Characterization of the dynamics of the successional stages of the Amazon forest using Google Earth Engine

Abstract

Aim of study: This study evaluates the potential of the Google Earth Engine tool, supported by fine-scale information obtained by Unmanned Aerial Vehicle, to apply and characterize the dynamics of the successional stages of the Amazon Forest in the state of Rondônia over ten years. Area of study: The study was carried out in the state of Rondônia located in the North Region of Brazil (Western Amazon). Material and methods: The data and its by-products were derived from the Landsat Level 1 - TOA collection of the United States Geological Survey, specifically Landsat 5 and 8. The mapping also used Phantom 4 Pro UAV images. We used the supervised classifier Random Forest to map the primary forest/advanced regeneration, medium regeneration, initial regeneration, and classes, and, subsequently, we crossed and quantified the successional advance and vegetation loss. Main results: It was observed that the state lost forest area even with the successional advance that occurred throughout the period, implying that the forest succession was insufficient in the face of forest deforestation. Research highlights: This study contributed to understanding the dynamics of the Amazon Forest, which goes through a process of deforestation and forest regeneration simultaneously.