Violent Tornadoes Are On The Decline In The US

By Paul Homewood


As we, fortunately, head towards the end of the third quiet tornado year in a row, let’s take a closer look at the intensity of tornadoes in the US.

We often hear claims that tornadoes are growing stronger as a result of global warming. But what do the facts tell us?

NOAA’s Storm Prediction Center maintains a database of tornadoes back to 1950. However, it is generally accepted by tornado experts that data is unreliable from the 1950’s and 60’s, which were known as the “tornado growth period”, as observation practices began to develop.

Another problem during these early decades was that, according to the SPC’s Greg Carbin, there were too many higher-rated tornadoes because of post rating.

Therefore, any analysis can only be reliably started from 1970.

It is also well accepted that many more small EF-0 tornadoes are spotted nowadays, that would have gone unnoticed previously, so these too need to be excluded from any analysis.

[Quick note –  The EF-Scale was introduced in 2007, to replace the F-Scale. All above references to EF-Scale refer to both.]

From the SPC data, we can therefore plot the annual number of EF-1+ tornadoes.



Figure 1

As we see, the number of tornadoes has been on a declining trend.

But within this total, are stronger tornadoes on the increase? We can plot the percentage totals of each category.


  Figure 2


Figure 3


Figure 4


Figure 5

The results are perfectly clear – the weakest EF-1 tornadoes have been increasing as a proportion, while all the other categories have been declining. (I have not shown EF-5’s separately, as these are such a small number, about one a year, as to make any trends meaningless).

Can we rely on tornado classifications being consistent over time? There is no easy answer to this, but when the Enhanced Fujita system was introduced in 2007, it was stressed that the new system was designed to ensure compatibility with the original Fujita scale.

In any event, NOAA themselves still publish official analyses using these classifications, stating:

To better understand the variability and trend in tornado frequency in the United States, the total number of EF-1 and stronger, as well as strong to violent tornadoes (EF-3 to EF-5 category on the Enhanced Fujita scale) can be analyzed. These tornadoes would have likely been reported even during the decades before Doppler radar use became widespread and practices resulted in increasing tornado reports. The bar charts below indicate there has been little trend in the frequency of the stronger tornadoes over the past 55 years.


Analysis of long term climatic trends can often be fraught with these sort of data issues, whether tornadoes, hurricane or temperatures. Nevertheless, the official data is the best we have got. And its message is clear.

Tornadoes since 1970 have been declining both in number and intensity.


All data from the SPC

51 thoughts on “Violent Tornadoes Are On The Decline In The US

    • +1
      The watermelons at the BBC seem, still, to be pushing the ‘irrefutable’; ‘illimitable’; ‘unrecoverable’; ‘unprecedented’; unanimous; unavoidable memes, but with little or no evidence [and an occasional ‘not for 1400 years’ – so it was, in fact, warmer in Roman times, and they had horse farts, I guess, not SUVs].

      There is a lot to be said for living in the UK/Britain/GB/England ( and a lot of people want to come here) – but some of our politicians are a tad trying, shall we say!

      And tomorrow – it’s Monday morning, and back on the 0647 up to Town – seats available at the start of the journey.


    • Not at the Guardian.
      They have 6 graphs that show we are doomed from the IPCC.

      Just try commenting there on about the impacts of emissions before 1950 and the higher emissions after 1950 – the relative trends. Fingers in the ears while the backroom boys hide the evidence.

      Graphs speak no words to the willingly deaf. It isn’t about the facts to the true believers.

  1. There should be some test of significance applied.

    You don’t explicitly say what method you used to determine the trend, so I am assuming ordinary least squares. A better choice to examine the trend would be Kendall’s tau. The latter test is not dependent on an assumption of linearity in the trend, does not require assuming a normal distribution to the data, and is also far less susceptible to anomalous values.

    My preference would also be to evaluate the numbers of various strength tornadoes rather than the proportion of the total, especially since detection of the weaker 0,1, and 2 storms has improved with the development of radar. Evaluating them all by proportion allows the change in the ability to detect the weaker storms to affect every analysis.

    • Not necessarily. If the raw data is fairly tight, OLS is a great choice. If the data is noisy than I agree with you.

      • Look at the EF3 data (the last chart, unnumbered). High of around 130 with two data points at around 100 and an eyeball average of around 30. Too noisy for least squares. Dominated by 1974.

        Look at figure 3, peak of 5, an average below 2, dominated by a possibly anomalous 1974.

        Look at figure 1. There are two values around 900, a few at 700. If anything the one in 2011 will obscure a declining trend.

        There are too many possibly anomalous values that would obscure an underlying trend to use least squares. If then you are going to use Kendall for those you should do it for all to be consistent. even though Figures 2 and 3 appear to be free of anomalies.

        Kendal would be far more convincing in detecting the underlying trends with this data.

    • Chris 4692: you are over-thinking this.

      The hypothesis is a simple, linear hypothesis: across time, the global warming is making there be more tornados, and more of the bad tornados.

      All that is needed to disprove this is a histogram – if the visible trend is in the opposite direction.
      The visible trend is in the opposite direction,
      No regression line is needed.
      No significance test is needed.

    • A tight fit has to do with residuals (how far away or discrepant the raw data is from an imaginary perfect linear fit). The residual sum of squares (RSS) is the sum of squares of residuals. It is also known as the sum of squared residuals (SSR) or the sum of squared errors of prediction (SSE). It is a measure of the discrepancy between the data and an estimation model (a perfect linear fit with regard to OLS). A small RSS indicates a tight fit of the model to the data.

      A residual plot is a graph that shows the residuals on the vertical axis and the independent variable on the horizontal axis. If the points in a residual plot are randomly dispersed around the horizontal axis, a linear regression model is appropriate for the data; otherwise, a non-linear model is more appropriate. Residuals that occasionally fall beyond 2.5 standard deviation is okay. Too many and you may not be able to draw a straight line with a straight face. If the residuals plot up in obvious curves, a linear model is not the right choice.

      If weather pattern variations are better understood and observed as multi-decadal with oscillations above and below the statistical “normative” period, we over-use OLS in describing temperature change.

    • FYI…When doing our tornado research we generally used the Mann-Whitney-Wicoxon test to compare different EF populations. It is a ranked non-parametric test and does not assume a normally distributed population as a t-test does. There is no reason to assume tornado intensities are normally distributed.

    • When I cut a board to replace a bad one on my house, there is not a lot of leeway in measuring the length of the board, whether in inches, centimeters, a broom, or with a piece of string, the board is the right length or it is not.

      Climate data trends depends on how and what data is analyzed, there is no length to the board, the length of the board is like the mystical sound of one hand clapping. The severe weather claims are being made by climate alarmists, using no scientific method but using instead compound speculation. Make a speculataion, presume it’s proven, build a specualtion on top of that one, presume it’s proven, make a speculation based on that, presume it’s proven, etc. The point being this data can be manipulated up and down until the cows come home, it is literally nothing to build local, regional, national or global policy on.

  2. The data is conclusive:
    Global warming decreases tornadoes.
    But this is empirical data and so it does not count with the modeling confraternity and their adherents.

    • Here is Dr. Spencer on this issue.

      Dr. Roy Spencer – April 29th, 2011
      MORE Tornadoes from Global Warming? That’s a Joke, Right?
      …..If there is one weather phenomenon global warming theory does NOT predict more of, it would be severe thunderstorms and tornadoes.

      Tornadic thunderstorms do not require tropical-type warmth. In fact, tornadoes are almost unheard of in the tropics, despite frequent thunderstorm activity……

      For example, the poster child for active tornado seasons was the Superoutbreak of 1974, which was during globally cool conditions……..

      • So this begs the question…is it the gradual warming of the middle Atlantic that has caused the trend? Can we expect more storms after the AMO shift?

    • Time and Time again it’s the same old story. More tornadoes and drought for extras.

      Time Magazine – Monday, June 24, 1974
      Another Ice Age?
      In Africa, drought continues for the sixth consecutive year, adding terribly to the toll of famine victims…….

      ……. Scientists have found other indications of global cooling. For one thing…….

      As the winds swirl around the globe, their southerly portions undulate like the bottom of a skirt. Cold air is pulled down across the Western U.S. and warm air is swept up to the Northeast. The collision of air masses of widely differing temperatures and humidity can create violent storms—the Midwest’s recent rash of disastrous tornadoes, for example. …….,9171,944914,00.html

  3. We often hear that warming the atmosphere will increase the energy for these sort of systems. Is this true?

    Actually, it’s the complete opposite. Warming the planet, with more warming in the higher latitudes provides LESS energy for violent tornadoes.

    Isolated cases of violent tornadoes have been reported in some intense hurricanes but the majority of them, are associated with a powerful jet stream that is linked to a great contrast in air masses.
    The greater contrasts are usually defined by the cold side.

    The greater temperature gradients cause bigger pressure differentials and stronger winds. Winds redistribute energy. When you have global warming of higher latitudes, less energy need to be redistributed.

    This real science is what they should be teaching in our elementary schools, not that humans burning fossil fuels are creating more extreme weather.

  4. I would expect an active tornado season in Spring-Summer 2015.
    2 reasons:
    1. weak El Nino, pushing some modest moisture across Mexico into Texas-Oklahoma tonado alley.
    2. cooler Arctic —> higher delta T between northern and mid latitudes.

  5. Yeah, but the decline in tornadoes is due to the increase in tornadoes due to global warming, which is exactly what the models predicted.

  6. I don’t see any news coverage on this or agency press releases on this or the pause in hurricane landfalls. I guess the political payola system only works in one direction.

  7. Children in the future won’t know what a violent tornado looks like…..along with snowy UK winters and wild poley bears in the arctic…its all very sad…

    • I’m hoping the Arctic will warm so much that we’ll have wild pole dancers instead of polar bears.

  8. Tornado ratings are given by weather service employees who survey the damage, and it is very subjective. It is difficult for them to tell the difference between F0, F1, or even strong straight line winds. The increased percentage of EF1 tornados could simply be the changing culture of the people doing damage surveys.

    • In climate science there are significant issues with both the data and the methodology. Your specific example points out one of the many issues with the data side of Climate science.

      At times I wonder if climate forecasting is really at the same level as gypsy fortune telling except with mathmatics and super-computers.

  9. Tornadoes since 1970 have been declining both in number and intensity.
    That states the facts succinctly, with good supporting graphics and analysis as well.
    Thanks Paul – Nice work!

  10. Reblogged this on Public Secrets and commented:
    According to theologians of the Cult of Anthropogenic Climate Change, we should be experiencing more storms (and more severe ones). Empirical facts, on the other hand, show otherwise. Dan those inconvenient truths.

    • LOL

      I think you’ll find the Sharknado chart, next to the one for increased umbrella sales dues to global warming.

  11. “Violent Tomatoes Are On The Decline In The US”. It’s scientific because we’ve got graphs for it. Ooops, sorry. I really shouldn’t read these things before the first coffee in the morning.

  12. One does not have to go very far back in history—just a few hundred years will do—to note that nature goes through cycles. Cycles of violent storms, volcanic eruptions, earthquakes, and yes! even warming and cooling. (Fancy that!) This is evident even without the help of graphs and statistics. Question is: Why won’t climate “ex-spurts” acknowledge as much?

  13. RH, as a former NWS operational forecaster who did the work of surveying damage that you referenced, I have mixed feelings on your comments. It’s actually relatively easy to discern tornadic rotation in a damage survey, even when straight-line winds cause a bit of curvature to the damage. And, assessing a damage category can be tricky, but again it’s not that hard, especially with the software used on site now. But, I knew many meteorology geeks in the NWS who thought they knew more than they really did, or who wanted to have a nice sexy tornado splashed on the evening news for themselves. I’ve even seen a NWS meteorologist visiting from out of the area who wanted to perform a survey, and since there was a radar signature of weak rotation aloft, it was labelled as a tornado despite little other evidence, in fact evidence to the contrary. On a related note, a Meteorologist-In-Charge called in verification of a severe thunderstorm warning with one downed tree limb on the side of the road in an effort to pump up our lagging severe weather statistics. It was midterm performance appraisal time; he probably needed his bonus.

  14. tomado toMado say Figure 1 1970-2003

    r+n=m or doesn’t…”m” just may = m


  15. Why does the NOAA chart of “US count of strong to violent hurricanes…” not show data later than 2012? Have they stopped tracking that data?

Comments are closed.