A two-step iterative data assimilation and calibration approach for improving large-scale hydrological processes in the Ganga basin

In recent years, the increasing frequency and magnitude of Extreme Hydrological Events (EHEs), such as floods and droughts, have highlighted the need for a more precise understanding and early identification of these events. Land surface hydrological models are a popular choice for quantifying EHEs. However, their performance is strongly influenced by (1) limited availability of accurate, region-specific, and compatible input data, (2) inherent model assumptions and initial conditions, and (3) challenges in model calibration, particularly in regions where in-situ observations such as streamflow, used to match modeled and observed variables, are limited.<br/>In this study, we propose a two-step iterative data assimilation (DA) and calibration framework to reduce the influence of uncertainties in the Variable Infiltration Capacity (VIC) model within the Ganga basin. In the first step, the model state is updated by merging top surface soil moisture (SSM) simulations with satellite-based daily SSM data using a sequential Ensemble Kalman Filter (EnKF). Herein, model uncertainties are represented by a VIC ensemble run with perturbed meteorological forcings such as precipitation and temperature. In the second step, model performance is evaluated against observed in-situ streamflow data at multiple locations across the basin using the Nash-Sutcliffe Efficiency (NSE) and sensitive model parameters are adjusted. This two-step DA and calibration process is repeated iteratively until the model converges to satisfactory performance. An independent assessment of the new two-step approach on key model outputs such as streamflow, evapotranspiration (ET), and water storage changes is conducted to examine model performance at monthly, seasonal, and multi-year scale. The findings aim to support the development of hydrological monitoring systems with improved representation of regional to large-scale hydrological processes, thereby contributing towards building more efficient disaster mitigation strategies.