A novel remote sensing-based calibration and validation method for distributed hydrological modelling in ungauged basins

Abstract

Distributed hydrological models allow spatial estimates of the main components of the water cycle. Consequently, they have been widely used in various applications. However, in-situ runoff observations are usually required to calibrate and validate these models, which largely limits their application in ungauged or poorly gauged basins. Satellite remote sensing (RS) provides temporally and spatially continuous data of water-related information, which makes it hold great potential to improve hydrological modeling. Against this background, we developed a novel RS-based calibration and validation method in this study for distributed hydrological modelling in ungauged basins. This method was demonstrated with the Soil and Water Assessment Tool (SWAT) model in Hala Lake basin, a closed watershed in Qinghai Province of China. The SWAT model was calibrated with RS-based terrestrial evapotranspiration (ET) products and validated with the lake water storage change (LWSC) retrieved from multi-mission satellite data and the basin water storage change retrieved from the Gravity Recovery and Climate Experiment (GRACE). The model calibration results demonstrated a Nash-Sutcliffe Efficiency (NSE) of 0.7 or higher in most sub-basins, proving the usability of RS products. The simulated ET results showed good agreement with two RS products, with an R² value of 0.8. Additionally, comparisons with GRACE data further validated the reliability of this method. This study demonstrates the significant potential in using multi-source RS satellite data for calibrating and validating models, as well as estimating monthly or annual runoff time series in data-scarce or ungauged basins.