SBI: A sandbar extraction spectral index for multi-spectral satellite optical imagery

Abstract

Satellite imagery allows for large-scale monitoring of dynamic coastal processes, with shoreline tracking being the most widespread application. Nearshore wave-generated sandbars influence coastal dynamics by acting as natural buffers that reduce beach erosion through wave energy dissipation and sediment exchange with the aerial beach. Despite their importance, they are often overlooked in satellite-based studies. This paper addresses this oversight by introducing the SandBar Index (SBI), a new methodology designed to optimize the detection of wave breaking pixels induced by the underlying sandbar while minimizing the SBI value pixels from the surrounding environment such as sand, land and water. Wave breaking pixels refer to image pixels where breaking waves generate foam, increasing reflectance in optical satellite imagery. Since wave breaking typically occurs over submerged sandbars, these pixels act as proxies for their detection. By integrating this index into an automated processing framework, long-term time series of sandbar positions are generated alongside shoreline positions. To validate our methodology, Sentinel-2 images are used to compare satellite-derived sandbar positions with in-situ bathymetric data from the Field Research Facility (FRF) in Duck, North Carolina (US), over a period of nearly ten years. Validation results show good agreement (STD $=$ 23.2 m - i.e. 2 Sentinel-2 pixels), demonstrating the ability of the method to capture the onshore and offshore migration of sandbars. The flexibility of the SBI allows implementation on different satellite platforms, including Landsat and VEN$\mu$S, demonstrating its transferability. This application lays the groundwork for future studies using over 40 years of historical satellite data to further investigate long-term sandbar dynamics, but also high-frequency dynamics with the concomitantly increasing revisit and resolution of satellite missions. The integration of multiple observable metrics from satellite data allows for a more nuanced characterization of the coastal system as a dynamic entity.