Denoising Raw Photon Counting Atmospheric Lidar Data Using Autoencoders

Photon counting detectors are used in a number of atmospheric backscatter lidar instruments. These detectors are highly sensitive and allow for the detection of optically thin aerosols and clouds. However, during the day, the solar background signal can be much larger than the signal from clouds and aerosols, making these features difficult to detect. Averaging the data to coarser horizontal resolutions has been the standard way to increase SNR and thus allow clouds and aerosols to be more easily detectable. Recent work has demonstrated success in boosting SNR without decreasing resolution using advanced filtering techniques. However, rapid advancements in Deep Learning based image denoising algorithms can further improve the SNR. A state-of-the-art Deep Learning autoencoder is trained to remove noise from daytime lidar data. Cloud Aerosol Transport System (CATS) lidar data is the training data source, with artificial noise being added to pristine SNR night data to create clean/noisy image sample pairs. Visual and quantitative SNR results are shown. Implications for data retrievals and instrument design are discussed.