How well do U.S. National Water Model short-range forecasts predict flood event timing and magnitude?

Continental United States (CONUS). The National Water Model (NWM), operational since 2016, provides real-time, continuous hydrologic forecasts across the CONUS. Prior studies have evaluated NWM performance, but comprehensive assessments across multiple regions, lead times, and settings remain limited. This study evaluates NWM version 2.1 short-range forecasts for multiple flood events during 2021–2023, analyzing performance across varying watershed characteristics, flood magnitudes (return periods), and lead times. We used data from 306 U.S. Geological Survey gauges across 16 study areas, several with multiple floods per site, to assess forecast accuracy in terms of hydrograph skill, peak discharge, flood volume, and time-to-peak bias. Results show that forecast accuracy improves with shorter lead times and smaller floods but varies by watershed traits and climate. Systematic underestimation of peak discharge and flood volume occurred across all basin types. Urban, regulated, arid, and low-order watersheds tend to show higher forecast errors. We qualitatively inspected hydrograph shapes, finding that while many forecasts captured rising and falling limbs, some exhibited systematic anomalies, such as consistently declining, delayed forecasts, and failure to detect sharp flood peaks, highlighting structural issues in model response. Findings provide insight into the strengths and limitations of NWM short-range flood forecasts and offer a baseline for evaluating future NWM versions and the emerging NextGen modeling framework. • The NWM underestimates peak discharge and flood volume across watersheds. • Forecast accuracy improves with shorter lead times but varies by basin traits. • Peak forecast timing errors shift from early at long lead times to near-zero at short leads. • Urban, regulated, arid, small basins and high-magnitude floods show larger errors. • Certain locations showed systematic anomalies in the shapes of forecast hydrographs.