Cerrado Native Vegetation Mapping in a Region of Três Lagoas – MS with Google Earth Engine

Abstract

The Cerrado is one of the biomes existent in Brazil and the second largest of the South America. It has a great importance due to its biodiversity, ecosystem and mainly to be like a tank or “sponge”, that distribute water to the others biomes around it. Besides, it is a local of sources for some largest watersheds in South America. However, due to the anthropic activities practiced (with emphasis for the pasture and silviculture) and the reduction of the native vegetation, this biome is threatened.  Considered a biodiversity hotspot, the Cerrado could not exist in 2050. With the preservation necessity, this study aimed at investigation of machine learning algorithm for the mapping of Cerrado native vegetation in Três Lagoas city, using the cloud plataform Google Earth Engine. The process was performed using a Landsat-8 OLI image, dated on 10 October 2018, and the Random Forest (RF) and Support Vector Machine (SVM) algorithms. In the classification validation, the kappa index resulted in 0.94 and 0.97, respectively. RF algorithm, when compared to SVM, presented noisy classification. It was verified the existence of native vegetation area about 2,556 km² when was adopted RF and 2,873 km² when SVM was adopted.