| Summary: | The great potential of NASA's Ice, Cloud and Land Elevation Satellite-2 (ICESat-2) to retrieve sea ice heights has been proven. However, a large number of noise photons in the ICESat-2 data make it more challenging to accurately monitor sea ice changes. In this paper, an adaptive clustering and kernel density estimate-based method (AC-KDE) to estimate sea ice heights in ICESat-2 photon clouds is proposed. First, sea ice signal photons are effectively detected by the adaptive clustering method. The input parameters of the method are determined by ATLAS parameters and the LiDAR transmission equation. Then, the adaptive-count signal photon aggregates are used to estimate sea ice heights and obtain variable along-track resolution by kernel density estimate method. By applying the AC-KDE method to the MABEL and ICESat-2 data, we compare it with other denoising algorithms, including the HBM, the DBSCAN, and the OPTICS algorithms. The results show that the proposed method is better in extracting signal photons with higher accuracy scores and F-scores, which are 0.97&0.97, 0.92&0.90, and 0.89&0.82 under high-medium-low signal–noise ratio conditions, respectively. In addition, the retrieved sea ice heights were compared with the ATL07 heights. Comparison against the coincident ATM heights indicates that the AC-KDE heights are remarkably correlated with a lower RMSE value (0.066 m) than that of ATL07 heights (0.104 m) and a vertical height precision of 0.01 m over flat leads. The proposed method can effectively extract signal photons and accurately estimate sea ice heights in the polar regions.
|
|---|