Guest post by Paul Homewood
The tornado season remained a quiet one in September. According to data from NOAA’s Storm Prediction Center, during September, there were 43 preliminary tornado reports. This is less than the 1991-2010 September average of 74 tornadoes, and marks the least active September in terms of tornado activity since 2009, when eight tornadoes were confirmed. The majority of the tornadoes were weak and associated with the remnants of Hurricane Isaac as it moved through the Lower Mississippi River Valley and into the Midwest early in the month.
There were also no tornado-related fatalities during September. The below-average tornado activity was similar to the rest of 2012 to-date. The preliminary number of tornadoes during the January-September period was 843 with 119 tornado reports still pending for July, August, and September, marking the lowest January-September tornado count since 2002.
Preliminary reports suggest that 2012 will finish with the lowest number of tornadoes since at least 2005. Up to July, the running total of EF1+ events was 301, and I would anticipate a final year total of just under 400.
Just how unusual would this be? Well, it turns out not very! Since 1970, there have been six years when EF1+ tornadoes totalled less than 400.
| Year | No of EF1+ Tornadoes |
| 1978 | 352 |
| 1985 | 376 |
| 1987 | 316 |
| 1994 | 388 |
| 2000 | 352 |
| 2002 | 311 |
In other words, a quiet year like this one comes along about every six years.
Confirmed numbers for May/June/July
NOAA have now issued final numbers for May through to July, as shown below.
| Month | EF0 | EF1 | EF2 | EF3 | EF4 | EF5 | Total EF1+ |
| May | 88 | 16 | 6 | 3 | 25 | ||
| June | 85 | 23 | 5 | 28 | |||
| July | 25 | 7 | 7 | ||||
| May-July | 198 | 46 | 11 | 3 | 60 | ||
| 1970-2009 Average | 314 | 162 | 54 | 14 | 3 | 233 |
The analysis shows that, not only were total tornado numbers well down on the long term average, but also that, as a proportion, there was a sharper decline in the stronger categories. EF0 storms were 63% of normal, whereas EF1+ were much lower at 26%.
This brings YTD numbers for EF1+ up to 301, as at the end of July, compared to the long term average of 383.
Next month I am planning to take a closer look at a state by state breakdown.
Remember that all tornado reports can be found on their own separate tornado page here.
How were tornados reliably recorded in the 70s? Apparently they cannot be spotted with radar, and it took a couple of decades for the NWS to get its full network of 230,000 tornado spotters in place. Populations were lower in the 70s and instant mobile communication (cells, internet) didn’t exist, so many tornados in remote areas likely weren’t reported. Whereas today, all are captured. Could there be a bias in the historical data?
As an example of possible bias the data, the years from 70-73 inclusive are lower than any other year since, and the period 70-78 is lower than any comparable period since. Improving detection?
Looks interesting. But links to data please.
Since you show some numbers you seem to have direct access to the data , how about a link ?
A link to the NOAA graphs at source would be good.
Also your final graph is bad science. This is similar to plotting annual thermometer temps on the same graph as ice core proxies that average over decades.
Perhaps if you put “error” bars on the decadal averages to show the extent of annual variations in each decade and coloured the annual data differently it would make more sense but plotting apples and oranges on the same graph is always misleading.
2011 looks massively higher but there is no way to see whether this is within the range of earlier decades. The reader will automatically start to be mislead by this graph.
It appears to show 2011 as exceptional but finally there is no way to compare it to the earlier period, so the impression may or may not be misleading. The graph is at best uninformative.
Good summary and commentary. Except. “least active … since 2009” sounds silly. That’s all of 3 years.
Were there really 300 tornadoes in September 2004?
If so what what so different about 2004 that there were 6 times the average number.
@Jason Wilkins
How were tornados reliably recorded in the 70s?
A NOAA study by McCarthy & Shaeffer gives a good summary of the changes over the years They say
” There was a significant increase in tornado occurrence during two periods in the last 33 years – in the early 1980s when National Weather Service (NWS) warning verification began, and in 1990 when the WSR-88D became operational.
The increase in reported tornado frequency during the early 1990s corresponds to the operational implementation of Doppler weather radars. Other non meteorological factors that must be considered when looking at the increase in reported tornado frequency over the past 33 years are the advent of cellular telephones; the development of spotter networks by NWS offices, local emergency management officials, and local media; and population shifts. The growing “hobby” of tornado chasing has also contributed to the increasing number of reported tornadoes.”
http://www.spc.noaa.gov/publications/mccarthy/tor30yrs.pdf
NOAA also say
” This disparity between tornado records of the past and current records contributes a great deal of uncertainty regarding questions about the long-term behavior or patterns of tornado occurrence. Improved tornado observation practices have led to an increase in the number of reported weaker tornadoes, and in recent years the number of EF-0 and EF-1 tornadoes have become more prevelant in the total number of reported tornadoes. In addition, even today many smaller tornadoes still may go undocumented in places with low populations or inconsistent communication facilities.
With increased national Doppler radar coverage, increasing population, and greater attention to tornado reporting, there has been an increase in the number of tornado reports over the past several decades. This can create a misleading appearance of an increasing trend in tornado frequency. To better understand the true variability and trend in tornado frequency in the U.S., the total number of strong to violent tornadoes (EF3 to EF5 category on the Enhanced Fujita scale) can be analyzed. These are the tornadoes that would have likely been reported even during the decades before Doppler radar use became widespread and practices resulted in increasing tornado reports.”
http://www.ncdc.noaa.gov/oa/climate/severeweather/tornadoes.html
This is why I only use EF1+ for long term comparisons.
@ P Solar
Points taken! Previous years’ data is all available from the SPC, who offer CSV files for analysis.
http://www.spc.noaa.gov/wcm/#data
I appreciate the comparison of 2011, with previous decadal data is not meaningful. It was really intended to show how things have changed year on year. To put 2011 into context with earlier years, the report below is worth a read.
https://notalotofpeopleknowthat.wordpress.com/2012/03/31/us-tornado-report-2011/
Data source: http://www.spc.noaa.gov/wcm/#data
counting annual tornadoes of EF1 or greater , we see once again the 60 year pseudo cycle. Here tornado count is compared to AMO. Subtracting annual count form a rough “neutral” value of 500 shows the strong inverse correlation between “global warming” and tornadoes. Warmer temps, less tornadoes.
The shift of 3y is probably because the atmosphere reacts quicker to the common driver than ocean temps to. Although AMO feeds hurricanes, which in turn impact tornadoes.
http://i50.tinypic.com/pxfsn.png
Yet another index showing a change of direction around the millennium.
AMO, ACE, Arctic ice, tornadoes : an irrefutable fundamental change in direction that had sod all to do with CO2.
And this is not looking like a “plateau” folks. Oklahoma would be well advised to prepare for 30 years of increasing tornado activity.
@ AC Osborn
Were there really 300 tornadoes in September 2004?
If so what what so different about 2004 that there were 6 times the average number.
Yes, 299 actually! Most of them seem to have hit the east coast in two episodes, 5th to 8th Sep and again from 15th to 17th.
http://www.spc.noaa.gov/climo/online/monthly/0409_summary.html#
Six were killer tornadoes, with 9 deaths in total. Some were associated with Hurricanes Ivan and Jeannie, so that may be a factor.
Paul, thanks for the response.
Paul Homewood says: “To put 2011 into context with earlier years, the report below is worth a read.”
Indeed the caveats about increased logging of EF0 are significant. However, when I look at EF1+ EF2 + and EF 3+ they all tell pretty much the same story. (Noting the pre 1970 difference in EF1 due to sampling).
http://i46.tinypic.com/nyial1.png
There is a very clear large scale pattern which is shows a definite negative correlation to AMO. The pattern is very similar in all three plots.
The conclusion of that report looks very questionable in view of this graph.
“When the weaker tornadoes are excluded, it is clear that there is very little trend since the 1980’s. ”
HUH??
“Conclusions
Whilst nobody can predict what 2012 will bring, there fortunately seems to be no evidence to suggest that there is any trend towards an increase in numbers or severity of tornadoes in the US.”
I would say there is very clear evidence of a reduction in tornadoes during the 1970-2000 warming phase and the beginning of a reversal of that trend since 2000.
The comment under figure 5 , looking at intensity seems equally dubious.
“… the pattern over the 10 years as a whole does not seem to indicate any real trend, simply going up and down around the historical averages.”
Firstly with 60 years of data why are we being blinkered by being show only the data since 2002 ?? Whenever, I see that sort of thing I known I’m being guided away from something. The so-called “historical” averages” are not that historical if we only look at 10 out of 60 years available.
Despite only choosing to present the flattest section of the data in order to report no “trend”, there is a clear curvature to EF2 and a corresponding opposite curvature in EF3 . This does show a variation in storm intensity even over this period, the least remarkable of all the data available.
To suggest there is no evidence of any trend is misleading to the point of being dangerous. It seems very likely that . contrary to the 1970-2000 drop (that they don’t seem to want to see either) we are seeing the beginning of increasing tornado numbers which will likely be similar to the pre-1970 trend.
In view of the record of 1975 , it would seem important that those states most at danger be warned now rather than be told “there is no evidence”.
Of course that would mean accepting linkage to temperature and noting that this is not a “plateau” or a “pause”, so the AGW mantra prevents the population from being warned of coming mortal danger.
It took me a couple of hours to find this relationship.
“When the weaker tornadoes are excluded, it is clear that there is very little trend since the 1980’s. ”
So if we exclude the trend before 1970, the most significant variation in the *whole record* from 1970 to 1980 and focus on a period were a lesser rise is followed by a downturn, we can pretend there’s “very little trend”.
That looks like little short of deception by omission.
P. Solar says: “.. so the AGW mantra prevents the population from being warned of coming mortal danger. ”
RE reading this , it looks like most of these questionable conclusions are Paul’s, not NOAA. It was not clear at first how much of that was Paul’s commentary and how much came from the NOAA report.
@ P Solar
RE reading this , it looks like most of these questionable conclusions are Paul’s, not NOAA. It was not clear at first how much of that was Paul’s commentary and how much came from the NOAA report
NOAA state
“The bar chart below indicates there has been little trend in the frequency of the strongest tornadoes over the past 55 years.”
The chart referred to, if anything, suggests a decline over that time. (See below)
http://www1.ncdc.noaa.gov/pub/data/cmb/images/tornado/clim/EF3-EF5.png
@ P Solar
So if we exclude the trend before 1970, the most significant variation in the *whole record* from 1970 to 1980 and focus on a period were a lesser rise is followed by a downturn, we can pretend there’s “very little trend”.
It is generally accepted that no meaningful comparisons can be made with years prior to 1970, as recording and observational practices were still being developed. As such it is difficult to compare the 1950’s even with the 1970’s.
e.g. McCarthy & Shaefer
http://www.spc.noaa.gov/publications/mccarthy/tor30yrs.pdf
Paul Homewood says:
>>
NOAA state
“The bar chart below indicates there has been little trend in the frequency of the strongest tornadoes over the past 55 years.”
The chart referred to, if anything, suggests a decline over that time. (See below)
http://www1.ncdc.noaa.gov/pub/data/cmb/images/tornado/clim/EF3-EF5.png
>>
Thanks for the link. I think NOAA making misleading, erroneous statements is a lot more serious than a guest poster here. That’s why I was pretty critical.
“The chart referred to, if anything, suggests a decline over that time” Well, if you insist on trying to summaries 60 years of climate with a major cooling and warming cycle, in one word like “decline” or “trend” I don’ t think it can be of any use.
I would agree there are generally less in the latter half than pre-75 , it is also warmer now.
If, on the other hand, you apply some light filtering, the longer term patterns start to emerge and they are not linear in either sense.
http://i46.tinypic.com/nyial1.png
Also, unless there is a good reason (analytically not convenience) I don’t see the value or validity of only looking at certain months rather than the full year.
The similarity I noted to AMO (inverted) is quite important. Almost opposite to your conclusions, I found that warming periods see less tornadoes and the recent increase matches definite cooling trends in other climate data such as AMO, ACE etc. It is interesting that tornadoes have the opposite dependence on temps that tropical storms do and a few years advance on sea temps and sea temp dependant climate like hurricanes.
I seem to recall Anthony posted on the meteo conditions that cause tornadoes some time back and explained why they were not a sign of warming.
I think if you take another look at the EF2 vs EF3 with a wider time span you may conclude that there is a clear shift. Your 2002-2011 plot shows they are almost a mirror image, suggesting there is indeed a shift first towards EF2 then back to EF3 as cooling sets in.
It was worth bringing this up, now the 2011 data is published since this is yet another climate indicator showing cooling rather than the “pause” in over manipulated surface records.
@ P Solar
I am puzzled at your tinypic graph. It seems to suggest tornadoes at their lowest in the 70’s. Am I missing something?
re. McCarthy & Shaefer
It is well and proper that sampling and other recording effects should be considered but that paper is light on assessing whether the numerous factors did have an effect. They talk of extra monitoring coinciding with an increase in recorded tornadoes in early 1980s be don’t seem to comment on the fact that there was also an increase in fatalities at that time.
both the 80’s and early 90’s show strong changes (falls in my inverted plot) that are almost identical even when looking at EF3+ only , which NOAA states were probably accurately reported anyway. Both those oscillations correlate strongly with inverted AMO .
there seems to be mindset to minimise any possible climatic link in official climatology.
Seeing the post 2000 turnaround and its implied signal of global (at least NH) cooling that is perhaps not surprising.
Paul Homewood says:
October 22, 2012 at 10:33 am
@ P Solar
I am puzzled at your tinypic graph. It seems to suggest tornadoes at their lowest in the 70′s. Am I missing something?
You have probably not noted the legend. eg. ” 500- EF1+ ” these data are _inverted_ tornado frequencies.
This was done in the initial plot to compare to AMO variation.
I should probably have emphasised that in the title once I took out AMO . The main point was the similarity of the different sub-categories but sorry for not making that clearer.
Same thing in the conventional sense. 😉
http://i50.tinypic.com/15d2h78.png
As a side note it is interesting to see that there is a slight phase lag in the EF3+ category during the warming period at the end of the century. As a guess I would say it was the later response of tropical storms to whatever is the ultimate driver of both, since they lag by about 2y.
This may delay the build up of the conditions required to form tornadoes and hence the onset of more powerful events.
This lag appears to be reversing as we move into cooling.
Strange there wasn’t more comment on this. The inverse linkage of tornado frequency looks clear and significant. Especially since it seems to be one of early responders of the climate system.
http://i50.tinypic.com/15d2h78.png
P Solar (and anyone else!)
I have developed a database of tornado stats going back to 1970. If you need any data, just log into my site.
http://notalotofpeopleknowthat.wordpress.com/
Paul, what’s going on here?
I decided to have a look at this question of intensity change and came back to refer to what you had posted. It seems there has been some silent retractions going on both here and at the copy on your site.
While it is good to see that you took note of my comments and criticisms, I don’t think it is very transparent to do so without comment and without showing the changes.
Many of my comments now refer to nothing that is visible. Worse still, the graph where you had looked at variations with EF category was valid and DID actually show something interesting about climate. It is simply your comments that did not seem coherent.
The “conclusions” I commented on are gone, figure 5 is gone and none of the figures are numbered any more.
Sorry but this sort of think smacks of septicscience and is not what I am used to seeing on WUWT.
Are the EF2 ansd EF3 plots you posted still available somewhere?
Thanks.
P Solar
I’m puzzled! I have not amended anything.
I think you might be referring to the 2011 Report, not the one in this article. I gave you the link for this (see below) and I know you made several comments about. (And it has a Figure 5!!)
Let me know.
Thanks
Paul
http://notalotofpeopleknowthat.wordpress.com/2012/03/31/us-tornado-report-2011/
Paul, sincere apologies. It is indeed that link you provided at October 22, 2012 at 3:44 am
That is the figure 5 I was thinking of. The report was of similar layout and I was confusing the two.
Anyway, you have raised my interest in tornado count, which I think is probably an early indicator or the current cooling trend. This negative correlation does not seem to get much (any) official recognition so I think it’s worth digging into.
Sorry for not paying more attention.
My impression of complementarity of EF2 and EF3 in figure 5 was correct.
In fact this is not too surprising since, having eliminated EF0 and EF1 the vast majority of the remaining events will be EF 2 or3 . However, as I suggested , looking at the full range of data is more informative than 2002-2010.
http://i50.tinypic.com/2mhhqjd.png
Here I have used non detrended AMO as an indicator of NH surface temps to get the big picture, not to suggest AMO is a driving force of continental US tornadoes.
Several important climate characteristics can be noted.
1. EF2 and EF3 make up the majority of events, thus their variations on both long and short time scales are very similar.
2. During the warming period 1975-2000 the decadal scale variations in SST show a strong anti-correlation to the proportion of EF3 events ( hence +ve correlation to EF2 proportion). ie the HOTTEST part of these fluctuations shows considerably LESS of the stronger events.
3. On the multidecadal scale, during the same period, there is a lesser, opposing tendency towards a higher proportion of EF3 with increasing temperature. The total magnitude of this drift is about a quarter of the short term peak to peak: 0.03 vs 0.11
4. The smaller decadal variations outside this period tend to show the inverse relationship.
5. The pre-75 cooling period shows some down turn , clearer in EF2 plot, but data does not go back far enough to catch the beginning of this tendency.
6 The underlying , long term trend seems to peak around 1995 and drifts down since.
7. The coolest period in early to mid 70’s was the most extreme in all EF categories. The warm peak around 2002-2003 was the least active.
The complex and opposing response to short and long term trends in temperature may have confounded earlier attempts to draw a simplistic conclusion as to whether there were more tornadoes “due to global warming”.
In relation to 3. above the variation is 0.22 +/- 0.055 ie +/-25% , so there is a substantial change in the proportion of EF3 vs EF2 on the decadal scale that is _negatively_ correlated to temperature.
Here is the inverted EF2+ and EF3+ (inverted) tornado counts compared to HadCRUt v3 and v4
http://i45.tinypic.com/oavqea.png
This big picture is a very clear inverse correlation between tornado count and surface temps.
Again, as with shifts from EF2 to EF3 above, the apparently inverse effect during the oscillations of the warming period 1975-2000 , however, closer inspection suggests this is more likely a 2 to 3 year lag in the ocean dominated surface response.
Flatter periods around 1960 and post 2000 are quite tightly in phase. Larger swings starting with the 1970 dip show a lagged response.
This lag also suggests that the two physical phenomena are responding to the same driver rather than tornadoes being driven the globally averaged temperature.
Comparing tornadoes to hadCRUT3 we see the long term downturn in temps is again lagged by a few years. Tornadoes would appear to give an earlier indication of the change in direction.
Recent attempts to warm up the post 2000 period in hadCRUT4 tend to look spurious when compared to the independent physical climate signal.