Scientific publications
Decreasing water availability reduces productivity in Swiss forests along an altitudinal gradient
12-05-2026
A 30-year study of 18 Swiss forest sites revealed a decline in forest productivity, especially since 2015, across all altitudes, age classes, and tree species. This decrease cannot be explained solely by forest ageing or stand density, but is also linked to reduced soil water availability and nitrogen deposition. Tree growth and regeneration rates were identified as the main drivers of productivity decline. Even at similar stand densities, forests in recent years showed lower productivity than in previous decades. The findings suggest a reduced long-term growth capacity of forests and highlight the crucial role of water availability under increasing drought and heat conditions caused by climate change.
European forest disturbance alerting using Sentinel-1
28-03-2026
This study presents a near-real time satellite-based system (RADD Europe) that detects forest disturbances across Europe throughout the year. The system identifies canopy cover loss greater than 50% in 10 x 10 m pixels with a minimum event size of 0.1ha, within a 27 days on average. The data is updated weekly and is accessible here
Remote sensing across scales and platforms: monitoring Castelporziano nature reserve forest insect outbreaks
18-03-2026
This study investigates the decline of Mediterranean stone pine forests caused by climate stress and insect outbreaks, focusing on Castelporziano Estate in Italy. It proposes a multi-source remote sensing approach combining high-resolution (Pléiades, DigitalGlobe), Sentinel-2, and PlanetScope data to detect and monitor disturbances. Sentinel-2 enables near-real-time yearly mapping using an improved unsupervised algorithm, while PlanetScope reveals an average detection delay of 8.5 days. The method achieves high accuracy (85–92%) when validated against detailed imagery. Overall, integrating multiple data sources improves the timeliness and reliability of monitoring, supporting effective management of forests increasingly threatened by climate change and insect infestations.
European forest disturbance alerting using Sentinel-1
18-03-2026
This study presents a near-real-time forest disturbance alert system for Europe using Sentinel-1 radar data, which works reliably under all weather conditions and throughout the year. By integrating temperature data (ERA5-Land) and forest type information, the method accounts for seasonal and environmental effects on radar signals. Validation shows high accuracy, especially when forest cover data errors are excluded. Disturbances are typically detected within 27 days, with potential reduction to near real time. The system improves detection of small-scale events and reveals seasonal disturbance patterns across Europe. It offers a valuable tool for forest management, conservation, carbon monitoring, and law enforcement.
Remote estimation of harvested timber volume in the Congo Basin using Sentinel-1 data
18-03-2026
The study examines whether reported timber harvest volumes can be reliably estimated using canopy gaps detected by Sentinel-1 radar data across 93 concessions in the Congo Basin. Results show a strong linear relationship between harvested volume and disturbance area, outperforming existing forest monitoring systems. Aggregating data over multiple years further improves accuracy while maintaining consistent trends. The method also enables detection of illegal activities along logging roads. Overall, it provides a scalable, cost-effective tool for verifying logging operations, supporting sustainable forest management and informing international policies like FLEGT and the EU Deforestation Regulation.
Bridging spatio-temporal gaps in ALS data using Landsat time series and forest disturbance-recovery metrics via multi-task neural networks
07-11-2025
European forests help mitigate climate change, but increasing disturbances like fires and droughts call for better monitoring. Current forest data from National Forest Inventories (NFIs) are inconsistent across countries, and satellite or laser data alone have limitations. This study developed a neural network model combining Landsat satellite data with airborne laser scanning (ALS) metrics to predict forest height, growing stock volume, and basal area. Tested across five European regions, the method showed strong accuracy (R² up to 0.78). By predicting ALS-based metrics from satellite data, this approach enables scalable, affordable, and frequent forest monitoring to support climate-smart forest management.
