b) Machine learning proves its effectiveness for large-scale DAS data processing. Cong et al. propose an attention-guided denoising convolutional neural network (ADNet) to denoise seismic images and retrieve weak signals. The network adapts different kinds of convolution layers to enlarge the receptive field size and extract the global feature of the input signal. Meanwhile, the attention mechanism is introduced to extract the hidden noise from the complex background. Wang et al. propose a multi-scale interactive convolutional neural network (MSI-Net) to denoise DAS data. The proposed MSI-Net incorporates both coarse-scale and fine-scale features to advance inherent serial convolution to multi-scale parallel convolution. As demonstrated by field examples, the MSI-Net can effectively attenuate a variety of unique DAS background noise and also effectively recover weak signals. Dong et al. develop a novel multi-scale dense-connection denoising network (MDD-Net) to achieve high-accuracy processing of complex DAS-VSP background noise. MDD-Net utilizes a widened convolution block to capture the multi-scale features of the analyzed data. Both synthetic and field DAS-VSP records are processed to verify the effectiveness of MDD-Net, even for the weak upgoing signals. Tang et al. establish a rank reduction U-net (RRU-net), an integration of an encoding structure for down-sampling and a decoding structure for up-sampling, to reconstruct and denoise the DAS-VSP seismic data simultaneously. Zheng et al. propose a deep-learning approach for DAS signal extraction from the borehole-flow noise during hydraulic fracturing. The authors utilize a combination of Convolutional Neural Networks (CNNs) and Bidirectional Long Short-Term Memory Networks (BiLSTM) to extract the spatial and temporal features from the DAS data. The field data results prove its robustness for intelligent monitoring and real-time evaluation for hydraulic fracturing. Ye et al. develop a deep-learning method based on the YOLOv5 object-detection algorithm to estimate traffic flow and vehicle speed from the DAS signals acquired via a 500 m fiber segment in suburban Beijing. The authors evaluate the temporal traffic trend of the road section and explore vehicle classification via the fast-beamforming technique.

c) The Research Topic also includes a number of DAS case studies. Staněk et al. focus on reflected S-waves and use microseismic events as sources to image induced hydraulic fractures, which act as reflectors close to events and monitoring fiber. The proposed algorithm is capable of real-time fracture mapping and tracking fracture evolution in space and time. Nakajima et al. present a case study of a walk-away DAS-VSP survey applied to monitor geological CO2 storage at a pilot site in Japan. The authors obtained DAS data using a single-mode fiber via a casing installation. The authors apply conventional VSP processing workflows to the acquired DAS data and obtain promising images. Zhang et al. develop optimized algorithms for accurate tracking of the train position, speed, and number of trains traversing the position of the DAS system. The authors transform a 6-day set of continuous DAS data sensed by a 2-km cable into the time-velocity domain using beamforming and automatically extract train position and velocity from the time-beampower curves. It serves a supplementary role to conventional temporally and spatially discrete monitoring systems in railway traffic monitoring.