River discharge monitoring sites are scarce in high mountain basins, especially on the Tibetan Plateau where rivers are widely distributed. The potential for estimating river discharge using remote sensing has been widely utilized for large continental rivers, while its application to smaller regional rivers has been limited. To address this issue, we investigate the reliability of calibrating a hydrological model for two small subbasins lacking discharge data in the upstream region of the Heihe River located on the Tibetan Plateau, based solely on multiple pixel ratio (MPR) data derived from near-infrared satellite images. Using six independent calibration schemes based on discharge or MPR data, the value of MPR data for hydrological modeling is explored. We first calibrate a model using observed discharge data and Nash-Sutcliffe efficiencies (NSEs) based on non-dominated sorting genetic algorithm II (Q-NSE scheme). We then use estimated discharge values derived from MPR data and identify different parameter sets that result in the highest NSEs in a Monte Carlo-based uncertainty framework (Q-NSE-1, Q-NSE-2 and Q-NSE-3 schemes). Spearman rank coefficients between MPR data and observed discharge data are then used as an alternative way to calibrate the model (MPR-SR scheme). The MPR-SR scheme is also applied with a simple water-balance filter, which removes parameter sets, which result in either underestimated or overestimated total discharge volumes (MPR-SRF scheme). The Babao River Basin and Yeniugou River Basin are used to compare the different calibration schemes and to perform parameter sensitivity analyses. The MPR-SRF calibration scheme achieved better performance in both river basins, with NSEs of 0.63 and 0.62, respectively, during the validation period. This study demonstrates the potential of the MPR-SRF calibration scheme to enhance models of small ungauged river basins lacking ground discharge observation data, and provides insights into how remote sensing data can be more effectively combined with hydrological modelling.
Read full abstract