Implementing and Evaluating National Water Model Ensemble Streamflow Predictions Using Postprocessed Precipitation Forecasts

• DOI: 10.1175/jhm-d-24-0111.1
• Publication Date: 2025-02-24T17:30:11Z
• Authors: Erin Towler; Diana Stovern; Nachiketa Acharya; Mimi Rose Abel; William Ryan Currier; Joseph Bellier; Rob Cifelli; Kelly Mahoney; Carolien Mossel; Michael Scheuerer; Andrea Thorstensen; Francesca Viterbo
• Keywords: Hydrometeorology; Quantitative precipitation forecast; Ensemble forecasting; Hydrological modelling

Abstract Improving probabilistic streamflow forecasts is critical for a multitude of water-oriented applications. Errors in water arise from several sources, one which the driving meteorology. Meteorological are often statistically post-processed before being input into hydrologic models. Shifts towards ensemble weather prediction systems have propelled advances post-processing, providing an opportunity to enhance forecasting. This study’s purpose implement and evaluate impact coupling state-of-the-art precipitation post-processing techniques with process-based, spatially distributed National Water Model (NWM). The has two steps: first, calibrated using censored, shifted, gamma distribution approach, second, reordered copula technique. NWM focuses on flood forecasting, but date only been run time-lagged forecasts. We medium-range (∼7 day) forecasting mode rain-dominated catchments northern California during extremely wet year, when advanced warning heavy could useful. Post-processing enhances terms spread accuracy, improving underestimation. Precipitation (streamflow) was generally skillful out day 4 (7), including (>75mm) relatively high flow thresholds, less consistently most extreme streamflow. These results suggest that ensembles be warranted priority basins predictable phenomenon, though there tradeoffs hydrological model complexity study can inform NOAA-led Next Generation Resources Modeling Framework, will need consider how integrate meteorological techniques.