Denoising of ICESat-2 photon-counting data based on the improved DENCLUE algorithm.

Area of Science:

• Oceanography
• Remote Sensing
• Geospatial Analysis

Background:

• ICESat-2 lidar data offers potential for oceanographic applications like bathymetry and coral reef monitoring.
• Existing noise reduction methods for ICESat-2 data are often parameter-sensitive, leading to insufficient noise suppression.
• Precise oceanographic investigations using ICESat-2 data are hindered by significant noise contamination.

Purpose of the Study:

• To develop an advanced density-based clustering algorithm for enhanced noise mitigation in ICESat-2 data.
• To improve the accuracy and reliability of shallow water bathymetry and coral reef monitoring using ICESat-2.
• To establish a robust methodological framework for remote sensing-based bathymetric exploration.

Main Methods:

• Determined sea surface elevation via characteristic analysis of surface point clouds and systematic data classification.
• Implemented an improved DENCLUE algorithm featuring regional density partitioning and simulated annealing for optimized denoising.
• Applied a refined denoising approach based on elevation distribution patterns and conducted refraction correction for seafloor data.

Main Results:

• The improved DENCLUE algorithm demonstrated superior noise mitigation capabilities for ICESat-2 data.
• Denoised ICESat-2 data showed exceptional accuracy compared to in situ bathymetric measurements (R²=0.91-0.99, MAE=0.22-0.51m, RMSE=0.26-0.73m).
• Validated the effectiveness of the algorithm for shallow water bathymetric exploration.

Conclusions:

• The developed improved DENCLUE algorithm effectively addresses noise contamination in ICESat-2 data.
• This methodology provides a reliable framework for utilizing ICESat-2 data in shallow water bathymetry.
• The findings support the advancement of remote sensing applications in marine science and coastal management.