possible that ships today discover more storms, because of increased traffic or changed routes)
i am aware of this problem.
the problem i see with the older data is this: if we count storms, that were encountered by a single ship only, there is a chance that the data is wrong. for the reasons i gave above, i think that an overcount (in storms reported by ships) is more likely than an undercount.
with modern surveillance, the probability of such an error is massively decreasing.
no. i am applying the same principal, that you are using while looking at the named storms numbers.
thanks. patience normally isn t my strong side. i guess i am just trying to show respect to our host.
so if a 5% increase in water temperature would lead to a 5% increase in storm number or intensity or range (or some combination), we might never be able to show the connection. we surely would be unable to show it in landfall storms for another century.
but it would still be a connection. missing it would be pretty unimportant. unless the the connection was not only proportional.
DAV makes good points.
nice, we posted 5% at the same time.
this is a position that i can agree with.
i know much too little about the subject, to argue over the numbers.
i (and i guess some others) did not know, that this was your position, before i asked the questions.
sod, you have to read the threads. that is homework #1.
homework #2 is this: install R and run this code:
It is easily modified to do poisson regression on the full time-series.
But be sure to replace the “ampersand lt” symbols with less-than signs. I will try again:
#Landfall hurricanes 1851-2005 (Source = HURDAT)year
argh. hopefully you get the picture.
An interesting post by SpenceUK related to this thread:
http://www.climateaudit.org/?p=2471#comment-170412
Sly sod references aside (Steve M, I realize and understand why you deleted my post(s) and not those of Mosher) I have a real problem, not with sod’s remarks, per say, but with what I see as a general wasting of bandwidth discussing evidenced-based science on a par with what I see as arm waving. This equivalency happens too frequently, in my judgment. I had earlier in the thread suggested that sod bring some evidence to the discussions to support Judith Currys comment suggesting historical misses being balanced by historical over counts in NATL tropical storm counts and the Dumb Ships theory. While I think David Smith and others here have addressed these counter arguments with evidence, I suppose it is difficult to think that real climate scientists would not have something in the way of evidence more readily available to back what appears to me as something otherwise to be considered conjecture. I will make a concerted effort to follow their references as completely as the internet links will allow.
sod says
nowadays storm are detected by satellites much eralier than by ships.
The statement that no valid information can be determined from landfalling hurricane countss because the restriction is too limiting seems like a mathematically testable statement. Do you have a quantitative explanation or a reference to a paper that does? Or is it just so much hand waving?
I can see one argument myself, that the increase in hurricanes is among short lived ones that don’t make landfall and that “normal” hurricanes are increased in intensity by some small amount, but I have no idea how one would prove it. And even if one could, it just does not seem that important.
I agree and in this case it’s partly my fault. sod was behaving like a troll, but I sensed that he was just being lazy, that he would come around if he were spoon-fed the data, code, and graphs. Rewarding for bad behavior is bad practice; that’s how people with troll-like tendencies grow into trolls. SI I fully agree, Ken.
The NHC maintains archives on many storms, as noted above. One storm from the modern era is Tropical Storm Beryl (1982).
Beryl was not a Tiny Tim (one ship reported 35kt winds). It lasted five days and had an ACE of over 5, which make it a seemingly decent storm.
But, the archives contain this interesting paragraph . “Had it not been for satellite pictures little would have been known of Beryl’s existence…”.
There was one ship report of 35kt winds but, as the reporter notes, in pre-satellite days the forecasters would have looked for corroborating surface evidence (low pressure, westerly winds, etc) which Beryl did not possess, so the ship report would have been tossed out and Beryl would be unknown.
sod says
I have to say that I agree with you 100% that the number of counted storms (pre 1945) is probably to high. The logic is as follows:
If a storm is counted then that count is either true or false. If it is true then there is no problem. If it is false then it is an overcount. The truth or falsehood of a counted storm does not allow for an undercount, only an overcount. Therefore the number of counted storms should be reduced by a percentage reflecting the probability of a counted storm being false.
However, 100% of uncounted storms are undercounts.
Now the real question is: Is the number of uncounted storms greater than the overcounted storms?
In the case of storms the size of Katrina shipping would find it difficult to miss so it would be counted, On the other hand, storms the size of Emesto could have easily been missed.
The further back in time you go the more “Tiny Tim” storms are going to be missed. This is because the further you go back, the smaller the area of the atlantic covered by shipping.
Re: #85
I have seen this tested twice, once by Kerry Emanuel and once by Roger Pielke, Jr. The results were contradictory, but I am not certain they used the same data base. Guess which tester obtained which result. Answer: Emanuel concluded that null hypothesis could not be rejected at the p equal 0.05 or less because of the smallish sample size.
I think that landfalling is only part of the evidence that as one uses progressively more easily detectable storm categories the counts over time tend to get flatter. I suppose one could combine all of the evidence in a single test.
Sod & Terry,
On what basis can you state that an observation, even if it is only from one ship, is false, without any evidence?
If you think it is reasonable to make such an argument, than I can claim that many of the modern day observations are also false, resulting in an overcount in modern times, then concluded that global warming reduces hurricanes. If you really believe that there is the possibility of an overcount in the historical record, then you must believe the same holds true for the modern record, inwhich many storms have reached their peak classification based on one observation! If we arbitrarily invalidate data without evidence, we can pretty much conclude what ever we want!
There is simply no possibility of systematic overcounts in the historical tropical cyclone record. On the other hand, undercounts must exist and be larger the further back in time we go.
Jim,
I dont say an observation is false, I say there is possibility of an overcount of counted storms. This probability lies between zero (no chance) and one (certainty). On the other hand every single unobserved storm is an undercount. The net result of this is that pre 1945 storm counts is probably (near certainty) too small
Bender (and others) — to get less than signs into web pages, use ampersand-l-t-semicolon. <this works fine, so does the equivalent for gt: >
But here are two other hints that I just worked out through a bit of googling and experiment:
1) The <code> tag is not what you think. It simply formats inline text with a fixed font. No proper help with pesky special characters (see below). Just
very plain text.2) Of much more interest is that WordPress has nice support for a different tag that appears to do close to 100% of what you want: Surround a block of code with <pre> and </pre> tags, and you’ll get pretty code, with indentation, most special characters and everything else you want! See below for MrPete’s code conversion rules…
#text tests (NOTE: you can copy/paste all this code, even if it overflowed to the right!) LTasMath = (a < b); // This is fine. But convert < to < to be safe (see below) keyboard="!@#$%^&*()_+-={}|[]:;<>?,./~`"; // This matches the special characters on my keyboard # backslashes and quotes, with all < converted to < : backslash= ' \\ '; // getting a backslash can be tricky. backslash-backslash gives a single backslash. quote= ' " ' '; // Plain quotes inside singlequotes show as-is... quote= " " ' "; // And so too, inside doublequotes... quote= " \\" "; // Often, dquote inside dquote is prefaced with backslash. Just double the backslash. quote= ' \\' '; // Same for single quotes url="http://some.dom.ain/folder/file.php"; //I hope this will NOT be turned into a link smileys=" ;) :) 8) and more..."; // Oh, and as you see, emoticons still get interpreted :( #indentation test x= 2x; if (a < b) { y=mx + b; // if only it were that easy... } else { x= 0; } # final test. An unconverted taglike < in text: produces typographical quotes in all following text: mytext= ' test ' + " more ";3) True inline code isn’t reliable. The pre tag breaks a line; the code tag inserts blanks after less than signs:
InlineCode= (a < - b);4) MrPete’s rules for code quoting:
Terry,
Your logic makes no sense.
1) You claim that it is likely that a storm encountered by a ship will be falsely identified as a hurricane. How??
Are you claiming that the readings by ships are unreliable, and that they are uniquely unreliable to the high side? On what basis do you make this claim?
2) You totally discount the possibility that small storms may be missed entirely.
Re #93 MarkW says
I’m not claiming that it is likely, if fact I personally think it is unlikely. Sod is claiming that the counted storms was to high due to a possibility that the ships captain may have lied or been mistaken. Sod may put this possibility at say 10% (I’ve no idea what percentage he thinks are wrong he has never given a figure) whereas I’d put it as approaching zero. The reason I would put it so low is because I have never come across any piece of research that calls into question any of the storms observed by ships using conflicting ship reports. Eg. ship A sees storm at X on date Y and ship B sees no storm at X on date Y.
In fact I say that every single uncounted (ie. missed) storm is an undercount.
The number of missed storms will far outnumber any possible overcount of observed storms.
David Smith, excellent analysis and write-up. It blows me away to think this analysis has not been done before.
The trend in where the storms are first discovered indicates how observation quality has changed over time. It’s moved more than 10 degrees away from the US. That’s about 600 n. miles.
Here’s a time series showing the discovery trend for all 1363 storms through 2006.
Here’s a map of where they were discovered and their intensity at the time. No storms have been discovered already at hurricane speed in the last 50 years, yet you can see there are many in the past. The colors indicate their intensity at discovery. More than 30 of the storms were only observed once. About half of them already at hurricane speed. So the likelihood of a past over-count is essentially zero and under-counting is virtually certain. The problem is quantifying it.
Forgot to mention. The lines in the charts are the starting points of the air recon and satellite periods.
Bob, look at the red graph again. The distribution changes at 1971 from gaussian/unimodal to flattened/bimodal. Is the difference significant?
Hurricane hysteria is clearly promoted by the popular press. See here for an example from the journal, Scientific American. In the July 2007 issue they have an illustration of a future hurricane that is at least 20 times larger than Katrina. You can see a side by side comparison of Scientific Americans future hurricane and Katrina on the same scale.
thanks for a reason finally to install R. i had seen the pictures before on the old thread.
very weak correlation for landfall storms.
looking at our example of a 5% increase, with 200 landfall storms over a century, the difference would be 10 storms. if the increase in storm activity is not fully in “extra storms”, but in stronger storms or an increased storm range, we might be looking at just 3 or 4 extra storms.
it will be pretty tough to notice this increase in numbers.
i know. i was playing along the tiny tim hypothesis and ignored all storms only discovered by satellites. even the number confirmed by ships might still be too high today. (i think the claim that even the numbers excluding tiny tims is too high, is sort of a valid one)
mostly hand waving, but see above.
perhaps this is a systematic fault in the tiny tim approach: how many ships AVOID getting close to a storm, because they have satellite data of it?
like a truck driver would use modern traffic surveillance information to avoid traffic problems in modern times?!?
i think i am being misunderstood. again.
i do not believe, that captains were inventing storms while being in calm waters. i am talking about honest faults. very local phenomenons. and a tendency to over blow things. especially when it is in your favor.
and the fact that there are no contradictions is the catch of my idea: with modern surveillance, it is much more difficult to make such an error or attempt.
i am using exactly the same reasoning, that you guys are using against the named storms count. sorry.
Good show, sod. Did you get the code running?
immediately. R is pretty impressive and so are your graphs!
They are Judith Curry’s graphs. She and Willis Essenbach did the hard work. All I contributed are the error bars and the trend statistics.
Bender,
I was wondering if someone would pick up on the difference once the satellite observations started. Good eye. 1968 is when they started entering subtropicals into the database. 22 of them up 2006. All of them are above of 20N. Only half of them eventually came within 30 miles of land. So weak, unlikely they’d have been noticed pre-satellite. If they weren’t included that linear trend line would slope even more. Here’s a map of the locations.
The big difference is in the apparent extra 600 miles swath of longitude they have to start detecting storms with air recon since 1946 and satellite 1968. The dispersion isn’t quite as noticeable in the longitude chart, but still evident were recon starts. At an average travel rate of roughly 11 knots. That adds more than two days to the effective observation time for storms east of 70W. That certainly accounts for some of the supposed difference in activity.
Note: The lines are actually at about 1946 and 1968. That’s just the way the x-axis was displayed.
i just played around a little with R. very impressive tool! if only i had a little more time… (and my knowledge wasn t that rusty..)
i just picked up this graph via a link from the 2007 blown of track discussion:
and here is some text that came with it:
the region is showing the exactly opposit effect: a decline in weak storms. (tiny tims?)
and the explanation is the one i provided far above: modern equipment might prevent an overestimate of storm power.
some lefty running the Australian BoM?
http://www.climateaudit.org/?p=2471#comment-169530
David,
Very interesting study.
I do not know the nature of ship reports but are there also changes in the nature of their reporting? For instance, if a ship gets satellite data that a storm is ahead I assume that is the ship’s report. The consequence of this may be that ships now miss storms they would otherwise have reported on. Also do most ships of a certain size or classification report storm data? Given there are so many more vessels at sea now, do some ships simply not report their own obserations of storm data because they recieved data from other sources?
But again, with so many more vessels, there is a greater possibility of recording more storms.
It seems historical comparisons with current storm data may be difficult even if the satellite, aerial and other methods aren’t considered.
Re #107 Keith those are interesting questions. A study of those issues would be worthwhile on its own merits but I doubt that it would resolve things. Regardless, I’ll do some digging.
A ship would likely steer away from hurricane-force winds but a typical Tiny Tim lasts only 40 hours and has 45 knot winds, so would a ship change course and burn extra fuel and arrive late for such marginal weather? I suspect that any avoidance decision involves sea condition, too, which will lag the wind increase. Seagoing vessels also sail the colder North Atlantic and North Pacific, where gale winds aren’t uncommon, so I’m not at all sure that reports or forecasts of gale winds scare many captains.
Regarding ship reports from the pre-recon era, my impression is that the forecasters did not rely on single ship reports but rather looked for a recognizable pattern (strong wind, low barometer reading, west wind confirming a closed circulation) before declaring a disturbance to be a tropical cyclone. Tiny Tims are often ill-defined so they would not be particularly easy to deduce from surface evidence.
In another post it’s been mentioned that non-tropical cyclones may have been accidentally counted as tropical. Possibly, but among the evidence used by forecasters was whether the strongest wind corresponded with the lowest pressure. In tropical cyclones that is almost always true while in non-tropical cyclones the strongest winds are usually a noticeable distance from the pressure center.
A dose of reality, from both land and sea: 39 kt may be the minimum standard for such storms, but one must realize that such a gale hardly qualifies as dangerous or scary to a reasonably experienced sailor.
Here is some info on the practical ocean and land implications of various wind speeds. Even this page gives descriptions that are a bit overblown with respect to winds on land.
A month ago I was sailing with friends in 47 kt (54 mph) wind. We had our sails reefed to 1/3 normal size, and just kept chugging along. Yes, it was exciting. Given time to build up on the open ocean, yes such a storm is nothing to laugh at. But smaller than that? No big deal at all.
On land… we regularly get wind here that in other places would be considered unbelievably strong. 60 mph winds with gusts much stronger than that. If you’re a building contractor, you learn to be careful: in such winds, a framed-but-not-trussed home can become a pile of toothpicks in a moment. But other than that, we consider such winds to be a nice vacuum cleaner to remove accumulated dust and dirt from the roads 😉
I personally think it unlikely that oceangoing professionals would mistake a Tiny Tim for a real storm. A Tiny Tim is to laugh about whether the ride was exciting. A real storm brings dread.
In the very early record they recorded nothing below 40 knots until 1871 when they added 30 knot. All winds are in increments of 10 until 1886 when they went to 5 knots. Assuming due diligence, they were recording the observations as close as they could with a reasonable degree of confidence. Evidently they felt confident enough to think they had reduced their error margin sufficiently to move from 10 knot to 5 knot records. They certainly wouldn’t have been recording at coarser resolution than they could justify. It’s also unlikely they would record at an unjustifiable finer resolution. Unless there is some known quantifiable error in their thought process or equipment that was unknown at the time, I think you have to take the wind records at face value.
Entirely missing storms and tracks is another story. On average storm duration appears to have increased by a couple days over the course of the record. I think it’s an observational artifact. An accurate statistical adjustment to past storm counts may not be possible. But I think a case can probably be made for an upward adjustment to the duration of past storms and thus their overall intensity. Storms are unruly but not chaotic. It’s probably too mundane for the high command to do though, and would detract from the “we’re living in unusually scary times” theme.
I received some corrected storm counts for my Easy Detect category of NATL storms — thanks to Bob Koss and David Smith. Below I present the calculations that I made using the original data and the corrected data. The calculations using the corrected data follow (in parenthesis) those using the original data. I have recopied a post from another thread here with the calculations from the original and corrected data. The corrected data does not materially change the results or conclusions previously reported.
1860-2007: Mean in counts = 5.99 (5.97); Trend in storms per century = 0.41 (0.36); p for fit to Poisson distribution = 0.052 (0.056).
1885-2007: Mean in counts = 6.21 (6.19); Trend in storms per century = -0.56 (-0.62); p for fit to Poisson distribution = 0.029 (0.160).
From Post #72 at the thread Mann and Sabbatelli:
I have done essentially what I described I intended to do in my preceding post. I used the Mann nino 3.4 index to divide the Easy Detect storms counts (using the David Smith criteria as described previously) into 2 categories depending on whether nino 3.4 was positive or negative for the months of DecJanFeb following the storm season. Under those conditions I calculated the following p for the chi square goodness of fit test for a Poisson distribution:
Nino 3.4 index was negative: p = 0.40 (0.54); Mean = 7.01 (6.91).
Nino 3.4 index was positive: p = 0.92 (0.95); Mean = 4.92 (5.00).
I think these results demonstrate that one can obtain similar results to those determined by Mann without using SST when the process is applied to Easy Detect storms. Following further on this line of thinking offers evidence that the variable SST with its trend increasing with time can easily be confounded with improvements over time in storm detection capabilities.
Since Manns use of the nino 3.4 index for the months following the storm season seemed rather an unphysical connection without some explanation, I looked at the differences that would result in the using the same months for categorizing the storms counts for the following season (as opposed to the preceding one as Mann evidently did). Those results for the goodness of fit test for a Poisson distribution were as follows:
Nino 3.4 index was negative: p = 0.01 (0.10); Mean = 6.15 (6.09).
Nino 3.4 index was positive: p = 0.90 (0.99); Mean = 6.05 (6.16).
It is interesting to note a couple relationships here. Firstly the Easy Detect storm counts for the years with a DJF nino 3.4 index that is positive yields a very good fit to a Poisson distribution whether applied to the following or preceding storm season. Secondly, when the nino 3.4 index is negative, the fit to a Poisson distribution becomes significantly less probable for the storm count years preceding the DJF indexed months and, when the negative DJF nino 3.4 indexed months precede the storm count years, one can reject the null hypothesis that the distribution fits a Poisson one at the 0.05 level.
I did one more analysis by removing the cyclical component from the Easy Detect storm count series (1860-2007) using the Willis E derived cycle for the Total count series of a sin wave with a peak-to-peak amplitude of 3.2 and a period of 58.8 years. I would caution that this procedure removed a cyclical component based on the Total count series and not the Easy Detect series and should be redone using the Easy Detect series. Removing the described cyclical component increased the p value for a chi square fit for a Poisson distribution from 0.01 to 0.59 (0.056 to 0.50).
Sod #106
Interesting graph and interpretation. It would be interesting to look at the actual data and see if there is anything special about a pressure of 970 mb. That seems to be roughly the equivalent of a category 3 hurricane.
The nearest I can come to getting actual data is here.
Unfortunately, the data look like they don’t match (see the 1970 data for instance). I’ve tried correcting for longitude, and also for storms too weak to be categorized as typhoons, but the graph you linked to seems to use a different database. Also, some years have 4 or 5 storms of exactly 970 mb minimum pressure. Depending which bin you throw those into, a single year’s bar could look a lot different.
Before I go to the trouble of graphing the data with a finer splitting out of storms (approximate cat 1,2,3 etc, as opposed to simply greater than or less than 970 mb), does anyone have a suggestion as to where those data actually came from? I didn’t see them in the original report referenced by the link in comment 106, but I may have missed them.
And sod, relative to the discussion at hand, I think the relevant data are prior to 1970, and would involve a comparison of weak storms to typhoons. I am wondering if the chart you show is only for Australian “hurricanes”, and not tropical storms (i.e. I’m not sure “minor storms” includes storms of only tropical storm strength), even though they use the word cyclone. We can’t be sure without the source data though.
John M, you will find the complete BoM historical cyclone data set here:
ftp://ftp.bom.gov.au/anon2/home/ncc/cyclone/cyclones_newformat.zip
Re # 113, John, an alternate listing is here . I haven’t looked to see how this differs from your list. Regarding the pressures given on the list you reference, those are probably best guesses, especially for the older storms.
Re #112. Nice work! I’ve been trying for a while to grasp, in physical terms, what the findings are telling us. This
I can grasp easily while this
just blows my mind, as we used to say. More coffee needed.
Re: #112
After studying some more the definition of and timing involved with a nino 3.4 index (as used by Mann and Sabbatelli in their 2007 paper) I remain perplexed why MS would use the index for the DJF months following the NATL tropical storm season. Can anyone here give any hints why they would use those months?
I found the nino 3.4 index for all months of the year for the period of 1871-2006 and divided my Easy Detect storm years into 2 categories (as described previously using DJF) by using the nino 3.4 index criteria of negative or positive. Instead of using the DJF months following the storm season, I used the months ASO in the same seasons as the storms. I then did a chi square goodness of fit test to a Poisson distribution for both categories. These calculations gave a more balanced measure of the two categories than when DJF months were used. That balance is shown in the following results:
Nino 3.4 index was negative for ASO: Number of years = 68; Mean annual storm count = 6.94; p for Poisson fit = 0.67.
Nino 3.4 index was positive for ASO: Number of years = 69; Mean annual storm count = 5.19; p for Poisson fit = 0.70.
Would it be fair to conclude that using the Easy Detect storms with compensation for el nino effects gives storm counts that fit a reasonably well behaved Poisson distribution? And furthermore that it gives evidence that the trends of increasing SST and detection capabilities can be confounded?
Re #116 Kenneth, I think that Nino 3.4 for ASO and the following DJF are correlated, due to simple persistence of trend. They’re only a few months apart. For example, the ASO 2007 Nino 3.4 is negative and I bet that DJF 2008 will also be negative.
I guess MS used those months because
1. ASO Nino conditions do affect storm activity (as you show in your subsequent calculations)
2. ASO and the following DJF are correlated due to persistence of trend
3. therefore it works, even though there’s no predictive value to the exercise.
My two cents on your final two questions are yes and yes.
Re: #117
David, I will list below the the percentage of years that had different signs in the nino 3.4 index. Also the fits to a Poisson distribution change when using ASO of the season in place of DJF that follows the season.
Below is listed the percentage of years from 1871-2006 when the nino 3.4 index has a different sign (negative/positive) for the reference periods as listed and compared. Season refers to the annual period that covers the occurrence of NATL tropical storms.
DJF months following season when compared with the ASO months in the season: 14% do not agree.
DJF months following season when compared with the annual average for the year of the season: 20% do not agree.
ASO months in the season when compared with the annual average for the year of the season: 15% do not agree.
DJF months following season when compared with the DJF months preceding the season: 39% do not agree.
sod:
I won’t speak for Steve or Dave, but my own answer is this: What becomes increasingly clear from reading climateaudit.org, realclimate.org, the IPCC reports, and a lot of other material is that no one is 100% sure of any large-scale conclusion as yet when in comes to piecing together historical data. The fair quesion is: which sets of data are the cleanest given everything we know and given the best logic that we can currently apply? And also, of course, which methodologies used to interpret the data are the most accurate and contain the least number of known errors?
Dave’s presentation calls into question a conclusion that has already been offered by the IPCC: namely, that hurricane frequency and/or intensity has increased in recent years due to global warming. That claim sets the context for this dicsussion, and that claim has the burden of proof. If problems can be found with the data used to support that claim, then he who points it out is improving the state of the art by helping to make it more ojective, even if said critic doesn’t have the whole story himself.
For all intents and purposes…YES!
Consider this analogy. You are tasked with recording the high temperature each day, but you only have a regular thermometer and you do not have the time to watch it all day long. In fact, you can only check on it randomly every several hours or so. Nonetheless, you faithfully record the highest temperature you actually observe each day.
Is there any chance that you will record a temperature that was higher than actual warmest air temperature of the day? NO
Is there any chance that you will miss the actual high temperature and record a reading that was somewhat less than the actual warmest temperature of the day? YES! In fact, it is likely almost every day. You may occasionally get lucky and look at the thermometer just as the peak temperature for the day occurs, but you will never record a temperature that was warmer than the actual high.
This is a similar scenario to the historical hurricane record. Fewer observations mean that the peaks were usually missed. There is no way to observe something greater than what actually occurred.
Perhaps Sod is asserting that there is/was a concerted effort to distort the record by systematically providing false information. Or perhaps he is asserting that there are/were problems with the observation techniques that resulted over sampling in the past and under sampling in the present. Such assertions without evidence are less than helpful. If evidence is not required, why bother with science at all
How do you expect to make a hockey stick without some hand waving on the stubborn facts? Landfalling hurricanes are stubbornly immune to these kind of “hanging chad” type “recount til you get the right answer” techniques. Guess what? No trend.
There are many puzzling aspects to the hurricane record. I don’t think that this possibility can be precluded and I’m a little surprised that it hasn’t been raised by the Webster, Curry camp. Their failure to do so suggests to me that they also are aware of the possibility and perhaps greater likelihood of past storms being downgraded e.g. what are now measured as Cat 3, 4 storms might have been recorded as mere hurricanes in the past. But that’s merely a surmise.
http://www.sailwx.info/shiptrack/shiplocations.phtml
Stupid question. Were the tiny tims reported by any ships?
Click to access ObservingHandbook1_2004_508_compliant.pdf
Some guidelines for ship reports. Nice photos of various wind conditions.
For the curious.
Part of the definition of a Tiny Tim is that storm-force winds (34 knots or higher) were not reported by a ship. The disturbance was so weak, short-lived, of small size and/or remote that no ship, or shore, encountered 34kt+ winds.
The records for the 1920s and 1930s indicate that forecasters used corroborating evidence before declaring a disturbance to be a tropical storm. They used windspeed, wind direction, wind shifts, low barometer reading, west wind (to indicate a circular system), ship latitude and indications of storm path in their assessments.
As an example here’s the September 1920 monthly summary ( link ) which illustrates the types of information forecasters pulled from ship reports in order to classify something as a tropical storm. There are plenty of reports of storm-force winds in these early years which did not result in a tropical storm being declared.
RE 126. Sorry david, I should have been more precise. NOAA have this VOS program.
The VOS ships are suppose to make repots every 3 hours.
My question is this: ( maybe you already have it in your analysis ..so color me stupid,
but use a nice color )
Were any tiny tims ( in the modern period) that were reported by satillite
Also catputred by the VOS ships?
I looked for VOS historical records but found nada..
Same with the NAvy weather observation ships from 1940-1980 No online data..
Re #127 hi steve, I didn’t get the gist. My data sources are the US NHC storm reports over the last 20 years. They make a point of looking for surface confirmation (ships, buoys, islands, mainland) and record such in their reports. The NHC has a motive (validation of their other analyses) in their search.
I don’t know how thoroughly the NHC searches the ship report data but I suspect it is thorough and, thanks to ease of access, better than it was in, say, 1935.
By the way, I saw this report tonight, wherein the NHC may upgrade an area of showers to something subtropical or tropical. I used your link to check nearby ship and buoy reports and find little to justify the bulletin, plus surface pressures are running 1015-1020mb and SST are around 25C, which normally lead one to say “forget it”. But, they may yet name another micro-storm for 2007.
I finally got around to following up on weak Australian storms (comments 113 and 106).
Thanks to Carl (#114) and David (#115) for suggesting additional data sites (the Smith Bros?).
The data Carl links to appears to be the raw data from which the BoM summary I linked to is drawn. David’s site also has storm track data in more or less raw form. Lacking the time or gumption to sort through it, I used the data I found, even though it doesn’t match the bar graph in #106 exactly. The original questions I raised were
1) does a finer split change things?
2) what about older data and the relevance to “Tiny Tim” storms.
In the case of 1, it turns out this had already been done about ten years ago. My treatment of the data seems to add nothing to the analysis of Nicholls, Landsea and Gill (NLG). The trends are not materially different if one breaks out intense, moderate, and weak storms, as shown in Figure 2 of NLG. When I updated the graphs and plotted up to 2005, the only difference was that the intense storms show a very slight decrease, instead of a very slight increase. The more recent conclusion sited in comment #106 of “no change” is probably accurate.
For question 2, unfortunately, the Australian data is simply not of high enough quality to draw any conclusions. This quote from NLG is informative.
Another problem with the BoM data is that it doesn’t really address “tropical storms”. (“Weak” storms in NLG are less than 995 Mb, which is the break for ca. cat 1 storms.) According to the definitions here, it appears that the equivalent of tropical storms in the BoM database are storms with a minimum pressure of greater than 995 Mb. For what it’s worth (and with every caveat you can imagine about the quality of the data), here is my plot for storms of greater than 995 Mb from the BoM summary data.
What we see is a noisy data set with a sharp spike in the 60s. This is presumably due to increased detection because of advances in technology. Then in the 70’s the weak storms simply stopped being tabulated. What seems to have happened is that the BoM just didn’t want to be bothered with weak storms any more.
This is opinion, course, but in my view, the relevance to “Tiny Tim” storms is the spike in the 60s due to better detection.
The quality of TC data in the Australian region continues to confound efforts to extract long term trends (see the more recent summary here). Data from the Atlantic basin, with all their warts, are statistical marvels in comparison.
Re: #129
Sorry, that last link (more recent summary) should be this.
I think that the most reliable global data, including SPAC, is from 1983 forward. This uses satellite image reanalysis and should be apples-to-apples across time. Kossin’s global study is here .
Kossin also offers this November 2007 paper for free and has an intriguing paper
in the works.
On one of these storm threads I posted a time series which contained a mistake. I inadvertently included subtropical storms (which are a pest in the database) in the “tropical storms with ACE less than 1.0” category.
The corrected time series is here . This corrected version has a more-pronounced hockey stick shape whereas the one in error had an odd mid-stick bump.
My apology for any confusion this might have created. The conclusions (large upswing in storms which were usually too weak/short-lived/small/remote to detect in earlier decades) are unchanged or, if anything, somewhat strengthened by this correction.
John M, during the 1970s the BoM shifted from being primarily a weather agency that kept detailed records and issued weather forecasts to become more focussed on public safety due to some quite serious cyclone impacts (Althea 1971 Townsville, Tracy 1974 Darwin, Joan 1975 Port Hedland) which initiated a period of intense cyclone research from which emerged the BoM cyclone classification and warning systems we have today.
Before the advent of satellites, the BoM cyclone record is reconstructed from various streams of data, such as BoM records, ship records, historical documentation, etc., and should be considered as a provisional dataset rather than a quality product, as in spite of the fact that funding applications have been made many times for projects to review and homogenize the full database, aside from one project that reviewed and homogenized much of the post early-1980s data before funding ran out, the govt. has not seen fit to further fund this process.
Further, due to cyclone definition differences, many marginal systems that would be considered as short-lived tropical storms under the JTWC definition (and ‘Tiny Tims’ here) remain as either ‘tropical lows’ or ‘east coast lows’ that are not recorded in the BoM cyclone database (and it is well recognised that some of these are still in the pre-satellite part of the dataset) – to me this is an important omission as it makes studies like David’s more difficult to verify using independent datasets.
BTW, this year the BoM has changed it’s cyclone classification from ‘gales >34 knots (10-min-av) in all 4 quadrants’ to ‘gales >34 knots (10-min-av) extending more than half way (>180 deg) around the centre’, so there is yet another change in classification researchers need to be aware of – and there are likely to be a few more named cyclones as a result – I am aware of several un-named marginal systems in recent years that would be classified as named cyclones under the new system.
One good thing that might come from this in light of all the AGW hype is that it increases the pressure for a full revision of the BoM historical cyclone data in order to apply the new standard throughout for climate studies – one hopes someone at the BoM is writing a funding application to our new govt. right now!
It’s Christmas and a good time to visit Tiny Tim. The visit compares Atlantic storms in two periods and uses Tiny Tims to help in the comparison.
One is a modern period (the last twenty years, 1988-2007). This is a period of good (and ever-improving) detection tools, like advanced satellites, improved recon devices, denser buoy networks and so forth.
The modern period also matches the 1988-2007 list of Tiny Tim storms. Tiny Tims are storms so weak, small, remote and/or short-lived that there’s no record of ships or land experiencing storm-force winds, yet they were classified as tropical storms. By historical standards these modern Tiny Tims would have been regarded as depressions or disturbed weather, not tropical storms.
The second period is 1925-1944, which immediately precedes the start of aircraft recon and has reasonably similar AMM/AMO characteristics to the modern period. I’ll call this the “pre-recon” period. This pre-recon period had only ship and landfall information for knowledge of the existence and strength of tropical cyclones. It also had what I describe as little more than educated guesses about a storm’s strength, mainly based on ship data which was often sparse and distant from the center of the stronger storms. The weathermen of the day were detectives, as are those who reconstruct storm history, and they made the best of what little data they had.
Why look at 1925-1944? The 1925-1944 period was at or near the prior peak in Atlantic storm activity, so an examination of that period could be useful in comparison to the present elevated activity.
The data I use are the “ACE” values of the individual storms. ACE is a function of storm duration and intensity. The quality of storm ACE data for the modern period is pretty good but it is of highly questionable quality for 1925-1944. That cannot be emphasized enough. It’s important to not put too fine a point on any comparison.
So, with that background, here are several questions which I try to answer with graphs:
1. What do the distributions of storm ACEs look like in the modern era and the pre-recon era? How do they compare on a normalized basis?
3. How do they compare on an absolute basis?
4. How do the storm counts compare if Tiny Tims are removed from the modern period?
Modern and pre-recon are bar charts which show the total count of storms in each ACE value (0 to 1, 1 to 2, etc). The overall visual impressions are (1) that the bulk of storms have fairly low ACE in both periods, (2) that there are more storms in the modern period than in pre-recon and (3) the pre-recon period seems to have an odd shortage of the weakest (ACE less than 1.0) storms.
On a normalized basis (see normalized plot ) the distributions look similar but with a couple of notable differences. The pre-recon “shortage” of very weak (0 to 1 ACE)storms is notable on the left side. Also, the pre-recon distribution seems to tail off more slowly (“fatter” around 15 and “thinner” around 30). Perhaps that is real or perhaps it reflects better ability to measure extreme events in the modern era or perhaps it’s a combination of effects – no way to know.
(Another item is the hint of a peak around 20 – maybe real, maybe noise, a topic for another day).
For questions #3 and #4 I’ll use one chart ( link ). For this one I put the ACE values into buckets (0 to 2.99, 3.0 to 5.99, etc). The blue line is the as-is storm count data, which shows that the difference between 1925-1944 and 1988-2007 is largely concentrated in the weakest systems. This is not a surprise for those of us who think that improvements in detection of weak systems have played a major role in the increase in storm count.
What if the Tiny Tims are removed from the modern period, to put things on more of an apples-to-apples basis? The answer to that is the dashed red line. The removal of Tiny Tims makes a remarkable reduction in the weak-storm difference.
That’s the main thing I was exploring – the impact of the Tiny Tims. But, the curve also shows a tendency for the modern period to be somewhat more active at ACEs above 5.
It’s my conjecture that some weak-to-moderate storms (5 to 10 ACE) in the remote eastern Atlantic were missed in the pre-recon era, due to the likely low density of ship traffic in that region in the Great Depression and WW2 era. There is also a possible issue with the extra-tropical classifications in the pre-recon period which I won’t get into here.
Even with those, though, there still appear to be about 15 more strong storms (ACE above 10 to 15) in the modern era than in the pre-recon era. That’s 0.75 storms a year of moderate or strong ACE. Maybe that’s real and due to higher SST, maybe that’s real and due to a partial mismatch of periods, maybe it’s another measurement artifact, who knows.
As mentioned near the start of this, historical ACE values need to be used with great caution as they were almost always based on best-guesses rather than measurements.
The interesting thing to me in this exercise is the notable impact of removing the Tiny Tims in the fourth plot, which puts things more on an apples-to-apples basis.
Has anyone plotted small storms (i.e. tiny tims) at sea over time? Is the slope of this trend equal to the slope of the trend in non-landfalling hurricanes? If so, then I think you’ve nailed it. The next question is how do tiny tims vary in autocorrelation structure and as a function of quasiperiodic forcings, ENSO etc. Next year CA sould provide separate forcasts for tiny tims vs. real storms. THAT will force the issue.
buckeyes, get ready for a beating
2 Trackbacks
[…] http://www.climateaudit.org/?p=2474 […]
[…] der Hurrikan-Daten von vor/nach Beginn der Satelliten-Ära stark eingeschränkt. Auch auf Climate Audit widmete man sich schon dieser Problematik. Die kleinen, früher wahrscheinlich unbemerkten, Stürme bekamen den […]