Drivers of Forest Fires: Identification and Strategies in the Colombian Amazon

Abstract

In recent years, due to the increase in number and intensity, forest fires in the Amazon rainforest are becoming crucial for biodiversity conservation, respect of indigenous rights, and sustainable local development. They are affecting the most species-rich subcontinental ecosystems jeopardizing the crucial roles of tropical ecosystems as carbon sinks and regulators of the water cycle. In Colombia, anthropogenic pressures are increasing and expanding in the Amazon, associated with different factors and patterns; it is responsible for an annual deforestation rate that ranges from 0.38% to 0.77% in the most affected municipalities. Different scholars partly studied those drivers and patterns in Colombian Amazon by highlighting important advances in research; however, integrated analyses able to include the multiple drivers, scales, actors, and their relationships are scarcely developed in scientific literature. For this reason, this study aims to identify and determine the importance of climatic, social, and economic factors using scenarios to understand the role that socio-environmental conflicts are playing in the development of forest fires in the Colombian Amazon. Due to its socio-environmental importance as well as for the climate injustice flourishing in the zone, the present research is focused on the Cartagena of Chiará Municipality. Spatial and temporal analyses were carried out over 8 years (2013-2021); meteorological data were acquired from ERA5, socioeconomic data from the governmental agencies, Landsat 8 satellite images, and CMIP6 climatic scenarios were used as input on Machine Learning (ML) tool to identify the importance and role of each variable as a driver of forest fires through socioenvironmental and climatic scenarios. Preliminary results highlight that forest fires occur during boreal winter at the border between the rainforest and the agricultural frontier and mainly in conditions with higher temperatures and lower relative humidity compared to climatology. In the context of climate change, the temperature and relative humidity climatology would be close to the forest fire conditions, which means that the probability of wildfire occurrence would increase. Socioeconomic variables such as rural population, displacement, and GDP may contribute to the increase of fires in the area. The ML model showed its potential as an evaluation tool to address stakeholders to identify inclusive mitigation and adaptation strategies for contributing to climate justice.