Decoding tropical forest change at fine scale with multi-sensor satellite data and artificial intelligence

Tropical forests are of crucial importance, but are rapidly disappearing and degrading due to activities like agriculture, logging and mining. To manage and protect forests, we need fast and precise information on forest loss and its causes. While today’s satellite monitoring systems can detect where and when forest loss occurs, they cannot tell why it happens. This limits their use to support climate policies, forest management, conservation, and law enforcement. In addition, many fine-scale changes, such as road construction and selective logging, remain a blind spot, as they are difficult to detect from space.

This thesis developed new methods based on artificial intelligence and multiple types of satellite imagery to rapidly identify the specific drivers of forest loss across the pantropics. It also introduced a method to precisely track road construction in the Congo Basin, to reveal the extent and impact of widespread logging activities. Together, these advances provide timely and detailed forest change information, supporting more effective forest governance.