SD-BSI : An Improved Method for Brine Shrimp Slick Detection in Salt Lakes Using Multispectral Satellite Imagery

To address the limitations of the traditional Brine Shrimp Index (BSI) in detecting brine shrimp slicks in salt lakes, this paper proposes an improved detection method, the Spectral Difference BSI Method (SD-BSI). Objectives: The objective of this study is to enhance the accuracy and robustness of brine shrimp slick detection in salt lakes by incorporating spectral differences from adjacent water pixels, which are overlooked by the traditional BSI method. Methods: The SD-BSI method was developed by adapting the traditional BSI to include spectral data from adjacent water pixels. This study utilized Landsat-8 Operational Land Imager (OLI) data across different salt lakes to conduct experiments. The performance of SD-BSI was compared with the traditional methods through accuracy assessment and robustness analysis under complex water backgrounds. Results: The experimental results demonstrated that SD-BSI achieved an average accuracy of 0.951, which is a 15.5% improvement over the traditional BSI method. The recall rate was significantly enhanced by 23.4%, effectively addressing the extraction failures in slicks of medium to very low density brine shrimp that are prevalent with traditional BSI. Furthermore, SD-BSI effectively mitigated disturbances from sunglints, thin clouds, and water with high turbidity and high chlorophyll concentration in complex salt lake environments. Conclusions: The SD-BSI method offers a significant improvement in detecting brine shrimp slicks across different salt lakes. This method not only enhances the precision and robustness of detection under various water conditions but also demonstrates balanced and stable performance, and could serve as a valuable tool for ecological monitoring and sustainable management of brine shrimp resources in salt lakes.