SWAT-Based Hydrological Data Assimilation System (SWAT-HDAS): Description and Case Application to River Basin-Scale Hydrological Predictions

This paper presents the development and application of a physically based hydrological data assimilation system (HDAS) using the gridded and parallelized Soil and Water Assessment Tool (SWATGP) distributed hydrological model. This SWAT-HDAS software integrates remotely sensed data, including the leaf area index (LAI), snow cover fraction, snow water equivalent, soil moisture, and ground-based observational data (e.g., from discharge and ground sensor networks), with SWATGP and the Parallel Data Assimilation Framework (PDAF) to accurately characterize watershed hydrological states and fluxes. SWAT-HDAS employs high-performance computational technologies to address the computational challenges of high-resolution and/or large-area modeling. Multiple observational system simulation experiments (OSSEs), including soil moisture assimilation experiments, snow water equivalent assimilation experiments, and streamflow assimilation experiments, were designed to validate the assimilation efficiency of various types of observations within SWAT-HDAS using an ensemble Kalman filter (EnKF) algorithm. Both the temporal and spatial correlations in the trend/pattern and the magnitudes of improvement between the simulated and true states (i.e., for soil moisture, snow water equivalent, and discharge) were satisfactory using the integrated assimilation, which suggests the reliability of SWAT-HDAS for regional hydrology studies. The streamflow assimilation experiment also showed that the observation location dramatically influences the assimilation efficiency. The quantity and quality of observations have effects of varying degrees on the streamflow predictions. SWAT-HDAS is a promising tool for hydrological studies and applications under climate and environmental change scenarios.