Background noise suppression for DAS-VSP data using attention-based deep image prior

Distributed Acoustic Sensing (DAS) has emerged as a widely used technology in various applications, including borehole microseismic monitoring, active source exploration, and ambient noise tomography. Compared with conventional geophones, a fiber optical cable has unique characteristics that allow it to withstand high-temperature and high-pressure environments. Moreover, the installed optical fiber in the well-bore can be used as an integral receiver, which provides high-density and high-resolution seismic records. However, due to its high sensitivity, the obtained seismic records are often corrupted with unavoidable background noise, which introduces more uncertainty in the subsequent seismic data processing and interpretation. Therefore, it is vital to develop an effective DAS data denoiser. In this work, we propose a ground-truth-free method for strong background noise suppression in Distributed Acoustic Sensing Vertical Seismic Profiling (DAS-VSP) data. The proposed method consists of four stages: training set extension with a patching scheme, feature selection with a kurtosis-based method, denoising with a deep image prior (DIP)-based unsupervised neural network, and an unpatching approach for denoised data reconstruction. Numerical experiments conducted on several profiles from the Utah FORGE project demonstrate that the proposed method can effectively suppress most of the background noise. Furthermore, the unsupervised learning approach allows for flexibility in dealing with field data from various regions because it does not require training data generation.