A Cloud-based Coastal Earth Observation Framework for Regional Seagrass Environment Mapping Across The Eastern African Coastlines

Kurzfassung

Seagrasses play an important role in global coastal seascape configuration and extensive blue carbon sequestration through their connectivity with other seascape habitats. Unfortunately, their population keeps on dwindling down due to climate change and unsustainable human activities. Furthermore, the lack of seagrass distribution data and adequate level of protection hampers efforts to conserve these key ecosystems. Our study focuses on mapping seagrass distribution, bathymetry, and water quality along the East Africa coastline using Sentinel-2 satellite images on the Google Earth Engine cloud platform. We perform the seagrass mapping between the depth of 0-15 m in the country scales of Kenya, Tanzania, Mozambique, and Madagascar with a combination of large-scale in-situ and human-annotated data. The presented framework consists of big satellite data analysis, turbid zone masking, machine learning classification, and satellite-derived bathymetry (SDB) estimation. The overall accuracy of the seagrass mapping ranges between 73-89%. The SDB explains the variation in more than 60% of the validation data and features an error of less than 10% of the full mapped depth range. Our country-scale seagrass, bathymetry, and water quality inventories can support integrated science and management efforts pertaining to seascape connectivity, blue carbon spatial variability, resource conservation, and drivers of change in these optically complex natural architectures.