JERAJournal of Electronic Research and Application2208-35022208-3510Bio-Byword Scientific Publishing Pty. Ltd.10.26689/JERA.v10i8.15206ArticlePrediction of Grain Yield in Anhui Province Based on Multi-Source Remote Sensing Data Introductionhttps://artdesignp.com/journal/JERA/10/8/10.26689/JERA.v10i8.15206JiMengxin,JiangHaifeng20261082026-09-03Grain output forecasting is an important means to ensure national food security, and accurate output forecasts can provide scientific support for government decision-making. This paper takes Anhui Province as the research area, builds a Stacking ensemble learning model based on multi-source remote sensing data, and predicts grain output in Anhui. The study uses annual structured features (arable land area, sown area, electricity consumption, fertilizer usage, etc.) and monthly time series features (NDVI integral, NDVI mean, sunshine hours, etc.) to build a Stacking ensemble learning prediction model, integrating XGBoost, MLP, and RandomForest base models. XGBoost uses annual features to capture structured information at the annual scale, MLP uses monthly features to capture temporal variation patterns within the growing season, and RandomForest uses fusion features to capture interactions between features. By adaptively learning the weights of each base model through the Ridge regression meta-learner, effective integration of multi-source information is achieved. Experimental results show that the stacking ensemble model performs well under the retain-one cross-validation method: the determination coefficient reaches 0.9575, RMSE is 140.90 kg/ha, and average absolute percentage error MAPE is 4.08%. Compared to the XGBoost benchmark model that uses only annual features, the stacking model improves by 0.0039, and RMSE decreases by 6.21 kg/ha, validating the advantages of multi-source data fusion. Feature importance analysis showed that sown area and arable land area were the most critical factors affecting yield forecasting, together accounting for 60.4% of the characteristic importance. Meta-learner weight analysis shows that XGBoost is the main contributor (weight 1.0224), while MLP and RandomForest serve as auxiliary contributors (weights 0.0309 and 0.0230, respectively), providing dynamic growth season information that annual characteristics cannot capture. The research results of this paper provide a new method for grain output forecasting in Anhui Province, validating the application value of multi-source remote sensing data fusion in yield forecasting, which can offer scientific support for government decision-making and promote sustainable agricultural development.Grain production forecasting, Multi-source remote sensing data, Stacking integrated learning, XGBoost, MLP

[1]       Gao X, Dong Y, Xu W, et al., 2025, Analysis and Prediction of Spatiotemporal Changes in Grain Production in Central Asia Based on ARIMA Model. Journal of University of Chinese Academy of Sciences, 42(4): 472–486. (in Chinese) [2]       Zhang W, Sun D, Wang Y, et al., 2019, Prediction of Grain Yield in Liaoning Province Based on Support Vector Machine. Economic Mathematics, 36(1): 96–99. (in Chinese) [3]       Dong M, Miao L, 2025, Research on Grain Yield Prediction Based on ARIMA-X-RFR Model. Journal of Jiamusi University (Natural Science Edition), 43(11): 21–24. (in Chinese) [4]       Jia M, Cai Z, Hu J, et al., 2021, Research on Grain Yield Prediction Model Based on Machine Learning. Journal of Hebei Agricultural University, 44(3): 103–108. (in Chinese) [5]       Shen Y, Mercatoris B, Cao Z, et al., 2022, Improving Wheat Yield Prediction Accuracy Using LSTM-RF Framework Based on UAV Thermal Infrared and Multispectral Imagery. Agriculture, 12(6): 892. [6]       Wang X, Huang J, Feng Q, et al., 2020, Winter Wheat Yield Prediction at County Level and Uncertainty Analysis in Main Wheat-Producing Regions of China with Deep Learning Approaches. Remote Sensing, 12(11): 1744. [7]       Zhao Y, Fan L, Qin Z, et al., 2025, Analysis of Influencing Factors and Combined Prediction of Grain Yield Based on SVR and CNN-LSTM-ATTENTION Model. Journal of the Chinese Cereals and Oils Association: 1–12. (in Chinese) [8]       Ma D, Zhao J, Yan W, et al., 2025, Grain Yield Prediction in Southern China Based on Stacking Ensemble Algorithm. Hubei Agricultural Sciences, 64(5): 155–159 + 184. (in Chinese) [9]       Zou Y, 2024, Research on Maize Yield Prediction in China Based on Machine Learning, thesis, Nanjing University of Information Science and Technology (in Chinese) [10]     Basso B, Ritchie J, Pierce F, et al., 2011, Spatial Validation of Crop Models for Precision Agriculture. Agricultural Systems, 68: 97–112. [11]     Liu J, He X, Wang P, et al., 2019, Early Prediction of Winter Wheat Yield Using Long Time Series Meteorological Data Combined with Random Forest Method. Transactions of the Chinese Society of Agricultural Engineering, 35(6): 166–174. (in Chinese) [12]     Panov D, Sakharova E, 2022, Using Radar Data for Grain Crops Yield Forecasting in the Novosibirsk Region. Russian Meteorology and Hydrology, 47(6): 473–478. [13]     Zhang L, 2025, Research on Grain Yield Prediction Model Based on Deep Learning, thesis, Changchun University of Technology (in Chinese) [14]     Jeong J, Resop J, Mueller N, et al., 2016, Random Forests for Global and Regional Crop Yield Predictions. PLOS ONE, 11(6): 56–57. [15]     Brown J, Hochman Z, Holzworth D, et al., 2018, Seasonal Climate Forecasts Provide More Definitive and Accurate Crop Yield Predictions. Agricultural and Forest Meteorology, 260: 247–254. [16]     Chen C, Lin Z, Sun X, 2007, Dynamic Prediction and Suggestions for Grain Yield in Shandong Province. Journal of Anhui Agricultural Sciences, 35(28): 8773–8775. (in Chinese) [17]     Tian Q, Li J, Ding Y, et al., 2025, Machine Learning-Driven Grain Yield Prediction and Influencing Factor Analysis in Heilongjiang Province. Grain and Feed Science and Technology, 2025(2): 13–15. (in Chinese)