Million dollar online radar idea

Dec 4, 2003
Looking at this squall line coming into north Texas tonight and not really knowing when it will get here, it occurred to me that a neat product to see would be a current radar mosaic for a region merged with an identical mosaic from 3 hours earlier. This would make it easy to estimate the approximate time when a squall line will reach a given spot.

An improvement on this would be to scan the image from east to west to find the eastern edge of a thunderstorm line to generate isochrones. Overlay six hours worth of these isochrones onto a single map, study it, and you can probably nail the storm arrival to within half an hour.

When I did Air Force weather I used to actually keep some isochrones on the PUP screen with a water-based marker. My commander at Dyess used to grimace at this, as it was a brand-new system, but he knew that I was able to nail the storm arrival times dead-on which was important as the flightline needed this info down to the minute.

I do rule out using any of the RPG storm tracking products (such as cell motion), as storm centroids are notoriously problematic in lines, and cell motion is not the same as the propagation of an entire storm complex.

I could do the 3-hour image (and likely the isochrone algorithm) with Digital Atmosphere, but I'm not programming right now due to the Chase Hotline. Until then I bet some university or online data vendor could do it.

Sounds like a really good concept. I just recently got the Storm lab software and have been using the centroid/pathcast feature to track storms, and found that it is somewhat accurate but there's still room for improvement. The storms usually move into other areas instead of the original pathcasted areas.
How would the software compensate for environmental changes as time goes on? Is that something that the isochrone algorithm takes into account? Talking about things like adjustments that would be need to be made as the storm modifies its own environment and heading and turns right (or splits left), slows down, or later differentiations in steering winds ... stuff like that - would think that a million different things could happen in three hours. It may work better for squall lines - but even then seems like you might end up with some weirdness, like the line hitting a boundary and falling apart, for instance.
There are already a number of applications that take current and past radar imagery and do some kind of correlation to predict a future radar image, typically out to 30-60 minutes. There are even some applications that can begin to blend numerical weather guidance for the later forecast periods (1-6 hours). Most of these have been developed by universities and governments (not just the U.S.), and then some have been marketed by private companies. One example:

Lakshmanan, V., R. Rabin, and V. DeBrunner, 2003c: Multiscale storm identification and forecast. J. Atm. Res., 367-380.

A good idea, in general

In general, that's a good idea for a young-to-mature squall line that has established its behavior and for locations that are downstream of the squall line in question. I know that I've done that with the 50-60 minute loops available on NWS websites.

For areas south of "downstream", a modification may be in order, since squall lines often like to develop along a line roughly perpendicular to the line's net motion. Simply extrapolating an eastbound squall line along I-39 between Madison and DeKalb would indicate that Chicago's south suburbs would be missed, even though they would have an excellent chance of meeting Tail-end Charlie. If CAPE is supportive, tracing a further southward development of the gust front would be helpful.

Also, last month metro Chicago was been tantalized by several MCCs coming in from Iowa fronted by squall lines -- just to watch them die somewhere around I-39. We need the rain pretty badly south and east of the city.
Neat idea. Now if only I could finish my radar viewing program so that I could play around with all of these spiffy ideas. What language is Digital Atmosphere programmed in, im assuming C/C++?