Abstract: Non-point source pollution is caused by rainfall or snowmelt moving over and through the ground. As runoff moves, it picks up and carries away natural and human-made pollutants, finally depositing them in lakes, rivers, wetlands, coastal waters, and groundwater. Agricultural non-point source apportionment is the premise for preventing non-point source pollution. Crop planting is an important source of agricultural non-point sources. However, traditional non-point source apportionment methods cannot quantify the nutrient loads originating from different types of crops. The apportionment accuracy of the traditional methods does not satisfy the demand for more precise environmental management. This study selected the Shahe River Basin as the study area, and used remote sensing and hydrological models to apportion the total phosphorus (TP) load to establish a high-precision agricultural non-point source apportionment method. The results indicated that the classification accuracy based on the Google Earth Engine (GEE) was higher than 88%, the kappa coefficients were higher than 0.81, and the classification results for different crops were credible. The major crops in the Shahe River Basin include winter wheat-summer maize, chestnuts, fruit, and other crops. The winter wheat-summer maize system has the largest planting area, accounting for 44%–67%, and the planting area of fruit tree was the second largest, accounting for 11%–29%. Winter wheat-summer maize planting area is generally declining, chestnut planting area is rapidly rising trend. The Generalized Watershed Loading Function (GWLF) model of the Shahe River Basin performed well in the simulation of river runoff and total phosphorus loading, with the NSE of the model calibration and validation periods above 0.59 and the R² above 0.79. Farmland was the largest non-point source of TP load in the Shahe River Basin, accounting for 61% of the TP load. Among the farmland winter wheat-summer maize system accounted for 52% of the TP load from agricultural sources, while chestnuts planting contributed to the second largest share (20%), but considering that the planting area of chestnut has been increasing in recent years, the total TP load from surface sources in the Shahe River Basin is still at risk of increasing in the future.