Forecasting tornados an hour in advance?!?

[Mike Smith]

Mike U.,
I agree that better remote sensing is a key.

In fairness, the TDWR's need to be added in so NYC, CLT, PHL, STJ and a few other areas are added to the 1km and below coverage.

I would like to see come 5cm radars or even the CASA 3cm's put in some of the data-sparse areas like northeast Missouri and lower Red River areas of Texas and Oklahoma.

Mike

 

[Rich Thompson]

The "fun" part of extending lead times for warnings is dealing with new storms and complex storm interactions. On 10 May, you would have to issue a forecast envelope with at least low tornado probabilities as soon as cumulus clouds developed SW of Norman to get even 45 minutes of lead time. That's a much different problem than extrapolating the future location of an efficient tornado producing storm. If you consider all of the complex storm interactions we often see, these sort of warnings will have to be updated on a 5-10 minute cycle.

I think we can do a better job right now in issuing more continuous warnings with more even lead time. These discrete polygons that still trace out counties don't add much of anything to the process. I'd prefer to see almost all warnings adjusted every 10-15 minutes, with warning areas going out no more than 30 minutes. The goal would be more fluid warnings with more even lead time across all areas in the warning. Such information can be conveyed rather easily via TV/radio, but could be more of a challenge with the current NWR setup. More frequent warning adjustments, smaller warning areas (in most situations), and more even lead time would seem to help. Ultimately, good warnings for severe storms should look something like the storms themselves. If you loop radar and warnings right now, you get a distinctly different impression.

Accomplishing those goals would take us a step closer to the future warning goals, without introducing any new products.

 

[Greg Stumpf]

More radars would be ideal, but there is one big problem. They are expensive. Even the CASA X-band gap filling radars have only about a 40 km range, so you would need many to fill all the gaps. But I'm hoping that their expense can somehow be justified. Gap-filling radars are wonderful, as I have can attest to through our Hazardous Weather Testbed experiments.

BTW - lead time and accuracy go hand in hand. As I said earlier, greater lead time really means less uncertain forecasts at greater forecast intervals. That also should mean less uncertain forecasts at the very short intervals (e.g., "warnings").

They're both important!