Thursday was the last day that we actively used the experimental forecast tools. Our experiments took us back to the DC Metro, and just like on Tuesday, we had several individual storms to look at.
Thursday was the final day for the forecasters, myself included, to use the experimental products in areas where severe weather was expected to develop. In our afternoon meeting with the guys from the other half of the Experiment (the Experimental Forecast Program, which mostly consists of Storm Prediction Center meteorologists), they had pinpointed the DC Metro area and points to our north and south as the areas to watch for severe storm development. So, that's where we centered our focus as the afternoon wore on.
The computer data that we've been using this week comes from several different sources, and each program had a different goal. One program is designed to differentiate between storms that will become severe, and those that will not; another program pinpoints areas where convection (tall cloud development) is expected; yet another is designed to alert the user when the frequency of lightning strikes increases rapidly, and so on. (By the way, the lightning tool is in the image below. It's called lightning flash density because it measures lightning strikes that hit the ground and those that stay in the cloud!) The three other forecasters and I have found that each program can be useful in certain situations, but nothing that we used this week was the end-all, be-all for severe weather forecasting.
Part of the reason for this is because not all severe storms are the same. Some are prolific hail-makers, while others produce strong wind gusts, and others can produce funnel clouds and tornadoes. Some storms produce two of those results, or even all three. While using the products today in the DC area, I noticed that most of the materials at our disposal were not tailor-made for the type of storms we had on Thursday. The storms produced gusty winds and even knocked down a few trees, but for the most part, the lightning was not very prolific, and there were zero reports of hail.
I've discovered through talking to my NWS colleagues that estimating wind gusts in a storm cell is one of the toughest things about severe storm forecasting. I talked to the software developers, who were there to teach us about their products, how helpful it would be to add a parameter to their forecast models that predicts peak wind gusts in a storm. The whole reason for these experiments is to see how real forecasters would use the tools in real-life severe weather situations. The input and constructive criticism that I and the other forecasters are giving could shape the next generation of severe weather forecast models!