The big problem with much of the discussions about trends in hurricane activity is that the databases that everyone is working from are known to have significant inhomogeneities due to changes in observing practice and technology over the years. So it’s not surprising that a new re-analysis (Kossin et al, published yesterday) has been generating significant interest and controversy among the hurricane research community (see e.g. Prometheus or Chris Mooney). However, rather than this study being taken for what it is – a preliminary and useful attempt to make homogeneous a part of the data (1983 to 2005) – it is unfortunately being treated as if it was the definitive last word. We’ve often made the point that single papers are not generally the breakthroughs that are sometimes implied in press releases or commentary sites and this case is a good example of that.
Kossin et al develop an algorithm based on North Atlantic data that can be theoretically used with the coarsest data available from the earlier parts of the record and in more remote regions. While the technique works well in the North Atlantic (picking up almost all of the storms seen in the standard data), it doesn’t work as well in other basins – possibly because the characteristics of tropical cyclones are not universal, or because the coarse early remote sensing data are still not sufficient. The poorer performance in the other basins is surely a reason to anticipate that further work will be necessary to refine these estimates, and should serve as a caution to those wanting definitive conclusions.
How does this fit in with some of the previous work? Well, it confirms the large trend in the North Atlantic (seen in Emanuel, 2005), but doesn’t show significant trends in the other basins (from 1983). This isn’t directly comparable with Webster et al (2005) though, since their trends start in the 1970s, and the shortness of the new reanalysis (only 23 years) emphasizes interannual and decadal variability associated with e.g. El Nino. The Kossin et al study is therefore unlikely to shed much light on the potential global warming/hurricane intensity link.
In summary, read the papers and the comments but don’t believe the hype.
We’ll start off the discussion with a few comments we have already received on the provocative study:
Based upon the new results of Kossin et al. (GRL, 2007), it looks like the IPCC SPM just barely covered itself in its proclamations on observed hurricanes:
There is observational evidence for an increase of intense tropical cyclone activity in the North Atlantic since about 1970, correlated with increases of tropical sea surface temperatures. There are also suggestions of increased intense tropical cyclone activity in some other regions where concerns over data quality are greater. Multi-decadal variability and the quality of the tropical cyclone records prior to routine satellite observations in about 1970 complicate the detection of long-term trends in tropical cyclone activity. There is no clear trend in the annual numbers of tropical cyclones.From the results presented in Kossin et al. the “suggestions” of increases in intense tropical cyclone activity in regions other than the Atlantic basin are not really so well supported, at least for the last 23 years.
We’ve tried to make this point a bit more clearly here, despite the implications of the headline of Kossin’s press release.
-Chip
(disclosure: I have, to some degree, been funded by the fossil fuel industry since 1992)
1) The methodology is trained on the Atlantic. It has no parameters to allow for different structures or size of storms, and there is no good reason why it should work well on storms in other basins. Given the different land-sea configurations and the different role of ENSO in the different basins, and the fact that disturbances in other basins do not form from easterly waves from off of Africa, there is every reason to expect that storms in other basins have different characteristics. For instance, there is greater activity in the Pacific Northwest, and the tropopause is higher in the western Pacific, and this affects brightness temperatures at tops of clouds. If the size of storms differs then the fixed form of EOFs will not be able to capture that form. The analysis must be able to account for differences among basins in order to have confidence in variability or trends. It would be easy enough to test whether the storms in other basins had different characteristics by also performing an EOF analysis for each region. This basic test was not done. It should be.
2) The results are suggestive of these problems. In the SIO where the method gives 0, 1 or 2 storms vs up to 6 in the best track data, there is a serious bias. Similar large biases exist in the SPAC (up to 2 vs 5 in best track). Obviously the threshold is effectively different and it is a comparison of apples and oranges.
3) In addition, this version of the paper deals with PDI. The earlier version of the paper dealt with intensity of storms and that was abandoned because the results were not very good. In particular, the presumption is that the older results were the problem because operational methods have improved. But the Kossin et al results showed bigger and greatest discrepancies with those from best track in recent years: there is no convergence over time. This is harder to see with PDI, because the biggest storms are emphasized, but the question of why is there not good agreement in recent years is not answered.
3. Dr. Judith Curry:
The most vexing thing about the tropical cyclone data sets is the uncertainty that analyst subjectivity contributes to this. The Dvorak scheme for determining tropical cyclone intensity is notoriously subjective, see the recent BAMS article on this. The importance of what Jim Kossin has done is to take this subjectivity out of the analysis.
Kossin’s method matches well the historical data in the North Atlantic (NATL) and East Pacific (EPAC). The method was trained using North Atlantic data, and the East Pacific regime in terms of dynamical and thermodynamical conditions is very close to the North Atlantic conditions. However, Kossin’s method diverges substantially from the established data sets in the Western Pacific, South Pacific, and Indian Oceans. Does this mean that the established data sets or in error, or that Kossin’s method (trained in the Atlantic) does not translate well to the other ocean regions?
Owing to problems with dealing with historical satellite data, Kossin’s study was extended only back to 1983 (the period for which the satellite data are well calibrated), and it is almost certain that this data set cannot be extended back prior to 1977. By itself, this data set is too short to say anything about a trend in intensity. But it can in principle be used to assess uncertainties in the established data sets.
My own analysis of the discrepancies has focused on the Western Pacific (WPAC) data, where 40% of the global tropical cyclones form. During the period 1983-1987, Kossin’s data overlaps with aircraft reconnaissance data; I would expect the WPAC TC data during this period to be of comparable quality to that in the NATL and WPAC, but large discrepancies are seen. I would also expect the agreement to be better in more recent years with the advent of more sophisticated satellite systems, but the discrepancies are largest during the most recent period. Its would not be surprising for this method trained in the NATL not to work well in the other regions. The method does not allow for different structures or sizes of storms in different basins. The NATL cyclones form primarily from easterly waves, while those in other regions do not. The role of ENSO is different in the different basins. The tropopause height is higher in the WPAC. etc.
Many people in the tropical cyclone community have questioned the Emanuel and Webster et al. papers owing to uncertainties in the data sets. Other than anecdotal analyses, little has been done to quantitatively assess the uncertainties. Kossin’s paper is arguably the first important word on this subject, but it certainly won’t be the last word. To establish the credibility of Kossin’s data set outside the NATL, considerably more analysis is needed to understand discrepancies in individual basins and the nature of the discrepancies on a storm-by-storm basis.
Looking at the 1995-2004 temperature anomaly:
image
The anomaly is mainly on the Northen hemisphere and on the land masses. Oceans seemed to have warmed very little.
Also, I’ve read somewhere (sorry no ref) that GW may decrease the temperature differential between the poles and the equatore hence reducing Hurricane intensity.
I thank Kevin Trenberth for his remarks and for his efforts in advocating for improved resources for data acquisition and analysis. The points raised about the bigger picture are good ones and are all too easy to forget when we get involved in the specifics of one study or one debate.
I just want to comment very briefly on the concern there is no convergence over time between the UW/NCDC and best track records. This might simply demonstrate the repercussions of making estimates over many years with virtually no validation. Ground truth measurements are exceedingly rare outside the financially privileged Atlantic basin, where we routinely have low- and high-level aircraft that drop sondes and measure surface wind with active microwave sensors. So although the satellite data has gotten better in the other basins, with no validation to calibrate the estimates against and uncover the biases, there may be no reason to expect better intensity estimates with better data. The biases and the fidelity of the variability might in fact become worse with time. The Dvorak technique is a subjective method, which is to say that two forecasters could form very different estimates from the same data. This happens routinely and is well-documented. Our algorithm is an objective form of the Dvorak technique. With these things in mind, perhaps it is not that surprising that the results don’t converge. This rather depressing situation truly underscores Dr. Trenberth’s call for improving current practices.
–Jim Kossin
Re #51: Because of the large thermal mass, the ocean is expected to warm more slowly than the land. But the time period for your graph includes the 1940 to 1970 period of slight cooling between warming episodes before and after it.
I reproduced your graph here (basically the same except it has a coarser smoothing algorithm), then produced a graph covering only the recent warming period: (1975-1985) to (1995-2004), shown here. You can see there is relatively more warming in the ocean, but still less than on land.
Global warming is indeed expected to reduce the temperature differential between the poles and the equator. I understand the effect of this on hurricanes is minimal, but maybe some of the experts that we are lucky enough to have here can clarify this. However, the differential may act to reduce the intensity of temperate zone storms. Of course this is only one factor affecting such storms.
Re 45
RC folks, I know you do not like to comment on political documents, but this document by the Senate Republican Policy Committee has to be about the worst case of cherry picking of published papers I have ever read. I agree with Lynn – it merits some comment.
Jim Kossin:
This to me implies a strong need for aerial recon in other basins, particularly the NW Pacific, which has the highest frequency of the category 4 & 5 storms that primarily fuel the debate.
Gavin, re 49: Of course many if not most of the serious scientists didn’t claim Katrina et al as proof of AGW. But there was a loud clammering for such from many others. I was just fighting ridiculous with silly since Ray lambasted all of us “denialists” for the reverse in 2006.
As someone else noted, the debate is fascinating. What I find somewhat odd is that there is no explicit consideration of simply variations in solar activity as a primary source of the energy for hurricanes and hence variations in their number and their intensity.
Second, I deal with numbers a lot and I get real nervous and skeptical when someone argues that a relative infrequent set of conditions that has to date an average of 7 associated events means that when 10 occurs that this is a “substantially” higher number. There must be something that I am missing or whoever thinks this should definitely stay away from Las Vegas.
RE#44, Thanks for the link to the SST paper. In return, here is a complete discussion of the development of the Dvorak technique: http://ams.allenpress.com/archive/1520-0477/87/9/pdf/i1520-0477-87-9-1195.pdf
RE#53, 51, Kerry Emanuel wrote a great short piece on hurricane physics, available at http://www.physicstoday.org/vol-59/iss-8/p74.html (“Hurricanes: Tempests in a Greenhouse”) which might help address that issue. The temperature gradient that matters for hurricanes is the difference between the sea surface and the top of the troposphere, and if the vertical structure breaks down due to wind shear the hurricane dissipates or won’t form. Thus, a decreasing pole-equator temperature gradient shouldn’t have much effect on hurricanes. Also, just because the average pole-to-equator temperature gradient is decreasing doesn’t mean that the seasonal variation won’t still be in place, and then there’s the whole issue of the hydrologic cycle intensification – a moister atmosphere carries more latent heat and thus may generate more intense mid-latitude storms as well. Such effects on mid-latitude storms, however, seem to be more uncertain than the effects on hurricane intensity and frequency. (see https://www.realclimate.org/index.php/archives/2006/12/on-mid-latitude-storms/ )
Regarding the issue of Katrina, note that in Sept 2004 Nature ran a brief news blurb entitled “Hurricane Ivan highlights future risk for New Orleans”. There is a relevant quote: “But scientists say that another disaster was only narrowly avoided – Hurricane Ivan missed the deeply vulnerable city of New Orleans by a tiny margin. In the face of future such storms, they are calling for action to restore the area’s wetlands, to act as a barrier against flooding.” Katrina could have been another Ivan – but throw enough darts at a coastline, and sooner or later you will hit a major city.
This news article ended with the claim that the hurricane season was at the peak of a 30-40 year cycle, however – a statement that seems poorly supported at best. For more on how the media has covered this issue, see http://www.cnn.com/2005/TECH/science/09/23/hurricane.cycle/index.html . Is the rather definitive statement that “The increased activity since 1995 is due to natural fluctuations (and) cycles of hurricane activity driven by the Atlantic Ocean itself along with the atmosphere above it and not enhanced substantially by global warming” at all supportable?
Re: 56. Rod, just for the record, although I am only a physicist and not a climate scientist, I was among those who counseled that one cannot infer trends from a single storm or a single season in 2005. When the initial studies came out suggesting an increase in hurricane intensity over time, I viewed it as intriguing, not compelling. Likewise, the current study. The thing is we in no way need an increase in storm activity to provide evidence for climate change. Such an increase is not expected to be evident at this stage. We have plenty of evidence from the places where the models tell us to look–at the poles, overnight low temperatures, dates of first and last frost, etc. I would continue to counsel that anyone who tries to make climate policy on the basis of the weather is unwise.
Re: 57, bjc. As someone who deals with Poisson fluctuations on a regular basis in my day job, I share your skepticism over drawing any rigourous conclusions. However, if one sees a trend in the Poisson mean over time, that may indicate that something is going on. Increases in insolation are easy to measure, so the trends can be compared.
Statistically, I would think that the Likelihood would be the proper tool, as it would be easy to compare various models and see if the increased model complexity for a particular trend in the mean actually yields increased information. To pursue your Vegas analogy, if you see a trend in payouts of slots over time, you might conclude to pursue your studies of applied probability (i.e. gambling) elsewhere.
Re Mike’s comments in #48, “In fact, 2006 was slightly above average w/ 10 total named storms, and substantially above average for an El Nino year (for which the typical number of named storms is closer to 7). Part of the explanation is that SSTs over the main development region were anomalously warm, favoring development even in the face of an unfavorable shear environment (associated with the incipient El Nino during Fall ’06).”
Are the studies controlling for these non-GW factors that would account for decreased and increased TCs? That, it seems, might yield better results. Also the other factors Kevin Trenberth (#47) mentioned.
[Response: Well, firstly tropical Atlantic SSTs are believed to be rising, in large part, as a consequence of anthropogenic climate impacts, so they are not necessarily unrelated to global warming. But there is a substantial body of research that has been done looking at how various factors (such as El Nino, the North Atlantic Oscillation, etc.) may influence Atlantic tropical cyclone numbers. James Elsner of Florida State University has, in particular, done quite a bit of interesting work in this area. – mike]
I guess if it weren’t for GW, the field wouldn’t be developing as rapidly as it is.
I’m even very impressed with the 3 dimensional doppler our weatherman used to show us Hurricane Emily. Perhaps that technology could get at the total size, though I imagine it would be quite expensive to use for each and every TC around the world.
Another thing about Emily, which I’ve raised here before, it brought a shower of hailstones to Brownsville, TX, which our weatherman said was unusual (he had never heard of that during a hurricane). I think all these idiographic factors need to be stored in data banks, so scientists can eventually explain them.
#57, RE small numbers, I’m no climate scientist, but I do know statisticians have methods, such as Chi-square and log-linear analysis (based on odds ratios), that are quite successful on data sets with small numbers of observations.
BJC, who told you that nobody considers solar variation? Who are you relying on for this? They’re not telling you the truth. You can look this up for yourself.
“… the Sun’s output fluctuates … by about 0.1 percent between maximums (1980 and 1990) and minimums (1987 and 1997) ….. Temporary dips of up to 0.3 percent and a few days’ duration are due to the presence of large sunspots …”
Read the numbers along the left side of the picture; http://ase.tufts.edu/cosmos/pictures/CambEncySun/Sun_ency_figs_3/Fig8_17_org_comp_7899f.jpg
#61
The number of observations has to be a function of the complexity of the phenomena being studied. A rough analogy: How many rolls of a dice do you need before you can determine the nature of its bias. The more causes or sides of the dice you hypothesize the more data you need. You can do some things with small numbers but you need very strong theories or constraints. For how many El Nino’s years do we have reliable counts of hurricanes?
#62
Many thanks for the nice clean images and the reference but I was actually referring to the discussion to date on this thread. Did I miss something?
#57 Solar activity (if you mean TSI) has no clear trend from 1980s (if any, less pronounced than the difference bewteen a minimum and a maximum in a cycle, and I’m not informed of a statistical link between solar cycle variation and cyclone activity). On the other hand, I don’t know if there are correlation studies of the cyclone activity with the surface insolation trends in past decades and for each tropical bassin. Wong et al. 2006 find for example an upward trend (decrease of outgoing SW)for the mean tropical zone, but with no information on the regional signature of this trend.
Re #61. Lynn, as someone who often has to work with sparse datasets where things vary Poisson-wise about some mean, that also may be changing, I can attest that drawing rigorous and meaningful conclusions is not easy. First, if you look at the entire dataset of hurricanes, your mean is under 10 per year that hit the US. A Poisson distribution with mean less than 10 is nongaussian, so things like chi-square etc. do not scale properly. Now if you split your dataset–looking for example only at Cat. 4 and Cat 5 storms, or only years where there was no El Nino or La Nina, the situation gets much worse. Likelihood-based techniques are probably among the most powerful for assessing different models, but they get a little squirrelly with sparse datasets. This is a tough problem, and it would be difficult to draw meaningful conclusions at high confidence. However, if you establish what your model is telling you a priori, and look to see if you find those trends, you can increase confidence in your model.
No statistical technique can squeeze more info out of a dataset than is there to begin with, and since we’re dealing with a small signal in a complicated and noisy dataset, this is a really tough problem.
Mike, et al: a really late and simple question. I don’t comprehend how we believe the Atlantic SST is rising (over the long-haul) due to recent AGW. (Or is maybe “recent” not the appropriate word…) The IPCC thinks the global average temp has increased about 3/4 degree C. the past rough century, which is about 0.0075°C per year. Given the annual/periodic swings of SSTs, how is it we can conclude that SSTs are on a rising trend (O.K., maybe a decent tentative conclusion) caused by AGW (a great leap in logic??)?
[Response: See Santer et al (PNAS, 2006) for a fairly thorough investigation of precisely this question, using a standard detection and attribution approach. – mike]
Here are a few references on the various ocean/atmosphere oscillations that could influence hurricanes:
http://www.sciencemag.org/cgi/content/full/309/5731/41 (the AMO)
http://oceanworld.tamu.edu/resources/oceanography-book/oceananddrought.html (ENSO, AO/NAO, AMO, PDO)
The one that is cited most frequently and authoritatively as the cause of the increasing hurricane activity is the AMO (for example, by William Gray, Phil Klotzbach, and Roger Pielke Jr. – see Storm frenzy is not an anomaly, but a phase, Sept 13 2005 )
So, on the one hand you have the claim that Atlantic hurricane intensity is controlled by the AMO, whose mechanism is poorly understood but which has something to do with the meridional overturning circulation, which is influenced by the sinking of water off of Greenland. The argument put forth by Gray and Klotzbach, among others, is that in 1995 the AMO entered a warm phase and that explains the increased Atlantic SSTs and the increased hurricane activity.
The argument is then that the reason trends outside the Atlantic are weak is that they aren’t being influenced by the AMO; the other explanation is that the other regions are already over the threshold, so that the Atlantic basin is more sensitive to changes in SSTs and atmospheric moisture than the other regions… or the data may be poor.
At the same time, the same groups (Pielke, Landsea, etc.) claim that the hurricane trend data is too questionable to support an increasing trend in intensity, and that no links can be made to global warming – Pielke, June 2005.
What they pointedly ignore is the question of what effect global warming will have on these natural oscillations. Attention has been focused on how El Nino will change in a warming world, and climate models do reproduce El Nino. If the AMO is tied to conditions off Greenland, then it is obvious that global warming will have some effect on it. Considering that the mechanism of the ‘natural AMO’ is so poorly understood, there’s no justification for immediately blaming increases in hurricane activity on it while entirely ignoring global warming effects on sea surface temperatures (and atmospheric moisture), for which very clear mechanisms do exist.
As in other climate phenomena, there may be a multitude of factors responsible for the observed trends – but are Gray, Pielke, Klotzbach and Landsea really claiming that global warming has no effect on SSTs? Do they really believe that there will be a decreasing trend in Atlantic SSTs as the AMO enters a ‘cool phase’ sometime in the future? This is a question that Roger Pielke Jr. doesn’t seem to want to answer, despite numerous requests.
What is fairly hypocritical is to publicly claim that the data over the past few centuries is good enough to support this AMO explanation for hurricane activity, while at the same time claiming that the data is too poor to produce a statistically relevant trend over the past few decades.
Incidentally, there are a few other people who are pointing to the high hurricane potential of the 2007 season (and this is also a good description of the factors involved in the 2006 season): Why has the 2006 Hurricane Season been so calm?
I also tried to find an estimate of the net effect of hurricane activity on upper ocean heat content; there are some reports on individual hurricanes ( http://www.aoml.noaa.gov/phod/cyclone/data/pubs/Opal.pdf ) but I couldn’t find any global estimates.
Hey folks, it looks like the skeptics have their own TV show. Apparently Drudge is linking to an upcoming documentary on British(?) Channel 4: The Great Global Warming Swindle. From the description, it appears to feature the usual suspects.
[Response: Indeed. There were sensible rebuttals in the Independent and Observer. -gavin]
[Response: Actually, George Monbiot wrote an even better piece debunking this (and the rather infamous Director Martin Durkin that is behind it) in the Jan 30 Guardian. -mike]
Causality is on tenuous ground here, unlike geologic temperature records, extreme weather event statistics are available for an infinitesmally small time period. It is impossible to know whether extreme weather events (drought,floods,hurricanes) are more or less prevalent in previous times of higher CO2 and temperatures. If ocean temperatures were deterministically generating them, they should be more easily forecast year by year using ocean thermometers than they are. It is entirely plausible that high enough global average temperatures may suppress more extreme events than are caused.
One quantifiable effect, which is spreading of severe weather events to places where they have not previously been experienced regularly, may not necessarily translate into continuing increase in probability of that occuring. It may be just the current temperature signature, which may well change with continuing increases in average temperature.
I am not saying that this research is invalid in general, but no weight should be given to this aspect with regards to policy. If climate change’s quantifiable effects other than hurricanes are not enough for a policy shift in themselves, they shouldn’t be with the consideration of hurricanes.
Marco, causality is actually on fairly firmer ground on this issue then you are assuming. A very interesting paper on causality that addresses the issue of the causes of the increases in Atlantic hurricanes is available at Evidence in support of the climate change-Atlantic hurricane hypothesis, James B. Elsner, GRL Aug 2006 (pdf). (thanks to mike for the pointer)
Here, the author draws causality relationships between global mean near-surface air temperatures and Atlantic sea surface temperatures and hurricane power dissipation indexes using statistical causality tests. This is one of the very few papers that directly addresses the issue of whether the observed trends in hurricane intensity can be blamed on an Atlantic Multidecadal Oscillation, or whether they are due to global warming. The causality path is different for each hypothesis, to quote:
“Figure 1 illustrates the two competing hypotheses concerning Atlantic hurricanes. The climate change hypothesis asserts that changes in radiative forcing resulting from increased greenhouse gas build up in the atmosphere increases GT and causes Atlantic SST to rise at least during the hurricane season months of August through October. On the other hand, the AMO hypothesis asserts that natural changes in the deep water circulation of the Atlantic Ocean drive hurricane season SST resulting in changes to both hurricane activity and GT. Under both hypotheses local SST plays a direct role in helping to power hurricanes by providing moist enthalpy and instability. Thus the point of departure for the two competing hypotheses is the causal connection between GT and Atlantic SST. The climate change hypothesis suggests the causality goes from GT to Atlantic SST whereas the AMO hypothesis implies it is the other way around.”
Based on the results of the causality tests, the author concludes that it is global near-surface air temperature that influences sea surface temperature, and not the other way around – which supports the global warming-induced increase in hurricane intensity.
Being a parochial West Australian I had to read the original article by Kossin et al to find out what it had to say about the S. Indian Ocean – basically that the dataset is suspect and no trend is particularly observable. Fair enough.
But a couple of anecdotal observations seem in order. This summer cyclone activity has been low and the season has been late. But this seems not to be due to El Nino but more localised conditions, specifically the SST have been unusually cold on the west coast of Oz and high off the east coast of Africa.
About two weeks ago, we had simultaneous formation of two cyclones(Gamede and Humba) in the Indian Ocean and currently we have George tracking across the Kimberley and a yet-to-be-named cyclone forming near Christmas Island. These double cyclone systems interact with one usually ending up dominating(Gamede). But they presumably also work against each other in terms of kinetic energy. So this raises the question of how double systems like these affect conclusions about cyclone intensity and cyclone numbers. Kossin et al limit their observations to cyclones above a certain category. So, are two simultaneous Category 2 cyclones( one of which goes to Cat 3 and the other dies) more or less important than one category 4(say) in terms of describing the removal of heat from the ocean or in terms of reaching conclusions about the relationship between cyclones and GW?
[[Second, I deal with numbers a lot and I get real nervous and skeptical when someone argues that a relative infrequent set of conditions that has to date an average of 7 associated events means that when 10 occurs that this is a “substantially” higher number. ]]
Depends on the standard deviation, doesn’t it? You seem to be assuming it’s large compared to the mean.
Re #60, thanks, Mike. I guess the non-global warming factors I was thinking of are the things that reduce hurricanes even when SST are rising, such as wind factors (I don’t actually know what I’m talking about here). But then it occurred to me that these, too, may be impacted by GW.
I don’t have time to read those articles now, but I’ve saved the links. I’m thinking some might have to do with how GW is impacting el ninos, which involves warming waters. I know RC has addressed this before and it’s complicated.
All I know is many decades ago I learned el ninos occurred roughly every 7 years. Then in the early or mid 90s there were more frequent and more intense el ninos, and also the weatherman blamed all the weird weather on el ninos.
It seemed to me that GW was likely impacting the el ninos (warm waters were also found responsible for die out of plankton & the fish that thrive on it). It seemed sort of like blaming a hired gun for killing a person and letting the big culprit off the hook. Prior to those weather assurances by the weathermen, people were beginning to think GW may be responsible for the weird weather, but the U.S. weathermen IDed el nino (like Perry Mason IDing the actual killer, only it was probably a hired killer and not the guy (GW) behind the crime).
So once the (supposedly natural) el nino cause was established, Americans forgot about GW and went back to business as usual….and now our GHGs are expected to rise 20% by 2020, over our 2000 emissions. And the rest will be history.
Re 73: Barton, higher moments–especially the skew can also be important when you are dealing with Poisson fluctuations, as this can introduce systematic errors into the analysis. Basically in statistics, we can handle things vary normally and maybe lognormally, but anything more complicated than that gives us trouble. I’ve recently played around with re-deriving the Chi-square distribution assuming Poisson distributed fluctuations vs. normally distributed. How important these fluctuations are depends on the Poisson mean, the size of the effect you are looking for and the level of confidence you require. For a Poisson distribution with small mean, the standard deviation is always large (square root of the mean). Note that if you want to play around with this stuff, Excel gives you most of what you need.
#73
My basic point is that with a small number of datapoints SDs tend to be driven by the next piece of data. Since we don’t “know” what that might be, making judgements as if we did are, as I said originally, troubling. Ray Ladbury’s more thorough explanation make sense to me.
Ike Solem:
Note there is an out. The data in the Atlantic are so much better, it would not be a stretch to claim the data in the Atlantic is good enough to detect a trend, but the global data is not. I don’t know that any of Gray, Pielke, Klotzbach, or Landsea are taking it, however. (Related – much of the media has been claiming ‘Global warming is honing in on the Atlantic – but this almost the opposite of the truth, which is that hurricane researchers, due to what might be called historical accidents, are honed in on the Atlantic. ) Although I am sure you have already read it, I think your set of AMO-related links is not complete without Atlantic Hurricane Trends Linked to Climate Change (Emanuel, Mann, 2006) (supplemental material here ) which may be the best available explanation for the AMO. In that paper, Kerry and Mike show the combination of global SSTs and regional aerosol forcing are a good predictor of Atlantic tropical SSTs and thereby Atlantic TC counts, and that there is very little Atlantic tropical SST variation left for a multi-decadal cycle to explain. This implies (though the paper does not directly state) that the AMO is just a case of mistaken pattern recognition (an idea Urs Neu has been promoting in comments here for a very long time).
The AMO seems related to the same phenomena that result in claims like ‘it was awfully warm back in the 1930s and 1940s’ . It seems the aerosol induced cooling between the 1950s to the 1970s has served the ‘its not happening’ camp so well, despite having been explained some time ago.
[Response: Thanks for mentioning this. Actually, we don’t go as far as arguing that the AMO itself is an artifact. I for one have published a number of papers very much arguing for its existence (e.g. Delworth and Mann, 2000 and Knight et all, 2005, you can find reprints of both here), and must confess that I actually coined the term “Atlantic Multidecadal Oscillation” in an interview with Richard Kerr of Science in 2000. I’m happy to let Kerr (2000) take the credit nonetheless :). Kerry Emanuel and I do point out in our piece that past work indicates that the AMO is unlikely to project substantially onto tropical Atlantic SST, and that the attribution of recent trends in tropical Atlantic SST to the AMO is indeed probably an artifact of the impact on the tropical Atlantic of competing anthropogenic forcings. -mike]
Re 47:
“The physical expectation is that TCs are apt to become
more intense and bigger with warming. But because one big storm takes much more heat out of the ocean than two smaller storms, the expectation is that numbers may decline.”
Dr. Trenberth,
While I understand the logic behind this idea, I think it is worth noting that, at least anecdotally, it does not seem to agree with what the history is telling us. For example, 2005 Atlantic had exceptionally high SSTs, the largest concentration ever of very strong storms, and also, by far, the largest number of named storms.
Further, looking back at hurricane records, there does not appear to be an inverse correlation between strong storms and number of storms, as one might expect if the assumption you refer to is correct. There may be a variety of explanations to account for this, and my observation is hardly rigorous, but it still strikes me as interesting. I wonder if the fact the higher SSTs will lead to somewhat longer seasons (and 2005 certainly had that), and larger sea-surface area where TCs might form, would serve to counteract the effect you describe, and still lead, net-net, to an increase in the number of tropical cyclones with higher SSTs.
Thanks.
Dan
RE: #35 #45 #54
Hi chaps,
This paper is smoothly seductive and well written in the sense that it has the look and feel of an authoritative document. Strip it to the bare bones and it’s the usual controversy-debate-uncertainty biaised tripe.
Also spotted is Lindzen, Ross McKitrick and the “Cooler Heads Coalition” being referenced. Well there’s a red rag waving right in front of you.
Maybe a word from the gurus? ‘Course it’s a full time job debunking deniers but silence implies consent.
Keep up the good work
Mike
[[My basic point is that with a small number of datapoints SDs tend to be driven by the next piece of data. Since we don’t “know” what that might be, making judgements as if we did are, as I said originally, troubling. Ray Ladbury’s more thorough explanation make sense to me. ]]
I’m sure it does. Is there a known time series for this number, so we can calculate the mean and standard deviation? I mean, Poisson corrections and so on aside, if we know what the standard deviation is, then the fact that this kind of problem tends to lead to large standard deviations is rather beside the point, isn’t it?
Thank you, Mike, for correcting my misunderstanding of your paper.
[Response: Sure thing L, actually you had the gist of it right. -mike]
Well, if AGW does cause an increase in North Atlantic SST then this might be another positive feedback. This is a link to a Science Now story about research into the winds around Greenland. http://www.sciencedaily.com/releases/2007/03/070305085344.htm It mentions offhand about a 2003 melting event on the icecap caused by the passage of a tropical storm. A warmer ocean will allow storms to retain their tropical characteristics as they venture into the north Atlantic and dump copious amounts of warm rain onto Greenland more frequently. This effect of warmer SST will occur whether or not tropical cyclones become more frequent or more intense. Not to mention the increased chance that a tropical storm will dump 20 or 30 inches of rain onto northern Europe. I recall that Portugal saw a tropical storm last year and that a couple of other storms just wouldn’t call it quits as they wandered northward due to very warm SST.
Re 80: Barton–by all means if you know the standard deviation, but how do we know the standard deviation unless we estimate it from data. The thing about the Poisson is that there is only one parameter to the distribution–the mean, so if you determint that, you know all the moments. In this case though, the mean is assumed to be a function of time, and there are other variables that affect it as well (e.g. El Nino), and we don’t even know if we have a reliable dataset (e.g. has our ability to determine hurricane strength improved with time). So this is not an easy problem. Finally, there’s the question of the confidence level for the results–you don’t want to be fooled by statistical fluctuations. I am by no means an expert here. I’m just going by how I’d look at the problem based on how I’ve solved others.
#83
My reaction was the same. The absence of data is the problem. Making statements about the relative distinctiveness of the number of events with no knowledge of its mean and distribution is very odd. I am sure the person just mispoke. EOD
Re: #71 On the other hand, the AMO hypothesis asserts that natural changes in the deep water circulation of the Atlantic Ocean drive hurricane season SST resulting in changes to both hurricane activity and GT. Under both hypotheses local SST plays a direct role in helping to power hurricanes by providing moist enthalpy and instability. Thus the point of departure for the two competing hypotheses is the causal connection between GT and Atlantic SST. The climate change hypothesis suggests the causality goes from GT to Atlantic SST whereas the AMO hypothesis implies it is the other way around.”
This analysis is useful in comparing two theories of “the past”. Whether this has anything to do with future higher GT’s, we are assuming that the relationships(between GT and SST’s and between SST’s and Hurricanes/cyclones) continues and that it is linear when we “extrapolate” outside the range of historical data. I am not claiming to have an alternate hypothesis that debunks these two, but extrapolating for one set of extreme events that we suspect may increase, and ignoring other kinds of extreme events that may reduce in risk, is cherry-picking, and presumptious. This is different to the case for predicting GT’s in the first place. We have geologic records of temp and CO2 – we don’t have geologic records of temp and hurricanes full stop.
The very late moderate El-Nino of 2006-07, is turning to La-Nina extremely fast! Implications for hurricanes are obvious. But the changeover speed is a bit fast? Or is this normal?
[Response: Not that unusual (see e.g. here). There is a tendancy for the tropical Pacific to ‘rebound’ out of an El Nino into a La Nina, as the underlying dynamics are fundamentallly oscillatory. -mike]
[[My reaction was the same. The absence of data is the problem. Making statements about the relative distinctiveness of the number of events with no knowledge of its mean and distribution is very odd. I am sure the person just mispoke.]]
I fear you may be making a mistake when you refer to the small number. The number referred to is the number of data points; I interpret your original post as implying that you believe the value of the number is the small number in question — i.e., because 7 or 10 is a small number, the distinction is obscure. But that 7 or 10 is only one data point. By changing the units we can make the value as small or as large as we like; the value of the point means nothing as to the ability to interpret a number of such points. How long has the number of named storms, or the number of large storms, been kept track of? I would be surprised if it were less than 30 years, and that is enough to give 95% confidence, other things being equal. If the typical numbers reported are as, let us say, 2,7,9,4,3,8 and so on, then a value of 10 is not very surprising. But if they are more like 7,6,7,8,8,7, then a value of 10 is surprising indeed. The person who mentioned the large value is someone who works professionally with the analysis of such storms, is he not? If so, I would be inclined to take his word for it when he says 10 storms rather than 7 indicates a meaningful deviation.
[Response: TC numbers are essentially Poisson distributed, so the standard deviations are approximately the square root of the mean rates. For El Nino years, that gives 7+/2.6 , i.e. 10 storms is a bit more than one standard deviation above the mean rate for El Nino years. Very roughly (keeping in mind that the Poisson distribution is heavy tailed), 10 or more counts should happen about 1 in 6 times randomly in an El Nino year. Clearly not outside the range of possibility, but relatively rare. The main point is simply that any claims that the the 2006 Atlantic TC season was in any way ‘inactive’ by historical standards, is simply silly. -mike]
[Response: I just now performed a quick calculation using the total number of named Atlantic TCs from 1870-2006 (the 1870-2004 data are available here, I updated w/ the latest two years numbers, 28 and 10). The mean rate for all years is 9 named storms, the mean rate for El Nino years (that is, years where an El Nino event was building over the course of the storm season, based on DJF Nino3.4 index > +1) is 6.3. Fitting a Poisson distribution to the series of El Nino year TC totals gives an 11% probability of equally or exceeding 10 storms. So my estimate above was too conservative. 10 or more storms in an El Nino year really is fairly unusual. -mike]
Mike, thanks for the link. The data look pretty Poisson, even though there’s a hint of bimodality. A maximum likelihood fit yields a mean annual number of TC of 9, with 90% confidence that the mean is actually between 8.5 and 9.5. The data for El Nino and La Nina years are much sparser, but I basically get a Max. Likelihood mean of about 7.5 with a 90% confidence interval of 5.75 to 9.75. So we can’t really say with high confidence that 2006 was way high, but it would certainly be flat wrong to say that it was low by historical standards for an El Nino year. Has anyone been looking at the possibility of reconstructing which years prior to the ’80s were El Nino, La Nina and “normal”? That might help us out in reconstructing any trends.
RE #3: There are several folks working on paleotempestology as it’s called. The ones I’m most familiar with is Kam-Biu Liu at LSU and Andy Reese at USM. The focus is using pollen records from lake sediment cores. I had a few courses with Professor Liu at LSU…intresting stuff.
A list of Liu’s work is at:
http://www.ga.lsu.edu/liu.htm
Cheers!
Royce
Ray Ladbury:
Klaus Wolter’s MEI El Nino page (scroll to bottom) has links to several online El Nino indicies.
[[I updated w/ the latest two years numbers, 28 and 10]]
I assume you mean 8 and 10? 28 would indeed be off the scale. Or did I miss what you were saying?
In answer to the Republican’s screed:
“Global Warming: The Settled Versus the Unsettled Science.”
http://rpc.senate.gov_files022707GlobalWarmingPG.pdf
Seems to me the debate about AGHG global warming and increasing TC frequency/intensity/duration boils down to the fact that as sea surface temperatures, as well as deeper water temperatures rise, the wallop of any TC over warmer seas without mitigating circumstances like wind sheer and dry air off land masses entrained in the cyclone will likely be much more devastating.
Do total numbers matter to most people if a large percentage of storms veer away from coastlines? Do any of the parameters one way or the other count for much if just one storm exacerbated by warmer oceans results in catastrophic effects?
Of course having the science nailed down is essential. But focusing the debate within societies on overall frequency or intensity or degrees of this or that in this or that geographical basin as a way to evaluate the relationship between climate change and tropical storms is confusing the issue. How many Hurricane Katrinas does it take to make this point?
No one can say global warming caused this hurricane. But the intensity of Katrina a day before landfall was category five. This was due to warm water in the GOM loop current. As warmer waters are entrained in currents throughout the gulf isn’t it just common sense to expect that as warm deep waters inevitably occur closer to land the intensity of violent storms way well be sustained as they make landfall?
The debate in the US is about whether or not capping emissions is necessary. Warmer water means more evaporation and a bigger heat engine offshore. I’d say without attempts to reduce emissions to mitigate impacts we’re playing Russian Roulette with violent cyclones whether there are other weather/climate patterns like El Niño or the NAO influencing occurrence or not.
Just monitoring the science will be insufficient to avoid economic consequences over-riding the consequences of doing nothing as our elected representative seek to avoid economic consequences to big oil, big coal, transportation and utility campaign contributors. Perhaps the Congress should meet in New Orleans or Galveston or Biloxi, Miss, from now on?
Off-topic, but I need the data to prepare for a ‘debate’ here:
During the Paleocene-Eocene Thermal Maximim (PETM) carbon dioxide was at about 2000 ppm and the Arctic temperatures rose about 40 K compared to today.
I need an estimate for the global surface temperature.
Thank you.
Re 91: Barton, He’s talking about 2005–a record year that was indeed off the scale–and 2006, an El Nino year. Hence the difference. The data are interesting.
If it hadn’t been for a large amount of dust and dry air from Africa during the middle of last summer, there would have likely been a few additional tropical storms. Some of the tropical waves that left the African coast were quite “promising”, but the dry dusty air did them in. With the increasing SSTs and forecast of more drought in Africa, I would imagine that we would likely see a decrease in the number of tropical storms (particularly in the eastern Atlantic because of dry dusty air), but the ones that develop will more likely be in the cat 4/5 range since there will not be as many storms to mix the waters. Even with dust reflecting some radiation back into space, the dust will also act as a blanket still allowing SST to increase. With warmer SSTs, I would imagine there will be more storms like Wilma that will increase their intensity from TS to cat 4/5 level in a very short period of time. This has to be causing some nightmares for the forecasters at the NHC and emergency preparedness officials. For example, the 5 pm advisory says the storm has 65 mph winds and a hurricane watch is issued for a heavily populated portion of the coast. Residents go to sleep thinking the next day they will have a cat 1 storm. The next morning, the residents find they have a cat 3 heading to cat 4 that will be making landfall. Andrew did a increase just before landfall as well as Hugo. Hugo picked up energy from the gulf stream just before landfall that increased its winds by about 20 mph. I lived on the east side of Columbia, SC (100 miles inland) and we took a nice hit from Hugo (lost power for 3 days). I now live 150 miles from the coast and I am beginning to think it is too close. I hope someone can shoot some holes in the above line of thought.
Re #92, the Russian Roulette analogy is good. That’s a one in six chance something bad will happen. I think that’s about the right standard for false-negative avoiders (people trying to avoid harms — which should be nearly everyone). We don’t play Russian Roulette; we mitigate at an 18% or greater certainty GW is happening, or GW is enhancing hurricanes (given the good theories and logic, which have been around 100 years, that we could expect GW with increased GHGs). Which means we should have started seriously mitigating at least by 1990.
Another point is that the extra amount of hurricane intensity or precipitation caused by GW, even if small, could be the “umph” that could blow the house down or breach the levee (the straw that breaks the camel’s back). The last “umph” will likely be much more devastating than the first “umph” of equal amount (can someone tell me what the unit of TC intensity is called?), and the last few inches of precip, than the first few inches. So the impact of GW on TCs or precip does not have to be very much to cause a tremendously greater amount of damage and harm.
Re #96, I know your feeling, Jim. We live in about 50 miles from the Gulf, in the Rio Grande Valley. As we moved down here the signs were pointing in the opposite direction, “Hurricane Evacuation Route.” They still talk about Hurricane Beulah (1967), and getting flooded 3 feet in their homes, and how it spawned a lot of tornados, as well. I’d only been thinking about getting on GreenMountain wind energy and growing tropical plants, & didn’t even consider the hurricanes. That’s the problem, mostly they don’t come, so we forget them rather quickly.
Re #3 and #89:
Thanks for the pointer Royce. Without a doubt Paleotempestology is the most fascinating new word I’ve learnt in a long time!
These pages provide good descriptions of the discipline.
* The Wikipedia entry on Paleotempestology
* Paleotempestology Resource Center
In summary:
1) Proxies with promise that are mentioned include;
– physical analysis of sedimentary deposits in coastal lagoons and marshes.
– isotope and physical analysis of rings in corals, trees and cave .
– isotope and physical analysis of layers in lakes especially those with varves (annual layers) and in arid areas.
2) It seems to be a relatively young discipline with lots to be done and breakthroughs to be made.
Sedimentary deposits are probably limited by the fact that the sea level has only been at roughly it’s current level for 6 or so thousand years, before which it was rising and therefore erasing deposits as it went. But some of the others offer amazing possibilities. For example check out the paper of Frapier (2005) describing the testing of a method for identifying tropical cyclone events using isotope ratios in the cores of stalagmites. They used a modern 23 year period with a known cyclone history and were able to discern 8 out of 10 cyclones but found only one false positive out of 1200 sample points – impressive.
A very interesting (and a wonderful piece of detective work) paper by Miller et al (2006).
It shows that late 20th century Atlantic hurricane frequency is not at all unusual.
I suggest that this paper finally ends the speculation that recent hurricane activity is driven by/linked to AGW.
What a pity that this seminal paper will not be included in the latest IPCC assessment.
Miller, D.L., C.I. Mora, H.D. Grissino-Mayer, C.J. Mock, M.E. Uhle, and Z. Sharp, 2006. Tree-ring isotope records of tropical cyclone activity. Proceedings of the National Academy of Sciences, 103, 14,294-14,297.
[Off topic]
I was wondering if you guys could comment at all on the study that just came out in EOS by Lau and Kim regarding a possible explanation for the non-extreme 2006 hurricane season. Their main hypothesis was that an increase in Saharan dust blown over the Atlantic decreased solar radiation, lowering SSTs. Interesting.
[Response: Interesting, but unconvincing. To anyone who was observing the TCs as they were organizing (or attempting to organize), it was fairly obvious that the development was being hindered by strong westerly winds aloft — this is reasonably attributable to the substantial incipient El Nino conditions that were in place last fall. A good test will be what happens this next season. We still have historically high SSTs, and it looks like a substantial La Nina will have taken hold, which should lead to decreased wind shear. This should provide a similarly favorable environment for TCs to that which was in place during the 2005 season, and a forecast of more than 15 total named storms would seem reasonable at this point. The authors of the Eos article should their own forecast. Let the best hypothesis prevail! -mike]
[Response: One thing that bothers me about the study is that no mention is made of what the covariability is between Nino3 SST and the Saharan dust index used. It it possible that ENSO impacts are masquerading in their study as dust impacts simply because of the covariability between the two. I would want to see that addressed before taking the results more seriously. The article is provocative and interesting, but at this point should be considered somewhat speculative. Hopefully, it will spur others to look at this connection more carefully. -mike]