Seagrass mapping of north-eastern Brazil using Google Earth Engine and Sentinel-2 imagery
Emma Deeks, Karine Matos Magalhães, Dimosthenis Traganos, Raymond D. Ward, Iran C. Normande, Terence P. Dawson, Pavel Kratina
Abstract
Seagrass ecosystems are globally important blue carbon sinks and support significant marine and terrestrial biodiversity. However, human-induced climate change coupled with other anthropogenic pressures have substantially reduced seagrass distributions, making them one of the most threatened ecosystems on Earth. The challenges associated with seagrass conservation include substantial data gaps and limited low-cost, near-real monitoring methods. To address these challenges, we used 507 Sentinel-2 satellite images, filtered between August 2020 and May 2021, in the Google Earth Engine cloud computing environment for regional scale seascape habitat mapping in north-eastern Brazil, a region where conservation efforts are particularly hampered by data limitations. We mapped 9452 km 2 of coastline up to a depth of 10 m. We identified 328 km 2 of seagrass ecosystems, providing vital open access positional information for a variety of research applications. We also assessed the capability of Sentinel-2 in monitoring temporal changes in coastal habitats, and revealed up to 15.9% declines in seagrass meadow coverage in specific areas over a five-year period in north-eastern Brazil. Our results demonstrate that Sentinel-2 is an effective tool in mapping seagrass distributions at a regional scale. The resulting maps are critical for supporting the conservation of Neotropical coastal biota, including the endangered Antillean Manatee. Our study emphasises the importance of replicable and systematic monitoring methods in the race to conserve threatened coastal ecosystems globally.