The important task of tracking seismic activity requires both sensitive detection and accurate earthquake location. Approximate earthquake locations can be estimated promptly and automatically; however, accurate locations depend on precise seismic phase picking, which is a laborious and time‐consuming task. We adapted a deep neural network (DNN) phase picker trained on local seismic data to mesoscale hydraulic fracturing experiments. We designed a novel workflow, transfer learning‐aided double‐difference tomography, to overcome the 3 orders of magnitude difference in both spatial and temporal scales between our data and data used to train the original DNN. Only 3,500 seismograms (0.45% of the original DNN data) were needed to retrain the original DNN model successfully. The phase picks obtained with transfer‐learned model are at least as accurate as the analyst's and lead to improved event locations. Moreover, the effort required for picking once the DNN is trained is a small fraction of the analyst's.

No abstract

No abstract

The important task of tracking seismic activity requires both sensitive detection and accurate earthquake location. Approximate earthquake locations can be estimated promptly and automatically; however, accurate locations depend on precise seismic phase picking, which is a laborious and time-consuming task. We adapted a deep neural network (DNN) phase picker trained on local seismic data to mesoscale hydraulic fracturing experiments. We designed a novel workflow, transfer learning-aided double-difference tomography, to overcome the 3 orders of magnitude difference in both spatial and temporal scales between our data and data used to train the original DNN. Only 3,500 seismograms (0.45% of the original DNN data) were needed to retrain the original DNN model successfully. The phase picks obtained with transfer-learned model are at least as accurate as the analyst's and lead to improved event locations. Moreover, the effort required for picking once the DNN is trained is a small fraction of the analyst's.Plain Language Summary Seismic sensors are widely used to monitor many energy-related systems. To monitor these systems effectively, we need to process a very large amount of data, which is very labor intensive. A few deep learning models have been developed to perform these tasks for earthquake-generated signals. We adopted one of these deep learning models developed for kilometer scale and updated it for signals recorded from a meter-scale project. This process not only allows us to overcome the significant spatial and temporal scale difference between our data and the data used by the original deep learning model but also significantly reduces the amount of required training data. Our results show that the updated model matches human performance but with a much faster speed. A workflow that combines the deep learning algorithm with existing imaging technologies enables improvements for both monitoring small earthquakes and studying subsurface structure.

No abstract