Multisensor analysis of discrepancies between vegetative vigor and grain yield in precision maize crop management

Abstract

Imagery from sensors embedded in satellites enables low-cost crop analysis and has been the subject of correlation studies between vegetation indices and productivity. Vegetation indices obtained from orbital platforms and crop maps have been important tools in the context of popularizing precision agriculture. However, there are many factors that affect maize yields and the resulting harvest maps. As a result, correlations between vegetation indices and yields are not always obtained. This leaves a gap for methodologies to identify areas of non-correlation and investigate the possible causes in a targeted manner. The aim of this study was to use freely available satellite images, together with yield data from a maize harvester, to identify regions with and without a correlation between yields and vegetation indices. In areas with correlation, a linear model of yield as a function of NDVI was obtained. A map of discrepancies was calculated, in which most of the crop was correlated, with yields varying by around 2 Mg ha⁻¹ in relation to the model. Areas with discrepant yields were identified, both negatively and positively in relation to the model, enabling a localized investigation into the possible causes of the phenomenon and crop management.