Storm-based warning verification: a new era in warning verification

On October 1st, 2007, the National Weather Service (NWS) officially began issuing short-duration warnings using latitude / longitude points to define the warned area, otherwise known as storm-based warnings. Previously these warnings had been issued using county boundaries to define the warned area. This change was made to more accurately define the threatened area, while reducing the population unnecessarily warned.

With this change in how warnings are issued, the NWS's Performance Branch was presented with the challenge of developing a new suite of verification metrics used to track the accuracy of these storm-based warnings. Unlike the traditional county-based warning verification methods which had been used to monitor performance since the early 1980s, these new metrics would rely heavily on geospatial tools to analyze each warning and event. The result was a suite of verification scores more representative of the service provided to the public for all severe thunderstorm, tornado, flash flood, and special marine warnings.

This suite of verification scores is available to all NWS forecasters via an interactive, web-based interface titled “Stats-on-Demand.” This interface uses Google™ Maps to produce graphical images outlining performance for every warning and verifiable event. From this interface, forecasters can analyze data and more easily find areas in which they may improve their warning accuracy and quality.

This presentation will describe the new geospatial techniques used to produce verification scores and demonstrate the Stats-on-Demand interface used to monitor performance. In addition, future storm-based warning verification development efforts will be discussed.