Improving Wheat Yield Prediction with Multi-Source Remote Sensing Data and Machine Learning in Arid Regions

Authors: RAZA, AAMIR - University Of Agriculture, Faisalabad; SHAHID, MUHAMMAD - University Of Agriculture, Faisalabad; ZAMAN, MUHAMMAD - University Of Agriculture, Faisalabad; MIAO, YUXIN - University Of Minnesota; Huang, Yanbo; SAFDAR, MUHAMMAD - University Of Agriculture, Faisalabad; MAQBOOL, SHERAZ - University Of Agriculture, Faisalabad; MUHAMMAD, NALAIN - University Of Agriculture, Faisalabad

Submitted to: Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/21/2025
Publication Date: 2/23/2025
Citation: Raza, A., Shahid, M.A., Zaman, M., Miao, Y., Huang, Y., Safdar, M., Maqbool, S., Muhammad, N.E. 2025. Improving Wheat Yield Prediction with Multi-Source Remote Sensing Data and Machine Learning in Arid Regions. Remote Sensing. 17(774). https://doi.org/10.3390/rs17050774.
DOI: https://doi.org/10.3390/rs17050774

Interpretive Summary: Accurate and timely prediction of wheat yield is crucial to crop production management in arid regions. This study used multisource remote sensing data to predict wheat yield at different growth stages with different machine learning models. With the work the predictive accuracy of the machine learning models were evaluated and the appropriate time frame was determined for wheat yield prediction in arid regions. Also, in this study the impact of climate parameters on model accuracy was investigated. This study indicates that combining multi-source data and machine learning models is a promising approach for wheat yield prediction in arid regions.

Technical Abstract: Wheat (Triticum aestivum L.) is one of the world’s primary food crops, and accurately predicting crop yield is crucial for regional trade and national food security. With the growing importance of integrating multi-source remote sensing data and machine learning techniques, there is a need to develop a simple, timely, and accurate model for predicting wheat yield at the administrative unit level. Many previous studies have focused on remote sensing data, climate data, and their com-binations. However, the best indices and exact time window for wheat yield prediction are still not clear in arid regions and have not been fully explored. The specific objectives of this study were to 1) assess the performance of multiple remote sensing indices (NDVI, EVI, ARVI) to predict wheat yield at different growth stages; 2) evaluate the predictive accuracy of four advanced machine learning models; 3) determine the appropriate time frame for wheat yield prediction in arid regions; and 4) evaluate the impact of climate parameters on model accuracy. In this study, three remote sensing indices, Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), and Atmospheric Resistance Vegetation Index (ARVI), and four machine learning models, Decision Tress (DTs), Random Forest (RF), Gradient Boosting (GB), and Bagging Trees (BTs), were selected to assess their predictive accuracy for wheat yield in the arid region by separating whole wheat growth period into three-time windows. A modeling framework was developed using the Google Earth Engine (GEE) platform, integrating climate and remote sensing data to predict wheat yield. The results showed that the RF model, particularly with ARVI, could accurately predict wheat yield at the grain filling and the maturity stages in arid regions with an R2 > 0.75 and yield error less than 10%. RF model with ARVI provided the most accurate wheat yield predictions. GB model with EVI yielded slightly lower precision but still outperformed all other models. It is concluded that combining multi-source data and machine learning models is a promising approach for wheat yield prediction in arid regions.