Improving Precipitation Forcings for the National Water Model

The National Weather Service's Office of Water Prediction (NWS/OWP) produces operational hydrologic products, many of which are generated by the National Water Model (NWM). NWM analysis cycles (also known as "near-real-time" or "update" cycles) are of key importance, since the land surface states and fluxes they produce are used to initialize all forecast cycles. Among all forcing fields (which include precipitation, temperature, humidity, radiation, and wind), precipitation is particularly important. Currently, NWM precipitation forcings for analysis cycles are generated by combining hourly radar-derived precipitation products from the Multi-Radar, Multi-Sensor (MRMS) system with short-term quantitative precipitation forecasts (QPF) from the Rapid Refresh (RAP) and High Resolution Rapid Refresh (HRRR) systems. Short term QPF is used in analysis cycles to fill coverage gaps in MRMS products, and its inclusion is necessary due to the short latency associated with NWM analysis cycles relative to the availability of other operational precipitation analyses. This presentation will describe the methodology used to remove QPF bias and to spatially merge MRMS, HRRR, and RAP into hourly forcing inputs for NWM version 2.0, expected to enter into operations in late 2018. The accuracy of version 2.0 precipitation forcings relative to reference data sources, and the degree to which these forcings will represent an improvement over those used to drive the previous NWM version (1.2), will be described.