Detection of FAD Position and Distribution with Sentinel-1 Synthetic Aperture Radar (SAR) Data with Constant False Alarm Rate (CFAR) algorithm optimized by Random Forest Classifier

The management of Fish Aggregating Devices (FADs) in Indonesian waters presents a significant challenge due to the vast marine territory and the numerous fishing fleets. Effective FAD management relies on accurate and reliable information, which is essential for informed decision-making. Synthetic Aperture Radar (SAR) technology, particularly from the Sentinel-1 satellite, offers a robust remote sensing solution that operates under various weather conditions, ensuring consistent data collection. This study leverages SAR data for FAD detection using an adapted Constant False Alarm Rate (CFAR) algorithm, commonly employed in ship detection. The CFAR model relies on five critical parameters target window, guard window, background window, and probability false alarm rate, whose optimization is enhanced by a random forest classifier. The optimized parameters are a target window (TW) of 10 meters, guard window (GW) of 50 meters, background window (BW) of 200 meters, and a probability false alarm rate (PFR) of 7.0. The model achieved a Mean Squared Error (MSE) of 86%, though validation on a validation dataset revealed an accuracy of 66%, with challenges in detecting non-metallic floating type of FADs and those under 10 meters in length. Despite these limitations, the findings underscore the potential of Sentinel-1 SAR data as a valuable tool for FAD monitoring and regulation. This research contributes to broader efforts to integrate remote sensing technologies into marine resource management, offering a scalable solution for global FAD detection.