Using satellite and real-time weather data to predict maize production

Abstract

Large-scale assessments of crop conditions prior to harvest are critical for providing early estimates of production. Satellite and weather information provide the opportunity for near real-time crop monitoring. The objective of this research was to develop an operational assessment system for crop production utilizing data from these sources. Maize (Zea mays) production was assessed in 42 Crop Reporting Districts (CRDs) across the United States Corn Belt, which produce 60% of all maize grown in the United States. Satellite, climatolocal, and agricultural data were collected for 8 years, 1985–1992, and aggregated into CRDs. A model predicting the normalized maize yields for each CRD was developed that included as independent variables a satellite data variable, the Vegetation Condition Index, and a climatological variable, the Crop Moisture Index. This model explained approximately three-quarters (R 2=0.73) of the variation observed in the normalized yields, and was examined both for its accuracy and its timeliness in providing production estimates. Predicted seasonal yields were summed to provide a maize production estimate for the entire Corn Belt study region. Production estimates deviated from the final USDA statistics, which become available several months after harvest, by less than 10% for all eight growing seasons. In addition, the production estimates were available approximately 2 months prior to the completion of the maize harvest. This system has the potential for providing timely in-formation to organizations monitoring regional or global agricultural production for humanitarian or economic benefits.