Analyzing the Impact of Different Filtering Methods on Satellite Altimetry Full Waveform Decomposition

Filtering is an essential step in the denoising of satellite altimetry full waveform data, since any deformation and distortion in the shape of the waveform can cause errors in range estimation and further waveform decomposition will also be adversely affected. This paper evaluated comprehensive performance of the popular filtering approaches like Gaussian filter, Taubin filter, Wavelet filter, and EMD based filter by simulated waveform data and ICESat/GLAS waveform. Firstly, according to the principle of each filter, the optimal parameters of filtering algorithm by ergodic tests were selected, then the Gaussian function using Levenberg-Marquardt method was used to fit full waveform to exact waveform parameters (i.e. peak amplitude, position, and half-width). Thirdly, through comparing SNR, RMSE of the pre and post filtering simulation waveform, and the consistency ratio, the average error of peak amplitude, position, and half-width in each Gaussian components of the fitted simulation waveform, verified the effectiveness of these filters and analyzed their influence on decomposition accuracy. Both the simulation experiments and ICESat/GLAS experimental results suggested that the Taubin filter had superior performance with the lowest peak position error, which turns out it has advantage in full waveform denoising and contributes to better full waveform decomposition. However, it introduces more parameters needed to be selected. The self-adaptive EMD based approach has the highest consistency, which shows EMD-based method is more suitable for the denoising of satellite altimetry full waveform decomposition