Cloud-native Remotely-sensed Seagrass Mapping and Blue Carbon Estimation in Mozambique

Abstract

The interest in the blue carbon sequestration has been increasing in the past few years due to their potential in climate change. Despite only covering less than 0.2% of the global seafloor, seagrass meadows store 10% of the world’s atmospheric carbon deep in the soils. The extent of seagrass meadows has been declining at an alarming rate of 1.5%/year on average. Field data collection of the underwater habitats can be costly due to the environmental challenges. Our goal in the Biodiversity of the Coastal Ocean: Monitoring with Earth Observation (BiCOME) project is to develop analysis-ready products that can be used for Essential Biodiversity Variable (EBV) quantification and downstream impact. In this study, we mapped the seagrass extent and stored blue carbon across the entire Mozambican coastlines. We processed 9,089 multi-temporal Sentinel-2 image tiles across 34,193km2 of the coastal area in Mozambique and mapped the satellite derived bathymetry (SDB) and seagrass extent by using in-situ bathymetry data, and both in-situ and self-annotated habitat data. The seagrass extent was mapped with the Random Forest image classification, resulting in 1,779.3 km2 of seagrass meadows, located between depths of 1.6 and 9.2 m. We estimated the seagrass blue carbon stocks by multiplying the mapped nationwide seagrass extent with their corresponding Tier 1 and Tier 2 carbon stock estimates. Based on our Tier 1 carbon stock assessment, the national seagrass blue carbon stock was estimated at 1.78-147 million Mg and our Tier 2 carbon stock assessment results in 1.64-4.35 million Mg.