Unpaired training: optimize the seismic data denoising model without paired training data

With the development of seismic exploration technology, distributed acoustic sensing has recently received attention in geophysics. However, owing to the complexity of the layout techniques in distributed acoustic sensing systems, and the unknown or harsh exploration environment, seismic data acquired by this technique usually contain the noise of diverse components, which increases the difficulty in the subsequent data analysis and interpretation. This study (1) trained a deep-learning model that effectively suppressed noise with augmented noise datasets to obtain a high signal-to-noise ratio in distributed acoustic sensing vertical seismic profile records; (2) introduced an attention module to enhance the extraction and recognition of signal features to recover effective signals under substantial noise interference; and (3) introduced adversarial loss and cycle-consistent loss to replace the commonly used L1 norm or L2 norm to train the network. The obtained hybrid training set containing unpaired synthetic and unpaired field datasets for model pre-training and fine-tuning effectively improved the denoising performance of the seismic field data. In summary, this study proposed an unpaired training-based distributed acoustic sensing seismic data denoising method that transformed noisy distributed acoustic sensing vertical seismic profile data into noise-free data. By analyzing the noise suppression results of other methods, including qualitative and quantitative analyses, it was demonstrated that the proposed method successfully suppressed multiple types of noise in distributed acoustic sensing vertical seismic profile data. The study showed clear and continuous signals in the denoising results and improved the denoising performance on the seismic field data.