Land use/land cover (LULC) analysis (2009-2019) with Google Earth Engine and 2030 prediction using Markov-CA in the Rondônia State, Brazil.

Abstract

The Amazonian biome is important not only for South America but also for the entire planet, providing essential environmental services. The state of Rondônia ranks third in deforestation rates in the Brazilian Legal Amazon (BLA) political division. This study aims to evaluate the land use/land cover (LULC) changes over the past ten years (2009-2019), as well as, to predict the LULC in the next 10 years, using TerrSet 18.3 software, in the state of Rondônia, Brazil. The machine learning algorithms within the Google Earth Engine cloud-based platform employed a Random Forest classifier in image classifications. The Markov-CA deep learning algorithm predicted future LULC changes by comparing scenarios of one and three transitions. The results showed a reduction in forested areas of about 15.7% between 2009 and 2019 in the Rondônia state. According to the predictive model, by 2030, around 30% of the remaining forests will be logged, most likely converted into occupied areas. The results reinforce the importance of measures and policies integrated with investments in research and satellite monitoring to reduce deforestation in the Brazilian Amazon and ensure the continuity of the Amazonian role in halting climate change.