VAE4OBS: Denoising ocean bottom seismograms using variational autoencoders

<p>Data from ocean-bottom seismometers (OBS) are inherently more challenging than their land counterpart because of their noisy environment. Primary and secondary microseismic noises&#160;corrupt the recorded time series. Additionally, anthropogenic (e.g., ships) and animal noise (e.g., Whales) contribute to a complex noise that can make it challenging to use traditional&#160;filtering methods (e.g., broadband or Gabor filters) to clean and extract information from these seismograms.&#160;<br><br>OBS deployments are laborious, expensive, and time-consuming. The data of these deployments are crucial in investigating and covering the "blind spots" where there is a lack of station&#160;coverage. It, therefore, becomes vital to remove the noise and retrieve earthquake signals recorded on these seismograms.<br><br>We propose analysing and processing such unique and challenging data with Machine Learning (ML), particularly Deep Learning (DL) techniques, where conventional methods fail. We&#160;present a variational autoencoder (VAE) architecture to denoise seismic waveforms with the aim to extract more information than previously possible. We argue that, compared to other&#160;fields, seismology is well-posed to use ML and DL techniques thanks to massive datasets recorded by seismograms.&#160;<br><br>In the first step, we use synthetic seismograms (generated with Instaseis) and white noise to train a deep neural network. We vary the signal-to-noise ratio during training. Such synthetic&#160;datasets have two advantages. First, we know the signal and noise (as we have injected the noise ourselves). Second, we can generate large training and validation datasets, one of the&#160;prerequisites for high-quality DL models.<br><br>Next, we increased the complexity of input data by adding real noise sampled from land and OBS to the synthetic seismograms. Finally, we apply the trained model to real OBS data&#160;recorded during the RHUM-RUM experiment.<br><br>We present the workflow, the neural network architecture, our training strategy, and the usefulness of our trained models compared to traditional methods.</p>