Evaluation of Meteorological Ensemble Forecast Processor (MEFP) using inputs, parameters, and outputs at various times of day

The Meteorological Ensemble Forecast Processor (MEFP) is a statistical post-processor within the Hydrologic Ensemble Forecast Service (HEFSv1) of the U.S. National Weather Service (NWS). The MEFP consumes raw precipitation and temperature forecasts from a range of different sources and produces bias-corrected forecasts as an input to hydrologic models. The MEFP depends on the time-of-day in several ways, including the issue time of forecasts used in parameter estimation (or calibration), the issue time of operational raw forecasts used as an input to MEFP, and the issue time of the MEFP forecasts. We evaluate the quality of the MEFP precipitation forecasts for various combinations of the issue times by conducting hindcasting with the MEFP using the NWS Weather Prediction Center (WPC) Quantitative Precipitation Forecasts (QPF) for basins across the US. Three questions were posed: Q1) When the issue time of the MEFP calibration dataset is different from the issue time of the MEFP forecast, what is the benefit of using more recent operational raw forecasts; Q2) When the issue time of the MEFP calibration dataset is the same as the issue time of the MEFP forecast, what is the benefit of using more recent operational raw forecasts; and Q3) When the same issue time is used for all three datasets (MEFP calibration, raw forecasts, and MEFP forecasts), how is the quality of the MEFP forecasts related to the time of day? Results suggested that, regarding Q1 and Q2, using a more recent operational QPF as an input to the MEFP generally produced a more skillful MEFP forecasts, and that, regarding Q3, within HEFSv1, the 12Z WPC QPF produced slightly better forecasts than 0Z WPC QPF. This presentation summarizes and discusses findings from hindcasting and validation experiments.