Verifying WPC Excessive Rainfall Outlooks across three hurricanes of varying rainfall impact

Atlantic basin hurricanes drive serious meteorological hazards for the southeast United States, including rainfall-induced flooding. The well-known Saffir-Simpson Hurricane Wind Scale communicates only wind hazard, creating the potential for low wind, high rainfall storms to be perceived as mild. Researchers have emphasized the need to center rainfall hazard in hurricane warning messaging, and prerequisite to this is verifying the accuracy of hurricane rainfall forecasts. This paper investigates the characterization of forecasted and observed rainfall for three Atlantic hurricanes with varying rainfall impact: Florence (2018), Michael (2018), and Ian (2022). We first quantified, visualized, and compared the distribution of daily and total rainfall for each storm to establish the concept of varied rainfall impact. By a Mann-Whittney U Test, we found that the distributions of total rainfall for Florence/Michael and Florence/Ian were significantly different while differences between Michael/Ian were not significant. We then compared Excessive Rainfall Outlooks (EROs) against a flash flood proxy both visually and through fractional coverage calculations. We used these to calculate fractions brier scores (FBS) and fractions skill scores (FSS) using old and new ERO definitions. New ERO definitions improved FSS scores across all hurricanes. Florence had the highest FSS, followed by Michael, then Ian, though all scores were deemed skillful. To the authors’ knowledge, this is the first instance of FSS scores used for ERO verification. These findings provide a framework for a wider study on forecast verification across hurricane rainfall scenarios, a key step towards improving how we communicate and characterize these hazards.