cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Check out the JMP® Marketplace featured Capability Explorer add-in
Choose Language Hide Translation Bar
scwise
Staff
Tornado Trend Challenge: Critical Thinking in Three Dimensions – Part 1: More Tornadoes?

F2 Tornado - Round Rock, TX - 2022F2 Tornado - Round Rock, TX - 2022

In the classic 1960’s Twilight Zone show, Rod Serling starts every science-fiction episode with the words: “You unlock the door with the key of imagination.  Beyond it is another dimension.”  However, we as humans often have trouble seeing beyond two dimensions, especially when presented with subjects where we may hold personal biases.  To complicate the matter, trying to visualize in three dimensions often overwhelms simple graphs and can even trick our brains into seeing patterns that don’t exist. 

 

So, to test this and come up with better ways to think in three dimensions, we will analyze data around our belief in recent tornado trends!  For our data we used the NOAA Storm Events Database, filtered for all tornado incidents and fatalities from 1950 to 2022 that took place in the United States.  You can find this public data at the following link: (https://www.ncdc.noaa.gov/stormevents/ftp.jsp).  We will also use the  JMP Graph Builder platform to make interactive visuals that can provide solutions to our questions.

 

Question 1: “Do you believe the total number of yearly tornadoes in the US is increasing?”.  Many believe that all types of weather are being negatively impacted by changes brought about due to the warming of our planet from the effects of climate changes.  If like me you have recently been in an area that has experience recent tornado activity (in my case in Central Texas) then you might be thinking that yes, there must be more tornadoes now!  But what does visualizing the data tell us?

 

Our first graph shows a simple Line Chart of the sum of yearly tornado count over the years.  The results show a rising slope line over time, so this two-dimensional view seems to be reinforcing our bias that tornadoes are indeed increasing.

Q1-A.png

However, we can bring in an extra factor to get the third dimension.  In this case we can improve our investigation by adding the type of tornado F Scale to create an Overlaid Line Chart.  The F Scale helps us figure out the windspeed of the tornado based on the severity of the damage it leaves behind.   For example, an F2 tornado will produce winds of 113 mph to 158 mph and will leave in its path several roofless structures.  See the different F Scales below. 

Screenshot 2023-06-03 at 6.16.55 PM.png

 

By creating separate overlaid graph lines by F Scale tornado type, we can see that the increase in the tornado count is due to F0 and F1 tornado types (see the blue and red lines)!  These less damaging tornado types both seem to really increase around the early 1990’s which coincides with the introduction of the widespread use of Doppler Radar.  This more advanced radar allowed for better indication and warning of tornadic activity within thunderstorms.  Now more and more F0 and F1 tornados started being reported that previously were chalked up to just strong storm wind damage.

Q1-B.png

If we remove the F0 and F1 tornado type counts with an interactive filter from our graph for a fairer comparison, we can now see the true trend behavior of the more serious F2 to F5 tornadoes.  The flatter trend behavior (or even decreasing F2 activity) now leads us to the conclusion that overall tornado account is not necessarily increasing over time!

Q1-C.png

 

Additional research shows that while many types of weather patterns like hurricanes and droughts are showing increasing count trends due to climate change driven warming temperatures, there is not yet convincing proof that this is also happening with respect to tornadoes.   For tornadoes to form they need both fuel (from big thunderstorms) as well as wind shear to cause the tornadic activity (such as from the jet stream).  However, while a warming planet is increasing the chance of getting stronger thunderstorms (adding more fuel to form potential tornadoes), some Scientists feel the warmer temperatures is also diluting the average power of the jet stream (reducing the chance of wind shear occurring that causes the tornadic activity).

 

So, our first best practice for “seeing in the third dimension” is to explore our data with interaction and filters.  See the following Tornado Trend Challenge JMP Public link if you would like to view and interact with these graphs and data yourself: https://public.jmp.com/packages/XyMCFvqxR9Dftkqhvvt6l

 

In our next blog we will address Question 2: “Do you believe that tornadoes in the US are getting more destructive?”. 

Last Modified: Jul 7, 2023 5:18 PM
Comments
P_Bartell
Level VIII

Great post Scott! A perfect example of Statistical Thinking where plotting the data and asking questions in an iterative fashion can help separate fact from fiction. Thanks for sharing.

Peter_Hersh
Staff

This is great Scott thanks for sharing.  It would be cool to see a CUSUM or EWMA chart on this data to see if the long term trend has actually shifted.  I am with you it looks like the number of F2 tornados dropped after wider use of Doppler radar in 1990, but that also seemed to correspond with a increase in F1 and F0 tornados.  Could this be a classification issue with improved instrumentation we do a better job of properly categorizing the tornados?