Produtividade do milho por meio da avaliação espacial temporal de imagens orbitais

Abstract

Due to the new sensors able to determine new wavelengths in orbital platforms, the need for new studies on vegetation indices (VIs) and their correlation with grain productivity arises. Thus, the objective was to estimate the corn crop productivity linked to the spatial-temporal evaluation of orbital images in three agricultural years in a commercial property. The indexes used were: BNDVI - blue normalized difference vegetation index, NDVI - normalized difference vegetation index, NDRE - normalized difference vegetation index with red edge (Red-Edge), GNDVI - green normalized difference vegetation index, and PNDVI - pan normalized difference vegetation index with green, red and blue, with proposed changes in the formulas of the GNDVI and PNDVI indexes, replacing the red wavelength (Red) by the red edge (Red-Edge). As a data source, to calculate the IRs, the multispectral bands B2, referring to the blue wavelength, B3, to the green, B4, to the red, B5, to the edge of red (Red-Edge), and B8, to the near infrared, provided by the Sentinel-2A and Sentinel-2B imaging instruments, were used. Productivity data collection was obtained by grain harvesters consisting of satellite navigation systems, panel and on-board sensors. The soil texture was determined by clay content for all sampled points, and the precipitation rates were determined by weather station. In the year with the lowest hydric regime the clay content influenced the productivity. It was possible to observe a high correlation between vegetation indexes and productivity. It was verified that the replacement of red wavelength (Red) by red edge (Red-Edge) improves the correlation between the VIs, and that the BNDVI and GNDVI indexes quickly saturate, not allowing the distinction among classes. Thus, the vegetation index that best represented the average annual productivity estimate in corn crop were NDRE and GRENDVI.