Data Mining Tornado Emergencies

Last night I posted a call for help from fellow meteorologists and weather nerds to help me acquire the tornado emergencies that I was missing. Wow! Thanks to the help of Jason Kaiser, Daryl Herzmann from Iowa State, and Rick Smith from National Weather Service (NWS) forecast office Norman, I now believe I have all tornado emergencies in my possession! Now the fun part of analyzing them begins! Since many people are already aware of my project, I decided to go ahead and post two quick graphics that I think will be of interest.

First, a bit about my method. For this analysis I treat the tornado warning and all subsequent severe weather statements associated with the tornado warning as a single episode. Since a tornado emergency can be issued in the text of either a tornado warning or a severe weather statement, all tornado emergencies issued within a single episode is consider a single tornado emergency. Consider the following scenarios:

Scenario 1: A tornado warning is issued for 2 counties and does not contain the tornado emergency text. Subsequently 4 severe weather statements are issued. The second severe weather statement contains the tornado emergency language for county A but not county B. The fourth severe weather statement contains the tornado emergency for county B but not county A. How many tornado emergencies will this count as?

Using my definition, even though 2 tornado emergencies were technically issued, they are both contained within a single tornado warning and thus this is treated as a single tornado emergency.

Scenario 2: A tornado warning is issued for county A and does not contain the tornado emergency text. Subsequently 2 severe weather statements are issued. The second severe weather statement contains the tornado emergency language for county A. As the tornado moves into county B, the NWS issues a new tornado warning, without the tornado emergency language, for county B. Subsequently, the NWS issues three severe weather statements for county B, all three containing the tornado emergency language. How many tornado emergencies will this count as?

Using my definition, even though 4 tornado emergencies were technically issued, they are all contained within the framework of only two tornado warnings. Thus this is treated as two tornado emergencies — even though this is for the same tornado.

Scenario 3: A tornado warning is issued for county A and does contain the tornado emergency text. Subsequently 2 severe weather statements are issued, both of which contain the tornado emergency language. How many tornado emergencies will this count as?

Using my definition, even though 3 tornado emergencies were technically issued, they are all contained within the framework of a single tornado warning. Thus this is treated as one tornado emergency.

Does that make sense?

With that said, there have been 143 distinct tornado emergencies issued since the first one on 3 May 1999 for south Oklahoma City, OK. The chart below breaks them down into yearly counts:

Tornado Emergencies By Year

Examining tornado emergencies by NWS forecast office, the top three are

  1. BMX (Birmingham, AL)
  2. HUN (Huntsville, AL)
  3. JAN (Jackson, MS)

This might be a bit surprising to many who typically think of “tornado alley” as being in the central United States. Unfortunately, or fortunately, you decide, many things go into the decision to issue tornado emergencies. Some of these things include population potentially impacted and local office policies. Remember, the tornado emergency is not a formal product. It wasn’t until the last year or two that local forecast offices has official guidance in place. Bottom line, not every office uses tornado emergencies. (Also, keep in mind the magnitude of the 27 April 2011 tornado outbreak and where that was located! However, to be fair, I will say that all of the top three offices had issued tornado emergencies prior to 27 April 2011.)

The rest can be found in the chart below:

Tornado Emergencies By WFO

In the coming days and weeks I will be doing more analysis on tornado emergencies and will post findings as they are completed. If there is something you would like to see, leave me a comment and let me know. If I don’t already have plans to examine your request, I’ll add it to my list!

  • Kim

    Neat!! What an interesting distribution. Could you do either of the following:

    (a) Remove the April 27 outbreak and replot the results in the graph above?

    (b) Create a map showing the WFOs & their relative #s of tornado emergencies to facilitate spatial pattern identification?

  • http://www.patricktmarsh.com pmarshwx

    Both are on my to-do list. In fact, (b) was what I was originally going to do. However, it requires slightly more programming than I felt like doing tonight. I’ll try to get these up ASAP.

  • http://jntk.freehostia.com Jason Kaiser

    Glad I could help! :)

  • Tim Brice

    Very nice work. I will be interested in seeing the rest of your analysis.

  • Tim Supinie

    Adding to what Kim suggested, perhaps you could redo the “Tornado Emergencies by WFO” graph, removing all the ones from 2011. I suppose we don’t have enough samples to call it a true statistical anomaly, but there are definitely extenuating circumstances surrounding the higher numbers of tornado emergencies this year.

    As an example, my guess is that a lot of the TE’s issued by WFO RAH (Raleigh, NC) were during the 16 April event from this year.

    A graph of (# of TE’s) / (# of tornadoes) by year or (# of TE’s) / (# of violent tornadoes) by hear might be interesting and might provide another method of normalizing the large numbers of TE’s this year.

  • Rick Smith

    Great stuff!

    It might be interesting to see how the TE fit within the overall flow of information for the event, and look at what information preceded the TE – SPC outlooks/risk categories, watch type, other warnings and statements, etc. If the TE was tied to a sig tornado, did the ramp up to the event help prepare people in the area for such an event?

  • Kevin Scharfenberg

    I’ll second what Rick Smith says. Also, do TEs have better skill in a PDS TOR Watch? A High Risk area?

  • http://www.patricktmarsh.com pmarshwx

    Rick, I have ideas to include watch information as well as atmospheric fields (CAPE, ESRH, CINH, etc.) from SPC Surface Objective Analysis data. All of this would be made available in the resulting database.

  • Tim Coleman

    I also would like to see the data with 2011 outbreaks removed. Very good work.

  • http://www.patricktmarsh.com pmarshwx

    Thanks, Tim. I’ve gone ahead and added a chart for the WFO counts sans 2011 (see post immediately after this one). I’ve also gone ahead and added a list of all the tornado emergencies that I have in my database. That’s 2 posts after this one.