Mapping Mediterranean seagrasses with Sentinel-2 imagery. (September 2018)

• Record Type: Journal Article
• Authors: Traganos, Dimosthenis; Reinartz, Peter
• Is Part Of: Marine pollution bulletin. Volume 134(2018)
• Page Start: 197
• Page End: 209
• Publication Date: 2018-09
• Subjects: Posidonia oceanica -- Cymodocea nodosa -- Mediterranean seagrasses -- Thermaikos Gulf -- Sentinel-2 -- Water column correction; Marine pollution -- Periodicals; Marine Biology -- Periodicals; Water Pollution -- Periodicals; Mer -- Pollution -- Périodiques; Publications périodiques; Pollution des mers; Lutte antipollution; Electronic journals; 363.7394
• DOI: 10.1016/j.marpolbul.2017.06.075
• Languages: English
• ISSNs: 0025-326X

Abstract

Abstract: Mediterranean seagrasses have been hailed for their numerous ecosystem services, yet they are undergoing a decline in their coverage. The major complication with resolving this tendency is the sparsity of data on their overall distribution. This study addresses the suitability of the recently launched Sentinel-2 satellite for mapping the distribution of Mediterranean seagrass meadows. A comprehensive methodology is presented which applies atmospheric and analytical water column corrections and compares the performance of three different supervised classifiers. Remote sensing of the Thermaikos Gulf, northwestern Aegean Sea (Greece, eastern Mediterranean Sea) reveals that the utilization of Support Vector Machines on water column corrected reflectances yields best accuracies. Two Mediterranean seagrasses, Posidonia oceanica and Cymodocea nodosa, cover a total submerged area of 1.48 km 2 between depths of 1.4–16.5 m. With its 10-m spatial resolution and 5-day revisit frequency, Sentinel-2 imagery can mitigate the Mediterranean seagrass distribution data gap and allow better management and conservation in the future in a retrospective, time- and cost-effective fashion. Highlights: We demonstrate Sentinel-2 suitability for mapping two Mediterranean seagrasses. Posidonia oceanica and Cymodocea nodosa are mapped between 1.4–16.5 m of depth. Support Vector Machines perform better than Random Forests and Maximum Likelihood. Atmospheric and water column corrections aid discrimination of the two seagrasses. A satellite-derived bathymetry map of Thermaikos Gulf, NW Aegean Sea is developed.