Premium
Vegetation Indices for Discrimination of Soybean Areas: A New Approach
Author(s) -
Silva Carlos Antonio,
Nanni Marcos Rafael,
Teodoro Paulo Eduardo,
Silva Guilherme Fernando Capristo
Publication year - 2017
Publication title -
agronomy journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.752
H-Index - 131
eISSN - 1435-0645
pISSN - 0002-1962
DOI - 10.2134/agronj2017.01.0003
Subject(s) - normalized difference vegetation index , enhanced vegetation index , remote sensing , moderate resolution imaging spectroradiometer , vegetation (pathology) , spectroradiometer , environmental science , leaf area index , vegetation index , geography , agronomy , satellite , reflectivity , medicine , pathology , biology , physics , engineering , optics , aerospace engineering
Core Ideas Automation of mapping of soybean areas. Use of remote sensors in the recognition of summer crop. Development of exclusive vegetation index for soybean.The aim of this study was to map areas cultivated with soybean [ Glycine max (L.) Merr.] in Paraná state, Brazil, using mono‐ and multitemporal MODerate‐resolution imaging spectroradiometer (MODIS) images. We applied the vegetation index perpendicular crop enhancement index (PCEI) and threshold determination for the automation of soybean area discrimination by geo‐object (GEOBIA). For this mapping, vegetation indices (normalized difference vegetation index [NDVI], enhanced vegetation index [EVI], and crop enhancement index [CEI]) and the development of the PCEI were used with the aid of time‐series images from the TERRA/MODIS system‐sensor. A support analysis, based on geo‐objects and a decision tree based on data mining, was used to determine the new vegetation index. “Classification” and “merge region” algorithms and feature extraction were used for classification. To evaluate the precision of the classifications, the Kappa (κ) and overall accuracy (OA) parameters were applied. Regarding the ground line, R and R 2 were above 0.92 and 0.84, respectively ( p < 0.01). The test results indicate that the proposed methodology is efficient for mapping soybean distribution, with 0.80 for the Kappa parameter, an appropriate crop spatial distribution, and no over‐ or underestimation of areas. Thus, this study allows automated mapping of areas cultivated with soybean crops at large scales.