Comparative assessment of environmental variables and machine learning algorithms for maize yield prediction in the US Midwest

Abstract

Crop yield estimates over large areas are conventionally made using weather observations, but a comprehensive understanding of the effects of various environmental indicators, observation frequency, and the choice of prediction algorithm remains elusive. Here we present a thorough assessment of county-level maize yield prediction in U.S. Midwest using six statistical/machine learning algorithms (Lasso, Support Vector Regressor, Random Forest, XGBoost, Long-short term memory (LSTM), and Convolutional Neural Network (CNN)) and an extensive set of environmental variables derived from satellite observations, weather data, land surface model results, soil maps, and crop progress reports. Results show that seasonal crop yield forecasting benefits from both more advanced algorithms and a large composite of information associated with crop canopy, environmental stress, phenology, and soil properties (i.e. hundreds of features). The XGBoost algorithm outperforms other algorithms both in accuracy and stability, while deep neural networks such as LSTM and CNN are not advantageous. The compositing interval (8-day, 16-day or monthly) of time series variable does not have significant effects on the prediction. Combining the best algorithm and inputs improves the prediction accuracy by 5% when compared to a baseline statistical model (Lasso) using only basic climatic and satellite observations. Reasonable county-level yield foresting is achievable from early June, almost four months prior to harvest. At the national level, early-season (June and July) prediction from the best model outperforms that of the United States Department of Agriculture (USDA) World Agricultural Supply and Demand Estimates (WASDE). This study provides insights into practical crop yield forecasting and the understanding of yield response to climatic and environmental conditions.

Assimilation of soil moisture and canopy cover data improves maize simulation using an under-calibrated crop model

“…Despite this, DA SM+CC still significantly improves yield estimates in almost all years (nRMSE = 15.74%) with good accuracy (Jamieson et al, 1991). The lead time and prediction accuracy are comparable to other yield prediction studies (Peng et al, 2018;Kang et al, 2020). Here only the assimilation of data within 90 days after planting (around 3 months before harvest) was tested.…”

Cereal Yield Forecasting with Satellite Drought-Based Indices, Weather Data and Regional Climate Indices Using Machine Learning in Morocco

“…Recently, several studies have examined the performance of machine learning algorithms such as Support Vector Machine (SVM), Random Forest (RF), eXtreme Gradient Boost (XGBoost), Artificial Neural Network (ANN) and Long-Short Term Memory (LSTM) for yield forecasting at county or province scales. They have used multi-source data as predictors, and they found that the non-linear machine learning methods showed a better performance for yield forecasting than the linear approach [48][49][50][51][52][53][54]. Schwalbert et al [55] have used different machine learning algorithms (linear regression, RF and LSTM) to predict soybean yield at the municipality level in Brazil by using remote sensing data (NDVI, EVI, LST) and precipitation as predictors.…”

Machine Learning-Based Modeling of Spatio-Temporally Varying Responses of Rainfed Corn Yield to Climate, Soil, and Management in the U.S. Corn Belt

“…Understanding how different factors (e.g., climate, soil, managements) affect crop yield has critical values to scientific research and practical applications. Identifying and quantifying the relationships between crop yield and various factors allows for better ways to close the yield gap, increase yield potentials (Lobell et al, 2009;van Ittersum et al, 2013), and improve the predictive capability of crop yield for both short-run commodity market and long-term climate change adaptation (Schlenker and Roberts, 2009;Cai et al, 2017;Peng et al, 2018;Li et al, 2019;Kang et al, 2020). The U.S. Corn Belt produces about ∼30% of the total global corn production and plays the most critical role in the global corn export.…”

“…How environment and managements affect corn yield is a classic question that has been studied extensively by agronomists, plant biologists, economists, and recently earth system scientists (Cassman, 1999;Al-Kaisi and Yin, 2003;Kucharik, 2003;Schlenker and Roberts, 2009;Subedi and Ma, 2009;Lobell et al, 2014;Kang et al, 2020). However, there are still a few gaps.…”