Something I Didn’t Know About The Tornado Numbers

A change of methodology means tornado reports are inflated when compared to prior years.

I was reading Reddit when I came upon this. A change in methodology means tornado reports are inflated when compared to prior years.

[T]he NWS changed its policy regarding duplicate severe weather reports this year.

It should be noted that duplicate preliminary reports are NO LONGER purged from the data set. As a consequence, it is not possible to compare this preliminary data to previous years’ data. In the past, possible duplicate reports were culled based on time and proximity. This is not the case anymore.

The old rule of thumb was ~15% of preliminary reports were duplicates. Based on the preliminary v. final numbers so far this season it is possible that up to 65% of these preliminary reports are duplicates.

On March 8, 2011, the SPC removed space/time filtering on incoming National Weather Service (NWS) Local Storm Reports (LSRs). This filtering had been used by SPC in an attempt to reduce duplicate reports and limit artificially inflated initial estimates of severe weather events when many reports arrived for the same event. Space/time filtering is no longer being applied to decoded NWS LSRs and this approach is consistent with NWS storm-based verification methods.

You probably already know the number of weak tornadoes reported has gone up over time as ‘chasing’ has made fewer of them unobserved.

14 thoughts on “Something I Didn’t Know About The Tornado Numbers”

    1. Not exaggerating, just changing how the counts are compiled. Usually the preliminary numbers were higher than the final count. Often a single tornado is reported in different locales. Now that will remain as multiple tornadoes instead of being instantly combined as one.

      1. No – the number of dead is unfortunately usually very accurate. This only has to do with the number of separate incidents.

  1. Yeah I just realized this as well. The preliminary data is still just unofficial you generally have to wait 6 months for the verified and official data to come out through “Storm Data”.

    I wouldn’t say the numbers are inflated they just post the raw reports as they come in from the NWS and then later verify them for their official publication.

  2. So Geoff, I am not really sure what you are saying here. Are you saying the reports of tornadoes are not accurate? What about all the people who died, there isn’t a way to make that number incorrect. I am confused by what you wrote here and don’t quite get the message. Maybe my brain is malfunctioning today.

  3. It think what Geoff is saying is that multiple reports of the same tornado skew the total numbers of tornadoes. If there was a tornado in Hartford and people in West Hartford, East Hartford, Bloomfield and Newington all reported seeing it, then it may be reported that there were 5 tornadoes when there was really only 1.

    This inaccuracy doesn’t negate the damage caused by that one storm.

      1. Unfortunately, doesn’t “storm severity” play into the response and reimbursement decisions of the (gulp!) insurance companies?? All I have to do is turn on my 36″ flat screen to see what horrific damage was done to the affected areas… not to mention lives and families altered forever 🙁

  4. Yes Micah, exactly what he is saying. Perfect, they even reported on CNN that the tornado that did such damage in one town did damage in several other towns because it stayed on the ground for so long. It was the same tornado, not a different one, but unfortunately they are reported multiple times as they touch down in different spots.

Leave a Reply

Your email address will not be published. Required fields are marked *