12 pages, 7 figures, 3 tables, supplementary material https://doi.org/10.5194/se-11-63-2020.-- Data availability http://dx.doi.org/10.20350/digitalCSIC/8966, During September-October of 2013 an intense swarm of earthquakes occurred off the east coast of Spain associated with the injection of the base gas in an offshore underground gas storage. Two weeks after the end of the injection operations, three moderate-sized earthquakes (Mw 4.0-4.1) occurred near the storage. These events were widely felt by the nearby population, leading to the indefinite shut-down of the facility. Here we investigate the source parameters (focal depth and mechanism) of the largest earthquakes in the sequence in order to identify the faults reactivated by the gas injection and to help understand the processes that caused the earthquakes. Our waveform modeling results indicate that the largest earthquakes occurred at depths of 6-8 km beneath the sea floor, significantly deeper than the injection depth (∼1800 m). Although we cannot undoubtedly discriminate the fault plane from the two nodal planes of the mechanisms, most evidence seems to favor a NW–SE-striking fault plane. We propose that the gas injection reactivated faults in the Paleozoic basement, with regional orientation possibly inherited from the opening of the Valencia Trough, This research has been supported by the Ministerio de Ciencia e Innovación, Spain (grant no. CGL2017-88864-R), With the funding support of the ‘Severo Ochoa Centre of Excellence’ accreditation (CEX2019-000928-S), of the Spanish Research Agency (AEI)

10 pages, 4 figures, supporting information https://doi.org/10.1029/2020GL092038.-- Data Availability Statement: The associated data is available at the repository DIGITAL.CSIC (https://digital.csic.es/handle/10261/216863, The offshore Castor Underground Gas Storage (UGS) project had to be halted after gas injection triggered three M4 earthquakes, each larger than any ever induced by UGS. The mechanisms that induced seismicity in the crystalline basement at 5–10 km depth after gas injection at 1.7 km depth remain unknown. Here, we propose a combination of mechanisms to explain the observed seismicity. First, the critically stressed Amposta fault, bounding the storage formation, crept by the superposition of well‐known overpressure effects and buoyancy of the relatively light injected gas. This aseismic slip brought an unmapped critically stressed fault in the hydraulically disconnected crystalline basement to failure. We attribute the delay between induced earthquakes to the pressure drop associated to expansion of areas where earthquakes slips cause further instabilities. Earthquakes occur only after these pressure drops have dissipated. Understanding triggering mechanisms is key to forecast induced seismicity and successfully design deep underground operations., The authors would like to acknowledge Álvaro González for sharing the catalogs that were used in Cesca et al. (2014). Funding: Víctor Vilarrasa acknowledges funding from the European Research Council (ERC) under the European Union’s Horizon 2020 Research and Innovation Program through the Starting Grant GEoREST (www.georest.eu) (grant agreement No. 801809). Antonio Villaseñor acknowledges funding from Spanish Ministry of Science and Innovation grant CGL2017‐88864‐R. IDAEA‐CSIC is a Center of Excellence Severo Ochoa (Spanish Ministry of Science and Innovation, Project CEX2018‐000794‐S). ICM‐CSIC is a Center of Excellence Severo Ochoa (Spanish Ministry of Science and Innovation, Project CEX2019‐000928‐S)., Peer reviewed

A recent publication by Cesca et al.1 reanalyzes and expands seismic data to identify hypocenters of observed seismicity induced by the Castor Underground Gas Storage (UGS) operations. Their results confirm those of previous studies2,3 that earthquakes occurred below the storage formation on a fault dipping opposite from the Amposta fault, which bounds the reservoir. However, two important sets of disagreements require revising the conclusions by Cesca et al.1: the depth of hypocenters and the processes leading to seismicity. Inaccurate estimates of hypocenters location and partial consideration of the physical mechanisms that induce seismicity may imply endangering future deep underground projects., V.V. acknowledges funding from the European Research Council (ERC) under the European Union’s Horizon 2020 Research and Innovation Programme through the Starting Grant GEoREST (www.georest.eu) under grant agreement No. 801809. A.V. acknowledges funding from Spanish Ministry of Science and Innovation grant CGL2017-88864-R. IDAEA-CSIC and ICM-CSIC are Centres of Excellence Severo Ochoa (Spanish Ministry of Science and Innovation Grants CEX2018-000794-S and CEX2019-000928-S, respectively, funded by MCIN/AEI/ 10.13039/501100011033), Peer reviewed

14 pages, 9 figures, 3 tables, Picking arrival times of P and S phases is a fundamental and time‐consuming task for the routine processing of seismic data acquired by permanent and temporary networks. A large number of automatic pickers have been developed, but to perform well they often require the tuning of multiple parameters to adapt them to each dataset. Despite the great advance in techniques, some problems remain, such as the difficulty to accurately pick S waves and earthquake recordings with a low signal‐to‐noise ratio. Recently, phase pickers based on deep learning (DL) have shown great potential for event identification and arrival‐time picking. However, the general adoption of these methods for the routine processing of monitoring networks has been held back by factors such as the availability of well‐documented software, computational resources, and a gap in knowledge of these methods. In this study, we evaluate recent available DL pickers for earthquake data, comparing the performance of several neural network architectures. We test the selected pickers using three datasets with different characteristics. We found that the analyzed DL pickers (generalized phase detection, PhaseNet, and EQTransformer) perform well in the three tested cases. They are very efficient at ignoring large‐amplitude transient noise and at picking S waves, a task that is often difficult even for experienced analysts. Nevertheless, the performance of the analyzed DL pickers varies widely in terms of sensitivity and false discovery rate, with some pickers missing a significant percentage of true picks and others producing a large number of false positives. There are also variations in run time between DL pickers, with some of them requiring significant resources to process large datasets. In spite of these drawbacks, we show that DL pickers can be used efficiently to process large seismic datasets and obtain results comparable or better than current standard procedures, The authors gratefully acknowledge the computer resources at Artemisa, funded by the European Union ERDF and Comunitat Valenciana as well as the technical support provided by the Instituto de Física Corpuscular, IFIC (CSIC-UV). Antonio Villaseñor acknowledges funding from Spanish Ministry of Science and Innovation MCIN/AEI/10.13039/501100011033 grants CGL2017-88864-R, PID2020-114682RB-C31, and the Severo Ochoa Center of Excellence accreditation CEX2019-000928-S to ICM-CSIC, Peer reviewed