Attributing single events to background conditions

The problem with this is that if you read accounts of life in the Midwest and South in the 19th century, severe floods and tornados seem to occur all the time. It looks more like we are coming out of a period with an unusual lack of severe weather events, instead of entering a period with unusually severe weather events.

Big river valley systems in the middle of continents are vulnerable to floods and tornados. Again, some Americans are pretending that the unusual (and unusually lucky) decades after World War II was the normal condition.

And if tornadoes haven’t particularly increased in the last 30 years? See here for example.

The short version is that tornado *warnings* are up due to a combination of the rise of tornado spotters, WSR-88D, and other weather watching improvements. Tornado frequency and intensity appears to be pretty much level.

Much of news reporting of course is about when tornadoes interact with towns as opposed to when they interact with open land. Which is why 1974 is called the year of the “Super Tornado Outbreak” (with the most deaths), even though 1973 had by far a larger number of tornadoes reported. Whether or not they interact with towns or open land, thus far, seems to be global warming independent.

Which is not to say that global warming couldn’t increase tornado frequency. Of course it could, or might not (as it might decrease the number of times that intense cold fronts interact with warm fronts by causing a general increase in temperature).

But the state of the science at the moment appears to be something like “global warming doesn’t appear to have increased tornado frequency because tornado frequency doesn’t appear to actually have increased”.

Whether or not they interact with towns or open land, thus far, seems to be global warming independent.

Until people start moving off the coasts and into Kansas…

Anyway, while it does seem to be global warming independent, it’s *not* independent of everything — in particular, the number and size of towns, i.e. population density. More towns -> more tornadoes in towns. (This is also the source of the trailer park tornado-magnet effect; tornadoes get more human attention when they do more damage to humans and human possessions, and they do more damage to trailer parks.)

Two secular trends going on at the same time in the same direction are correlated, but not causally related.

P.S. Why would global warming lead to drier air? Warmer air holds more moisture, and a warmer ocean evaporates more water vapor per square mile per day, doesn’t it? (Lakes and rivers too, but their surface area is much less than the ocean’s.) I’m curious about what kind of mechanism could not just offset that, but completely reverse it. Less movement of airmasses from ocean to continent and vice versa?

This seems like a good time to link John McPhee’s The Control of Nature: ATCHAFALAYA about the incredible hubris of the US regarding the Mississippi river. Short version, when the Mississippi river finally breaks out and floods Louisiana, killing thousands or more, it won’t be global warming. It will be because the US has sought to control one of the largest rivers in the world in increasingly dangerous ways.

Scientists, social and otherwise, being a naturally diffident group, are loath to claim that we are now seeing a trend. However, since it is now generally established and accepted (at least amongst the reality based) that Anthropogenic Global Climate Change is a fact and probably not A Good Thing, some of the less cautious are beginning to suggest that we will see more frequent (adjusting for reporting, which has been a given for at least 20 years or longer, unlike cutting edge medical diagnostic tools, say) and extreme climate events. And, indeed, that we might be seeing the beginning of such a trend. All this of course is exacerbated by very real social and cultural factors such as population movements and locations, poverty and Republicans.

Bill McKibben is an alarmist seems to be the new Michael Moore is fat and therefore we (left and right) can ignore him.

The usual formula I’ve heard is: we can’t say this particular extreme event is the result of global warming, but we can say we’re likely to get more extreme events like it as a consequence of warming.

I was under the impression that the science was not yet in on tornados and global warming. But I think a science of extreme events – especially the probability of observing many across multiple domains that may be related to a single cause or trend – would be a challenging and interesting field. Is it being done?

Saying “these events may not be unconnected and it’s okay to think about that” is not the same as saying “every extreme event can be concluded to be due to AGW.”

sheesh.

As for heat and light, our culture’s intellectual understanding of the threat posed by global warming seems to greatly exceed our willingness to take the steps necessary to combat it. Since it is emotions that motivate people, this suggests that it is heat, not light, that is in short supply, though more light is always welcome too.

@Martin, things are really more complicated than that. A major effect of higher global temperatures (as opposed to a temporary local heat wave) is that more water gets absorbed into the atmosphere. So more preciptation and more cloud cover are as likely (and really more likely) outcomes as more drought. Of course, for some parts of the globe the outcome will be more drought, but that is definitely not the outcome people should most expect.

The effect on small-scale phenomena like tornados is even harder. It’s like asking whether an increase in income will make you forget you keys more or less often. Some potential mechanisms will increase the conditions needed for tornados, others will decrease them. The tornado zone of the US is a very particular combination of geography and climate that leads to regular strong tornados, so the question is not even whether global warming in general makes tornado condition more likely, but whether it will in zones that are prone to have sizable ones at all.

As yet, observation does not show an increase in tornados at all. It might be politically tempting to boldly say that there is an increase, and that it is linked to global warming.

But it’s probably best that such claims are made by people without a link to science, so that their reputation won’t do too much damage to that of people who are trying to make reliable prediction of the effects of climate change. Really the last thing we want is a mud-fest of unreliable claims on both sides.

No, it’s not have your cake and eat it. The question is whether there is a statistical increase in severe weather events, including but hardly limited to tornadoes. On the other side is whether isolated incidents of cold disprove global warming. These are not symmetrical propositions. Some are disputing the fact of the first, but if the fact is granted, it does constitute valid evidence. The second does not constitute evidence regardless of the facts.

Zamfir, no, it is not like asking whether a change in income will make you lose your keys more or less often. There are definite causal mechanisms that are reasonably well understood. Making an analogy like that is like saying the workings of a car are unknowable because you don’t happen to understand them. That’s why NASA simulations of the effects of global warming 4 years ago specifically predicted increased tornado and heavy storm activity in the central and eastern United States

“The central and eastern areas of the United States are especially prone to severe storms and thunderstorms that arise when strong updrafts combine with horizontal winds that become stronger at higher altitudes. This combination produces damaging horizontal and vertical winds and is a major source of weather-related casualties. In the warmer climate simulation there is a small class of the most extreme storms with both strong updrafts and strong horizontal winds at higher levels that occur more often, and thus the model suggests that the most violent severe storms and tornadoes may become more common with warming.”

Zamfir, global warming produces greater humidity is some areas and less in others, true. But no one is claiming that all areas are seeing increased wildfires. Just the drier ones, which is precisely what you would expect.

“Researchers have predicted that some regions would have less humid air in a warmer climate and be more prone to wildfires as a result. However, drier conditions produce fewer storms. “These findings may seem to imply that fewer storms in the future will be good news for disastrous western U.S. wildfires,” said Tony Del Genio, lead author of the study and a scientist at NASA’s Goddard Institute for Space Studies, New York. “But drier conditions near the ground combined with higher lightning flash rates per storm may end up intensifying wildfire damage instead.”

And Torquil, you believe this month has been exceptionally mild for tornadoes? Well, this year has had almost double the usual number of tornadoes. This month certainly does not seem like a spectacular improvement. Got cite?

holden:

“These are clearly lone wolf climate events, not influenced or supported in any way by the climate in which they appear. Can’t plan for them and, no blame can be accrued to the larger climate whatsoever, no matter how it looks.”

surely you mean a few bad apples, right?

d

Torquil, indeed, May is not yet over, and the official death toll from recent tornadoes is likely to rise further. But look at the statistics you cited: YTD killer tornadoes: 54, over twice the three year average of 22. Tornado deaths: 504, more than 8 times the 3 year average of 64. The only statistic that might seem to support your sanguine attitude is the total number of tornadoes, but, as it says right below the table, other than the first two months, these are just preliminary reports based on newspaper accounts. The actual figures could well be higher: for January, they are 60% higher, for February unchanged. So far, then, we have an average of 30% higher, though the sample set is too small to make reliable inferences. And it says right there that you shouldn’t compare the preliminary to the actual figures for other years.

Martin,
The key determining factor on how big a wildfire is after it starts is the fuel load available. 100 + years of fire supression policy guarantees that once a fire starts these days, it’s going to be massive. Remember the Yellowstone fire? Since then, US has tried to reverse policy and do off season controlled burns and, in areas that have burned recently, the Forest Service is far more likely to let a fire burn so that fuel does not accumulate leading to a major conflagration. However, there’s a whole lot of fuel out there. And then when people start living near that fuel, they don’t want to have controlled burns because it messes up their swimming pools (See Mike Davis City of Quartz Chapter 1) or they fear that the controlled burn will get out of control if the weather does something funky (see the Los Alamos Forest Fire in 1999 (?) which started with a controlled burn gone awry. The forest fires you see on the news have nothing to do with global warming. In fact, globabl warming will probably lead to smaller less-newsworthy fires (although they may be more frequent) because fuel will not have time to accumulate.

As for “killer tornadoes,” the killers here aren’t the tornadoes. They are the lawmakers and rules-writers who caved into the housing lobbies and allowed standards for manufactured housing that they know will result in deaths and when tougher standards are imposed, they don’t grandfather those rules in. Similarly, the hurricanes in Florida (ie:Andrew) in the 80s revealed the inadequacy of Florida’s building codes and code enforcement because people could build houses that would not resist a big hurricane and then they could sell them as “hurricane safe.” By the mid-2000s, developers in Florida were busy trying to get the post Andrew codes repealed, cutting inspectors etc.

Even if you stopped climate change tomorrow and reversed it to 1950s or 1970s levels, the death and property destruction tolls from “Natural Disasters” would continue to rise because we continue to make policy that puts (poor) people and (rich people’s property) in harm’s way. These events are failures of policy that a carbon tax (or whatever your preferred remedy for climate change is) won’t solve.

Again, the literature on this is huge and the best starting place is Ted Steinberg’s Acts of God, updated since Katrina btw.

Martin, I’m not asking you to believe me. I’m asking you to believe Stephen Pyne, the guy who wrote the book Fire aka world’s leading authority on the history of forest fires. I’m asking you to believe all the historians who have studied flood control that the increase in 100 year floods has been caused by a) development in flood plains and wetlands and b) development upstream that increases run-off and c) the continued violation of the first rule of flood: control keep the bed deep and the banks low.

None of this denies that global warming is happening, just that the methods McKibben uses to measure it aren’t the right ones.

And the NASA site said fewer storms but more lightning strikes. Since lightning is what causes forest fires, more strikes = more fires. Since current policy is to let fires burn and to try to do controlled burns to eliminate fuel, the result (if current policy continues) is more but smaller fires. The NASA case made no mention of fuel levels and probably assumes current fuel levels (although it’s hard to know if they even accounted for fuel since they hedged their bet with “may”). If you’re not willing to read the books at least look at the page.

A fascinating thread. One would conclude that Crooked Timber is read by people who would mostly agree that Anthropogenic Global Climate Change is a fact and would (and apparently do with a few exceptions) deny that we are seeing any effects from said change. Most curious. McKibben is derided for saying the sky is falling. The sky fell in Joplin.

Big fires in Oregon or Wyoming are related to fire suppression in forests. Fires across grasslands in Texas probably aren’t.

RE: Australian flood control. Funny, the experts at this site say the rainfall that caused the record flooding was less than historic highs. Rather, the flooding was caused by 1) over-reliance on dams 2)increased non-permeable surfaces leading to more run-off and 3) more building in the flood plain. http://www.aussmc.org/2011/01/rapid-roundup-queensland-flood-crisis-%E2%80%93-experts-respond/

I especially liked this quote:
Likewise throughout Queensland we know the flood levels, some flood records were exceeded, most were not, so we can expect more floods and we can expect worse floods – it’s not if they will occur again, it’s just when.

So having these floods is no surprise, it’s something we have been waiting for. What is, however, regrettable is that knowing full well such flooding will occur again, and again and again – what do we do about it? Everyone with mud in their house lives on a floodplain – floodplains flood and will keep flooding. So if we want to keep the water and mud out of our houses we need to adapt both where and how we build and how we manage the river systems.
…
• Why was the Brisbane floodplain, devastated in 1974, allowed to be developed even more intensely – with no flood protection?

Western Dave, my understanding was that the rainfall that caused the recent floods was very close to the previous highs. And this is the problem: you’re focussing on household damage, which as you say is related to building codes, inequality, etc. But the phenomena themselves are trackable. Australia had two huge floods in a year (there were more in the South East, too, right?) At the same time China has a record drought, etc. The question is not whether the damage is linked to global warming; it’s whether the weather events that caused it are linked to global warming. And all the events I listed are related to higher temperatures and water vapour. At some point it stretches credibility to say that they aren’t driven by the climate.

Sebastian, if you believe there are only two ways that deaths from tornadoes can increase: either there are more tornadoes, or more tornadoes are migrating to inhabited areas, and there is no way to see what would cause the second, then there must be more tornadoes. The increase in both total deaths from tornadoes and total number of tornadoes that have killed people this year are empirical facts: any theory must be consistent with them. Therefore, by the argument you yourself have made, tornadoes must have increased (note: your argument, not mine. I do not accept this simple model). This is a straightforward statistical inference and is independent of causation: if you believe this bad weather is because we have offended the great god Enlil, fine. If there are more tornado deaths, and tornado distribution is consistent, Enlil must have created more tornadoes. This doesn’t even get to the problem of global warming as cause.

So have tornadoes increased? The NOAA site has no answer. The preliminary data is understated (we can be very confident that tornadoes mentioned in the newspaper exist, but it is very unlikely that all extent tornadoes are so mentioned, largely for the reasons that form the basis of your own argument), and the actual data is not there yet. The paper you pushed earlier is apparently close to a decade old. You say you have evidence of a lack of increase? What is that?

And let’s suppose the contrary: assume the number of tornadoes YTD is around the average. The 3 year average YTD number of tornadoes till the end of May is 700. Average deaths: 52. Average killer tornadoes: 15. 2011 YTD deaths: 517. 2011 YTD killer tornadoes: 66. Percentage of all tornadoes killer 3 year average: 2.1% Percentage of all tornadoes killer 2011, assuming normal number of tornadoes: 9.4%. Quite an increase, wouldn’t you say? Why are so many more tornadoes just happening to hit the inhabited areas? Of course, I don’t buy your model. I think the intensity of the tornadoes matters as well as the number, so I would not project an approximate 4.5 fold increase in the number of tornadoes. I expect stronger, not just more, tornadoes.

Is it perhaps an outlier? Well, if we take off Joplin, supposing that a bit of bad luck, it doesn’t change the picture that much. Reduce the number of killer tornadoes by 1. Reduce the death toll by 123. That still leaves a death toll of 394, close to 4 times the average. Like I say, one year doesn’t prove a trend, but that there is a dramatic increase this year is indisputable.

And trying to explain things like this is terms of multi-decade trends doesn’t work. If marijuana use among teenagers were to increase this year, could we credibly blame rock music? There is possibly a correlation between marijuana use and rock, or specific styles of rock. But rock music has been around for decades, and whatever influence it may have it had at the beginning or is long-term; it cannot explain a sharp recent change. And all of Western Dave’s arguments amount to arguing that sharp recent changes are the result of situations that have existed for decades.

Dave also argues that you cannot definitively attribute any specific weather event to global warming. That’s why I had the roulette dialog earlier. The statement is technically correct, but obtuse. You can attribute a pattern of weather changes to global warming, and that is what McKibbon is doing. You can argue that he is incorrect, that the statistical inferences he is making are unwarranted, but not that the project is fundamentally misguided. I personally would love to see a rigorous form of the argument McKibbon is just outlining.

One of the worst tornado outbreaks, in terms of deaths, damage, and apparent wind speed (one of them lasted 3 hours and more than 200 miles) took place in *1925*, see The Tri State Tornado.

See also the Tupelo–Gainesville tornado outbreak. That was *1936*.

Yes the recent April 24-28 outbreak really sucked. It was really big. We haven’t had a big one in a long time. But there were other really big ones in recent memory. And they were big enough to be comparable even though they didn’t have Doppler radar tracking the entire country at the time. They had some of the strongest, most intense tornadoes that we’ve seen. In 1925 and 1936.

Martin Bento @20: Making an analogy like that is like saying the workings of a car are unknowable because you don’t happen to understand them. That’s why NASA simulations of the effects of global warming 4 years ago specifically predicted increased tornado and heavy storm activity in the central and eastern United States

That is a simulation of a future in which the surface is 5 degrees warmer and there is twice the amount of carbon dioxide. It did not predict increased tornadoes and heavy storms in 2011.

And 2011 isn’t significantly warmer than 2007. In fact for the most relevant three month period for which I can see the mean for global land-ocean surface temperatures (DecJanFeb), 2011 is cooler.

My point is that studies like that – while suggestive and importance to assessing the long-term impact of climate change – should NOT be used to support the idea that we can attribute severe weather events right now to climate change.

With respect to wildfires, there are gems like this in the NASA release which you linked to:

“These findings may seem to imply that fewer storms in the future will be good news for disastrous western U.S. wildfires,” said Tony Del Genio, lead author of the study and a scientist at NASA’s Goddard Institute for Space Studies, New York. “But drier conditions near the ground combined with higher lightning flash rates per storm may end up intensifying wildfire damage instead.”

In other words, they don’t know.

Martin Bento @27: What McKibbon is pointing out, though, is that we are seeing myriad record-breakers in very close proximity. Two “hundred-year floods” in five years. Simultaneous record drought in multiple parts of the world, etc.

I struggle to separate what McKibbon wrote in that article from the perennial crop of apocalyptic seers who see a massive hurricane and note that this confirms what they would have predicted.

My understanding is that a higher likelihood of certain types of severe weather events in a given region is influenced most strongly by ENSO – not, at this point, climate change.

Over the long-term future, of course, that’s a different story.

As to the need for “more heat” in the public discussion in order to facilitate policy action, my view is different. I think these articles weaken credibility in the eyes of the public, since not only are these articles extreme, but they practically invite the public to question climate change on the basis of individual weather events that do NOT fit with climate change models.

And, if I remember correctly, a few years ago we had a significantly colder year than that of the ten years preceding – and polls showed a drop in public perception of climate as an actual phenomenon. Not conclusive obviously, but suggestive of the problem.

Sebastian, if you believe there are only two ways that deaths from tornadoes can increase: either there are more tornadoes, or more tornadoes are migrating to inhabited areas, and there is no way to see what would cause the second, then there must be more tornadoes.

But there *is* a way to see what would cause the second: proliferation of inhabited areas, i.e. population growth and/or sprawl.

You’ve also left out buildings becoming systematically less safe, which would result in more deaths per tornado/inhabited area interaction. I don’t have any specific information that building codes have been weakened in a way that increases danger from tornadoes, but given the politics of the most tornado-prone areas of the U.S., I wouldn’t rule it out.

That’s why you need to measure tornado activity directly: deaths from tornadoes are a very poor proxy for it, and the tornadoes:tornado deaths ratio may have a secular trend.

Andrew, your last paragraph at 42 is exactly the same logic as you’re decrying in your previous paragraphs. Does that make sense to you or validate your theory? I’m not sure.

Western Dave and Andrew, if people conclude that the odd rare cold year or mild season is conclusive proof against climate change then the chances are that it’s got something to do with the huge media attention on just that fact. It doesn’t have much to do with people pointing out – quite reasonably – that record hot temperatures are to be expected in a warming world, and that fires and floods follow a warming, humidifying trend. If the lay observer doesn’t get this, it’s because the media they’re observing from their lay position are conveniently ignoring the facts. That’s not McKibbon’s fault, quite the opposite in fact.

It’s high time that we started pointing out that all the predictions of AGW science are starting to come true. We can do this in a nuanced way, without saying “OMFG! It was 37C yesterday! AGW!!!1” because guess what! ordinary people aren’t stupid. But saying “oh no, all these strange events that keep happening, that are mysteriously very closely related to temperature, they’re all just coincidence so as you were folks! Wanna burn another seal?” is not going to help.

Chris, as I pointed out in #41, deaths from tornadoes have significantly declined in absolute terms since the 1950’s (at least if you take this year out). That’s not per capita, so that’s despite population increase, sprawl, more common trailer parks, etc. Also, building code changes. You’re trying to explain a fact not in evidence: there is no long-term increase in deaths, but the opposite. The increase is specific to this year. But to be clear, I don’t think the increase in deaths is primarily a result of more volcanoes’ I think it more likely stronger volcanoes (though it easily could be both), which are more likely to hit inhabited areas because they travel farther. I agree with measuring tornado strength directly, but that data won’t be available for a while yet apparently. I am willing to go out on a limb and say aggregate tornado strength so far this year will be found to have been at least triple the most recent 10 year average. And, if you cut off at the end of April and take Joplin out, at least double.

Again, what determines whether a tornado is likely to hit an inhabited area? If tornadoes were stationary objects, it would in principle be primarily be a function of what percentage of the land is inhabited (my guess is not quite, as a lot of buildings about probably breaks up the wind and discourages formation). But tornadoes move. They typically move several miles, while having a diameter much less than that. Therefore, if a tornado hits an inhabited area, it is very likely because the tornado moved there, not because it was spawned there, and this is more likely to happen the further the tornado moves. And the path length of a tornado is largely a function of its strength. If you have significantly more tornadoes hitting inhabited areas, it can be because you have stronger tornadoes or more inhabited areas, but if it is a one year jump, it has to be stronger tornadoes.

Andrew, I made that point about the NASA study specifically in regard to Zamfir’s argument that one could not in principle predict increased tornado activity based on global warming. That said, the study also overestimates the ability of vegetation to absorb CO2. It has recently been discovered that this is significantly less the previously supposed, and accelerated warming is held by the researchers doing the study to be a direct consequence.

As I have said, the tornado situation in itself proves nothing. Whether all the recent bad weather globally proves anything should be subject to rigorous analysis, but at first glance, it does look outside the range of normal variation. But decrying what McKibbon wrote is essentially shunting aside the question. You’re only going to ask this question and fund the necessary research if you believe it might be true, so someone needs to point out that it might be true, preferably pretty loudly.

“more volcanoes” – tornadoes, of course. I should have my coffee before I post.

You’re trying to explain a fact not in evidence: there is no long-term increase in deaths, but the opposite. The increase is specific to this year.

Hmm, I guess I just assumed it was a long-term trend because AGW is a long-term trend so it would make absolutely no sense whatsoever to try to tie it to a single outlier year. (A cluster of extreme years is another matter, of course. If we have one.)

A long term fall in fatalities could easily be explained by better prediction, better communication of predictions once they have been made, better shelters, or a combination, to name just a few. These are ad hoc, and determining whether any of them was *really* responsible would be quite hard, but that’s the point — whichever way the long-term trend in tornado deaths points, it’s a poor measure of tornado activity because of all the confounding variables. And measuring it directly is hampered by the fact that the technology is barely here *now*, let alone being able to go back in time a few decades and deploy it to measure past activity.

If you have significantly more tornadoes hitting inhabited areas, it can be because you have stronger tornadoes or more inhabited areas,

Over a short term it can also be luck. I’m not sure if a year is “short” or “long” in the relevant sense here. (Maybe the “significantly” was intended to rule that out? But if so, have we really seen a big enough cluster of hits to be *significantly* more and not just *noticeably* more? The sample size seems small to me.)

but if it is a one year jump, it has to be stronger tornadoes.

But if it’s *really* a one year jump, it can’t be AGW, which occurs gradually over decades or centuries. Higher tornado activity due to AGW would produce multiple high-tornado years over decades, which we haven’t really seen in past decades and don’t yet know if we’re going to see in future decades. “2011 — fluke or harbinger?” is not a question that can be sensibly answered, or even attempted, in 2011.

Is the one year jump really “jumpier” than the normal year-to-year fluctuation? Is that a sign of a long-term trend in mean activity, variance in activity, or just an unlikely event that is just as unlikely to happen again as it was unlikely to happen the first time? There are statistical tools to answer those types of questions, I think, but the data may be too sparse to draw any firm conclusion.

“This is why we need a science of extreme weather events.” Which is why I keep name-checking the environmental historians who actually write about this stuff not that anyone seems to want to go check them out (But one last time Ted Steinberg, Acts of God is the logical starting place for US “natural” disaster history excluding forest fires which is quite well covered on a global scale by Stephen Pyne). But apparently nobody wants to take the time to read them. And btw the drop in fatalities in tornadoes is due directly to the introduction and spread of better detection technology (which also explains the uptick in numbers of tornadoes since 1950s). Now will those of you who think you can explain weather phenomena by googling some stats please at least read the book reviews, find an article by these folks if you have JSTOR access or something beyond three years of tornado evidence and acting like it’s science or history or something. I’m thrilled McKibben scared you into realizing that global warming is an important issue, but if you want to save some lives, we need to quit quibbling about whether he is a useful or useless fearmonger and get with a legislative program. Some good starting places in the US: ending NFIP, permeable paving, reverse the gutting of the weather service, better building codes and better enforcement in hazard prone areas, etc. etc. etc. And a carbon tax wouldn’t hurt either. But it won’t save anybody’s life in the next ten years.

sg @46: Andrew, your last paragraph at 42 is exactly the same logic as you’re decrying in your previous paragraphs. Does that make sense to you or validate your theory? I’m not sure.

In my past paragraph I point out that polls showed a dip in public affirmation that climate change exists and is a problem during a year which was significantly cooler than those preceding. My intention was to illustrate the problem that arises when we encourage the public to judge vastly complex theories like climate change by single weather events.

What’s going to happen in the year that we have abnormally few tornadoes, for instance? We will see scientists try to explain that a single such year holds little significance. We will see scientists try to explain that weather is, itself, an extremely complex system, and that trying to predict what types of weather will result from climate change (complex in itself) is highly difficult and fraught with likely error.

And the scientists will be right. But they’ll be fighting an uphill battle thanks to articles like this one.

sg @46: It doesn’t have much to do with people pointing out – quite reasonably – that record hot temperatures are to be expected in a warming world, and that fires and floods follow a warming, humidifying trend. If the lay observer doesn’t get this, it’s because the media they’re observing from their lay position are conveniently ignoring the facts. That’s not McKibbon’s fault, quite the opposite in fact.

Fires and floods follow lots of different things. McKibbon is doing in this article precisely what climate change “skeptics” encourage: attempt to tie a small set of observations in a single season to climate change theory. This is a double-edged sword.

Martin @48: But decrying what McKibbon wrote is essentially shunting aside the question. You’re only going to ask this question and fund the necessary research if you believe it might be true, so someone needs to point out that it might be true, preferably pretty loudly.

I don’t think research into climate change or weather requires op-eds in the Washington Post to make alarmist claims and give bad advice to the public. I fully agree that it is worth studying whether certain frequencies of extreme weather would be predicted under current conditions in climate change models – but it appears to me that we’re a long way from solid research on that question.

Also, on the fire issue, there is a VERY plausible argument that multi-year awful history of US over-supression/fire policy is a much greater contributor to outbreaks of scary wildfires than global warming is. (See for example the last two San Diego fires). This is similar to the problem which will eventually come about with the Mississippi river. It may very well be that the precipitating event of the Mississippi river project breaking down and flooding huge parts of Louisiana will be global warming influenced. But the reality is that the project is a classic example of non-sustainability that the US isn’t willing to look at–it is going to break down eventually even if we only have normal flood fluctuations.

Sorry Martin didn’t meant to confuse the issue between long term and short term horizons. Could have written that clearer. The long term drop in tornado fatalities since the 1950s is largely due to better detection and was largely accomplished by the early 1960s. The introduction of Doppler radar helped with detection since then. The recent uptick in fatalities (esp. fatalities per tornado) is probably due to 1) sprawl 2) more people living in tornado unsafe housing (ie: trailers) 3) cuts in detection and warning funding at the local level by the NWS during the Bush administration.

Andrew, you implied that the dip in the polls during a cool climate was related to a media phenomenon connected to the cool climate. When someone points to an increase in tornados and suggests its due to an effect of a warming climate you dismiss it as nonsense. You can’t have things both ways.

Sebastian, the Volcano eruptions in Iceland aren’t necessarily statistically significant – depends on the distribution of eruptions. If they’re known to be very low probability events that cluster with high correlation, it’s to be expected that any one volcanic eruption will be followed by others. And most volcano experts I’ve read all seem to be saying that they were expecting this, or that we can expect more eruptions because they’re heavily correlated. So no we can’t agree that they’re statistically significant, or that if they were they had anything to do with global warming, and the suggestion you’re making is intended purely to be rude.

As for fires … i pointed out above that other countries with good forest fire management policies and 30 years of fuel control have experienced unusually large fires. Not everything that happens in the world happens in the US, you know.

Martin, what are you trying to argue? That a step-change in climate conditions has lead to sudden increase in tornados this year? That a step change has lead to particularly lethal tornados? That climate change has gradually made tornados as a whole more likely, making an extreme month also more likely?

Meteorologists seem to see no evidence for any of those, they say that this year falls within the normal range of tornado severity, with unlucky hits as the cause for the deaths. And they have not yet observed a longer trend towards more tornados.

Zamfir, let me qualify: “coincidentally” is doing some work above. If the tornadoes are stronger than average, they typically would have longer path lengths, making them more likely to hit an inhabited area. So it’s not just that a greater percentage of tornadoes hit inhabited areas, but that this still holds after you correct for path length. That would presumably be random.

“If the tornadoes are stronger than average, they typically would have longer path lengths, making them more likely to hit an inhabited area. So it’s not just that a greater percentage of tornadoes hit inhabited areas, but that this still holds after you correct for path length. ”

But what in the world do you mean by ‘stronger than average’? They are within the normal observed range for high intensity tornado outbreaks. Tornado outbreaks of this intensity have occurred previously at least once in the 1920s, twice in the 1930s, once in the 1950s, and once in the 1970s. So of course this year is ‘stronger than average’. But not out of the normal range or indicative of a trend, which is what you want to claim the statistics for. The really odd thing is that we had such a long period without torndaoes outbreaks of this intensity. If you used 1920-1940 as the baseline and compared to 1990-2010 you would be struck by how global warming seemed to have dramatically diminished the number of intense tornado outbreaks. Same thing if you used 1940-1960. And if you used 1960-1980 you would have found one major outbreak. The TREND was to fewer tornado outbreaks, not more.

Now I would have no problem arguing that it wasn’t even a real trend toward fewer outbreaks. I would argue that serious tornado outbreaks are pretty rare, so skipping two decades doesn’t show much. But if that is true, this particular outbreak looks like a return to normal, not a shocking event that needs AGW as an explanation. It only seems shocking because we went through such a slow period.

“If they’re known to be very low probability events that cluster with high correlation, it’s to be expected that any one volcanic eruption will be followed by others. And most volcano experts I’ve read all seem to be saying that they were expecting this, or that we can expect more eruptions because they’re heavily correlated.”

Quite. And look at what Martin does with tornadoes. A tornado outbreak is definitionally a large scale, short-term weather phenomenon which involves hundreds of heavily correlated tornadoes.

[re volcanoes]”So no we can’t agree that they’re statistically significant, or that if they were they had anything to do with global warming”

Spot on again. We can’t agree that this year’s tornado outbreaks are statistically significant. Therefore trying to attribute it to global warming tends toward non-science.

Sebastian, you missed the point. What we can conclude about tornadoes depends on the distribution. They don’t have the same distributional properties as volcanoes, and you don’t know what their distributional properties are, so arguing it “tends towards nonscience” is vacuous.