Using Probabilistic Rainfall Totals To Improve Flash Flood Messaging

Flash flooding is the second-deadliest weather-related cause in the United States, responsible for an average of 88 deaths each year. Because the power of water is often underestimated, many individuals are typically caught off guard by the rapid onset and intensity of flash floods. To help save lives and ensure that safety precautions are taken, it is essential that National Weather Service (NWS) meteorologists deliver proper Impact-Based Decision Support Services (IDSS) messaging to NWS Core Partners and elevate the threat levels of flooding rain events when severe weather grants it necessary. Because this need is constant and ever-evolving, enhancements to messaging are highly sought-after.

This study assesses whether probabilistic rainfall totals from the High-Resolution Ensemble Forecast (HREF) can improve external messaging and identify higher-end rainfall totals that would deviate from the widespread rainfall total forecast typically communicated by a Weather Forecasting Office (WFO). It utilizes a comparative analysis approach between the HREF forecasts and Multi-Radar Multi-Sensor (MRMS) rainfall totals observed during intense rainfall events. Specifically, the study focuses on Flash Flood Warnings labeled as “considerable” or “catastrophic” that occurred between January 2021 and May 2024 across several southeastern County Warning Areas (CWAs) including Huntsville, AL, Birmingham, AL, Memphis, TN, and Jackson, MS.

The study concludes that the HREF max 24-hour rainfall forecasts, followed by the localized probability-matched mean (LPMM), are the most representative of higher-end rainfall totals.