The Role of Baseflow Data Assimilation in Hydrologic Modeling and Peak Flow Prediction

Abstract Recent studies have revealed that subsurface water storage and flow pathways contribute a substantially larger share of streamflow than previously assumed. However, because representations of these subsurface contributions in most hydrologic models are uncertain and often biased, they can strongly influence simulated streamflow generation and partitioning. Here, we use Hydrologic Generative Ensemble Data Assimilation method to merge baseflow observed data with the outputs of a hydrologic model to update the states associated with lower‐zone water storages. We show that after performing data assimilation across the eastern U.S., runoff partitioning shifts mostly toward higher baseflow contributions. Results show that baseflow assimilation helps better characterize the full hydrograph, with peak flows detected more consistently and accurately. The enhancements demonstrated in this study are achieved without modifying the model structure, rather they result from updating the model states, providing an effective approach to enhance the representation of subsurface information in hydrologic modeling systems.