Biophysical variables of tree canopies in northern peatlands from hyperspectral EnMAP imagery

Northern peatlands are increasingly affected by climate change and human activities such as drainage and restoration, yet their tree layer dynamics remain poorly characterized over large areas. Hyperspectral satellite data, despite their potential, have not been used extensively to assess biophysical variables of tree canopies in peatlands. In this pilot study, we demonstrate a large-scale approach using hyperspectral EnMAP satellite imagery and national airborne laser scanning (ALS) data to estimate four key biophysical variables of tree canopies, height, canopy cover, effective leaf area index, and above-ground biomass, across boreal and Arctic peatlands in Finland. We utilized 16 EnMAP scenes of peatlands covering approximately 1.78 million pixels (with a spatial resolution of 30 m) along a 1000 km latitudinal gradient (60°N–69°N). ALS-derived biophysical variables served as reference data, and non-linear support vector regression was used to model these tree canopy attributes under three peatland drainage conditions: 1) both drained and undrained, 2) only drained, and 3) only undrained. Model performance varied by drainage status, with relative RMSE ranging from 27 % to 91 % and R² values between 0.73 and 0.85, with the lowest errors observed in drained peatlands. Our results confirm the feasibility of using hyperspectral satellite data for accurate, large-scale assessment of tree cover characteristic in peatlands. This approach offers a scalable solution for monitoring sensitive peatland ecosystems in support of carbon and biodiversity assessments, particularly in the context of future global hyperspectral missions.