Submerged Aquatic Vegetation (SAV)

Algorithm description

This is a prototype product for a new Copernicus Service element which is being developed in the course of FOCCUS.  The algorithm computes the presence/absence of submerged aquatic vegetation (SAV; seagrass and macroalgae) in subtidal regions of the Balearic Islands based on temporally aggregated Sentinel-2 imagery and EMODNET bathymetry. Sentinel-2 data are atmospherically corrected using ACOLITE and water pixels are identified using IdePix. Images are processed and aggregated using the Copernicus Data Space Ecosystem. Aggregate images are classified as SAV (0) or non-SAV (1) using machine learning methods. Results are clipped to water depths shallower than 25 m and cleaned up using a 3x3 pixel median filter.

Limitations

Data availability depends on the number of usable observations (clouds, water clarity, glint) and may therefore vary spatially and temporally. The quality of the products depends on proper flagging by IdePix classification and other masks applied. The algorithm does not differentiate between different types of submerged aquatic vegetation (e.g., macroalgae or seagrass) and does not consider vegetation health. A known issue is are misclassifications of optically deep water in some areas.