Análisis multitemporal de la cobertura vegetal a partir de la imagen satelital Sentinel-2 y Landsat-8 mediante el uso de algoritmos de clasificación, en la Reseva Ecológica El Ángel y Reserva Ecológica Arenillas

Abstract

The present research was based on the multitemporal analysis of the vegetation cover of Sentinel-2 and Landsat-8 satellite images of the El Angel Ecological Reserve from 2016 to 2017 and the Arenillas Ecological Reserve from 2015-2019 through the use of supervised classification. The methodology consisted of determining the study area followed by image acquisition through the United States Geological Survey platform, spatial data processing was performed by preprocessing (cropping, rescaling, band combination, radiometric correction, atmospheric correction), using the SNAP software for the Sentinel-2 sensor and the ERDAS software for the Landsat-8 sensor, followed by processing by means of two classification algorithms, Support Vector Machines and Artificial Neural Networks executed in the ENVI 5.3 software. The results obtained were validated by calculating the confusion matrix, for the present study it was determined that the best algorithm was MSV. The results obtained reflect the detection of changes in vegetation cover using the cross tabulation matrix executed in ArcGIS software. In the El Ángel Ecological Reserve, the areas that increased are: herbaceous moorland (138.61 hectares), cushion moorland (121.53 hectares); while the areas that decreased are: shrub moorland (189.84 hectares), frailejones moorland (99.38 hectares). In the Arenillas Ecological Reserve, the areas that increased are: shrub forest (513.04 hectares), dry forest (225.61 hectares); while the areas that decreased are: bare soil (305.09 hectares), urban area (280.13 hectares).