The Nature in Norway (NiN) system classifies nature types according to environmental variables. Large scale mapping of benthic nature types by water is challenging, as light is quickly attenuated in deep areas and shallow water is difficult to traverse. Existing data of benthic nature types are mainly based on pointmeasurements. To improve the accuracy and detail of NiN, extensive mapping of the Norwegian coastline is essential.

For the purposes of this study, shallow benthic nature types are classified according to dominant pigment groups. This allows for pigment based classification of hyperspectral photomosaics, and hence relation from image to the nature type system. This study applied an Unmanned Surface Vehicle (USV) and an Unmanned Aerial Vehicle (UAV), both carrying hyperspectral imagers, mapping the same area in Hopavågen, Norway. The class identification potential was described for each photomosaic. The hyperspectral photomosaics were classified using the Spectral Angle Mapper (SAM), and the classification accuracy was assessed for the photomosaics obtained with the USV and UAV.

USV and UAV are promising large scale mapping tools, but it was established that further studies should take place in the spring/summer to obtain sufficient light signal. SAM classification of the USV photomosaic had an accuracy of 63.4%, while the classified UAV photomosaic had an accuracy of 24.8%. Seagrass and brown algae were classified poorly, while red algae and sediment were classified quite well.