There have been lots of claims about hurricanes since Katrina, many of them linking “global warming” to hurricane frequency, which is of course flat wrong.
This study however is in my opinion, probable in its method and results. It is well known that hurricanes spawn tornadoes, lots of them. Creating a tool to predict how many from hurricane size and intensity is a valuable contribution to both meteorology and public safety. – Anthony
From the Georgia Tech Newsroom:
Tornado Threat Increases as Gulf Hurricanes Get Larger
Atlanta (September 8, 2009) —Tornadoes that occur from hurricanes moving inland from the Gulf Coast are increasing in frequency, according to researchers at the Georgia Institute of Technology. This increase seems to reflect the increase in size and frequency among large hurricanes that make landfall from the Gulf of Mexico. The findings can be found in Geophysical Research Letters online and in print in the September 3, 2009 issue.
“As the size of landfalling hurricanes from the Gulf of Mexico increases, we’re seeing more tornadoes than we did in the past that can occur up to two days and several hundred miles inland from the landfall location,” said James Belanger, doctoral student in the School of Earth and Atmospheric Sciences at Georgia Tech and lead author of the paper.
Currently, it’s well known that when hurricanes hit land, there’s a risk that tornadoes may form in the area. Until now, no one has quantified that risk because observations of tornadoes were too sporadic prior to the installation of the NEXRAD Doppler Radar Network in 1995. Belanger along with co-authors Judith Curry, professor and chair of the School of Earth and Atmospheric Sciences at Tech and research scientist Carlos Hoyos, decided to see if they could create a model using the more reliable tornado record that’s existed since 1995.
The model that they developed for hurricane-induced tornadoes uses four factors that serve as good predictors of tornado activity: size, intensity, track direction and whether there’s a strong gradient of moisture at midlevels in the storm’s
environment.
“The size of a tropical cyclone basically sets the domain over which tornadoes can form. So a larger storm that has more exposure over land has a higher propensity for producing tornadoes than a smaller one, on average,” said Belanger.
The team looked at 127 tropical cyclones from 1948 up to the 2008 hurricane season and went further back to 1920 modifying their model to account for the type of data collected at that time. They found that since 1995 there has been a 35 percent percent increase in the size of tropical cyclones from the Gulf compared to the previous active period of storms from 1948-1964, which has lead to a doubling in the number of tornadoes produced per storm. The number of hurricane-induced tornadoes during the 2004 and 2005 hurricane seasons is unprecedented in the historical record since 1920, according to the model.
“The beauty of the model is that not only can we use it to reconstruct the observational record, but we can also use it as a forecasting tool,” said Belanger.
To test how well it predicted the number of tornadoes associated with a given hurricane, they input the intensity of the storm at landfall, it’s size, track and moisture at mid-levels, and were able to generate a forecast of how many tornadoes formed from the hurricane. They found that for Hurricane Ike in 2008, their model predicted exactly the number of tornadoes that occurred, 33. For Hurricane Katrina in 2005, the model predicted 56 tornadoes, and 58 were observed.
The team’s next steps are to take a look to see how hurricane size, not just intensity (as indicated by the Safir-Simpson scale), affects the damage experienced by residents.
“Storm surge, rain and flooding are all connected to the size of the storm,” said Curry. “Yet, size is an underappreciated factor associated with damage from hurricanes. So its important to develop a better understanding of what controls hurricane size and how size influences hurricane damage. The great damage in Galveston from Hurricane Ike in 2008 was inconsistent with Category 2 wind speeds at landfall, but it was the large size that caused the big storm surge that did most of the damage.”
“The study predicted exactly the number of hurricanes seen for Hurricane Ike, 33. (Photo courtesy: NOAA)”
Type-o. Should read “number of tornados”.
Chris
Norfolk, VA, USA
REPLY: Good catch. Fixed. The typo exists on the Georgia tech Press Release also. – Anthony
I can’t question it, but I wonder at the timing.
A point of comparison – Hurricane Camille was a small but very intense storm. See
ttp://en.wikipedia.org/wiki/File:Hurricanes_Camille_and_Katrina_comparison.jpg and
http://en.wikipedia.org/wiki/Hurricane_Camille for comparisons.
What? No mention of global warming?
I expect some enterprising ‘journalist’ in the coming days will correct that serious omission so that the ‘proper’ spin can be placed on this study.
On the positive side, this study gives a more multi-dimensional picture of hurricanes than the usual Safir-Simpson scale description that we’re all too familiar with.
The official Hurricane Summary reports in the U.S. for both Ike and Katrina report 29 tornadoes for Ike and 43 for Katrina, not 33 and 58 respectively reported here. Where did they get their numbers?
This quote: “The great damage in Galveston from Hurricane Ike in 2008 was inconsistent with Category 2 wind speeds at landfall, but it was the large size that caused the big storm surge that did most of the damage.”
It was NOT just the size and breadth of Ike that contributed to the Cat 4 or 5 storm surges….not at all.
It was the extraordinary energy output from this storm that will go down in history as reclassifying storms beyond the limited Saffir-Simpson scale. A new classification where total energy output is more important than windspeed or barometric pressure.
Chris
Norfolk, VA, USA
Whoop di do! Can they predict when and where these tornados will hit? Of course not, so what’s the point?
Frankly, (and I hate say it…) but I’m skeptical about a paper which seems only to state the obvious: big storms with “opportunity” will do big damage by a variety of mechanisms. I’m also highly suspect about historical analysis with regard to storm size, tornado count, and other parameters that will increasingly have only anecdotal confidence the further back you go.
““The beauty of the model is that not only can we use it to reconstruct the observational record, but we can also use it as a forecasting tool,” said Belanger.”
Well, duh… A model built on the observational record can be used to reconstruct the observational record (but with at least a 4% error rate). I hope so! I’m not sure what the value of these kinds of models are, except as parlour games. The observational record already provides the scale data emergency response needs, and standard current meteorological tools and processes will provide the current scale data for an actual storm for emergency response. Any model with any error rate won’t improve the response envelope. That ultimately falls back to the direct observational data arising out of the event.
Reads like this was probably a fairly decent master’s or PhD thesis, but the media hyperbole is trying to turn cotton into silk (I was going to say dress up a pig, but that seemed a little harsh), which brings me to my last point: is anyone else getting a little annoyed at scientific press releases that present wunderbar tools and “settled” science, without attaching the complete foundational research?
We have already passed the peak day of the hurricane season (Sept. 10th) and this year looks to be a very mild tropical storm year.
This Georgia Tech group is known for data analysis (torture) techniques that produce global warming / hurricane hockey sticks. One will have to look at the base data to see what the real story is.
Haven’t we improved, dramatically, our tornado detection ability since 1995?
Tornadoes happen when a big wedge of hot, wet air gets up underneath a large cold air mass. That’s what we see every spring and fall in tornado alley when a norther comes through.
fta: “The size of a tropical cyclone basically sets the domain over which tornadoes can form. So a larger storm that has more exposure over land has a higher propensity for producing tornadoes than a smaller one, on average,” said Belanger.”
Bigger storms make more tornadoes. Wow, that’s a bombshell.
An error in his dataset – he’s assuming that reportage from rural areas about number of tornadoes was just as accurate in 1955, or even 1920, as now. He’s also assuming that storm size measurments from 1948, before the era of satellites, was as accurate as now.
Neither assumption is justified.
How many variables? I could make the elephant wiggle its trunk at you Watts with them.
I’m confused. Are they saying this is due to AGW or what?
Also:
“they could create a model using the more reliable tornado record that’s existed since 1995.”
So…the data comes entirely from the warm phase of the AMO. So there is no way to possibly test whether it has anything to do with AGW.
So, as now there are less hurricanes there are/ will be less tornadoes…and as hurricanes are directly proportional to SST, then we are cooling.
Another statement of the bleedin’ obvious which cost a fortune to produce.
If you read the paper and not just the press release, seems pretty clear where the tornado data came from. Problem with the official hurricane reports from the National hurricane center is that it’s a preliminary assessment of tornado frequency. The actual numbers don’t come out until Storm Data is published.
Also, this paper doesn’t even use the tornado data prior to 1998. It uses the reconstruction of the historical record based on statistical relationships with a set of predictors to reproduce tornado frequency. So wws is incorrect about the assumptions the authors make.
Also, I find it very interesting and consistent that tornadoes are forming farther from the center of hurricane which agrees with the change in size that they found earlier. Sounds like some of you should read the paper and not just the press release before making these comments.
OT: How to acknowledge mistakes? Just as RA Pielke did today in http://pielkeclimatesci.wordpress.com/2009/09/12/error-identified-by-urs-neu-in-one-of-my-interpretations-of-the-results-in-lin-et-al-2007/
That’s classy, clean and scientific, an example for all climate researchers.
The claim that they have a model that agrees with observations is a worthy news article if proven to be factual. This would be a first for climate science! In the past observations needed to be adjusted to fit the model output so this is a step forward.
Here come lots of ifs. If hurricane strength is tied to SST, and if SST anomalies of the hurricane breeding grounds are tied to the AMO, and if the AMO peaked around 2005 and has started its 30-year decline , then will the hurricane strength and tonado frequency decline, too? I guess I could have added a few more ifs if I’d tried.
Since their recent low in Nov 2008, SST anomalies in the Gulf of Mexico have risen again. But they are showing a decline over the past few years.
http://i32.tinypic.com/21l08sl.png
Hmmm. I hadn’t noticed the saw-tooth pattern in Gulf SST anomalies before. Interesting.
The other graphs in my post about SST anomalies in the hurricane breeding grounds haven’t been updated since June 2009, but for anyone interested, here’s a link:
http://bobtisdale.blogspot.com/2009/07/hurricane-breeding-grounds-sst.html
I don’t think they assumed that. They said they made “adjustments” to account for differences in data gathering. But, such adjustments can only be guesswork. You can’t make up data where there is none.
A 35% increase since 1995? When they detection became better? Imagine that!!
And, I don’t know the answer, but I’m sure Nexrad wasn’t instantly in place in all cyclone-prone areas, had to be a gradual process. Wouldn’t that account for a gradual increase in detected tornadoes??
So how does:
“… it’s well known that when hurricanes hit land, there’s a risk that tornadoes may form in the area. Until now, no one has quantified that risk because observations of tornadoes were too sporadic prior to the installation of the NEXRAD Doppler Radar Network in 1995.”, a good but minor and verifiable contribution to weather science, but which would have led to little publicity;
lead (leap) to:
” …since 1995 there has been a 35 percent percent increase in the size of tropical cyclones from the Gulf compared to the previous active period of storms from 1948-1964, which has lead to a doubling in the number of tornadoes produced per storm.”,
and
“The number of hurricane-induced tornadoes during the 2004 and 2005 hurricane seasons is unprecedented in the historical record since 1920, according to the model.”, which not verifiable but leads to headlines?
Comparing pre and post Doppler radar numbers is “apples and oranges” no matter how many adjustments you add to the appearance of the fruit.
Had they shown a increasing linear trend in size and power between 1995 and the present, that would have to be treated more seriously.
Richard111 (22:19:34) :
I can’t question it, but I wonder at the timing.
They need to talk up big hurricanes in theory because for all practical purposes there haven’t been any actual hurricanes this year.
We were all bone tired, soaked, and hadn’t seen the sky in three days. When we had finally cut the broken mainmast overboard, it had taken our best officer with it, and the last man able to sight a position – as if we could see anything of the stars anyway. The radio was not working, in or out, so God only knew where we were then or what lay in our path.
The wind was relentless as wave after wave broke over the battered hull. So loud it was that all conversation was impossible at anything less than a full shout. It would be days before our hearing would return and then came the rain. We had been taking on water even before.
Soon it would be “all hands to the pumps” again, my forearms still barking from the last shift. During the breaks I had been trying to keep the ship’s log so there would be a record of us, if even at the bottom of the sea.
“What do you figure” I yelled again, and this time he answered: “Its blowing like Hell, mate.”
So that’s how I wrote it down, somewhere in the Atlantic, knowing that it was only observations like this that would form the historical record of hurricanes.
I’m missing it . . . bigger storms cause more tornadoes . . . t
It’s like bigger blizzards cause more snow drifts ?
Have the hurricanes actually gotten “larger” , or have they changed the way they measure them ? In other words , are they simply adding significance to the fact that strong winds and high waves are noted eighty or one hundred miles on either side of landfalls , even with a cat 1 or 2 ? Is it also not possible that tornados that occured farther inland might not have been perceived as associated with a hurricane in the past , or possibly not even reported in remote areas ?
…CHECK NHC RIGHT NOW HUGE AREA IN THE GULF … THE BLOB IS BACK…!!!
P Walker (10:51:27) : Here’s a good question-how would you measure the size of a Hurricane at all before satellites, Doppler Radar, before Hurricane Hunter Aircraft???
Out of curiosity, how do we do that now?
Just curious really.
The tornado database (http://www.spc.noaa.gov/climo/online/) is at best a rough estimate of real tornadoes. Storm reports increased exponentially with the advent of the WSR-88D (doppler radar) during the 1990’s which has led to more tornado warnings. Any tornado warning issued by the National Weather Service will result in the local office digging around to verify their warning. This in turn leads to more reports.
Many tornadoes that occur within hurricanes are very weak, often masked by larger scale damaging winds within the eye wall. Given that most tornadoes associated with these tropical systems are 1) short-lived, 2)weak, 3)masked by larger scale processes, it’s impossible to accurately report, much less predict the number of tornadoes. More awareness is the primary reason for the ever increasing reports.
Additionally, tornado genesis is very complicated, and quite honestly not understood. Having said that, given that buoyancy and shear are required for tornadoes, a larger storm would generate a bigger area for potential tornado development. Although there are many cases of “small” hurricanes or even tropical storms that were prolific tornado producers.
wws (06:35:44) : “Tornadoes happen when a big wedge of hot, wet air gets up underneath a large cold air mass. That’s what we see every spring and fall in tornado alley when a norther comes through. ”
Not so true in hurricane spawned tornadoes. Most happen in the front right quadrant at or near land fall. You have a large mass of swirling air basically uninhibited as it travels over water which then runs into land with it’s uneven surface profile, eddies are formed which create these relatively small tornadoes. It has also been discovered that there is much more damage from micro bursts then previously thought.
Since most of us are hunkered down at that point or have already evacutated the area, this info is a nice fact to know but does little else.
All this based on an analysis period of 11 years (re: “this paper doesn’t even use the tornado data prior to 1998” so the assumption must be that we are considering data 1998 onward)? Doesn’t this seem to be a rather slim population of data from which to draw and form inferences?
Is this sufficiently long period to include the variable influnces such as La Nina, El Nino, or PDO and AMO? Not to mention any solar variablities; influnces due to moulation of GCRs (aglatactic cosmic rays) and possible influences on cloud formation?
As an analysis ‘snapshot’, as a case-sample-of-one if you will (stipulated to be over a the 10 yr study period), sure, it (the study) is probably ‘accurate’, that is to say it is “internally consistent” in its methodology … but is it representatve, does it have any correlation to the functions or laws of nature/the laws of science, does it have linkage back to first principles – ?
What can be done with the model(s) developed, besides given -exactly- the same input/existing climate/PDO/AMO/solar conditions replicate the output? Lacking linkage to nature, lacking linkage to first principles, what will a change in un-studied, unprepared-for parameters result in?
.
.
It is well known in the medical world that improved technology and diagnostic techniques lead to apparent increases in the prevalence of all kinds of diseases etc.
Whilst often the MSM make capital out of this the medics themselves have learnt to take a more sanguine view.
Would this were the same for climate scientists!
I’ll have to read the paper, not the press release, before forming an opinion. This may be a pretty good paper.
The interesting part looks to be the reported increase in storm aerial coverage and how that was determined.
I seem to remember that not until Hurricane Andrew went through Homestead FL did we even know that tornadoes were adding to hurricane damage. I think I saw a PBS documentary that had Dr. Fujita flying over destroyed Homestead and pointing out tornado signatures in the debris fields.
John in Florida (07:54:40) : “Also, this paper doesn’t even use the tornado data prior to 1998. It uses the reconstruction of the historical record based on statistical relationships with a set of predictors to reproduce tornado frequency.”
So, you mean, instead of using genuine garbage, they used a model to create new, improved garbage substitute? Wo-ho! I am SO not impressed!
“Tornadoes that occur from hurricanes moving inland from the Gulf Coast are increasing in frequency, according to researchers at the Georgia Institute of Technology”
I’m still waiting for a study that says: “After studying all available data, we conclude that nothing out of the ordinary is going on”. Now there is someone I would listen too.
Has the sun an influence on the number of tornadoes?
I made the following line of thought.
1. See the graph on http://wattsupwiththat.com/2009/08/09/noaa-lowers-hurricane-season-outlook/ . One can see in the graph (Tropical Cyclone ACE Northern Hemisphere May + June + July 1970 – 2009) an obvious solar cycle “impression”. The resemblance with the solar cycle is striking.
A possible scenario of the mechanism can be:
– the very small amount of extra energy of the sun during solar maximum causes more evaporation;
– this results in stronger storms and trade winds.
2. In this thread it is stated by Belanger: “a larger storm that has more exposure over land has a higher propensity for producing tornadoes than a smaller one, on average,” . This means, that, if there are less hurricanes during periods of solar minimum, less tornadoes will be seen.
Possibly, there is some proof of my reasoning in http://www.cimms.ou.edu/~schultz/pubs/verboutetal06.pdf , Fig. 7.
During the solar minima (possibly with a lag of 1 – 2 years) we notice a drastic decline of tornadoes (1966, 1978, 1986, 1999 (in this latest year not convincing)).
Something worth looking into?
http://en.wikipedia.org/wiki/Hurricane_Beulah
“It spawned 115 twisters across Texas, which established a new record for the highest amount of tornadoes produced by a tropical cyclone.”
I’m pretty sure this record has yet to be broken.
Scratch that. I was wrong. Ivan had 2 more than that:
http://en.wikipedia.org/wiki/Hurricane_Ivan
Of course, Beulah could have had more, as Ivan is from the period of more reliable records and is only slightly the record holder.
Wow that is really cool; they can actually predict the number of tornadoes in a storm.
Hey I dodged by just a few years, a storm in the midwest that went generally from St Louis to Chicago (in that direction) and wiped out farm house after farm house, and nobody predicted how many tornadoes were in that storm; hey nobody even observed how many tornadoes were in that storm, but I remember ed it piled dozens of school buses all over the football filed at some school or other; a memorable video I saw from a palne flying over the storm path showed the debris from destrroyed farm houses and it was all lying arcoss the planes track; you see the plane was flying across the storm front; not from St Louis to Chitown; they figured there muat have been hundreds of Tornadoes across that front.
So Nyet on the prediction credibility; more like shovel ready hay that has already been once through the horse.
George
For those of you interested in the possible relationship of AGW and hurricanes, I suggest this article:
Vecchi, G.A., Swanson, K.L., & Soden, B.J. (2008). Whither hurricane activity?. Science. 322, 687-689.
I am currently looking into this question and here is what I have found so far (there are literally <hundreds of articles in the past 5 years about this question):
I see no conclusive evidence that AGW, by itself, is causing an increase in the frequency of Atlantic hurricanes. It appears that ENSO and NAO are still the major factors controlling frequency.
I do see mounting evidence that increasing SSTs can cause more intense hurricanes (more wind, greater precipitation). Keep in mind that the power of wind increases as the cube of the wind speed. So the question then becomes how much higher are SSTs due to AGW, if at all? That question is very debateable.
Scott Mandia (03:24:00) : The key word is “can”. Which is not the same as will.
In fact, the Vecchi, Swanson, and Soden article is quite good. It shows that the increase in Atlantic PDI is as well correlated with the differential warming in the Atlantic compared to other basins as it is with the warming of the basin itself. Which implies to vastly different relationships to AGW.
My personal view has always been that it would come down to something like that. After all, the evidence that intensity is increasing in other basins is sparse.
I’m currently speaking with a graduate student at FSU who is interested in these issues and I’m going to have comments on the various hurricane indices when I get a file from him.
Anthony, if I sent you such an analysis, would you think it would make a good guest post?