Winter wheat yield prediction using convolutional neural networks from environmental and phenological data

• Published in: Scientific reports Vol. 12; no. 1; pp. 3215 - 14
• Main Authors: Srivastava, Amit Kumar, Safaei, Nima, Khaki, Saeed, Lopez, Gina, Zeng, Wenzhi, Ewert, Frank, Gaiser, Thomas, Rahimi, Jaber
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 25.02.2022; Nature Publishing Group; Nature Portfolio
• Subjects: 631/449; 704/158; Agricultural production; Correlation coefficient; Crop yield; Crops; Deep learning; Genotypes; Humanities and Social Sciences; Learning algorithms; Machine Learning; multidisciplinary; Neural networks; Neural Networks, Computer; Predictions; Science; Science (multidisciplinary); Seasonal variations; Seasons; Soil; Soil moisture; Triticum; Triticum aestivum; Weather; Wheat; Wind speed; Winter wheat
• ISSN: 2045-2322; 2045-2322
• DOI: 10.1038/s41598-022-06249-w

Crop yield forecasting depends on many interactive factors, including crop genotype, weather, soil, and management practices. This study analyzes the performance of machine learning and deep learning methods for winter wheat yield prediction using an extensive dataset of weather, soil, and crop phenology variables in 271 counties across Germany from 1999 to 2019. We proposed a Convolutional Neural Network (CNN) model, which uses a 1-dimensional convolution operation to capture the time dependencies of environmental variables. We used eight supervised machine learning models as baselines and evaluated their predictive performance using RMSE, MAE, and correlation coefficient metrics to benchmark the yield prediction results. Our findings suggested that nonlinear models such as the proposed CNN, Deep Neural Network (DNN), and XGBoost were more effective in understanding the relationship between the crop yield and input data compared to the linear models. Our proposed CNN model outperformed all other baseline models used for winter wheat yield prediction (7 to 14% lower RMSE, 3 to 15% lower MAE, and 4 to 50% higher correlation coefficient than the best performing baseline across test data). We aggregated soil moisture and meteorological features at the weekly resolution to address the seasonality of the data. We also moved beyond prediction and interpreted the outputs of our proposed CNN model using SHAP and force plots which provided key insights in explaining the yield prediction results (importance of variables by time). We found DUL, wind speed at week ten, and radiation amount at week seven as the most critical features in winter wheat yield prediction.