The Importance of Accurate Reporting

[Kiel Ortega]

Well, since we have the start of our severe weather season going on, I thought I would share some things that I have learned from using Storm Data for research purposes. The big thing is it does not make for a good research database...at all. There are too many factors that feed into the reliability of the database. However, I figured that I would try to start a grassroots campaign to help make it just a little better.

First, accurately report your position. While estimating is okay (like 5 SW of Podunk), if you report an intersection you are 100% better---so long as the person taking your report writes down the intersection in the comments (hint, hint NWS personnel).

Second, for hail, accurately report size. Use anything (specifically a RULER or a coin or a lighter or something common which a NWS office or future researcher can get the measurement), but don't guess-timate by eyeball. Here are some stats I found from a 2.5 year database of severe weather for a multi-sensor hail diagnosis algorithm study I did for NSSL:
the inclusive part is the lower end (so if x is the size then, 1" <= x < 1.5")
of 5671 hail reports,
< 1": 2650
1" to 1.5": 1653
1.5" to 2": 1069
2" to 2.5": 81
2.5" to 3": 172
3" to 3.5": 13
3.5" to 4": 1
> 4": 32

nickle = 0.88" got 886 hits (33% of its size category)
quarter = 1" got 1440 hits (87% of its size category)
golfball = 1.75" got 949 hits (89% of its size category)
baseball = 2.75" got 108 hits (63% of its size category)
softball = 4.5" got 16 (50% of its size category)

I highly doubt the atmosphere is so specific in forming hail of a certain size as the stats would lead one to believe!

I won't comment on wind since I have no experience using it (although, I am sure it is FAR worse than the Storm Data hail reports).

And for those writing down the reports, please write down all the information supplied to you (like road intersection; and you can put down a size estimate, but if the hail is compared to something off the wall, write it down too in the comments...I might find it useful )

Most of all, tell the truth! :wink: Hopefully no one lies about storm reports, but it would be a shame if they did.

Examples of possible errors in hail reports:

Here, 1.75" hail is reported in Archer City, though the highest hail is to the south by several kilometers. Error in the algorithm or error in report??? And if it is an error in the report, would it make a researcher's "truth" in the future?

Here, 2.75" hail is reported in Olney by the fire department. The algorithm says there is no way ~3" hail was failling anywhere near Olney! Which does one believe?

 

[Aaron Kennedy]

Storm Data is one of those beasts that can drive a man insane. I did my time with hail forensics... yikes. The most aggravating thing is the differences in how offices report hail. Even if a good observer writes down a street location, many times this data gets omitted when stuck into Storm Data. Instead, the office will simply put a generic number such as 5SSW of XXX town. Then there are the issues of offices only writing down enough reports to verify a warning... and omitting the rest. Blah!

Aaron

 

[Rob Satkus]

Originally posted by Aaron Kennedy
The most aggravating thing is the differences in how offices report hail.
Aaron

Exactly.... more and more offices are using a central point within a town ( how it's decided, I don't know) and reports are listed based on that...so if 1" hail was reported 4 miles south of that point, even if it's well within town limits, it will show up as 4 miles south of town X.

Rob

 

[Morgan Palmer]

How's this for another impact of inaccurate storm reporting...

An insurance actuary looks at hail reports and finds that the Dallas/Ft. Worth area is oddly susceptible to very large hail? Hence everyone pays a surcharge.

That couldn't really happen, huh?

 

[Kiel Ortega]

Originally posted by Aaron Kennedy
Storm Data is one of those beasts that can drive a man insane. I did my time with hail forensics... yikes. The most aggravating thing is the differences in how offices report hail. Even if a good observer writes down a street location, many times this data gets omitted when stuck into Storm Data. Instead, the office will simply put a generic number such as 5SSW of XXX town. Then there are the issues of offices only writing down enough reports to verify a warning... and omitting the rest. Blah!

Aaron

Yes, the office by office relativity of the reports is not easy to deal with. Most offices do only get point data on the hail. However, some offices will put the duration of the hail at a point in the report or, like in N AL, complete hailswaths are commonly put in as reports (time and space durations).

Also, in response to the insurance issue...that's a touchy subject for me. The reason why is because most insurance companies for events use some sort of "hailswath", which is usually a combination of low level reflectivity and MESH. I have seen some examples and, in a way, an example can be seen in Marshall, et al. 2004 (http://ams.confex.com/ams/pdfpapers/81089.pdf), using the radar for saying where hail of a certain size did or did not fall isn't the most accurate way of doing hail verification for insurance or scientific reasons.