"Our greatest responsibility is to be good ancestors."

-Jonas Salk

Saturday, April 30, 2011

Spinning Tornados

Adopting Andrew Sullivan's methodology I point you to interesting stuff elsewhere.

There's a huge kerfuffle about attributing severe weather in Alabama to climate forcing. Kevin Trenberth and Peter Gleick among others proclaim "this is the sort of thing"-ism.

It is irresponsible not to mention climate change.The environment in which all of these storms and the tornadoes are occurring has changed from human influences.
David Appel, who gets far too little credit as a pioneer of climate blogging, is, perhaps surprisingly, appalled.
You don't have to look very far to disprove this -- in fact, you don't even have to look farther than the Drudge Report, which today links to this story:
5 P.M. UPDATE: Hundreds treated at DCH
"The loss of life is the greatest from an outbreak of U.S. tornadoes since April 1974, when 329 people were killed by a storm that swept across 13 Southern and Midwestern states."
When are activists going to learn that they will never make their case by falsifying the science, and that, in fact, they only harm their cause when they do so? You cannot draw conclusions about climate based on weather. You can only do it via long-term (decadal or more) statistics.

Please tattoo this on your foreheads, so you don't ruin this for those of us trying to communicate actual, real science, with all its inconvenient unknowns and uncertainties.
Judith Curry, who has many good links, is somewhat more predictably appalled.

I think that we are seeing another instance of excessive attention to "attribution" in a statistical sense. The climate is changing with increasing rapidity. Some of the changes will be anticipated, some not. We shouldn't presume that changes will be locally monotonic. They won't be. Under the circumstances, we'll get extraordinary runs of just-the-sort-of-awfulness-we-get-around-here in various places as the system wobbles about. I mean, what did you expect?

On that basis, +1 Trenberth

h/t for Tuscaloosa F5 tornado video Dan Satterfield. This is right on Dan's turf. Check out his report.


Tom said...

Trenberth's name should be changed to mud merely because of his attempt to redefine the null hypothesis. This is just an extension of that.

Someone should tell him and others that the bad practices being brought to bear on the politics of global warming will stick around long after the food fights are over.

Unknown said...

Deaths are of course tornadoes divided by our response to it, so counting them does not prove much.

Still, I wonder how this graph would look, 15 years later.


It looks like the number of tornadoes at least is going up, on average.

Anonymous said...

Here's one way to look at it.

Humans: Holy shit, 300 people just got killed by tornado. What up with that, science?

Science: Well, the climate is changing so because this is interconnected, we must assume that everything will change within that system.

Humans: Yeah, but what about tornadoes, worse or better?

Science: Some conditions from a warming climate are favorable to fiercer tornadoes, some are not.

Humans: Huh?

Science: The answer is that we don't know, it is uncertain.

Some humans: Okay, let's wait it out until we know for sure.

Some humans: Shit, this is a terrible risk. We better do something.

Some humans: Scientists are advocates and should shut up and they shouldn't get anymore of my tax money.

Some humans: What are the ethics involved here?

Situations change, we learn more, more questions come about, answers remain constant.

Jonathan Gilligan said...

Do note that Trenberth also says at Revkin's place that we also can't look for a climatic signature in decadal (or more) tornado statistics because the long-term statistics are too poor and biased (this is relevant to Victor Eijkhout's comment):

"Statistics on tornadoes are unreliable and exhibit spurious upward trends that are known to correspond to more people being in more places to see them. We have comments on this in the IPCC report. There have been some attempts to adjust them but none are entirely satisfactory."

Tornadoes just don't seem a good system to look for climatic signatures regardless how we process the data because there's a whole lot about them that even the experts don't understand and because unlike lots of other meteorologic phenomena, there's not a really good observational record with which to test theories about climatic effects.

Grypo Saurus summarizes the whole thing very well.

Michael Tobis said...

In my determination not to write meandering and sloppy articles anymore, I have to admit that making the case the way I see it will not be easy, but I've asserted my opinion above, and it is stronger than Grypo's.

I believe that we will indeed see a higher incidence of severe events in sporadic bursts. These will come in two types: "stuff-we've-never-seen-around-here" and "the-usual-stuff-but-worse". Both come from the human habit of associating patterns with places, an association we call "the climate of a place". The climate system wobbles about as we push it harder and harder. So sometimes it wanders to where, at location X, X gets even worse episodes of the nasty stuff X is famous for. (Tornados in the US south and midwest.) Sometimes X gets weather that usually would be seen at Y.

Predicting the details is beyond the skill of current models, and to a large extent is scenario-driven anyway. Attribution is even harder than prediction.

In a sense, statistical attribution is a phenomenon of a stable climate. As we get into a period of ever more extreme changes, we will see one-off events, some of them only one time. The fact that you can't do a statistical attribution doesn't tell you anything.

Absence of evidence is not evidence of absence.

I think what we really need to test this sort of idea is some sort of objective weirdness metric. F5 tornados have existed before. Severe outbreaks, too. But they aren't ordinary events either.

People's perceptions of weirdness may be heightened, so it is not easy to separate the subjective from the objective.

But I believe that there is an increasing anthropogenic wierdness trajectory just the same.

Pending an objective measure, the people to ask are the best meteorologists. Jeff Masters had some interesting things to say about it last summer, you'll recall. (Search for him in my quote gallery, linked at right.)

dhogaza said...


"You don't have to look very far to disprove this -- in fact, you don't even have to look farther than the Drudge Report, which today links to this story:


5 P.M. UPDATE: Hundreds treated at DCH
"The loss of life is the greatest from an outbreak of U.S. tornadoes since April 1974, when 329 people were killed by a storm that swept across 13 Southern and Midwestern states.""

Of course, the death toll has now surpassed 1974 - Appel should understand that one needs to wait until all the wreckage has been searched before the death toll will be finalized. Even then, some people might not be located for awhile - if a pickup-truck has been lifted and then deposited 4 miles away, it's not hard to imagine that a few missing people might have been blown some distance from where they went missing.

Besides, the death, injury and damage metric is a horrible one regarding the rarity of the event. Depends on where they touch down, how much warning people have, and whether they have access to a shelter or are in well-constructed buildings (close to 50% of fatalities in tornados in this country are due to people being killed in mobile homes).

I like Jeff Masters' metrics better:

"The 4-day total of preliminary tornado reports of 346 from this outbreak is close to the 323 preliminary tornado reports logged during the massive April 14 - 16 tornado outbreak. That outbreak has 155 confirmed tornadoes so far, making it the largest April tornado outbreak on record, and 3rd largest in history. The numbers from this week's outbreak may be even higher, giving April 2011 the 3rd and 4th largest tornado outbreaks in history, and the deadliest outbreak in 75 years. According to a list of tornado outbreaks maintained by Wikipedia, only two other tornado outbreaks have had as many as 150 twisters--the May 2004 outbreak (385), and the May 2003 outbreak (401)."

So the fourth largest known outbreaks have happened in April 2011, April 2011, May 2004, and May 2003.

Of course, records are less precise the further back you go but all things tornado have been very unusual this year.

RPJr knows that warning systems are more effective, newer buildings built to be more robust than in the past (though not to the point of withstanding a direct hit by an EF5), etc etc. Tends to lessen the death toll, thank god. It's understandable that RPJr would focus on that rather than actual weather stats as Masters has done.

Adam said...

Who would argue that increasing sensible and latent energy in the atmosphere would not produce more energetic weather events?

Michael Tobis said...

Adam, that isn't conclusive. An engine delivers energy from a more to a less energetic subsystem: it's the difference that matters. And as the poles warm more than the tropics (especially in the Arctic where there is so much less ice) the horizontal temperature gradient goes down. So you might expect less large-scale organization.

So far that isn't panning out, so your intuition may be right. But the point that there is more energy in the system doesn't by itself necessarily mean the energy flows will be more intense.

My argument is that the climate is not well-trained beast, so when you start hustling it from one place to another it will have plenty of wobbles along the way. Basically good weather is "normal" weather more than it is comfortable weather. And the concept of normal weather is gradually going away.

Steve Scolnik said...

As the statistics continue being compiled, the extreme-ness of the event has increased. We don't need no stinkin' decadal stats to know this is off the charts:

2011 More Than Double Previous April Tornado Record, Highest Month of All Time, Highest Deaths of Modern Warning Era (http://goo.gl/tC8aA)

Caveats: Higher population density, better detection and reporting technology, chaotic phenomenon with multiple causal factors, yadda, yadda, yadda

Jim Bouldin said...

That day in 1974 obliterated the town of Xenia, OH, killed a whole bunch of people. I remember it well.

Jonathan, I'm thinking that Compo's recent 20th century reanalysis data might be used to address this--looking not at trends in tornadoes themselves, but trends in conditions favorable to their development. Seems to me it could be used for that.

Jim Bouldin said...

I also remember two other big outbreaks firsthand. The first was on Palm Sunday 1965; one destroyed a neighborhood nearby (north Toledo, OH) and killed maybe 15-20 people. The second was on Memorial Day 1985. Came face to face with a twister that day in rural central Ohio when I was inspecting corn fields and was first on the scene at a few destroyed farm houses. One woman, whose house had no basement, huddled with her 4 kids in their bathtub--the entire house was obliterated and that act saved them all. The humidity that day was unbelievable--you could swim in it. Won't ever forget it.

Adam said...

The humidity that day was unbelievable--you could swim in it. Won't ever forget it.

Rising absolute humidity is the factor that makes me suspect stronger convective events are inevitable with global warming. It is hard for me to understand how lifting more latent heat could do anything but produce more intense storms, unless some other factor were coincidentally changed.

Perhaps, as MT suggests, the flattening horizontal temperature gradient might, for instance, diminish the intensity of fronts that provide the lift, but is there any sign of that?

I agree that current knowledge seems insufficient to make statements about the climate change influence on a single flood or tornado outbreak. On the other hand, an awful lot of crazy shit has been happening lately.

Paul Daniel Ash said...

Perhaps the crazy-shit index could be used as a basis for a >climate prediction market...

Jonathan Gilligan said...

Jim Bouldin: I agree completely that analysis of trends in conditions would be very interesting and relevant.

I don't know the reanalysis you're talking about---could you post a link or citation?

One thing I just don't know about is how consistent measurements of the kinds of things I think of---CAPE, LI, SRH, SWEAT, etc.---are measured consistently enough from one station to another and one year to the next to use them usefully for studying climatic trends.

All of these things are derived from soundings and there have been problems in the past with calibration issues in soundings (think of the whole mid-90s hullaballoo over radiosonde vs. surface temperature trends). After all, the primary motivation of most soundings is short-term forecasting, not climatology.

I'd love to learn more about what people are doing on this because I'm completely ignorant but think it's an important question to ask and very relevant to the topic of this thread.

David B. Benson said...

Are tornadoes anti-tornados in the anti-hemisphere?

Brian said...

David Appell makes a Roger Pielke Jr-style mistake (except for Appell it's clearly just a mistake) when he says "You cannot draw conclusions about climate based on weather."

It's the other way around: you can draw (probabilistic) conclusions about the weather based on climate.

Whether it's correct to predict more/worse tornadoes from climate change is the specific question to focus on.

Jim Bouldin said...


Google NOAA-CIRES 20th Century Reanalysis. Gilbert Compo is the P.I.

Jim Bouldin said...


dhogaza said...

And, then, of course, we have Keith Kloor being a trivializing ass:


susan said...

So scientists have to say it is impossible to attribute any single weather event to climate change, though events like these are likely to increase yada yada yada

Then fake skeptics say - this definitely provides these weather events are not due to climate change.

Time to turn this one around. One side is being honest, the other is exploiting.