The Intersection

Global Warming and Tornadoes?

i-c3da8bf9dc6261b33cdfab1be6009cc5-tornado.jpg

Following the incredible recent destruction from tornadoes in Florida, it seems appropriate to do a brief post about whether there’s any significant global warming-tornado relationship, or at least, any relationship that we can confidently discuss at this point in time. I particularly want to address this topic because it’s one where, sadly, my own intellectual allies have left themselves vulnerable.

In An Inconvenient Truth, Al Gore threw in the 2004 U.S. tornado season in his laundry list of phenomena apparently related to climate change: “Also in 2004, the all-time record for tornadoes in the United States was broken.” Incidentally, for this quotation I am relying on the Competitive Enterprise Institute’s takedown (PDF) of Gore’s movie. I really hate to do that, but in this case–at least on this individual point–CEI seems to have found a weakness in Gore’s presentation.

Why would Gore throw in the mention of tornadoes, except to imply some sort of causal connection between the 2004 season and global warming? As I recall the movie, Gore didn’t make any specific causal claim, but again, why else include the topic if not to suggest the existence of one?

And yet now, here comes the IPCC Fourth Assessment Report Summary for Policymakers (PDF), which clearly states:

There is insufficient evidence to determine whether trends exist in the meridional overturning circulation of the global ocean or in small scale phenomena such as tornadoes, hail, lightning and dust-storms.

If there’s no documented trend in tornadic activity at this point, then of course we cannot claim that climate change is directly affecting tornadoes. There may well be theoretical reasons for thinking that there would be an impact (I’m not sure what those would be, but then, this isn’t my area). But in the absence of even a documented trend, much less a successful causal attribution, we need to be very cautious about how we discuss this subject. And I just can’t see how Gore’s statement in An Inconvenient Truth would fit this standard.

Again, I point this out because it makes me very unhappy when my own intellectual allies leave themselves vulnerable, as Gore has apparently done in this case. And while we’re at it, allow me to note that Barak Obama has also made a similarly incautious statement on tornadoes. In Obama’s case, though, I can’t help wondering whether the error is rooted in popular confusion over the difference between hurricanes and tornadoes.

Incidentally, and just to make things still more complicated, hurricanes also spawn tornadoes, so if global warming is affecting hurricanes….but, well, that’s another issue.

P.P.S.: Jeff Masters also invokes the IPCC report regarding any possible global warming-tornado relationship. He further notes a tornado-El Nino relationship, at least for Florida…

Comments

  1. #1 Scott Belyea
    February 5, 2007

    Ah, “climate change” vs. “recent weather.” And on the same tack, would it not be better to call it “climate change” vs. “global warming”?

    But that isn’t really my purpose – I recall reading somewhere a couple of years back (and of course I can’t recall where) an opinion from a scientist working in the field who suggested that severe tornado (or hurricane) seasons were not yet significant. He suggested a minimum of 10-20 more years for a reasonably decent assessment. And when asked what might convince him at the lower end (10 years) he responded, “5 bad years in a row.”

    He went on to say that in the popular media, the dramatic events were over-emphasized, and that less noticeable but perhaps more stable trends such as growing-season start, season length, and changes in bird migration patterns would end up being more useful “indicators.”

    Just curious whether my memory of this matches reasonably well with what you’ve learned …

  2. #2 ChrisS
    February 5, 2007

    Incidentally, and just to make things still more complicated, hurricanes also spawn tornadoes, so if global warming is affecting hurricanes….but, well, that’s another issue.

    Makes sense, logically, that if global warming is an increase in energy retained by the earth’s atmosphere (low-grade energy, but energy nonetheless) and it is not a consistent increase across the entire globe (i.e., differential heating is occuring), then the potential for inducing or amplifing the movement of air masses could come from that, and moving air masses are responsible for most tornadoes (excepting those generated by Bond villians and Cobra Commander).

    The earth is a pretty big place and necessitates a rather large addition of energy to the system to warm it even as slightly as it has. I’m sure lots of relatively little things are bound to occur as a result. The problem with linking hurricanes to global warming, iirc, in the 90s was due primarily to the fact the GCMs did not have the resolution to model the relatively small hurricanes.

    Based on the CEI pdf and Gore’s quote, it’s hard not say that Gore should refrain using that in his presentations. On other hand, the very nature of weather and climate makes it rather difficult for people to make confident assesments of trends while they’re in one. If over the next, say, five years, if we experience more and more tornadoes, Gore will look fine.

    But in the end, whether or not Gore overstated a couple of potential examples doesn’t imply that anthropogenic global climate change is not occuring.

  3. #3 bigTom
    February 5, 2007

    I think we will have to wait for a lot more data to find out about that one. Statements, that the (climate system is warmer) therefore theres more energy for storms seem to be just plain incorrect. Storms in some sense are heat engines, powered by differences in temperatures, not by heat itself. Storms have two sources of differences in heat to draw energy from. Vertical differences, i.e. ground/ocean/lower atmosphere, and the cooler air higher up. Horizontal differences, for example North South variations are also very important. The average vertical differencexs is probably increasing as a result of greenhouse forcing. The horizontal differences are expected to decrease as the poles warm more than the tropics due to snow/ice albedo feedback. So working out which factor is more important in a given instance must be carefully done. Then of course water vapor is an important working fluid for most storms, so there is yet one more way simple analysis could go wrong.

  4. #4 Gerard Harbison
    February 5, 2007

    Makes sense, logically, that if global warming is an increase in energy retained by the earth’s atmosphere (low-grade energy, but energy nonetheless) and it is not a consistent increase across the entire globe (i.e., differential heating is occuring), then the potential for inducing or amplifing the movement of air masses could come from that, and moving air masses are responsible for most tornadoes (excepting those generated by Bond villians and Cobra Commander).

    Ah, but AGW is strongest at the poles, so it actually decreases polar/equatorial gradients.

    Tornadoes/hurricanes are IMO one of the weakest arguments for AGW. They may indeed increase, but it’s not a straightforward relationship, and I wouldn’t be surprised if it was actually the other way round.

  5. #5 mgr
    February 5, 2007

    Concern with tornados and hurricane intensity and incidence can be more a natural hazards construct than a meteorological one.

    I would agree with Gerard that tornado incidence is a weak AGW causality, simply because tornadoes cannot be directly connected to increased residence time of energy. The problem is that tornados are primarily a local weather phenomena, linked to air mass climatology, topography, and the physics associated with the difference in air pressure and shear of air masses along a cold front. The effect of warming can go two ways here–1.) either warmer polar air masses interacting with warmer and moister air masses may imply a decrease in tornado incidence; or 2.) warming shifts the timing of tornado seasons, increasing incidence in some places, decreasing it in others, with an overall increase in total incidence (for USA, because there would seem to be shift northwards, including more continental mass).

    There is also a problem with the metric, do you include funnel clouds in your metric or not.

    As to Gore, one has to consider whether this example is of a predicted climatic outcome from global warming, or the meteorological phenomena associated with climate change.

    Mike

  6. #6 James Annan
    February 5, 2007

    If there’s no documented trend in tornadic activity at this point, then of course we cannot claim that climate change is directly affecting tornadoes.

    Chris, this statement is obviously wrong and neatly encapsulates much of the confusion between frequentist and Bayesian probability.

    Of course climate change is affecting tornadoes. The chances of every competing effect cancelling out to precisely zero is …. zero.

    What you can reasonably say is that the overall effect seems to be fairly small compared to natural variability, and/or that we currently have no good understanding as to whether the overall effect will be positive or negative. It is entirely possible that we will develop sufficient theoretical understanding to become fairly confident that the effect is a small increase (say) long before it can be detected in the standard way. In fact as a practical example of this we are already confident of a THC decline even though this is not formally detectable.

    As an analogy, consider tossing apples onto a lorry full of gravel. We have strong confidence that each apple will increase the total load by a few ounces, even though we are also quite sure that the effect will not be detectable until we’ve added a few thousand of them (using our weighbridge).

    Perhaps it’s a bit harsh to pick on a writer, since basically all the scientists involved routinely get it wrong too.

  7. #7 Chris Mooney
    February 6, 2007

    Thanks to everyone for the great comments. I hope to say more later, but to James, let me just say: Your point is taken, and I should have phrased more carefully.

  8. #8 Harold Brooks
    February 6, 2007

    Steve Bloom asked me to respond to this thread to give the current state-of-the-science of tornadoes and global warming.

    1) The reports within the historical database in the US are of insufficient consistency and quality to say anything about changes over time and space, with the possible exception that the late 1980s appear to have had very few tornadoes. Changes (whether officially-mandated or not) in data collection procedures over time make it very difficult to compare events over a long period of time. For example, the adoption of the Fujita scale for rating damage in the mid-1970s led to retrosptive rating of all events back to 1950. From a range of evidence, it appears that the tornadoes were overrated in the retrospective rating process, at least in comparison to the ratings applied in 1975-1999. Something else occurred to the system between roughly 1999 and 2002 that resulted in another sharp drop in the rated damage. While we can’t rule out that it was related to meteorological changes, it is exceedingly unlikely that the atmosphere in the US suddenly changed in such a way to cut the probability of a violent tornado occurring by approximately 75% in a short period of time. The recent change in the National Weather Service from the Fujita scale to the Enhanced Fujita scale will have unknown effects on the distribution of damage. Claims that there is a downward trend in the strongest tornadoes, as opposed to downward trends in the numbers of the strongest tornado reports, as done in the CEI document, are just as unsupportable, in terms of the meteorology, as the Gore quote. The dataset is ill-suited to address those questions unless it is subjected to serious manipulation.

    2) None of the databases in other countries are remotely adequate for analyzing time trends in meteorological changes in comparison with non-meteorological changes.

    3) As recommended at the IPCC Workshop on Extreme Weather in 2002, work is underway to identify favorable large-scale environments (typically combinations of convective available potential energy, vertical wind shear, and cloud base height) for severe thunderstorms and tornadoes and attempt to identify time trends in those. It appears that there may be reasonable hope for the severe thunderstorm problem, but the tornado problem will be much more difficult since it relies heavily on boundary layer information, which may not be handled well in the numerical models.

    4) We have no physical basis to make a prediction of the sign of any change in tornadoes in climate change scenarios. You can argue from simple first principles either sign (in the mean, CAPE should go up in a warming world, positive for tornadoes; deep shear should decrease, negative for tornadoes), but lots of other things, such as initiation of storms and the fact that we don’t really care about the mean conditions, but the juxtaposition of two extremes at the same time and place, means that those simple arguments may not be relevant. There’s likely to be more confidence in predictions of changes in very large hail than in predictions of tornadoes. Work that’s going on now should have reached a sufficient point of maturity that it can be included in AR V.

  9. #9 llewelly
    February 6, 2007

    Thank you, Harold Brooks, for a long an informative comment.

  10. #10 Mark Hadfield
    February 6, 2007

    “Work that’s going on now should have reached a sufficient point of maturity that it can be included in AR V.”

    Oh no, please no Roman numerals. AR5 will do quite nicely, thank you.

    BTW, great comment, James. Apples on a lorry. (Just one question: is a lorry similar to a truck?)

  11. #11 Chris Mooney
    February 7, 2007

    yes, let me just add, harold, that that was really excellent

  12. #12 kimi
    March 15, 2009

    “Work that’s going on now should have reached a sufficient point of maturity that it can be included in AR V.”

    Oh no, please no Roman numerals. AR5 will do quite nicely, thank you.

    BTW, great comment, James. Apples on a lorry. (Just one question: is a lorry similar to a truck?)