Blending PlanetScope and Sentinel-2 imagery to assess subtidal seagrass changes in turbid waters

Seagrass meadows provide important ecosystem services, acting as nutrient and sediment traps, enhancing water quality and environmental health. Their high sensitivity to environmental changes enables their use as bioindicators, playing a crucial role in buffering impacts in coastal lagoons, increasingly threatened by eutrophication. This study presents an open source Earth Observation methodology for detecting changes in small patches of subtidal seagrass meadows in shallow turbid waters. The PSeagrasS2 open source approach blends PlanetScope (Classic and SuperDove) and Sentinel-2 imagery into a 3-m resolution multi-band raster combining the advantages of both sensors. This method was tested in a subtropical coastal lagoon affected by the collapse of a wastewater treatment facility, which released effluents and sediments that impacted Ruppia maritima and Halodule wrightii meadows. Field surveys conducted three years before and after the event provided seagrass presence/absence data to train Random Forest classifiers. The method, developed in Google Earth Engine Python API integrated with ACOLITE processor, revealed an overall improved accuracy using the multi-sensor approach, detecting a total seagrass loss of 92.12 %. Results revealed the importance of coastal blue bands, the Depth Invariant Index and the inclusion of water quality parameters into the models. Spectral signatures indicated higher resistance of H. wrightii over R. maritima , alongside increased epiphytes, underlining the importance of red-edge bands for assessing aquatic vegetation health. Although mapping sparse subtidal seagrass in turbid waters remains challenging, this multi-sensor approach enhances the assessment of environmental impact severity and potential recovery capacity of these critical ecosystem-forming bioindicators from space. • Small subtidal seagrass patches can be identified at 3-m pixel by blending PlanetScope and Sentinel-2 imagery. • PSeagrasS2 Google Earth Engine Python API open code created for coastal turbid waters. • PSeagrasS2 multi-sensor approach features higher accuracies than stand-alone sensors to evaluate marine pollution impacts. • ACOLITE has been integrated into Google Earth Engine for Sentinel-2 imagery. • Methods provide a subtidal seagrass assessment tool for environmental health monitoring using seagrass as bioindicators.