A Density and Distance-Based Method for ICESat-2 Photon-Counting Data Denoising
Xuebo Zheng, Chunping Hou, Meiyan Huang, Dan Ma, Menglong Li
Abstract
The Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) dynamically monitors water depth in shallow waters around islands and reefs. Noise removal is a prerequisite for accurate reconstruction of seafloor topography based on ICESat-2 data products. To this end, we propose a density and distance-based method (DDBM) to extract seafloor signal photons from ICESat-2 data. The DDBM first separates the photons into three parts: above water, water surface, and water column. The water-column photons consist of seafloor signal photons and noise photons. The DDBM adopts a two-step denoising strategy to remove noise in water-column photons to obtain pure and complete signal photons. In the first step, an orientation-variable adaptive ellipse filter is developed, which can adaptively adjust the parameters according to the water depth to remove low-density noise photons. In the second step, a novel distance-based filter (DBF) is designed for stubborn high-density noise clusters. These noise clusters are far from the signal photons, and the DBF removes them by a distance threshold derived from the Otsu threshold method. We select three high-density ICESat-2 datasets to validate the DDBM. Compared with the reference data, the comprehensive evaluation indexes <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$F$ </tex-math></inline-formula> of the DDBM in all datasets are above 0.99, and the highest is 0.998, showing superior performance.