Tropical forests play a crucial role in mitigating global climate change as carbon sinks and are also major biodiversity hotspots. Despite this, they suffer from high rates of deforestation.
Remote sensing is being used to monitor tropical forests and to detect and report deforestation. Although valuable, such methods only react to deforestation that has already happened.
In this project, we aim to use explainable machine learning to extract insights from satellite imagery about drivers of tropical deforestation and patterns of landscape changes over time and on a global scale. Using this information, we then aim to predict future occurrences of deforestation before any trees are cut.