Warning Performance Analytics through the Lens of a Tornado's Lifecycle

Formally, the quality of National Weather Service (NWS) Tornado Warnings (TORs) is assessed via comparison with specific Government Performance and Results Act (GPRA) targets for probability of detection (POD), false alarm ratio (FAR), and lead time. Such metrics provide benchmarks for TOR performance; however, these metrics remain static regardless of tornado intensity and the associated impacts on life and property, which vary by tornado and within a given tornado’s lifetime. As such, GPRA goals and associated verification statistics may not accurately reflect the true value of TORs in the context of the NWS’s mission.

In this presentation, we will introduce supplemental analytics that assess the quality of TORs on a scale of 0 to 100 throughout the lifetimes of associated tornadoes by considering lead times and tornado damage ratings along their tracks. Such a “feature scaling” approach emphasizes the importance of lead time for the strongest stage of the tornado (based on the damage it produced), when casualties and damage to property are most likely to occur. This quantitative approach is more consistent with both a meteorologist’s qualitative considerations during severe weather operations and the practical impacts to NWS end users. This presentation will also demonstrate an automated method for calculating these supplemental analytics for past events, which could be valuable as a training tool for Science and Operations Officers at local weather forecast offices.