UAVs in Mapping and Monitoring Stonewort Meadows - Assessing the Use of Remote Sensing in Shallow Coastal Lagoons in the Gulf of Bothnia

Sammanfattning

Shallow coastal lagoons in the Gulf of Bothnia are unique habitats exhibiting the distinct effects of land uplifting. A regional type of coastal lagoon, flads, share high ecological value, many consisting of mud bottoms as well as dense vegetations such as charophyte meadows, and are considered an endangered habitat. These meadows provide essential habitat for birds, fish and benthic invertebrates, and store both carbon and nutrients. Given their importance for the ecosystem, it is of interest to map the distribution and coverage of these charophytes, and monitor both the species’ and the bay’s development. An UAV was used to take images of the flad and produce an orthomosaic, which in combination with ground truth points was used to make a classified map of the spatial distribution of the species present in the flad, with focus on Chara tomentosa and Chara aspera. Spectral data was extracted from the orthomosaic to explore the relationship between different dominating species, biophysical indicators gathered at the ground truth points such as substrate type and grade of sedimentation, and relative spectral brightness values. The classified map produced in the study had high accuracy when compared to true vegetation in the flad, correctly distinguishing different habitats and most types of vegetation. Studies on the spectral data showed high within-group variation while no significant difference was found between different dominating habitats. No biophysical indicators were found to affect relative spectral brightness of different dominating habitats. This indicates that while the method of using machine learning is effective in classifying stonewort meadows, the differences in relative spectral brightness for species needs to be further investigated.