چكيده لاتين
Understanding the water resources status, particularly in arid and semi-arid regions, is crucial for long-term water resource planning and land use management. This study investigates the effects of scale changes in landuse/land cover (LULC) maps and different classification schemes on runoff (in the upstream of Zayandehrood Dam, between 51°02′ to 51°12′ E and 33°11′ to 33°18′ N, covering 545419 hectares) using the SWAT model. To explore the impact of LULC map scale changes on runoff, twelve LULC maps were derived from free Landsat8, Sentinel2, and MODIS satellite images processed in Google Earth Engine (GEE) using the random forest classification algorithm. In general, images derived from Landsat 8 exhibited higher classification evaluation accuracy compared to images obtained from Sentinel2. Runoff simulated by the SWAT model at four hydrometric stations at the Zayandehrood Dam inlet, Qaleh Shahrokh, Pol Zaman Khan, and Cham Asman inlet for the 1992-2021 statistical period was calibrated and validated using a land cover map with a spatial resolution of 10 meters and the SUFI-2 algorithm available in the SWAT-CUP software. Calibration results indicated that the model (reference model) performed reasonably well in simulating runoff at all four hydrometric stations. The Nash-Sutcliffe coefficient for Zayandehrood Dam inlet, Qaleh Shahrokh, Pol Zaman Khan, and Cham Asman inlet were 0.54, 0.73, 0.78, and 0.55, respectively. Subsequently, the optimized parameters from the reference model calibration were applied to models using other LULC maps with spatial resolutions of 10, 20, 30, 60, 250, and 500 meters, and the calibration and validation processes were conducted for these models using the obtained parameters. Results showed that the lowest simulation error was observed in models with maps of 10 meters spatial resolution and then in models with maps of 30 meters (3% higher error). However, satisfactory simulation results were also achieved using maps with lower spatial resolution. Increasing the number of LULC classes positively impacts the performance of the rainfall-runoff model. Additionally, in areas with higher vegetation coverage, using models with more classes and higher spatial resolution images is recommended for rainfall-runoff models.
