Authors: Iman Maghami, Daniel P. Ames, Jacob Anderson, Easton Perkins, Jerson Garcia, Abin Raj Chapagain, Roja Najafi – Department of Civil and Construction Engineering, Brigham Young University, Provo, Utah, USA
Title: A Multi-Year Multi-Region Analysis of Flood Timing, Magnitude and Severity in Comparison to the National Water Model Forecasts
Abstract: Flooding, a significant global hazard, results in substantial human and infrastructure losses annually. Flash floods, caused by intense precipitation, pose particularly severe threats due to their rapid onset, allowing minimal time for response. Robust real-time flood forecasting systems facilitate timely and effective response measures to mitigate the adverse impacts of such catastrophic events in vulnerable areas. In the United States, the NOAA National Weather Service (NWS) Office of Predictions (OWP) has developed the National Water Model (NWM), providing near real-time streamflow forecasts and land surface simulations. While a growing body of research has started assessing the accuracy of the NWM for retrospective and forecast streamflow outputs, a comprehensive evaluation of its effectiveness in predicting floods from intense precipitation events is lacking. This study assesses the effectiveness of the NWM Version 2.1 in predicting floods resulting from extreme short-duration rainfall events across diverse regions of the Continental United States (CONUS) for the years 2021-2022. Through a comparison of NWM forecasts with USGS observational streamflow data, performance is evaluated under various temporal and spatial conditions, considering different lead times and basin characteristics. Twenty major flash flood events, primarily in urban areas, were identified for analysis, and historical NWM forecast data was retrieved. Performance metrics, including modified Kling–Gupta efficiency (KGE), were employed to assess peak flow magnitude, timing, and their relationship with flood forecast lead time. Basins were delineated and their characteristics were gathered using tools such as the USGS StreamStats Batch Processing Tool. The results also focus on correlations between basin characteristics and flood forecast performance metrics. The study’s methodologies and workflows, presented as Jupyter notebooks, aim to enhance accessibility and reproducibility for researchers and decision-makers. We finally discuss opportunities to use these results to enhance interpretation and application of the NWM forecasts in operational flood management.