Analysis of environmental variables and deforestation in the amazon using logistical regression models

Area of Science:

• Environmental Science
• Remote Sensing
• Ecology

Background:

• Deforestation in the Amazon poses significant environmental challenges.
• Understanding the impacts of land-use change on environmental variables is crucial for conservation efforts.

Purpose of the Study:

• To identify deforested areas in Rondônia, Brazil, using remote sensing data.
• To evaluate the environmental effects of deforestation on key variables from 2000 to 2022.

Main Methods:

• Multivariate logistic regression model applied to MODIS/Terra sensor data.
• Analysis of albedo, temperature, evapotranspiration (ETr), vegetation index (EVI), and gross primary productivity (GPP).
• Model accuracy assessed using Area Under the Curve (AUC), pseudo R², and Akaike Information Criterion.

Main Results:

• Deforested areas exhibited higher albedo (25%) and surface temperatures (3.2°C) compared to forested areas.
• Significant reductions observed in EVI (16%), GPP (18%), and ETr (23%) due to biomass loss.
• A model incorporating surface temperature and albedo achieved 91.6% accuracy in identifying deforestation impacts.

Conclusions:

• Deforestation in Rondônia leads to measurable changes in surface energy balance and ecosystem productivity.
• Surface temperature and albedo are key indicators of deforestation's environmental impact in humid tropical regions.
• Findings provide critical data for environmental management and policy in deforested zones.