DefinePK

Abstract

The monitoring of the Earth's surface and the dynamics of its vegetation using remote sensing techniques stands out in agricultural activities. The objective of this study was to estimate and map areas cultivated with soybean [Glycine max (L.) Merr.] by means of mono and time-series MODIS images in Paraná state through principal component techniques. For this mapping were used vegetation index (EVI and CEI) with the help of as time-series from images of MODIS sensor also was performed by supervised classification algorithms and partially unsupervised with use of principal component analysis. For statistical evaluation parameters were used Kappa and overall accuracy and their respective Z and t-tests. When analyzing the data obtained by the methods used in the estimates of soybean areas it appears that the ratings by the CEI index was highlighted with higher Kappa parameters (κ) and Overall Accuracy (OA), unlike the classifier K-means. For the principal component used five images including vegetation indices, presented to the Kappa 0.48 parameter. The mapping, discrimination and quantification of soybean fields in the state of Paraná was possible with the use of classifiers and MODIS images, which the systematization presented results of Kappa parameters and overall accuracy satisfactory.