by Judith Curry
With the tragic damage and loss of life in Joplin, the tornado madness continues unabated. Here is the latest from Roy Spencer, Bill McKibben, and Joe Romm.
Rather than spouting “gut feelings” and making back of the envelope arguments, Roy Spencer has actually analyzed the U.S. tornado data since 1950, focusing on the largest F3-F5 tornadoes.
The bottom panel of following graphic shows what most meteorologists already know: there has been a downward trend in strong (F3) to violent (F5) tornadoes in the U.S. since statistics began in the 1950s. As seen in the top panel, this has also been a period of general warming. For those statistics buffs, the correlation coefficient is -0.31. Obviously, the conclusionshould be that warming causes fewer strong tornadoes, not more. (Or, maybe a lack of tornadoes causes global warming!)
So, how reliable is the tornado data? I actually looked at the U.S. tornado data set in this paper, and inferred (like many others) that there is a substantial undercounting in the tornado database prior to 1990 and particularly in the earlier decades. The main issue in the pre-Doppler radar and pre Weather Channel days was simple lack of identification, particularly the weaker ones. Circa 1970, the F- tornado classification was introduced by Ted Fujita at the University of Chicago, who had a personal research interest in the strongest tornadoes. Classification of the strongest tornadoes was done operationally after 1972, but earlier ones were classified retroactively. So, there may be some undercounting of the the strongest tornadoes and potentially some misclassification prior to 1970. Such potential data problems in the earlier part of the record would make Roy Spencer’s negative trend even more pronounced!
At Climate Progress, Joe Romm has assembled a number of opinions on this, including Jeff Masters, who stated:
In summary, this year’s incredibly violent tornado season is not part of a trend. It is either a fluke, the start of a new trend, or an early warning symptom that the climate is growing unstable and is transitioning to a new, higher energy state with the potential to create unprecedented weather and climate events. All are reasonable explanations, but we don’t have a long enough history of good tornado data to judge which is most likely to be correct.
Looks like Jeff forgot to look at the U.S. historical tornado data set.
Bill McKibben ties it all together with this essay posted at Salon:
Caution: It is vitally important not to make connections. When you see pictures of rubble like this week’s shots from Joplin, Mo., you should not wonder: Is this somehow related to the tornado outbreak three weeks ago in Tuscaloosa, Ala., or the enormous outbreak a couple of weeks before that (which, together, comprised the most active April for tornadoes in U.S. history). No, that doesn’t mean a thing.
It is far better to think of these as isolated, unpredictable, discrete events. It is not advisable to try to connect them in your mind with, say, the fires burning across Texas — fires that have burned more of America at this point this year than any wildfires have in previous years. Texas, and adjoining parts of Oklahoma and New Mexico, are drier than they’ve ever been — the drought is worse than that of the Dust Bowl. But do not wonder if they’re somehow connected.
If you did wonder, you see, you would also have to wonder about whether this year’s record snowfalls and rainfalls across the Midwest — resulting in record flooding along the Mississippi — could somehow be related. And then you might find your thoughts wandering to, oh, global warming, and to the fact that climatologists have been predicting for years that as we flood the atmosphere with carbon we will also start both drying and flooding the planet, since warm air holds more water vapor than cold air.
It’s far smarter to repeat to yourself the comforting mantra that no single weather event can ever be directly tied to climate change. There have been tornadoes before, and floods — that’s the important thing. Just be careful to make sure you don’t let yourself wonder why all these record-breaking events are happening in such proximity — that is, why there have been unprecedented megafloods in Australia, New Zealand and Pakistan in the past year. Why it’s just now that the Arctic has melted for the first time in thousands of years. No, better to focus on the immediate casualties, watch the videotape from the store cameras as the shelves are blown over. Look at the news anchorman standing in his waders in the rising river as the water approaches his chest.
Cumulative catastrophic weather events are being used to support the case for global warming action. Sorry Bill and Joe, but we need to look at each type of extreme event, in different regional locations, and then interpret them in the context of the local historical records, and then cumulatively in context with the teleconnection weather regimes and multi-decadal oscillations. Once we’ve done that and then find an upward trend in frequency and/or intensity that cannot be explained by problems with the data record or natural climate variability or weather roulette, THEN lets talk about the potential impact of global warming.