Search

Timely and cost-effective crop yield prediction is vital in crop management decision-making. This study evaluates the efficacy of Unmanned Aerial Vehicle (UAV)-based Vegetation Indices (VIs) coupled with Machine Learning (ML) models for corn (Zea mays) yield prediction at vegetative (V6) and reproductive (R5) growth stages using a limited number of training samples at the farm scale. Four agronomic treatments, namely Austrian Winter Peas (AWP) (Pisum sativum L.) cover crop, biochar, gypsum, and fallow with sixteen replications were applied during the non-growing corn season to assess their impact on the following corn yield. Thirty different variables (i.e., four spectral bands: green, red, red edge, and near-infrared and twenty-six VIs) were derived from UAV multispectral data collected at the V6 and R5 stages to assess their utility in yield prediction. Five different ML algorithms including Linear Regression (LR), k-Nearest Neighbor (KNN), Random Forest (RF), Support Vector Regression (SVR), and Deep Neural Network (DNN) were evaluated in yield prediction. One-year experimental results of different treatments indicated a negligible impact on overall corn yield. Red edge, canopy chlorophyll content index, red edge chlorophyll index, chlorophyll absorption ratio index, green normalized difference vegetation index, green spectral band, and chlorophyll vegetation index were among the most suitable variables in predicting corn yield. The SVR predicted yield for the fallow with a Coefficient of Determination (R2) and Root Mean Square Error (RMSE) of 0.84 and 0.69 Mg/ha at V6 and 0.83 and 1.05 Mg/ha at the R5 stage, respectively. The KNN achieved a higher prediction accuracy for AWP (R2 = 0.69 and RMSE = 1.05 Mg/ha at V6 and 0.64 and 1.13 Mg/ha at R5) and gypsum treatment (R2 = 0.61 and RMSE = 1.49 Mg/ha at V6 and 0.80 and 1.35 Mg/ha at R5). The DNN achieved a higher prediction accuracy for biochar treatment (R2 = 0.71 and RMSE = 1.08 Mg/ha at V6 and 0.74 and 1.27 Mg/ha at R5). For the combined (AWP, biochar, gypsum, and fallow) treatment, the SVR produced the most accurate yield prediction with an R2 and RMSE of 0.36 and 1.48 Mg/ha at V6 and 0.41 and 1.43 Mg/ha at the R5. Overall, the treatment-specific yield prediction was more accurate than the combined treatment. Yield was most accurately predicted for fallow than other treatments regardless of the ML model used. SVR and KNN outperformed other ML models in yield prediction. Yields were predicted with similar accuracy at both growth stages. Thus, this study demonstrated that VIs coupled with ML models can be used in multi-stage corn yield prediction at the farm scale, even with a limited number of training data. [ABSTRACT FROM AUTHOR]

As the world's leading corn producer, the United States supplies more than 30% of the global corn production. Accurate and timely estimation of corn yield is therefore essential for commodity trading and global food security. Recently, several deep learning models have been explored for corn yield forecasting. Despite success, most existing models only provide yield estimations without quantifying the uncertainty associated with the predictions. Also, the traditional deep learning approaches typically require a large training set and are easily prone to overfitting when the number of samples in the training set is relatively small. To address these limitations, in this study, we developed a county-level corn yield prediction model based on Bayesian Neural Network (BNN) using multiple data sources that are publicly available, including time-series satellite products, sequential climate observations, soil property maps, and historical corn yield records. Using preceding years since 2001 for model training, the developed BNN model achieved an average coefficient of determination (R2) of 0.77 for late-season prediction across the U.S. Corn Belt in testing years 2010–2019, and outperformed five other state-of-the-art machine learning models. Detailed evaluation in three representative testing years demonstrated that the proposed BNN model could accurately estimate corn yield not only in normal years but also in abnormal years when extreme weather events happened. Moreover, the timeliness of the prediction was evaluated within the growing season with an R2~0.75 achieved by middle August, which is about 2 months before the harvest. We also assessed the predictive uncertainty, and more than 84% of the observed yield records were successfully enveloped in the 95% confidence interval of the predictive yield distribution. Our results also showed that the uncertainty level decreased steadily as time proceeded and stabilized around early August. Uncertainties in yield prediction were mainly induced by the observation noise and also related to the interannual and seasonal variabilities of environmental stress such as heat and water stress. This paper provides a robust framework for the within-season prediction of crop yield and highlights the need to obtain a deeper understanding of the effects of environmental stress on agricultural productivity and crop yield estimation. • A Bayesian neural network was developed for corn yield and uncertainty estimation. • The developed model outperformed five widely used machine learning models. • The near-optimal performance was achieved 2 months before the harvest. • Predictive uncertainty could estimate the confidence level of yield prediction. • RS data noises and environmental stress increased predictive uncertainty. [ABSTRACT FROM AUTHOR]