How skillful are high-resolution ensemble means in capturing warm-season extreme precipitation events?

Extreme precipitation from warm-season convection remains a well-known operational forecasting challenge. Oftentimes these events occur from high-intensity, short-duration rainfall that can lead to flash-flooding and related impacts detrimental to life and property. Deterministic convection-allowing models (CAMs) provide guidance that can be useful for anticipating these kinds of events, but can fail to represent the range of possible forecast scenarios. An ensemble of CAMs, like the NCEP High Resolution Ensemble Forecast (HREF) system provides both high-resolution guidance and forecast uncertainty information.

The HREF outputs short-range (0–48 h) probabilistic forecasts at high spatial (~3 km grid spacing) and temporal (1 h) resolution. It can provide useful guidance for high-impact and extreme precipitation events, for which the skill of individual deterministic CAMs can be limited. In this presentation, the performance of the HREF ensemble mean, probability-matched mean (pmm), and localized probability matched mean (lpmm) precipitation forecasts will be assessed in terms of how well they forecast various extreme rainfall thresholds and average recurrence interval exceedances. Rainfall totals over 1-, 3-, and 6-hour durations from these forecasts will be compared to observed totals from the Multi-Radar Multi-Sensor (MRMS) Quantitative Precipitation Estimates (QPE) Pass 2 during the warm-season convective months of June-August for 2021, 2022, and 2023. The results of this work are intended to inform how, when, and where to apply HREF ensemble-mean forecasts in anticipation of an extreme precipitation event.