On improving ensemble forecasting of extreme precipitation using the NWS Meteorological Ensemble Forecast Processor (MEFP)

Abstract:

Reservoir operations are sensitively dependent on unbiased forecasts of inflow volumes under extreme conditions. The quality of precipitation forecasts for large and extreme events is hence of particular importance in hydrologic ensemble forecasting. The Meteorological Ensemble Forecast Processor (MEFP), an integral part of the Hydrologic Ensemble Forecast Service being implemented at all NWS River Forecast Centers, is designed to input potentially biased quantitative precipitation forecasts (QPF) and produce unbiased ensemble precipitation forecasts that capture the skill in the conditioning QPFs. Operational experience suggests that, while generally unbiased for most events, the MEFP precipitation ensembles tend to underforecast extreme observed precipitation. In this presentation, we describe two recent efforts to diagnose and address the conditional bias using the GEFS reforecast dataset for selected basins in California: 1) a regionalization approach to MEFP calibration whereby data are pooled from other locations with sufficiently similar statistical properties to increase sample size for extreme events and 2) analysis of sensitivity of MEFP ensembles to differently prescribed precipitation forcing and parameters in MEFP calibration. Also presented are the findings and possible research directions.