Mapping the distribution of habitats is vital for successful management and monitoring of biodiversity. Within Switzerland, there is strong demand for a spatially explicit, area-wide habitat map to inform field surveys, provide base data for research and ecological infrastructure projects. The stakeholder group is broad including scientific researchers, government agencies, environmental advocacy groups. It is important for the habitat map to reflect not the potential distribution of habitats but the current distribution taking into account landscape management. Therefore, we take advantage of Earth Observation data, in particular high temporal resolution Sentinel-1 and 2 satellite imagery, digital aerial photography and high resolution 3D information derived from it. We map habitats to the second level of the Delarze, Gonseth et al. (2015) classification, which is the most widely used in Switzerland. Within the software eCognition, airborne ortho imagery (1m resolution) was segmented into ‘image primitives’ based on reflectance in the RGB and NIR bands and vegetation indices. We then brought together information from habitat distribution models, landscape models and land use maps to assign habitat types to these segments in a rule-based approach. The habitat distribution models used in the assignment procedure were developed via machine learning approaches trained with field data from large scale Swiss vegetation monitoring programmes, and spatially explicit, area-wide predictors derived from Earth Observation data from the Sentinel 1 and 2 constellations, climate, topography, and soil data. We undertook stakeholder consultation and workshops to ensure that the habitat map was developed to meet stakeholder needs and will be fit for purpose for the wide variety of end-users. The approach has been developed to be semi-automated so that it can be re-applied with updates of the base data at specified time intervals, enabling use for monitoring purposes.
Keywords: remote sensing, species distribution models, image classification, machine learning, vegetation indices, vegetation height model, Switzerland
Reference: Delarze R., Gonseth Y., Eggenberg S. and Vust M. 2015. Lebensräume der Schweiz. Ökologie – Gefährdung – Kennarten. 3 Edition. hep Verlag, Bern 440 p.
Bronwyn Price, Land Change Science, WSL
Link for online participation
Please send an Email to Rafael Wüest Karpati if you would like to get access to the live stream.
The Biodiversity Seminar Series (BD-Seminars) are organized by the WSL Biodiversity Center. Every two weeks, we aim to host a seminar speaker that presents research or outreach on topics relevant to the biodiversity community at WSL. The seminars are public and are usually broadcasted online.
Find out more about the WSL Biodiversity Center and a complete list of events here:
Please send an Email to Rafael Wüest Karpati if you would like to be updated on the activities of the WSL Biodiversity Center.