Modeling deforestation drivers in the Brazilian Amazon: a comparison of quantitative approaches

Deforestation in the Brazilian Amazon, with approximately 17% of the biome lost, remains a critical global issue. This study analyzes deforestation dynamics using various quantitative approaches, including linear regression, univariate time series models, and multivariate dynamic time series models....

Full description

Saved in:
Bibliographic Details
Main Authors: Alisson Castro Barreto, Tailon Martins, Adriano Mendonça Souza
Format: Article
Language:English
Published: Taylor & Francis Group 2025-06-01
Series:Big Earth Data
Subjects:
Online Access:https://www.tandfonline.com/doi/10.1080/20964471.2025.2510770
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Deforestation in the Brazilian Amazon, with approximately 17% of the biome lost, remains a critical global issue. This study analyzes deforestation dynamics using various quantitative approaches, including linear regression, univariate time series models, and multivariate dynamic time series models. While traditional models like OLS and ARIMA provided valuable insights, they were limited in capturing the complex temporal dynamics of deforestation. In contrast, the BVAR model demonstrated superior performance by effectively modeling lagged effects and feedback loops among variables. Granger causality tests confirmed that cattle ranching significantly precedes changes in deforestation rates. Impulse Response Function (IRF) analysis revealed that shocks in timber extraction, cattle ranching, and soybean production lead to significant increase deforestation, with effects emerging in the 1st, 4th, and 8th years, respectively. Variance decomposition showed that cattle ranching dominates deforestation in the short term, while timber extraction becomes the primary driver in the long term. These findings underscore the effectiveness of BVAR in capturing the complexities of land-use change dynamics in the Amazon.
ISSN:2096-4471
2574-5417