Thanks to recent technological advances, we are entering a new era of precision and scale in environmental monitoring. The influence of two major developments makes this possible:

• An increase in satellite data availability: Modern satellite constellations now capture high-resolution imagery of the entire planet daily, with spatial resolutions as fine as 3 meters or better. This unprecedented data availability enables the observation of forest properties that, until recently, could only be studied at a limited number of ground-based plots.
• Breakthroughs in Artificial Intelligence: In particular, deep learning models have unlocked new possibilities for analyzing vast volumes of high-dimensional, spatiotemporal data. These advances open the door to intelligent monitoring systems capable of detecting subtle patterns across space and time. However, despite recent progress, current AI-based approaches still fall short of delivering global forest maps at the necessary spatial and temporal precision.

Introducing AI4Forest

AI4Forest is a research project designed to overcome these challenges. It will deliver high-resolution forest maps of Europe—and eventually the entire earth- a level of detail and frequency never before achieved. By combining AI models with modern satellite and airborne data, AI4Forest will enable cost-effective, scalable analysis of massive environmental datasets.

The project is centered around two core objectives:

1. High-Resolution, AI-Driven Forest Mapping

AI4Forest will generate highly accurate forest maps for Europe and major global biomes, resolving details down to the level of individual trees. These maps will be updated monthly or even weekly, allowing the continuous monitoring of key forest variables such as tree health, biomass, and disturbances.

Leveraging Earth observation data and tailored AI methods, AI4Forest will significantly enhance the current capabilities in forest monitoring. This includes quantifying forest biomass and carbon dynamics.

2. Scalable AI for Global Data Processing

The volume of satellite time series data generated today poses a major challenge for AI applications, often requiring days or weeks of processing time even with substantial computing power. AI4Forest will develop efficient, scalable AI techniques designed specifically for large-scale environmental analysis. These methods will minimize computational and storage costs, ensuring faster and more accessible processing across a variety of platforms.

While the primary focus is forest monitoring, the developed AI techniques will be broadly applicable to other data-intensive domains in Earth and environmental science.