Spectral characteristics of northern peatland surface vegetation

Northern peatlands are crucial ecosystems, both for carbon storage and as hotspots for biodiversity. As climate conditions in northern ecosystems change faster than the global average, they are also increasingly threatened by global change. There-fore, detailed understanding and integrating their changing dynamics into, e.g., global climate models is both urgent and essential. The key indicator of a peatland site’s ecohydrology is its surface vegetation, and surveying species composition is an established way to investigate the site’s nutrient status and hydrological traits. Hyperspectral data offers a promising approach in retrieving information about vegetation attributes, such as the species, health status, or moisture. This approach has been applied to study the vegetation of northern peatlands but still to a very limited extent, as applying this technique requires fundamental understanding of the spectral characteristics of peatland vegetation. The aim of this dissertation is to measure extensive open-source spectral libraries of northern peatland surface vegetation, present a systematic examination of their hyperspectral properties, analyze them in relation to vegetation composition, habitat and moisture, and to discuss the possibilities of using hyperspectral data in peatland research. The results demonstrate the advantages of hyperspectral data in mapping plant functional types in peatland vegetation. Comparing models with different spectral inputs showed that use of the full spectrum (350–2500 nm) provided the most accurate predictions of plant functional cover. Furthermore, examining the spectral properties of Sphagnum mosses, a peatland-specific genus of bryophytes, showed that much of the moisture-related information is located in the shortwave -infrared wavelength region. Additionally, the spectral response of the Sphagnum moss species to desiccation was tightly connected to their respective growing habitats. Applying species-specific spectral information in vegetation mapping from peatland vegetation communities, as well as estimation of within-community plant diversity, yielded mainly weakly accurate results. However, litter cover could be assessed with moderate accuracy. The insight into the spectral characteristics of northern peatland surface vegetation in this dissertation highlights the advantages of applying hyperspectral techniques in peatland vegetation mapping. Furthermore, both the results and spectral libraries produced in this study provide directions and facilitate versatile possibilities for detailed future research of these vulnerable ecosystems.