Radar Images of Long Track Tornadic Supercell

[Mike Smith]

Aaron asked me to repost this item, so here goes.

A long track (more than 200 miles) tornadic supercell developed just east of Houston on the 23rd. Here is a series of images as captured by WeatherData's SelectWarn® which were annotated, in real time, by NWS LSR's.

To me, this storm illustrates a number of things:

The National Weather Service is getting good at drawing the warning polygons, which greatly decreases the avoidable false alarm area associated with each warning.

Having the LSR's available immediately helps the credibility of the warning.

This was the type of situation in which scores of people used to die. Even though this was a largely rural storm, the warnings obviously got out and the death toll was low.

Now, the images:


This image shows the developing supercell just east of Houston about the time it produced its firs tornado. We have indicated the future path of the storm.




The next image shows two tornadic storms. The long track supercell is moving toward the Louisiana border. Note how much smaller the warned area is with the polygons as opposed to the county warnings (the warned counties are outlined in red).



As the storm gets ready to cross into Louisiana, it is trailing a number of real time storm reports of the tornado.


Now in Louisiana, the supercell and tornado continue moving northeast.


Nearly three and a half hours after the first tornado from the supercell was reported, a tornado is on the ground in central Vernon Parish.


A hook echo is still suggested as the storm approaches U.S. 71 in Natchitoches Parish. The supercell is in the classic "supercell ahead of a line" which is a favored location for a tornadic storm. This is more than four hours after the first tornado was reported.

Hope everyone is enjoyed their holiday weekend..and, some calmer weather.

Mike[/img]

 

[Mike Umscheid]

Originally posted by Mike Smith
Aaron asked me to repost this item, so here goes.

A long track (more than 200 miles) tornadic supercell developed just east of Houston on the 23rd. Here is a series of images as captured by WeatherData's SelectWarn® which were annotated, in real time, by NWS LSR's.

To me, this storm illustrates a number of things:

The National Weather Service is getting good at drawing the warning polygons, which greatly decreases the avoidable false alarm area associated with each warning.

Having the LSR's available immediately helps the credibility of the warning.

This was the type of situation in which scores of people used to die. Even though this was a largely rural storm, the warnings obviously got out and the death toll was low.

This coming spring, our office (DDC) along with other offices, will be testing the new polygon warning verification system. This should help mitigate the county-shaped polygons that are still seen a lot of times with mainly SVR warnings. This will also make it much easier, (psychologically speaking) to draw appropriate polygons that encompass 3 or 4 counties... where in the current system, you have to verify each county that the polygon included. There will probably be growing pains with this new verification system with polygons for those "on the edge" reports... "is it in or out of the polygon??"... since we are trying to get away from the political boundary warning verification.

Mike U

 

[Greg Stumpf]

Originally posted by Mike Umscheid
This coming spring, our office (DDC) along with other offices, will be testing the new polygon warning verification system. This should help mitigate the county-shaped polygons that are still seen a lot of times with mainly SVR warnings. This will also make it much easier, (psychologically speaking) to draw appropriate polygons that encompass 3 or 4 counties... where in the current system, you have to verify each county that the polygon included.

Do you know if the WFO mets are being asked to not tweak the polygons to county borders? This is still very commonplace, even in the present polygon based warning system. In order to really do this test correctly, the county overlay maps should be turned off!


greg

 

[Joe Nield]

Originally posted by Greg Stumpf+--><div class='quotetop'>QUOTE(Greg Stumpf)</div>

<!--QuoteBegin-Mike Umscheid

This coming spring, our office (DDC) along with other offices, will be testing the new polygon warning verification system. This should help mitigate the county-shaped polygons that are still seen a lot of times with mainly SVR warnings. This will also make it much easier, (psychologically speaking) to draw appropriate polygons that encompass 3 or 4 counties... where in the current system, you have to verify each county that the polygon included.

Do you know if the WFO mets are being asked to not tweak the polygons to county borders? This is still very commonplace, even in the present polygon based warning system. In order to really do this test correctly, the county overlay maps should be turned off![/b]

Yes, mets are being asked to ensure that polygons are encompassing the actual threat area and not the county borders. That is policy already, but no one really got on anyone about it until the polygon warning test came down the line.

The interesting situation to see will be the first time a huge derecho moves through an area with a history of 60 knot winds all along its leading edge, and someone does one warning capturing about 12 counties with one polygon along the leading edge. Valid, but how will people feel about it? Only time will tell.

 

[Mike Umscheid]

Originally posted by Greg Stumpf

Do you know if the WFO mets are being asked to not tweak the polygons to county borders? This is still very commonplace, even in the present polygon based warning system. In order to really do this test correctly, the county overlay maps should be turned off!


greg

We are instructed to draw the polygons as meteorologically accurate as possible...meaning, greatest threat area. However, this is NOT strictly enforced... in other words, we don't get slapped on the wrist for drawing polygons to county lines. This is a local office policy, however, since all CWAs have different size/shape counties. Of course, in my CWA (and much of the western plains for that matter), all the counties are neat little qudrilaterals, so it is easier to get away with the drawing to county lines. With the low threshold for hail (3/4"), out on the high plains, you can get this kind of hail over a large area with any severe storm, so it's just more convienent to "box" the whole county. I understand that we may be raising the SVR hail threshold to 1", which opens up another whole can of worms... It's tough to teach an old dog new tricks... referencing some of the more seasoned radar meteorologists and "boxology" as you will....

You mention turning county outlines off... that's a good experiement. I may give that a try next spring just using my CWA border with cities & highways... that's a good idea (I know that wouldn't go over well with many other radar mets though!)...

Mike U

 

[Greg Stumpf]

Originally posted by Mike Umscheid
You mention turning county outlines off... that's a good experiement. I may give that a try next spring just using my CWA border with cities & highways... that's a good idea (I know that wouldn't go over well with many other radar mets though!)...

Why not? If the verification is no longer tied to counties, they shouldn't be concerned.

 

[Greg Stumpf]

Originally posted by Joe Nield
The interesting situation to see will be the first time a huge derecho moves through an area with a history of 60 knot winds all along its leading edge, and someone does one warning capturing about 12 counties with one polygon along the leading edge. Valid, but how will people feel about it? Only time will tell.

A reasonable way to verify warnings would be to use a similar method of verification that is used for SPC watch boxes, which is grid-based. However, for warnings, the grid sizes would probably have to be smaller (say 5 km versus 25 km). The onus would then be on the WFOs to verify each grid that intersects the warning polygon, which will be a lot more work especially for the larger polygons. But, IMHO, I am 100% in favor of this - more reports leads to improved warning applications and improved climatological database. In other words, the NWS should understand that their verification is not to only serve the purpose of warning verification but to also improve the science and understanding. Afterall, the NWS is the only one doing the verification.

[EDIT: in bold - bad typo!]

I've been trying to get more official information about this project through several WFO and WSH contacts with no luck. Can you point me (us) in the right direction (either you or Mike U.)?

 

[Joe Nield]

Originally posted by Greg Stumpf+--><div class='quotetop'>QUOTE(Greg Stumpf)</div>

<!--QuoteBegin-Joe Nield

The interesting situation to see will be the first time a huge derecho moves through an area with a history of 60 knot winds all along its leading edge, and someone does one warning capturing about 12 counties with one polygon along the leading edge. Valid, but how will people feel about it? Only time will tell.

A reasonable way to verify warnings would be to use a similar method of verification that is used for SPC watch boxes, which is grid-based. However, for warnings, the grid sizes would probably have to be smaller (say 5 km versus 25 km). The onus would then be on the WFOs to verify each grid that intersects the warning polygon, which will be a lot more work especially for the larger polygons. But, IMHO, I am 100% in favor of this - more reports leads to improved warning applications and improved climatological database. In other words, the NWS should understand that their verification should be to only serve the purpose of warning verification but to also improve the science and understanding. Afterall, the NWS is the only one doing the verification.

I've been trying to get more official information about this project through several WFO and WSH contacts with no luck. Can you point me (us) in the right direction (either you or Mike U.)?[/b]

Sounds like you have a pretty good grasp of it. I'm not really sure what else I could point you towards aside from what you already know.

A grid-based verification system, while interesting, may just be more than we can do. We get much of our reports from law enforcement, and many of these are after the fact, with us making calls out looking for reports. We have a hard time getting specific locations out of them, as they are busy as well. IMO, without a much larger reporting base than what we already have, grid-based verification is still too much to do operationally.