%0 Research articles
%T Ultra-high resolution remote sensing of northern boreal peatland vegetation and land cover
%A Räsänen, Aleksi
%A Virtanen, Tarmo
%D 2021
%J Suo - Mires and peat
%V 71
%N 1
%U http://suo.fi/article/10595
%X <p>Northern boreal peatlands have fine-scale spatial heterogeneity in land cover and vegetation patterns. We produced ultra-high spatial resolution vegetation and land cover maps in three different peatland sites in Finnish Lapland, using drone data, aerial images and lidar data, as well as satellite imagery. Our study was divided into four sub-studies. First, we compared two different vegetation mapping approaches in Kaamanen mire: floristically defined plant communities and pre-defined land cover types. We found out that land cover type mapping has higher classification accuracy but plant communities show the fuzzy pattern of peatland vegetation more realistically. Second, we tested the data and resolution requirements in land cover mapping. According to the results, multiple different datasets at different scales are needed, including spectral and topographic features, and in order to achieve the highest accuracy, ≤ 0.5 m spatial resolution data should be used. Third, we evaluated how these mires (Kaamanen, Lompolojänkkä, Halssiaapa) differ structurally and vegetationally, and showed that there are notable differences between mires in how well vegetation patterns can be detected, what vegetation properties should be mapped and what the optimal remote sensing datasets are. Fourth, we assessed the relative benefit of hyperspectral drone data in detecting biomass and leaf-area index patterns in Halssiaapa mire. Hyperspectral data boosted somewhat the accuracy of the different models. Overall, accurate vegetation and land cover maps can be produced with drone and other remote sensing data, and different datasets and approaches should be tested and combined to find out the suitable methods at different study sites.</p>