Noise Reduction for Ground-based Atmospheric Detection Lidar: A Universal Method Based on Signal Segmentation and Reconstruction

Abstract Atmospheric detection lidar is a powerful tool to achieve accurate detection of various elements of the atmosphere. However, noise is an unavoidable factor in the lidar signal, and it affects the analysis of the detection results. In this study, a universal de-noising algorithm based on signal segmentation and reconstruction is proposed to enhance the signal-to-noise ratio for ground-based atmospheric detection lidar. The proposed algorithm performs well in most cases The signal segmentation, which serves as the keystone of this universal algorithm, segments the signal into three parts: the noise-free signal, normal signal, and layer signal (if it exists). The two latter signal fragments must be processed by a de-noising method, which is a combination of the empirical mode decomposition (EMD) and the discrete wavelet transform (DWT) with a fixed wavelet base. The former one deals with the normal signal, and the latter one is designed to process the layer signal. The determination of the starting and ending bins of each signal fragment, the self-adaptive EMD, and a proper selection of DWT ensure that signal processing can be accomplished automatically without any artificial intervention. A series of simulation tests quantitatively shows the advantages of the proposed method. The dual field-of-view Mie lidar provides the possibility of de-noising evaluation based on real lidar data. Our method is proven as a universal and effective approach for de-noising processing of lidar signal.