Temporal seagrass mapping using machine learning and field-validated satellite imagery to inform restoration and management

Abstract

Seagrasses are vital coastal ecosystems, but their growth and ecological dynamics in response to seasonal and interannual changes are still poorly understood, limiting our ability to detect their responses to climate and human-driven impacts. Remote sensing is increasingly used to assess seagrass presence and extent to support management and restoration. However, most remote sensing approaches provide only a static snapshot of seagrass distribution making it difficult to identify seasonal and interannual trends in seagrass dynamics. In this study we mapped seagrass extent at a restoration site in Western Port, Victoria, using Sentinel-2 imagery calibrated with drone-derived ground control da..