Fewer storm days but tornado-clustering a trend? Implications? Plausibility longterm?

[Jason Harris]

Thought people might have some thoughts and speculations of interest on this article.
__

In the new study, published today in Science, Brooks and his colleagues tallied U.S. storms from 1954 to 2013, leaving out the small F0 twisters. Then they looked at the days on which those storms occurred. They found that the frequency of tornado days has declined over that time. In 1973, for instance, tornadoes formed on 187 days. By contrast, 2011 saw twisters on only 110 days—but nine of those days saw more than 30 tornadoes each.

“In effect, there is a low probability of a day having a tornado, but if a day does have a tornado, there is a much higher chance of having many tornadoes,â€￾ the researchers write. Now, about a fifth of a year’s cyclones occur on just three days of that year.



Read more: http://www.smithsonianmag.com/scien...climate-change-180953069/#IfsgEOXHRsCWszHJ.99
Give the gift of Smithsonian magazine for only $12! http://bit.ly/1cGUiGv
Follow us: @SmithsonianMag on Twitter

http://www.smithsonianmag.com/scien...nging-united-states-climate-change-180953069/

 

[Royce Sheibal]

In my undergrad degree, I'm half climatologist, but if we have any doctoral climatologists they might give a better answer. I'll try anyway.

There are two explanations for this. The first is technology. In the past, many tornado days with few tornadoes were likely not tornadoes, but misreported wall clouds with F1+ straight line winds. As reporting has gotten better, these are now correctly identified, therefore reducing the number of tornado days. Meanwhile reporting has also increased the number of total tornadoes that are visible, because of the more spread out populations in the suburbs, easily available digital recording, storm chasers, cell phones, etc. So tornado days go down, but reported larger tornadoes goes up, due to human perception.

Answer #2 is Anthropogenic Climate Change (which is the correct wording if you wish to use it). Of all the consequenes of climate change, more violent tornado outbreaks was one prediction. More hot vs cold, more moist vs dry, more ingredients essential for big tors. The inverse issue is that overall warming will likely cause it to rain on fewer days, due to capping and mixing. The end result is that fewer storm days will occur due to warming / drying issues, but on days that due break the situations may be more extreme. A moderately slow spring in Nebraska turned into 3 days of hell (6-16 to 6-18) with F4's on every day, for instance. The complete lack of tornado watches in tornado prone parts of OK and KS this year might be another good example, while only 3 years ago we had massive tors multiple days around OKC.

Or maybe its a mix of both?

 

[MClarkson]

They found that the frequency of tornado days has declined over that time.

I would think the statistical significance of such a claim is marginal at best.

 

[Warren Faidley]

I believe this is partially a result of increased inversions covering a larger area to the west and a fickle jet stream that likes to drop in for a big event then depart. There does seems to be fewer of the classic, individual supercells producing tornadoes. I'm guessing there are more set-ups that involve "loaded" days where the cap is breached by strong dynamics, thus resulting in multiple, violent tornadoes.

W.

 

[Rob H]

I tend to have a hard time believing that changes in reporting due to advances in the field, population sprawl, etc. don't play a significant factor in this variability but the article claims that to be the case. The study's abstract makes it seem like a "here's why we're having a hard time matching climate change and weather patterns" and less of a "climate change is driving weather patterns" which seems to be how the Smithsonian is presenting it.

I'd love to look at the actual paper, as I'm sure Brooks/Carbin/Marsh wouldn't put out a sloppy study.

 

[calvinkaskey]

The sun isn't shooting out more energy with global warming so with the warming that is occurring near the surface, etc. there should be cooling higher up. Any studies on this? If thunderstorms or weather gets involved in this cooler layer it seems like it could lead to trouble.

 

[Rob H]

The sun isn't shooting out more energy with global warming so with the warming that is occurring near the surface, etc. there should be cooling higher up. Any studies on this? If thunderstorms or weather gets involved in this cooler layer it seems like it could lead to trouble.

The cooling higher up is offset by all the sodium pentothal and fluoride being pumped into the atmosphere by the Reptoids that have infiltrated our government - long story short, nothing will change.

 

[rdale]

I would think the statistical significance of such a claim is marginal at best.

Did you read the article? Or the related paper earlier this year? The stats look pretty good (and the authors writing them don't support tornado walls )

I tend to have a hard time believing that changes in reporting due to advances in the field, population sprawl, etc. don't play a significant factor in this variability but the article claims that to be the case.

Remember the article says we are having LESS tornado days. With more chasers in the field, and more iPhones, if tornadoes were occurring on those "non-normal" days they are being caught. An EF1 in the middle of Podunk, Arkansas in 1966 on an otherwise quiet day never got reported (or if it was reported, it wasn't logged as a tornado.)

The sun isn't shooting out more energy with global warming so with the warming that is occurring near the surface, etc. there should be cooling higher up. Any studies on this? If thunderstorms or weather gets involved in this cooler layer it seems like it could lead to trouble.

Huh? None of that makes any sense.

 

[MClarkson]

Given the very large standard deviations of tornado occurrence data, the number of data points(years) involved is not that large. The statistical significance is thus reduced. I don't see the raw numbers reported anywhere thats not behind a paywall, but they would have to be very blatant to make up for this.

 

[Todd Lemery]

I read the article and it seems a pretty tight window to draw any conclusions from. Also considering that temps haven't increased in the last 15-17 years (depending on how measured) any fluctuations in tornadic activity would seem to reflect nothing more than natural variations. The article points out that larger storms ( f1 to f5) have remained constant and only the smaller storms (f0) have increased. That's easily explainable by the huge increase in trained storm spotters and the ease of which storms can be reported via the internet. The only thing I can pull from the article is that nothing has really changed. The article goes on to state that tornado days are decreasing. They make their point by cherry picking 1973 having tornados on 187 days while 2011 only had 110 days. You can't cherry pick two years to make a point. If there was a 20-40 year trend in increased violent tornados you could get my attention, but this seems to be much ado about nothing. There is nothing in the article that leads me to believe that there has actually been any long term changes in the frequency or intensity of tornados in the US. By that matter, I don't think you can come to any conclusions on the coming year from that article either.

 

[rdale]

I read the article and it seems a pretty tight window to draw any conclusions from.

Just to be clear - you read the news article about the research? Not the research? Because...

They make their point by cherry picking 1973 having tornados on 187 days while 2011 only had 110 days. You can't cherry pick two years to make a point. If there was a 20-40 year trend in increased violent tornados you could get my attention, but this seems to be much ado about nothin

...the research did not just cherry pick two years. There is a 20-40 year trend in decreased numbers of tornado days. Which cannot be explained by spotters, because (if anything) there should be more tornado days because of more spotters and more iPhones.

 

[rdale]

In addition, using a different method, an independent team also came up with the same conclusion.

http://link.springer.com/article/10.1007/s00382-014-2277-3

 

[Greg Blumberg]

Warren Faidley, that's an interesting hypothesis that could be tested. I seem to remember Howie Bluestein saying something several years back about drought in the NM, TX areas lends to stronger elevated mixed layers (EMLs) and he linked drought to higher probabilities of cap busts. One of the tenants of climate change that I've often heard is that "the wetter places will get wetter and the drier places will get drier", which ties the effects of climate change to Howie's observations regarding drought and the observed frequency of severe storms (keep in mind what I'm suggesting here is an educated guess ONLY.) A stronger EML could have two potential scenarios: 1) increased rate of failure for the storms to initiate due to the stronger stability (but only if parcel moisture content is constant) and 2) storms that do initiate will encounter steeper lapse rates and have stronger vertical accelerations, however there are no significant relationships between EML strength and tornado strength I am aware of. Since the EML is like a traditional synoptic scale airmass, perhaps the EML contributes to a larger area of tornado ingredients though (i.e. more horizontal area of tornado ingredients). Either way, because of these two things I mentioned, a stronger EML could be a culprit behind the decreased tornado days yet nearly constant number of tornados per year Brooks et al. observed.

I would not say that Brooks et al. are cherry picking those two days. They cite those two days as an example. They have a key plot that shows lines for a) the number of days per year with at least one (E)F1+ tornado and b) number of days per year with more than 30 (E)F1+ tornadoes over time (1955 - 2013). Line A surely shows a decrease between 1980 and now, line B shows an increase since 1955. While there is the chance that this signal is an artifact of non meteorological factors in the database (which I've cited the authors below), they come to the conclusion that it's unlikely due to the fact that there must be a combination of other factors all acting to bias the data in the same direction.

I hope this is okay to cite the paper here, but here's what they say (emphasis mine):

"The fact that the two series move in opposite directions is very difficult to explain as a result of changes in how they are reported. Increased likelihood of reporting [(my words here): related to the population increase over time and technological reporting] should lead to an increase in both the probability that a day has at least one (E)F1+ tornado and that a large number occurs on any particular day. It still is possible that the observed changes are the result of some nonmeteorological factors, but that would require a complex set of influences acting in opposite directions, making it exceedingly unlikely. Unfortunately, the nature of the database makes it impossible to be absolutely certain."

 

[Rob H]

Remember the article says we are having LESS tornado days. With more chasers in the field, and more iPhones, if tornadoes were occurring on those "non-normal" days they are being caught. An EF1 in the middle of Podunk, Arkansas in 1966 on an otherwise quiet day never got reported (or if it was reported, it wasn't logged as a tornado.)

Maybe I didn't explain what I was thinking well enough. You have a value that is not definite - we can't capture every tornado, we have to use reports. Less tornado days in 2000-2010 compared to 1980-1990 is not a very useful comparison because they're different time periods. The only way we could accurately measure the effect of exurban sprawl and better reporting is to replay 1980-1990s weather today and compare the reports.

So yes, we are having less tornado days and a higher relative frequency of strong tornadoes. How are they able to rule out non-climate related causes though? That seems like it would be a difficult statement to make.

edit: Ok, this is useful information cited from the paper, emphasis mine:

It still is possible that the observed changes are the result of some nonmeteorological factors, but that would require a complex set of influences acting in opposite directions, making it exceedingly unlikely. Unfortunately, the nature of the database makes it impossible to be absolutely certain.

Pretty sure we just had CDC people saying Ebola was exceedingly unlikely to spread in the US. Sometimes unlikely things happen.

edit: VVVV Ebola did spread to two healthcare workers. So that was something unlikely that certainly did happen. Besides the point, though..

 

[rdale]

tornado, we have to use reports. Less tornado days in 2000-2010 compared to 1980-1990 is not a very useful comparison because they're different time periods.

Given the externalities at play (i.e. more people watching and recording tornadoes) we can "infer" (and oftentimes when experts do it, that's reliable.)

I suppose you could test it... Ignore all cell phone video tornado reports and all reports gathered via social media, submitted online, or obtained electronically. That would give you a better idea? I doubt it would INCREASE the number of tornado days...

So yes, we are having less tornado days and a higher relative frequency of strong tornadoes. How are they able to rule out non-climate related causes though?

Not sure what you mean... What other causes could there be? (NOTE: They do not blame CO2 emissions for this change.)

Pretty sure we just had CDC people saying Ebola was exceedingly unlikely to spread in the US. Sometimes unlikely things happen.

But as we're seeing now, Ebola is not spreading in the US so oftentimes unlikely things don't happen