Detecting forest disturbance with Earth Observation

Detecting forest disturbance with Earth Observation

26 Apr 2024

Around 10 million hectares of forest are currently undergoing degradation worldwide. The process of degradation refers to the gradual decline in forests’ health, usually characterised by changes in the canopy structure, biodiversity loss, and diminished ecosystem services. The primary causes for forest degradation are anthropogenic or natural disturbances such as burning, logging, windthrow or insect outbreaks, which pose a significant threat to biodiversity.

EvoLand candidate prototype to detect and map forest disturbance

Despite the vital importance of forests for the environment and the extensive research conducted to safeguard forest habitats, mapping forest disturbance and attributing disturbance agents in Europe remains a significant challenge. This difficulty is related to several factors: In Europe, forest type, tree species composition, tree cover density, vegetation season, forest management regimes and observed forest disturbance agents and drivers vary considerably by biogeographic region. In addition, many forest changes in Europe occur on a small scale (< 0.2ha) or show only partial damage, such as a partial windthrow or tree type-specific insect infestation in a mixed forest and are thus more difficult to detect and quantify as opposed to a large-scale removal associated with deforestation. Whereas Earth Observation data has been extensively utilised for monitoring deforestation (read here about EvoLand prototype related to continuous monitoring of forests), its application to small-scale disturbance and disturbance type attribution remains technically challenging. 

Fortunately, technological, and methodological developments show promising potential for improved disturbance detection. One of the eleven prototypes developed in EvoLand is exploring and testing practical solutions for an Earth Observation-based disturbance detection methodology. By using a combination of Earth Observation and in-situ data, along with the latest Machine Learning techniques, the Horizon Europe-funded project EvoLand is developing a Copernicus Land Monitoring Service (CLMS) candidate prototype to identify some of the key disturbance agents driving forest degradation. These include wildfires, windthrow events and insect infestations, such as spruce bark beetle.   

An example: Bark Beetle Infestation

Let’s consider, for instance, one of the most dreaded threats for every plant lover: an insect infestation. Outbreaks of bark beetles can infect large areas of Europe’s conifer and temperate forests. These beetles impact tree health, resulting in smaller or fewer leaves, a reduced crown size, diminished photosynthetic activity, and some species – e.g. Ips typographus – can cause the death of the infested trees. In a few  years with favourable climatic conditions, mass outbreaks can cause widespread forest disturbance, leading to extensive ecosystem degradation and economic losses.  

Earth Observation imagery and datasets can help identify and analyse distinct spectral responses of affected trees at regular intra-annual update rates. EvoLand is exploring the extent to which bark beetle activity can be differentiated from other disturbance agents based on the unique reactions of trees after infestation, by analysing the spatial, spectral as well as the temporal pattern of the infested trees. To do so, innovative Artificial Intelligence and Machine Learning methods are used to analyse Copernicus Sentinel-2 data at high spatial (10m) and temporal resolution (5 day repeat cycle). The methods feed on in-situ reference datasets on forest disturbances for model training and time series-derived forest change information, developed in line with the candidate prototype on Continuous Forest Monitoring. An accurate timing of the forest disturbance event is crucial, as the agent classification needs to happen closely after the detection and before damaged trees are removed or sanitation cutting occurs.  

Detecting disturbances and identifying disturbance agents

The bark beetles example illustrates EvoLand’s envisaged approach for conducting long-term monitoring of dense tree covered areas for disturbance events.  

In the future, such a monitoring approach shall allow stakeholders to not only detect the start of a disturbance, but also its spatial extent, as well as to analyse the development and dynamics and specific patterns of the disturbance over time. In a subsequent step, harmonised training data on forest disturbance agents will be integrated in the workflow to assign the respective driver to a verified disturbance area. The latter information is particularly important in the context of EUDR (EU-Deforestation Regulation) or the recently proposed EU Regulation on a monitoring framework for resilient European forests. 

The goal of this candidate prototype is to be able to detect and map both sudden forest disturbance events like windthrow damage and more slowly evolving disturbances such as those caused by insect infestation. Options for a rapid detection, at minimum on a monthly basis, as well as a more consolidated annual forest disturbance prototype product are being tested.  

The project has selected prototype sites in Central Europe, Spain and Sweden to test the method in different biogeographic regions and over different years. 

Cover Image Credit: European Union, Copernicus Sentinel-2 imagery 

Would you be interested to follow the results of EvoLand and the development of this candidate prototype? Sign up for our newsletter (here below or via this form), follow us on LinkedIn, Mastodon or X (Twitter). 

Questions about the article or the project? E-mail us at 


This article is part of a series providing more details on all 11 EvoLand candidate prototypes 


Stay current! Subscribe to our EvoLand newsletter!