基于多源遥感与SWAT模型协同的径流模拟研究OA

To reduce the dependency of hydrological models on in-situ streamflow observations and mitigate simulation inaccuracies arising from sparse gauge distributions,this study utilized multi-source remote sensing data for the calibration and validation of a hydrological model to simulate runoff in the Ganjiang River Basin.Operated on the Google Earth Engine(GEE)platform,the research integrated Landsat-5/7/8 and Sentinel-1/2 satellite imagery.A high-precision time series of river width was extracted using the modified normalized difference water index(MNDWI)in conjunction with an adaptive threshold algorithm.Subsequently,a power-law function was employed to establish a river width-discharge inversion model.The discharge data derived from this model were used as input for the soil and water assessment tool(SWAT)model for integrated calibration and validation.Results demonstrate that the SWAT model,driven by the multi-source remote sensing-inverted discharge,achieves satisfactory agreement with observed values during both the calibration and validation periods,with Nash-Sutcliffe efficiency(NSE)values of 0.84 and 0.80,respectively,and coefficients of determination(R2)exceeding 0.85.These outcomes indicate that the proposed methodology can effectively support runoff simulation in ungauged or data-scarce basins.This study provides a novel methodological framework and practical pathway for simulating hydrological processes and managing water resources in the Ganjiang River Basin and other regions confronted with similar data scarcity challenges.Conventional hydrological modeling heavily depends on in-situ streamflow measurements for parameter estimation,which introduces considerable uncertainty in basins with limited monitoring infrastructure.Remote sensing technology offers a viable alternative by enabling continuous,spatially distributed monitoring of key hydrological variables.This study specifically focuses on the relationship between river width and discharge,a fundamental component of hydraulic geometry,to reconstruct streamflow dynamics temporally.Data processing is conducted within the GEE cloud platform,which is selected for its comprehensive satellite data archive and computational efficiency.Multi-temporal imagery from Landsat-5 TM,Landsat-7 ETM+,and Landsat-8 OLI/TIRS,as well as Sentinel-1 SAR and Sentinel-2 MSI,is integrated to enhance temporal resolution and minimize gaps caused by cloud contamination.The MNDWI is applied to optical imagery to accentuate water features while suppressing interference from vegetation and built-up areas.For Sentinel-1 SAR data,which is impervious to cloud cover,a threshold-based classification is implemented to distinguish water from land based on backscatter intensity.An adaptive thresholding technique,accounting for seasonal and spatial heterogeneity,is applied to derive binary water masks,from which river widths at a predetermined cross-section are systematically extracted.A power-law regression model,expressed as Q=aWb,is calibrated using limited synchronous gauge measurements to establish a robust empirical relationship between river width(W)and discharge(Q).This model is subsequently applied to the entire river width time series to reconstruct a continuous discharge record.This remotely sensed discharge series serves as the reference dataset for calibrating the SWAT model,a process-based and semi-distributed hydrological model implemented for the Ganjiang River Basin.Digital elevation models,land cover data,and soil data are incorporated into the model setup.Calibration and validation are performed against the satellite-derived flows,significantly reducing reliance on conventional ground-based data.The high model performance,as evidenced by NSE values of 0.84 and 0.80 during calibration and validation,confirms that the integration of multi-source remote sensing data for discharge inversion provides a reliable alternative for hydrological modeling in gauge-sparse regions.This approach presents a transferable framework for supporting hydrological analysis and water resource management in ungauged or poorly gauged basins globally.