The potential of an unmanned aircraft system for surveying lake and river vegetation

Abstract: Vegetation in and along aquatic systems has important regulatory and ecological functions. Aquatic vegetation is also frequently used as an indicator of environmental conditions. To enhance our knowledge of the complex natural processes in the littoral zone, i.e., the zone from the high water mark to the lower limit of aquatic vegetation, it is critical to assess plant occurrence and abundance at the species level. This assessment is traditionally performed with labour-intensive field methods. Recently developed remote sensing systems with unmanned aircrafts that take aerial images at low flying height, offer new possibilities for surveying aquatic and riparian vegetation. Unmanned aerial systems (UASs) produce aerial images with sub-decimetre spatial resolution and high spatial accuracy at low cost and are highly flexible in time and space. The goal of this thesis was to evaluate the potential of an UAS for surveying non-submerged aquatic and riparian vegetation, including the identification of plants at thespecies level and vegetation mapping. Based on produced vegetation maps and field sampling, we also assessed the total biomass of entire riparian zones (320-m river stretches) and the biomass and trace metal content of dominant species along a pollution gradient. In total, the UAS was tested in four aquatic systems, two lakes and two rivers, in boreal northern Sweden. Generated orthoimages were interpreted visually. The spatial resolution varied from 5–5.6 cm. At two test sites we identified plant species with high accuracy (94.6 and 80.4% for aquatic and riparian vegetation, respectively). Prior knowledge on locally occurring species was necessary for correct species identification. The time needed for manual vegetation mapping increased with increasing vegetation complexity. At the test site for biomass assessment, biomass and trace element (Cu, Cd, and Zn) contents varied considerably between species. Salix sp. (willows) comprised only 3% of the total dominant-species biomass but contained 73% of all Cd and 24% of all Zn. In contrast, Carex rostrata/vesicaria (bottle and blister sedge) comprised 80% of the total biomass and contained 85% of all Cu and 66% of all Zn. This result emphasizes the need for species-specific assessment, for example, in planning of phytoremediation measures. The tested UAS facilitates species-level surveying and mapping of non-submerged aquatic and riparian vegetation and can be used in combination with field sampling to accurately assess biomass and the amount of accumulated contaminants at the scale of entire riparian zones. The main future challenge will be to identify/develop automated methods for vegetation mapping which successfully cope with the inherent complexity of the orthoimages generated with the presented UAS.