Short-range Ensemble Streamflow Forecasting of the Upper Trinity River – Evaluation via Hindcasting Experiments

In this work, we evaluate the quality of short-range streamflow hindcasts for five headwater basins in the Upper Trinity River Basin in North Texas. The hindcasts are generated from the Hydrologic Ensemble Prediction Service (HEFS) of the National Weather Service (NWS). The HEFS operates as part of the Community Hydrologic Prediction System (CHPS) of the NWS. Currently, the HEFS models the input and hydrologic uncertainties using the Meteorological Ensemble Forecast Processor (MEFP) and the Ensemble Post-Processor (EnsPost), respectively. For generation of short-range ensemble precipitation forecasts, the MEFS models the conditional probability distribution of observed precipitation given the single-valued quantitative precipitation forecasts (QPF). For hydrologic uncertainty, EnsPost models the conditional distribution of observed streamflow given the simulated streamflow and the most recently observed streamflow. We design and carry out hindcasting experiments to evaluate the quality of both precipitation and streamflow ensembles, and to assess the value of short-range ensemble streamflow forecasts relative to the existing single-valued forecasts. For the above, we used the Ensemble Verification System (EVS) also developed by the NWS. We describe the hindcasting experiments, summarize the results, and identify issues and challenges.