In September 2018, NASA launched ICESat-2 (Ice, Cloud, and Land Elevation Satellite -2) which carried the unique spaceborne photon-counting lidar system ATLAS (Advanced Topographic Laser Altimeter System) in orbit so far. The lidar has the characteristics of multi beams, high repetition rate and high ranging accuracy, and has great potential in obtaining high-precision and high-resolution wave parameters. In this paper, an optimized algorithm based on point density with adaptive thresholds is proposed to extract sea surface signal photons, and a method based on sea surface geometry shape to calculate the significant wave height value of sea wave, which proves that the significant wave height value that is consistent with the NASA marine product’s can be obtained by using photon data with a length of 1 km, better than the NASA’s 3-7 km wave height data resolution. Significant wave height data is used to generate a wave height space distribution with 0.2°×0.2°grids in the South China Sea, and analyzes the spatial distribution characteristics of significant wave height and the law of its variation with time. It shows that the significant wave height values of the sea area around the Zhongsha Islands and the Luzon Strait are larger throughout the year, while the values of the Beibu Gulf, the Natuna Islands, the Sulu Sea and the Celebes Sea are smaller. The changes of wave height and wind speed with time in the South China Sea are highly synchronous, indicating that the wave in this area is mainly driven by wind. The quantitative relationships between the significant wave height and wind speed in nearshore shallow water and offshore deep water are given respectively in combination with the wind speed data of ERA5, and proves that the significant wave height of the deep water is greater than the shallow water under the same wind speed. With the help of ICESat-2 photon-counting lidar, higher resolution wave height space distribution can be obtained, especially accurate wave heights in nearshore shallow water area, which can fill the space and time gaps of other observation methods of wave, that is helpful to the optimization of wave numerical model and the oceanographic research.