Detecting multiple types of marine floating debris via satellite remote sensing informed by hyperspectral features

Marine debris has become a critical global environmental issue, with large areas of floating debris causing significant harm to marine environments and organisms. Satellite remote sensing technology offers promising opportunities for large-scale monitoring of marine floating debris. However, the diverse material composition of marine debris leads to complex and variable optical characteristics, posing significant challenges for remote sensing-based detection. This study conducted both laboratory experiments and theoretical simulations to analyze the spectral reflectance characteristics of multiple debris types, including foam boards, polyvinyl chloride buoys, polypropylene bottles, and wooden boards. Our results demonstrate that floating debris exhibits markedly higher reflectance than background seawater, particularly in the near-infrared and shortwave infrared spectral regions. Building upon these spectral characteristics, we developed a novel debris spectral index (DSI 1 ) to extract multiple floating debris from background waters. Furthermore, to mitigate the detection confusion between debris and large floating algae, we proposed a second spectral index (DSI 2 ) utilizing the strong red-band absorption feature of large floating algae. The optimal threshold values for DSI 1 and DSI 2 indices were determined through an integrated analysis of field-measured, simulated, and satellite-derived data. The developed remote sensing detection model was successfully applied to Sentinel-2 imagery to identify floating debris distributions across three representative scenarios. This study enhances the understanding of the spectral characteristics of multiple floating debris and establishes an operational satellite-based methodology for the rapid monitoring of floating debris, providing critical technical support for marine environmental management and protection. • Lab-measured spectra revealed high NIR reflectance shared by multiple debris types. • Two spectral indices were proposed to detect floating debris using Sentinel-2 imagery. • The model distinguishes debris from water and algae under varied conditions. • Case studies demonstrated effective debris detection across diverse marine settings.