Abstract:Food demand is expected to rise significantly by 2050 due to the increase in population; additionally, receding water levels, climate change, and a decrease in the amount of available arable land will threaten food production. To address these challenges and increase food security, input cost reductions and yield optimization can be accomplished using yield precision maps created by machine learning models; however, without considering the spatial structure of the data, the precision map's accuracy evaluation assessment risks being over-optimistic, which may encourage poor decision making that can lead to negative economic impacts (e.g., lowered crop yields). In fact, most machine learning research involving spatial data, including the unmanned aerial vehicle (UAV) imagery-based yield prediction literature, ignore spatial structure and likely obtain over-optimistic results. The present work is a UAV imagery-based corn yield prediction study that analyzed the effects of image spatial and spectral resolution, image acquisition date, and model evaluation scheme on model performance. We used various spatial generalization evaluation methods, including spatial cross-validation (CV), to (a) identify over-optimistic models that overfit to the spatial structure found inside datasets and (b) estimate true model generalization performance. We compared and ranked the prediction power of 55 vegetation indices (VIs) and five spectral bands over a growing season. We gathered yield data and UAV-based multispectral (MS) and red-green-blue (RGB) imagery from a Canadian smart farm and trained random forest (RF) and linear regression (LR) models using 10-fold CV and spatial CV approaches. We found that imagery from the middle of the growing season produced the best results. RF and LR generally performed best with high and low spatial resolution data, respectively. MS imagery led to generally better performance than RGB imagery. Some of the best-performing VIs were simple ratio index(near-infrared and red-edge), normalized difference red-edge index, and normalized green index. We found that 10-fold CV coupled with spatial CV could be used to identify over-optimistic yield prediction models. When using high spatial resolution MS imagery, RF and LR obtained 0.81 and 0.56 correlation coefficient (CC), respectively, when using 10-fold CV, and obtained 0.39 and 0.41, respectively, when using a k-means-based spatial CV approach. Furthermore, when using only location features, RF and LR obtained an average CC of 1.00 and 0.49, respectively. This suggested that LR had better spatial generalizability than RF, and that RF was likely being over-optimistic and was overfitting to the spatial structure of the data.

A Multiple Instance Dictionary Learning Approach for Corn Yield Prediction from Remote Sensing Data

Learning county from pixels: Corn yield prediction with attention-weighted multiple instance learning

A Deep Learning Approach to Conflating Heterogeneous Geospatial Data for Corn Yield Estimation: A Case Study of the US Corn Belt at the County Level.

Corn Yield Prediction based on Remotely Sensed Variables Using Variational Autoencoder and Multiple Instance Regression

DeepCropNet: a Deep Spatial-Temporal Learning Framework for County-Level Corn Yield Estimation

ESTIMATING CORN YIELD IN THE UNITED STATES WITH MODIS EVI AND MACHINE LEARNING METHODS

Optimal County-Level Crop Yield Prediction Using MODIS-based Variables and Weather Data: A Comparative Study on Machine Learning Models

A Temporal–Geospatial Deep Learning Framework for Crop Yield Prediction

Impact of Large Kernel Size on Yield Prediction: A Case Study of Corn Yield Prediction with SEDLA in the U.S. Corn Belt

Daily DeepCropNet: A Hierarchical Deep Learning Approach with Daily Time Series of Vegetation Indices and Climatic Variables for Corn Yield Estimation

A county-level soybean yield prediction framework coupled with XGBoost and multidimensional feature engineering

County-level corn yield prediction using supervised machine learning

Early Season Forecasting of Corn Yield at Field Level from Multi-Source Satellite Time Series Data

Multi-Stage Corn Yield Prediction Using High-Resolution UAV Multispectral Data and Machine Learning Models

Simultaneous corn and soybean yield prediction from remote sensing data using deep transfer learning

DeepCropMapping: A multi-temporal deep learning approach with improved spatial generalizability for dynamic corn and soybean mapping

Prediction of Maize Yield at the City Level in China Using Multi-Source Data

A Prediction Model of Maize Field Yield Based on the Fusion of Multitemporal and Multimodal UAV Data: A Case Study in Northeast China

Predicting County-Scale Maize Yields with Publicly Available Data.

Corn Grain Yield Prediction Using UAV-Based High Spatiotemporal Resolution Imagery, Machine Learning, and Spatial Cross-Validation

Predicting County Level Corn Yields Using Deep Long Short Term Memory Models