مختصر البحث:
Wheat is one of the most important staple crops worldwide, and accurate prediction of its yield is crucial for ensuring food security, optimizing resource allocation, and informing agricultural policies. However, wheat yield is influenced by a compl…
Wheat is one of the most important staple crops worldwide, and accurate prediction of its yield is crucial for ensuring food security, optimizing resource allocation, and informing agricultural policies. However, wheat yield is influenced by a complex interplay of spatial and temporal factors such as weather variability, soil conditions, and crop phenology, which pose significant challenges for traditional prediction methods. Remote sensing technology offers a promising solution by providing frequent, objective, and large-scale data, yet effectively extracting meaningful information from such multi-dimensional and heterogeneous data remains a challenge. In this study, we propose a novel hybrid deep learning approach that integrates Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks to capture both spatial patterns and temporal dynamics from multi-source remote sensing and environmental datasets. A rigorous feature engineering and selection process, supported by Random Forest algorithms, is employed to identify the most relevant bioclimatic variables, vegetation indices, and cyclical temporal features, which significantly enhance the model’s predictive capability while reducing data dimensionality. The proposed model is evaluated on a comprehensive wheat yield dataset from northern China, demonstrating superior performance compared to several traditional and machine learning-based methods, including linear regression, random forest, and gradient boosting. Our approach achieves notable improvements in accuracy, robustness, and generalizability, highlighting its potential as an effective tool for timely and reliable wheat yield forecasting. This research contributes to the advancement of smart agriculture by providing a scalable and adaptable framework capable of handling complex agricultural datasets, ultimately supporting more informed decision-making and sustainable food production.
