An Adaptive Photon Denoising Method Over Mountainous Forest Areas

The Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) that carries the Advanced Topographic Laser Altimeter System (ATLAS) photon-counting LiDAR (PCL) provides an astonishing opportunity for Earth observations. The precise identification of PCL signal photons is the prerequisite for ICESat-2 data applications in Earth observations. However, due to the influence of variable topography and vegetation structures, denoising PCL photons remains a significant challenge, particularly in mountainous forest areas. The overall goal of this study is to develop an adaptive photon denoising method tailored for mountainous forest areas. Focusing on study areas with Tahoe National Forest and two simulated datasets, the specific goals are to: 1) setting the neighborhood size based on elevation frequency histogram (EFH) and wavelet transform criterion, which considers the effects of noise and vegetation structure on signal photons; 2) setting the neighborhood direction using detrend operation, which could fit the terrain curve more accurately; and 3) setting the discrimination threshold according to the constraints of the signal and noise rates, which could identify noise photons with complex vegetation structures. The results from ICESat-2 data demonstrate that the terrain elevation and relative height of 95% (RH95) derived using the proposed method exhibit superior performance, with the coefficient of determinations (CODs) of 1.00 and 0.62, respectively.