In-Season Crop Mapping for the Continental United States

Credible predictive modeling of the spatial distribution of crops is a critical first step in near-real time crop monitoring. The United States (US), one of the largest food producing nations, currently lacks a publicly-available, high-resolution crop data product during the growing season. The US Department of Agriculture (USDA) provides the Cropland Data Layer (CDL), a crop-specific land cover map for the Continental United States (CONUS), for the current year only by the spring of the following year. The aim of this study was to generate in-season crop maps for the CONUS and estimate the earliest time by which major crop types can be mapped with sufficient accuracy. The study employed a scalable generalized classifier trained on multiple years of ground truth and annual trajectories of Normalized Difference Vegetation Index (NDVI) derived from the MODIS sensor aboard the Terra and Aqua satellites. A crop map generated early in the growing season in conjunction with seasonal climate forecasts can be used to estimate end-of-season crop yield over regional to national scales, which could act as a valuable decision support aid for multiple public sector and federal agencies.