Work

Depth Inversion Using SAR and Super-Resolution Enhancement: A Case Study on Case II Waters

Year: 2025

Type: article

Abstract: Depth inversion is crucial for ocean exploration, with synthetic aperture radar (SAR) technology being a key tool due to its all-weather capabilities. However, SAR-based depth inversion typically has limited resolution, ranging from hundreds to thousands of meters, which restricts its use in fine-scale underwater topography studies. Therefore, we introduce super-resolution enhancement (SRE) technology to improve SAR-based depth inversion. Using the Case II waters off California coastline as a case study, we pair low-resolution ETOPO data with high-resolution (HR) multibeam data, employing the residual channel attention network to construct a depth SR model. Our key findings demonstrate that in these Case II waters, the SR model successfully enhances resolution from 180 to 90 m, achieving SSIM$>$0.94, PSNR$>$16.0 dB, and significantly improved accuracy (average RMSE: 4 m, MRE: 11.9%) compared to SAR direct inversion (RMSE: 5.4–8.3 m, MRE: 16.8–21.3%). The method we proposed indicates the integration of SAR-based depth inversion and SRE technology effectively retains original information while enhancing both spatial resolution and accuracy, demonstrating its efficacy in complex marine environments and providing a novel and effective method for HR marine surveying, thereby expanding SAR’s application potential in marine sciences.

Source: IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

DOI: 10.1109/jstars.2025.3590037