Operationalizing earth observation based machine-learning models for field-scale crop yield forecasting

Accurately determining crop growth progress and crop yields at field-scale can help farmers estimate their net profit and finance part of their input purchases, enable insurance companies to ascertain payouts, and help in ensuring food security. At field scales, the troika of management, soil and weather combine to impact crop growth progress, and this progress can be monitored in-season using satellite data. In a public-private partnership between NASA Harvest and the Argentinian agricultural monitoring company SIMA, we use satellite derived metrics, from both optical and radar satellites, and machine learning models to model field-scale crop yields for over 3,000 Soybean, Wheat and Maize fields in Argentina. We compare yield forecast capabilities of several machine learning models, including lasso regression, generalized additive models, random forests and mixed effect random forest. Our results show the promise of combining mixed effect models with non-parametric models in improving yield modeling capabilities. We also demonstrate the utility of specific satellite derived metrics and extracted features in improving model performance and show that our approach can explain greater than 70% of the variation in yields (Figure 1) while remaining generalizable across crops and agro-ecological zones. We discuss our experiences in evaluating these model in Argentina and how to successfully transition a machine-learning model from the scientific domain to operational use.