Early Bark Beetle Detection through Satellite Time Series
Bark beetle outbreaks are one of the most serious threats to Central European spruce forests, and they are often discovered only after trees have already changed colour and the infestation has begun to spread. We developed models that use satellite imagery and deep learning to detect early signs of stress, so forest managers can act sooner and limit the damage.
By analysing how forest conditions evolve over time, the system highlights areas that are still hard to assess from the ground and turns complex environmental data into clear, usable insights for faster and more sustainable forest management.
Sentinel-2
Python
PyTorch
MLflow
QGIS
UX Research
Prototype Design
UI Design
European Space Agency (ESA)
Slovenian Forest Service
Slovenian Forestry Institute
Slovenian State Forests
Benefits
Early Detection
Identifies canopy stress while trees may still appear green, giving managers more time to act before an outbreak expands.
Large-Area Screening
Satellite coverage makes it possible to assess entire forest regions, including remote stands that are difficult to inspect on the ground.
Targeted Intervention
Pixel-level healthy/affected maps help focus sanitary felling and field surveys on the locations with the highest risk.
Repeatable Forest Health Reporting
Time-series monitoring supports regular, data-driven updates instead of one-off inspections.
By the Numbers
5+
Regions tested across Central Europe.
40M+
data points processed for model training and validation.
8+
years of satellite observations used to track outbreak progression.
