Estimation of reclaimed crop yield in the dump of open-pit coal mine in the black soil region based on unmanned aerial vehicle remote sensing

Land reclamation mitigates the impact of coal mining on grain production in cultivated land.Early prediction of reclaimed crop yields facilitates precision field management and ensures food security.Advances in unmanned aerial vehicle (UAV) platforms and sensor technology enable the acquisition of high spatial and spectral resolution data, making rapid yield estimation feasible.The analysis focuses on soybeans cultivated in reclaimed open-pit coal mine waste dumps within the black soil region. This study evaluates soybean yield estimation models using four methods—multiple linear regression(MLR), random forest regression(RFR), backpropagation neural network(BPNN), and support vector regression(SVR)—with vegetation index (VI), texture index (TI), and combined vegetation-texture index (VTI) as feature variables.The results show that: (1) Compared to using only VI or TI as feature variables, models incorporating both VI and TI demonstrated improved accuracy; (2) Support Vector Regression (SVR) exhibited the best stability and accuracy in yield estimation modeling, achieving R²=0.84, RMSE=0.003 9, and RPD=2.53.This research findings provide technical support for estimating soybean yields in the dump of open-pit coal mine in the black soil region and offer data-driven insights for precision field management and reclamation effectiveness evaluation.