Application of UAV in monitoring and mapping coastal habitats over São Vicente, Cabo Verde

Abstract

The coastal areas are characterized by constant change due to the dynamic interplay be-tween the ocean and the land surface. Waves, seas, and winds appear as crucial factors in a continuous erosion and sediment deposition process. In addition, the coastal areas are home to vegetated ecosystems, including mangrove forests, seagrass beds, and salt marshes which pro-vides many services to human. However, these coastal areas and their ecosystems are threat-ened by human pressure and the effects of climate change. These factors lead to the degradation of these coastal ecosystems and the loss of biodiversity. Therefore, there is a need to implement effective and less costly measures to sustain and manage the coastal habitats. With the emer-gence of uncrewed aerial vehicle (UAV) technology over the past decade, drones have found a successful role in research areas such as precision agriculture, forestry, and ecology. Their flex-ibility and ease of deployment allow cost-effective surveys to be carried out at a small scale and mesoscale. At this scale, UAVs can repeatedly produce high-resolution images similar to other remote sensing products. In this project, the use of a consumer-grade UAV was examined to map coastal habitats. The resulting habitat map served as an ecosystem management tool and a baseline for assessing future changes in habitat condition and distribution, complementing previous ecological mapping and surveys in Cabo Verde. It was indicated that it is feasible to use a drone to produce maps of the submerged coastal regions in São Vicente, in Cabo Verde, under suitable weather and lighting conditions. During 1 ½ month, five high-resolution ortho-mosaics were created from images acquired by a DJI Phantom 4 Pro V2.0, equipped with an RGB camera. The orthophoto mosaic of Baia bay was classified in Maximum Likelihood model. Though the training dataset was based on manual photo interpretation, the model clas-sifies the shallow water coral and algae and the terrestrial variables with an accuracy of around 70%. This study provided the basis for future UAV-based mapping and monitoring approaches in Cabo Verde.