The power of data

I watched, for my sins, a new Global Warming Policy Foundation (GWPF) video called Extreme Weather Events & Global Warming: How Good Is The Evidence? To be fair, it was more balanced than I had expected but really just seemed to be Roger Pielke Jr versus the rest. The basic message seemed to be that there was evidence for more heatwaves and heavier precipitation, but no real evidence for more extreme events like hurricanes, cyclones, tornadoes, and flooding. I will add, though, that the video very clearly mis-represents what Connie Hedegaard (from the European Commission) says. The narrator (David Whitehouse) says

she admits there’s little evidence for an increase in climate extremes, before claiming that there is an increase. Contradicting herself and the IPCC.

No, what Connie Hedegaard says is that it is difficult to establish whether a particular event is actually linked to climate change ……. but the overall pattern of events seems to be bearing out the forecast that climate change will bring more frequent and more extreme weather. This isn’t a contradition. It’s really just the very obvious point that an individual event is not a trend.

Anyway, I’m no expert at this, but a quick search of Google Scholar returned a paper by Holland & Bruyère (2013). This paper seems to develop an Anthropogenic Climate Change Index (ACCI) and then considers how the different categories of hurricanes vary with ACCI. This is shown in the figure below which seems to indicate that although the fraction of Category 1 hurricanes decreases with ACCI, the fraction of Category 4 and Category 5 hurricanes seem to increase. It also appears that Aslak Grinsted finds that there is a trend in extreme hurricane damage. A point that Aslak makes in his post, though, is that it is simply impossible for any normalization procedure to remove all non climatic influences / societal changes that has taken place over the 20th century.

Variation in hurricanes against ACCI (credit : Holland & Bruyere 2013)

Variation in hurricanes against ACCI (credit : Holland & Bruyere 2013)

Now, I don’t know enough about this to really know whether or not there is evidence for a trend, associated with global warming, in extreme events like hurricanes and cyclones. I was, however, going to comment on something related though. The GWPF video also included Jennifer Francis who says

I think what we’re seeing happening just in these last few years seems to be very consistent with what the models are projecting for the end of the century.

This may sound a bit wishy-washy but, to a scientist, I think this makes sense and maybe illustrates an issue with the approach that I suspect some take with respect to issues like this. I’m thinking, for example, of those economists who regularly claim to understand statistics much better than typical climate scientists. Essentially, I get the impression that there are some who think that data is paramount. For example, if you want to claim/suggest that there is a rising trend in extreme events, you need to find a statistically significant trend in the data. If not, then you really can’t say anything.

I suspect there are many who think that that is indeed the right approach. It, however, misses – in my opinion – something important. If you have models that suggest that extreme events should increase over the coming decades and that also suggests we should already be able to detect a weak trend (possibly masked by natural variability) then that is extra information. Finding a weak, statistically insignificant, trend when models predict such trends is different to simply finding such a trend in some data without any associated modelling. It doesn’t necessarily prove it beyond any measure of doubt, but does allow scientists to say consistent with what the models are projecting. I would add that analysing data, without having some kind of underlying theory or model, is really largely meaningless.

So, I don’t know enough about extreme events (both in terms of modelling or data) to really know what the evidence suggests or how consistent the models are with the data. It does seem, however, that there are a lot of people who want to treat the models and data as completely independent entities when, in fact, we should be using both to understand what’s happening. Many, of course, are insistent that models need to be validated using data, but seem to completely ignore that you also need models/theories in order to understand the data itself. It’s not a one-way street.

This entry was posted in Climate change, Global warming, IPCC and tagged , , , , , , . Bookmark the permalink.

23 Responses to The power of data

  1. chris says:

    A very interesting paper that makes a very thoughtful and logical analysis of the question of trends in hurricane intensity and the role of anthropogenic global warming is:

    J. A. CURRY, P. J. WEBSTER, AND G. J. HOLLAND (2006) “Mixing Politics and Science in Testing the Hypothesis That Greenhouse Warming Is Causing a Global Increase in Hurricane Intensity” Bull. Am. Meteorolog. Soc. 87(8), 1025-1037.

    The contrast between the beautifully scientific arguments in this paper (the conclusion is: Yes, the evidence supports a role for anthropogenic global warming and increased hurricane intensity) and Dr. Curry’s recent approach to scientific debate and representation is astonishing. In this paper she and her colleagues even describe the battle with climate change “deniers” (their word) in desiring to represent the science faithfully.

    Something very odd has gone on there.

  2. One should also not act as if heat waves and strong precipitation are a minor concern just because they are not covered in the media in the way hurricane Katrina was. The heat waves in 2003 (France) and 2010 (Russia) killed thousands of people.

    I think that what Jennifer Francis wanted to say is that such heat waves are now an extreme heat wave, but could well be “normal” summer weather at the end of the century. I still remember 2003, that was not my idea of summer fun. The death toll will be less around 2100 because we will be better adapted and because many old and sick will have died before in what is then a real heat wave.

  3. Interesting and, as you say, very odd. Maybe Judith thinks she’s seen the light, but it is hard to understand how someone can publish a paper with that conclusion and, today, seem to hold views completely at odds with that.

  4. I agree. It’s more than just about extreme weather in terms of hurricanes, cylcones, etc. It’s also about how what we see as extreme today, will be normal by 2100.

  5. BBD says:

    What frustrates me is the contrarian rhetorical trickery. We are at the *beginning* of a process which, by later this century, will see a very considerable further increase in the amount of energy in the climate system. This will inevitably lead to an increase in the frequency and intensity of disruptive weather events. How can it not?

    But contrarians like to look back over past decades and say “look – no catastrophe there, so you are being alarmist”. As if the past few decades can tell us anything much about the end of the century.

    Yet this is what they do, over and over again, and it is a glaring nonsense. Personally I would call it deliberate misdirection.

  6. Rachel says:

    She was abducted by aliens in 2007 and an alien impostor is now impersonating her.

  7. What you say about energy is a good point. If we keep adding energy to the system, we would expect to get more energetic events than we would have in the past. I appreciate that there are some subtleties (global warming could reduce wind shear which is a driver for tornadoes) but the idea that we can continue to add energy and not see more extreme weather events does seem absurd.

  8. dbostrom says:

    Vanishing energy. Tantamount to imagining no difference in kinetic energy as velocity increases. Just a short leap to “gravitation is only a theory; we don’t really know what the difference will be if we drop a mass 1 meter versus 5; the potential energy of a mass hanging at 1 meter is quite possible the same as 5.”

    What I’d like to see is some solid exploration to help explain how all this energy will vanish. We hear suggestions about no need to worry, no explanation of why. Presumably the assumption is that all the extra energy will be so perfectly homogeneous that we won’t see any differences, but how can that be? If the whole system is smoothly more energetic there somehow won’t be any difference in things such as convection, etc.? That sounds impossible.

  9. What you say probably highlights why I started writing about this. I know the climate is a complex system, but you can consider some aspects from a relatively simple perspective. Global warming, for example, is fundamentally about energy. We can’t keep adding energy without making changes to the system. The precise details of how the system will change might be difficult to predict but that certain events will get more energetic seems like a fairly obvious conclusion.

    It does seem like many who are highly skeptical (or who are spreading skeptical views) are not actually physical scientists and it does make me wonder if they just do not understand these fundamentals.

  10. dbostrom says:

    For once, it really –is– spherical. 🙂

  11. BBD says:

    Yes. Physics. Always physics.

    And this can be a problem, as it was for me. I’m too stupid to understand the physics of radiative transfer from the inside out, so I took my initial scepticism for a long walk through paleoclimate. If the physics is right, I reasoned, past climate behaviour will provide empirical evidence. And so it does. Lots of it. More than enough to confirm the efficacy of CO2 as a climate forcing. This helps compensate for not being able to understand the physics like a physicist. I have evidence I can understand and intellectual integrity is maintained. It’s a shame other sceptics and lukewarmers can’t muster up the intellectual curiosity to try this themselves, because it works a treat in clearing out the nonsense.

  12. I wouldn’t say that increases in extreme weather are guaranteed. In as much as the mean temperature and precipitation increases, you would expect that also the extremes increase. However, also the variability around the mean is important for the probability of getting an extreme event.

    Many types of variability are expected or observed to decrease. The temperature difference between the equator and the poles. The temperature difference between summer and winter (stronger trends for winter temperatures), and between night and day (stronger increases in daily minimum temperature). Also the year to year variability in the yearly mean temperature may get smaller. (Other types of variability do increase.)

    In the end, climatology expects that heat waves will get worse and that strong precipitation will increase stronger as mean precipitation. However, that is not fully trivial, you do have to study for every phenomenon what will happen. Studies on extremes are still in their early stages. A lot can still happen to our understanding a changes in extreme weather.

    If climate ostriches had one a little understanding of science, they would not make ludicrous claims about CO2 not being a greenhouse gas, they would not claim that the mean temperature is not increasing or only half, they would not claim that homogenization is just smoothing and all that other much too obvious nonsense that hurts their credibility, they would talk about all the uncertainties in projections for extreme weather and thus uncertainties in the societal costs of climate change.

    This naturally assumes that climate ostriches would keep their strange slogan that uncertainty means that nothing will happen.

  13. BBD says:


    Of course you are correct. To be clear, I chose the term disruptive weather events because I was thinking about the progressive impact on agricultural productivity of expanding Hadley Cells and an accelerated hydrological cycle. It seems plausible that the net impact will become increasingly negative over time.

  14. dbostrom says:

    I’ve got the “Father Guido Sarducci 5-minute University” tattered remains of undergrad physics to rely on when thinking about all this, so actual penetration of the surface of this problem is not genuinely possible for me.

    That said, my expectation based on “5-minute University” physics would be that with more energy available to drive events, weather events involving mass will be more kinetic and more dynamic. From my perspective, most such events involve convection, movement of air and water in its various forms. These seem to be powered by heat in the process of being homogenized; more available heat would seem to lend more power to these processes.

    So with that in mind and further stretching my neck, I’d expect to see “average” weather of the future to more resemble “extreme” weather of the present.

    But all this presupposes that there’s a heat sink somewhere in the system, ultimately down the road a -real- heat sink (not the ocean) that scales dissipation as input load increases. Is that true?

  15. I also studied physics, my intuition used to be the same as yours. But the situation is really a bit more complex. Your keyword dynamics is also an interesting example. If dynamics in the sense of movement becomes stronger, the risk of floods could decrease. The dangerous situation for many floods is slowly moving precipitating systems that locally produce a lot of rain.

    I aim not claiming nothing will happen with respect to extremes, for many extremes increases seem likely, just that the situation is not completely trivial and we do need to study it carefully.

  16. Maybe I’ll chip a comment in here (but it is getting late so maybe this won’t be all that clear). Increases in things like heatwaves and precipitation seem straightforward. Higher temperatures means more heatwaves and also that the atmosphere can hold more water vapour. If I understand what you’re saying, there are some situations where it is more complex and multiple effects can play a role. An example would be tornadoes where warmer air might mean stronger tornadoes, but warmer air could also mean less wind shear which plays a crucial role in the formation of tornadoes.

    So, as far as I see it, more energy would imply that if an event can actually take place there is a greater chance that it could be more energetic, however the complexities in the system could mean that certain events (tornadoes for example) could be less likely.

  17. dbostrom says:

    “But the situation is really a bit more complex.”

    Tactful understatement. 🙂

    Leaving aside impacts on various living things we care about, this is all pretty fascinating.

    Do we end up with boringly uniform warmth, or do we end up with a lot of atmospheric drama?

    Once the ocean gets over being shocked and is smoothly coupled with the action topside, it seems that the plot must lead upward. At the end of the day, the planet is going to reach radiative equilibrium for whatever timescale we’re interested in and that entropy leads up.

    Long ago I read an SF book that was mostly concerned with geopolitical and commercial outcomes of dealing with climate change some 200 years from the present. I wish I could find this book again because at the time I read I knew even less about this topic than I do now and I’d like to reassess it, In any case, the book conjectured frequent occurrences of what we call supercell convective storms, except these were energetic enough to transport their load much farther up in the atmosphere than what we see today. An interesting idea; I suppose the author was hypothesizing one means to quickly dispose of heat.

  18. BBD says:

    Perhaps the focus on extremes is in danger of becoming a red herring. My understanding of the ecologists’ concern is that rapid change in local conditions will disrupt ecosystems globally. Since we dine on top of the established Holocene order, this will hit us along with everything else.

    Pulling back a bit, arguments about ECS and TCR also tend to distract from the fundamental issue of ecosystem sensitivity to rapid climate change.

  19. I would expect that changes in extremes and variability are quite important. Also ecosystems respond to variability and not just to the mean temperature and precipitation.

    Palm trees grow where it does not freeze (too often), similar for many bugs. Heat waves also damage ecosystems. I would expect that the composition of forest also depends on how often a strong storm kills the less deeply rooted trees and storms create open spaces with lots of new niches and biodiversity in otherwise dark forests.

    But herring salad with red unions is delicious.

  20. gman64 says:

    Living in Boulder and having experienced about 85% of a year’s worth of precipitation in a week, including 9″ in a single 24-hour period (average is about 20″/year) the science around extremes is of more than an academic interest to me. A relatively small number of people died, but the infrastructure generally suffered greatly. Many mountain areas are still inaccessible and much work remains to be done and much expense remains to be paid. If this sort of event is going to become much more normal, we need to seriously rethink a great many things.

  21. Thanks for the comment. As someone who experienced what happened in Boulder, what’s your view of Roger Pielke Jr’s comment : I pretty much feel 1000 year estimates are in the realm of fantasy.

  22. gman64 says:

    I generally don’t take RPJr. very seriously, but given that the 9″ recorded on 12 Sep was nearly double the old 4.8″ record (over the 1897-2013 period) we’re talking lots of sigmas of deviation. Whether that qualifies as a 1/1000 precip event, I’m not an expert, but RPJr.’s standard insistence on downplaying a genuinely extraordinary event makes me dubious of his claims.

  23. Indeed. I think that there is evidence that it is a 1000 year precipitation event. What Pielke Jr was doing, I think, was focusing on the flooding only. I believe that this has the problem that it is hard to remove how our developments have changed how serious the flooding will be. So it might not technically be a 1000 year flood event but if it is a 1000 year precipitation event, nit-picking about whether or not it is technically a 1000 year flood event (especially as this might be because we’ve actually done things to make the flooding less severe than it could have been) does seem rather disingenuous.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.