Ground elevation retrieval from photon-counting LiDAR data based on Quadtree Isolation and Cloth Simulation

Abstract

The Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2), the first global photon-counting LiDAR satellite, can be used to draw global elevation maps and monitor their changes. However, a large amount of noise in the photon data poses a challenge to the surface elevation retrieval from the photon data. The purpose of this study is to propose and verify an effective algorithm framework that can accurately extract signals from the photon data and retrieve the ground elevation. The framework includes two key steps. First, the original data is isolated by a Quadtree Isolation to remove noise photons. Second, the ground photons are extracted from the signal by Cloth Simulation, followed by the retrieval of the ground elevation via interpolation. The results show that the algorithm framework can effectively extract signal photons and obtain high-precision ground elevation curves under different land cover and terrain conditions. Specifically, the noise removal algorithm not only reduces the negative impact of land cover and slope but can also adapt to changes in the data acquisition environment. The mean error and root mean square error of the retrieved ground elevation were -0.62 to -1.38 m and 1.74 to 2.74 m, respectively. This study provides an effective solution for estimating the ground elevation by using photon-counting LiDAR data.