Improving Wheat Yield Prediction Accuracy Using LSTM-RF Framework Based on UAV Thermal Infrared and Multispectral Imagery

Yield prediction is of great significance in agricultural production. Remote sensing technology based on unmanned aerial vehicles (UAVs) offers the capacity non-intrusive crop yield with low cost and high throughput. In this study, a winter wheat field experiment three levels irrigation (T1 = 240 mm, T2 190 T3 145 mm) was conducted Henan province. Multispectral vegetation indices (VIs) canopy water stress (CWSI) were obtained using an UAV equipped multispectral thermal infrared cameras. A framework combining long short-term memory neural network random forest (LSTM-RF) proposed for predicting VIs CWSI from multi-growth stages as predictors. Validation results showed that R2 0.61 RMSE value 878.98 kg/ha achieved grain LSTM. LSTM-RF model better compared to LSTM n 0.78 684.1 kg/ha, which equivalent 22% reduction RMSE. The considered both time-series characteristics growth process non-linear between remote data data, providing alternative accurate modern management.