Attention Mechanism-Combined LSTM for Grain Yield Prediction in China Using Multi-Source Satellite Imagery

Grain yield prediction affects policy making in various aspects such as agricultural production planning, food security assurance, and adjustment of foreign trade. Accurately predicting grain is great significance ensuring global security. This paper based on the MODIS remote sensing image data products from 2010 to 2020, adds band information vegetation index temperature form composite a dataset. Aiming at lack models for large-scale forecasting need human intervention traditional models, this proposes estimation model deep learning. First, cropping mapping techniques are used process generate training samples. Then channel spatial attention mechanism (convolutional block module, CBAM) added extract different bands improve efficiency model. Long short-term memory (LSTM) neural networks obtain feature time dimension. Finally, national-scale constructed. After study, it was found that LSTM using combination multi-source satellite images an can effectively predict China. Furthermore, proposed tested 2018 2020 showing average R2 0.940 RMSE 80,020 tons, indicating Chinese better. The extracts directly data, solves problem small-scale research imprecise end-to-end manner.