Evaluating ICESat-2 Seafloor Photons by Underwater Light-Beam Propagation and Noise Modeling

Ocean surveying is of great significance to mankind's development and utilization of the ocean. Island and reef area surveying is an important part of ocean surveying and mapping. The Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) has been proven to have a certain bathymetric capability. However, the precise extraction of seafloor signal photons in these regions remains a challenge. This study introduces a method for extracting seafloor photons that is water-depth adaptive and works at various depths. In addition, we propose a method to evaluate ICESat-2 seafloor signal photons by underwater light-beam propagation and noise modeling, using the decision tree method to classify signal photons into high-, medium-, and low-confidence levels. The results indicate that the method exhibits better signal continuity, better slope adaptability, and better signal-to-noise ratio (SNR) adaptability in seafloor signal photon detection, and remains more surface object signal photons in island signal photon detection than adaptive variable ellipse filtering bathymetric approach (AVEBM) method. The high-, medium-, and low-confidence seafloor signals exhibit consistencies (R-2 ) of 0.9954, 0.9926, and 0.9874, respectively. The root-mean-square errors (RMSEs) are 0.49, 0.66, and 0.93 m, and the mean absolute errors (MAEs) are 0.24, 0.44, and 0.86 m, correspondingly. Higher-confidence photons perform significantly better than lower-confidence photons. The confidence evaluation of seafloor photons will provide an important reference for users and will lay the foundation for further research into the use of ICESat-2 for offshore bathymetry.