Variabilidade da produtividade da cultura do milho utilizando imagens do satélite Landsat 8 em uma propriedade na região Noroeste do Mato Grosso

Abstract

The mapping of spatial variability of plants and soil is used as a fundamental tool for agricultural management in the field of precision agriculture. It was initially intended (Chapter 1) to investigate Pearson's linear correlation between maize crop productivity and NDVI generated from satellite images in a 64 ha plot with a mean yield of 101.16 sc.ha-1 and coefficient of variation of 11.31%, located in Tabaporã, northwest region of MT. The first evaluation process was carried out as follows: gross productivity map filtering, Landsat 8 satellite imagery selection, point-to-point NDVI pixel value extraction, based on productivity data and correlation analysis. The correlation result was 0.6 tested through the T test with significance of 1%. Considered by the authors a satisfactory result, a new work (Chapter 2) was conducted in which the objective was to evaluate the spatial variability of maize productivity, as well as the linear correlation between productivity and NDVI, in 3 consecutive harvests, 2015 , 2016 and 2017 on the same property in question. The Matlab program (2018A) was used to evaluate the correlation between variables. The coefficients of variation of productivity were relatively low (<10%) and productivity increased from 2015 to 2017. By 2016, 70% of the stands had correlation coefficients (between productivity and NDVI) above 0.3 tested at 1% significance (T test), where a greater number of stands were found in the ideal phenological stage (VT and R1) to identify the variability of productivity. The use of Landsat 8 satellite images was efficient in identifying the spatial variability of maize yield with a significant correlation coefficient of up to 0.78. However, satellite scanning time and cloudfree images limited the evaluation of plots with available yield data.