Supported by

Transforming Forest Monitoring Through AI and Satellite Data

We are entering a new era of precision and scale in environmental monitoring. Modern satellite constellations capture high-resolution imagery of the entire planet with unprecedented frequency, making it possible to observe forest properties that could previously only be studied at a limited number of ground plots.

Illustration of four green trees

At the same time, breakthroughs in deep learning unlock new ways to analyze vast, high-dimensional time series. AI4Forest connects these advances to deliver detailed and frequently updated forest information while reducing the computational and storage costs of global-scale analysis.

Our mission

Two objectives. One clearer view of forests.

Climate change is reshaping forests faster than conventional monitoring can capture. AI4Forest combines Earth observation and scalable artificial intelligence to turn massive satellite and airborne datasets into timely insight - from individual trees to global forest biomes.

Objective 01

AI-based high-resolution forest maps

AI4Forest creates and evaluates high-accuracy maps of European forests and other major forest biomes worldwide, at spatial and temporal resolutions reaching the level of individual trees. Tree health, biomass and carbon content can be monitored monthly or even weekly. Combining spaceborne and airborne observations also helps attribute forest change to climate extremes, management decisions and policy.

Individual trees Biomass & carbon Forest health High-resolution mapping
Objective 02

Scalable AI for global-scale analyses

Modern satellite constellations produce enormous time-series datasets whose analysis can otherwise take days or weeks. AI4Forest develops efficient, scalable AI methods that reduce computational and storage costs during both training and inference. These techniques enable global forest monitoring and can transfer to other data-intensive scientific domains.

Scalable AI Satellite time series Efficient inference Global monitoring
Global high-resolution canopy height map

Participating Research Institutions

AI4Forest brings together leading French and German institutes with complementary expertise in machine learning, remote sensing, climate science and forest ecology. By connecting methodological research with domain knowledge, the partnership turns large Earth-observation datasets into robust tools for understanding forests at scale.

From methods to impact

Research Foci

Eight connected research areas transform Earth-observation data into detailed, reliable and computationally efficient information about forests.

Biomass Estimation
Focus 01Biomass Estimation
Canopy Height Maps
Focus 02Canopy Height Maps
Disturbance Detection
Focus 03Disturbance Detection
Single Tree Segmentation
Focus 04Single Tree Segmentation
Tree Species Maps
Focus 05Tree Species Maps
EO Foundation Models
Focus 06EO Foundation Models
Sparse Neural Networks
Focus 07Sparse Neural Networks
Uncertainty Quantification
Focus 08Uncertainty Quantification

What People Say

Philippe Ciais

AI4Forest

AI4Forest is a research project that combines artificial intelligence and forest science to better understand and manage forest ecosystems.

Research Areas
Contact Principal Investigators
Prof. Dr. Philippe Ciais
Laboratoire des Sciences du Climat et de l'Environnement
Université Paris Saclay
Tél. : 01.69.08.95.06
philippe.ciais (at) lsce.ipsl.fr
Prof. Dr. Alexandre d'Aspremont
CNRS - ENS
45 rue d'Ulm
Paris, France
48 rue Barrault
75013 Paris
aspremont (at) ens.fr
Prof. Dr. Fabian Gieseke
University of Münster
Leonardo Campus 3
48149 Münster
Telefon: +49 251 83-38151
fabian.gieseke (at) wi.uni-muenster.de
Prof. Dr. Sebastian Pokutta
Einrichtung FG Mathematische Optimierung
Takustraße 7
14195 Berlin
Tel.: +49 30 84185-209
pokutta (at) math.tu-berlin.de
Prof. Dr. Cornelius Senf
School of Life Sciences
Hans-Carl-v.-Carlowitz-Platz 2
85354 Freising
Tel.: +49.8161.71.4371
cornelius.senf (at) tum.de