12 November 2013

Are Typhoon Disasters Getting More Common?

This morning I have been engaged in a Twitter debate with Jeff Sachs, of Columbia Earth Institute, motivated by his tweet as follows.
The reference is to a paper by Elsner et al. (2008) in Nature which shows an increase in the strongest tropical cyclones in some basins over the sub-climate time period of 1981-2005. Unfortunately for Sachs that paper does not show trends significant at the >90% level for the strongest cyclones in the western North Pacific basin (the world's most active and where Haiyan occurred). The lesson here is that if you are going to pick cherries, make sure that the fruit is not a lemon.

Fortunately, there is a more relevant study (Weinkle et al. 2012, here in PDF) which looks specifically at landfalls in the western North Pacific basin. Landfalls are of course what cause disasters. The data from that paper for the major landfalling tropical cyclones (i.e., Category 3+) is shown at the top of this post. The trend line is added by Excel, and shows a decline. However, the western North Pacific basin has been shown to exhibit very large variability, so I wouldn't put much weight into any claims of trends up or down (but don't believe me, check IPCC). That said, recent research has looked at the recent decline in activity in that basin.

Given this data, substantial research on it and a strong IPCC consensus does anyone really want to debate that typhoon disasters have become more common? If so, my comments are open to you.


  1. The real question:

    Are idiotic rants on Twitter becoming more common?

    I'd say yes.

  2. Roger, you say

    Unfortunately for Sachs that paper does not show trends significant at the >90% level for the strongest cyclones in the western North Pacific basin

    I would argue that what the Sachs paper actually shows is that one cannot out rule out no trend at the 90% level (if you consider Figure 2 at least). This is slightly different to what I think you're implying with the above statement. Even in the WNP basin, the Elsner et al. paper still suggests that a positive trend is more likely than a negative trend (or no trend). Also, if you consider Table 1 from Elsner et al., which goes beyond the 0.85 quantile, it seems to suggest that the trend is both positive and, in some cases, statistically significant.

  3. -2-wottsupwiththatblog

    Thanks ... The trends reported by Elsner et al. (2008) are not statistically (90%) different than a zero trend.

    That said, have a look at the graph at the top of this post. Start your trend analysis in 1973, 1983 or 200, and I'm pretty sure you'll find an upwards trend.

    In noisy systems finding trends is easy. Thanks!

  4. I think this is a lost cause like the Coriolis effect in bathtubs (http://en.wikipedia.org/wiki/Coriolis_effect) that got in to the popular science education and impossible to eradicate. Unless, hurricanes and typhoons stop forming entirely, the few that pops up will always be regarded as a proof of more extreme events.

  5. -2-wottsupwiththatblog

    So, do you still feel Sachs' tweet is accurate? I've read your blog and when it comes to skeptics, you have a pretty high bar. Well, can you apply it equally?

  6. Elsner linked it to sunspot numbers anyway.

    Elsner, J.B. and Jagger, T.H. 2008. United States and Caribbean tropical cyclone activity related to the solar cycle. Geophysical Research Letters 35: 10.1029/2008GL034431.

  7. Here is an exchange of letters in the FT on extremes, first I wrote on 10/21:

    October 21, 2013 12:27 am
    Claims at odds with IPCC report
    From Prof Roger Pielke Jr.

    Sir, Julian Hunt and Johnny Chan write, in their article “We must face up to the rising threat from coastal storms” (October 17): “We
    know that the frequency of heavy rain events, as well as rainfall intensity (amount of rain per unit of time), has already increased. This
    has led to severe flooding becoming more frequent.”

    This is factually incorrect. The Intergovernmental Panel on Climate Change concluded in its assessment report last month: “There
    continues to be a lack of evidence and thus low confidence regarding the sign of trend in the magnitude and/or frequency of floods on a
    global scale”.

    Lord Hunt and Professor Chan also cite a 2010 WMO report to claim: “The World Meteorological Organisation’s committee on the
    relationship between climate change and tropical cyclones expects global warming to cause the average intensity of tropical cyclones to
    increase up to 11 per cent by 2100.”

    This claim is also at odds with that of the recent IPCC report, which says: “ . . . it is likely that the global frequency of occurrence of
    tropical cyclones will either decrease or remain essentially unchanged, concurrent with a likely increase in both global mean tropical
    cyclone maximum wind speed and precipitation rates. The future influence of climate change on tropical cyclones is likely to vary by
    region, but the specific characteristics of the changes are not yet well quantified and there is low confidence in region-specific
    projections of frequency and intensity.”

    While, in principle, there is no problem with scientists holding outlier views, in this case the authors have actually made errors in their
    representation of the current state of the science.

    Roger Pielke Jr, Professor and Director, Center for Science and Technology Policy Research, University of Colorado,
    Boulder, CO, US

    They respond today with this:

    November 12, 2013 9:27 pm We have to keep examining data
    From Lord Hunt and Prof Johnny Chan.

    Sir, As meteorologists with some experience we were surprised by the accusation (Letters, October 21) by Professor Roger Pielke Jr (a
    leading policy academic whose thesis about vulnerability we agree with) that our FT article “We must face up to the rising threat from
    coastal storms” (October 17) was not based on good science, and contradicted the recent IPCC Working Group 1 Report.

    Paragraph B1 of the report concludes that extreme weather events are likely to have become frequent, severe and last longer, with the
    implication in other parts of the report (not stated very clearly) that these trends will continue, unless or until human influences on the
    global climate are mitigated. The published data on extreme events mentioned in our article (some of which, although from highly
    reputable institutions, had not been submitted or read by IPCC) provide details of where and how these events occur.

    The IPCC Working Group 1 Report has done a fine job, but scientists need to keep looking at data and providing explanations about
    trends in severe climatic and weather events, whether or not they coincide with the current IPCC consensus.

    Julian Hunt, Former Director, British Meteorological Office; Johnny Chan, Chair, Tropical Cyclone Panel, World
    Meteorological Organisation

  8. Roger, the AGW movement is far past the point of needing actual evidence to make claims about claimed impacts of CO2. Any AGW revelation, if it supports the general idea of climate disaster, is credible to the believers. Likewise all rebuttals contrary to the climate disaster dogma, no matter how well evidenced, are to be discarded by the good believer.

  9. This is obviously a difficult issue to speak to with certainty, primarily because disparities between the various storm track databases.

    The very latest literature does support the work by Elsner and what Sachs said, that globally the strongest TCs are getting stronger.

    From Kossin et al. (2013, in press in J. Clim.), my annotation in square brackets is from the paper's body text:

    "Our analyses using a new homogenized record of tropical cyclone intensity suggest that the stronger tropical cyclones, globally, have become more intense at a rate of about +1 m s− 1 decade−1during the 28 -yr period 1982– 2009, but the statistical significance of this trend is marginal [p-value = -0.1].
    Dramatic changes in the frequency distribution of Lifetime Maximum Intensity (LMI) have occurred in the North Atlantic, while smaller changes are evident in the South Pacific and South Indian Oceans, and the stronger hurricanes in all of these regions have become more intense. There are no significant changes noted in the eastern North Pacific, and negative changes are found in the western North Pacific, that is, the strongest hurricanes have become weaker. There are insufficient data to determine trends in the distribution of LMI in North Indian Ocean hurricanes."

    So what Sachs said is correct, but the data do not seem to support his assertion concerning TCs over the western tropical Pacific, at least for the period 1982-2009 using these data.

    Before accusations are made about cherry picking by certain "honest" brokers, the authors in Koppin et al. provide very sound reasons for going back to 1982 and not earlier.

  10. The fact remains that Haiyan is the strongest TC to make landfall on the planet to date.

    Also, (unofficial) data presented by Dr. Jeff Masters) indicates that 50% of the top 10 strongest TCs to make landfall have occurred since 2000, 70% since 1990.

  11. I have made a plot where I apply quantile regression to the JTWC WNP data. (I.e i am applying Elsner's method to the data used by Weinkle et al.). Should be useful in discussing Weinkle et al. vs. Elsner et al.


    Weinkle: http://sciencepolicy.colorado.edu/admin/publication_files/2012.04.pdf
    Elsner: http://myweb.fsu.edu/jelsner/PDF/Research/ElsnerKossinJagger2008.pdf

  12. -9-Albatross

    You write: "the data do not seem to support [Sachs] assertion concerning TCs over the western tropical Pacific, at least for the period 1982-2009 using these data"

    Ya think? Kossin et al.: "negative changes are found in the western North Pacific, that is, the strongest hurricanes have become weaker"

    There is always the hope of other data, other time periods ;-)

    Also from Kossin et al. 2013: "when interpreting the global and regional changes in tropical cyclone intensity shown in the present work, it is clear that framing the changes only in terms of linear trends forced by increasing well-mixed greenhouse gasses is most likely not adequate to provide a complete picture of the potential anthropogenic contributions to the observed changes.

    At present, detection and attribution of changes and trends in tropical cyclone activity remains a significant challenge ..."

    The IPCC doesn't always get things right, but on TCs the AR5 accurately reflects the current state of understandings. Thanks!

  13. From Geoff via email, a comment that made me laugh out loud:

    "I've having trouble to reconcile the Johnny Chan of the recent paper showing a decline in Pacific tropical cyclones, and the co-author of the WMO report showing no trends, and the Johnny Chan who is taking you to task for point out there is no discernible trend. Presumably he's reading his own papers."

  14. Roger,

    Oh dear, you need to please carefully read my post again and you also need to carefully read the tweet made by Sachs that upset you so...he said "disasters like Haiyan" (i.e., globally and the strongest TCs).

    Kossin et al agree with Elsner et al. that globally the strongest TCs are increasing in strength at the 90% level of confidence (more than that in some basins). The papers differ in one region, the WNP-- I made that clear in my post.

    And when you are not trying to score points, maybe you and your readers here could contribute something useful and positive by helping classifying over 30 years worth of TC data:



  15. -15-Albatross

    Thanks for the advice ... we've actually done peer-reviewed research on global tropical cyclone landfalls (i.e., the storms which cause "disasters"), and we did not find evidence to support claims of increasing disasters:


    "Using currently available historical TC best-track records, a global database focused on hurricane-force strength landfalls was constructed. The analysis does not indicate significant long-period global or individual basin trends in the frequency or intensity of landfalling TCs of minor or major hurricane strength. The evidence in this study provides strong support for the conclusion that increasing damage around the world during the past several decades can be explained entirely by increasing wealth in locations prone to TC landfalls, which adds confidence to the fidelity of economic normalization analyses."

    If you think our analysis or data is in error, then write it up and submit to a scientific journal. That would be far more valuable than snarky comments on a blog ... Thanks!

  16. Albatross, based on the comment I just rejected ... you are back on one comment per day, thanks.

  17. Am I right in thinking that the graph at the top relates to the *number* of landfalling TCs in the WNP? If so, it does not say very much about what we are interested in, since both theory and observations, (including Webster, Holland, Curry and Chang 2005) indicate that it is the intensity of TCs that will increase, while the frequency of regular TCs will be unaffected or even diminish.

  18. -18-Richard Lawson

    Thanks ... you are close. It is the number of the most intense storms (which are Cat 3+).

    A key point here is that if the projections of increased intensity are correct, we would not be able to detect such an increase until the latter half of the 21st century. So I do wonder why people think that we will be able to detect those changes today ... you have to conclude that the models are very wrong if you believe that.


  19. Just in time to take advantage of the Haiyan publicity, I got a solicitation phone call from the Environmental Defense Fund this AM, which after thanking me for past support (last contribution about 1990, FYI), the next sentence was, "With the increase in extreme weather events..."

    My response was that there has been no increase in extreme weather events, and you would have known that if you had read any of Roger Pielke Jr.'s peer reviewed articles.

  20. Tom, Roger Pielke Jr.'s peer reviewed papers do not actually say that. They mostly say that there has been no increase in the damage as measured by money, or that there has been no increase in cyclones that hit the shore. Those are very different things.

  21. Albatross (#9 and #10),

    You quote the Kossin paper "stronger tropical cyclones, globally, have become more intense at a rate of about +1 m s− 1 decade−1during the 28 -yr period 1982– 2009, but the statistical significance of this trend is marginal [p-value = -0.1]."

    In other words, the trend does not reach the usual measure of statistical significance. It's not valid to claim increasing intensity based on that. In addition, finding the trend from the low point in 1982 makes it unreliable, even if statistical significance at the 95% level is achieved. Trends that do not account for or cover full cycles of phenomena like the AMO, PDO, ENSO and so on conflate natural decadal variability with long-term trends. Nothing terribly useful can be said, even with statistical significance.

    You also respond by saying "Before accusations are made about cherry picking by certain "honest" brokers, the authors in Koppin et al. provide very sound reasons for going back to 1982 and not earlier."

    Even if there may be good reasons for going back only to 1982, it doesn't change the reality that going back only to 1982 is misleading and unrepresentative of long-term trends. Again, if going back further is unreliable, then the data are simply not capable of giving reliable information. Better to avoid it altogether than to claim trends that are meaningless.

    You also said "The fact remains that Haiyan is the strongest TC to make landfall on the planet to date."

    No, that's not a fact. The highest windspeeds were ESTIMATES from satellite data. No direct land-based windspeed measurements show that Haiyan was that strong, nor is there evidence of instruments being maxed out or broken, as happened with Camille. Haiyan was strong, yes. The strongest in the instrumental record? No. The strongest "to date"? That's a completely unscientific and unverifiable statement. Please refrain from such nonsense.

  22. Well Brian, given that p value, which way would you bet?

  23. Roger, sailors might disagree with the statement: "Landfalls are of course what cause disasters."

    Typhoon Cobra (1944) for example, cost 790 lives and a whole bunch of floating iron ships at sea.

  24. Roger... In your research you base your conclusions on data using the Saffir-Simpson scale. Right? Do you agree that the SS scale (and likewise ACE) do not fully account for the total kinetic energy contained within cyclones?

  25. "Well Brian, given that p value, which way would you bet?"


    Well, if I were confident that systematic errors were removed, I would bet in favor of more intense cyclones. Unfortunately, we know there are serious issues of systematic error, in part due to using the 1982 starting year (when cyclone activity was in a low portion of its natural cycle), so I wouldn't bet either way based on this data. It's just doesn't tell us anything meaningful about what cyclone activity is likely to do in a warming world. People should stop pretending or hoping that it does.

  26. Brian... We know with complete confidence that we are adding energy to the climate system through increased greenhouse gas concentrations. Would you actually bet on that NOT somehow producing more frequent and/or energetic and damaging cyclones?

  27. Rob (#27),

    As I said in #26, I'm not willing to bet either way at this point. But I think you are really asking whether I can envision a scenario in which we cause warming and cyclones do not increase or get more violent. The short answer is "yes, easily."

    The simplest view of cyclones says that the water (or land) temp is higher in a warmer world, providing more energy for cyclones. But cyclones are basically heat engines and, like all other thermal phenomena, operate on temperature DIFFERENCES. The problem is that standard climate science predicts that the lower troposphere should warm FASTER than the surface (the lapse rate reduces), causing LESS of a temperature difference from top to bottom. That should make cyclones weaker and less numerous, unless they manage somehow to grow taller.

    But then there's the issue of lateral wind shear, which tends to prevent cyclones from forming or blows them apart once they start. Wind shear could get stronger in a warmer world (more energy available) but it could also get weaker (smaller temp gradients from equator to poles because of a faster warming arctic).

    These effects are complex and it's really impossible to predict how they will combine to affect cyclones. I think the most up-to-date GCMs predict fewer but more powerful storms, but there's no reason to think the GCMs at this point are reliable enough to provide a final word. The bottom line is that even from the standpoint of basic climate science, a warming world could either create more powerful and numerous storms or not. Because of this complexity, the best approach is to look at what the data shows. Unfortunately, even after 150 years, the data is still ambiguous, other than favoring no major effect either way.

    Ultimately, then, there's no scientific basis for expecting one outcome over another, and there is not likely to be such a basis for many decades.

  28. There is no reliable evidence, based on the UN's own science, for the attempt by the UN, EDF, Greenpeace, etc. in Poland to place blame on developed countries for the Haiyan disaster. Yet that's exactly what they're attempting to do. Sachs seems to be pushing the same meme.

    I have to wonder if comment #3 by Roger which provides several possible cherry pickable dates showing a linear increase in hurricanes, was blind luck or are commenters like #9 that predictable? ;-)