Monitoring technoloigies

[MO-01] Development of microseismic monitoring system using deep learning P- and S-waves picker in geothermal fields

*Kyosuke Okamoto¹, Yusuke Mukuhira², Hiroshi Asanuma¹, Markus Häring³ (1. National Institute of Advanced Industrial Science and Technology (Japan), 2. Tohoku University (Japan), 3. Häring GeoProject (Switzerland))

Microseismic monitoring in geothermal fields is a fundamental tool to estimate the extent of geothermal reservoirs, geothermal activity, flow paths in real-time. However, precise hypocenter determinations, which are the first step of the subsequent microseismic analyses, are based on manual picking of P- and S-wave arrivals by human analysis. This manual process is time and cost consuming. In this study, we demonstrated an automatic picker that was specified to microseismic events in geothermal fields. We developed an automatic picker based on deep learning with Okuaizu Geothermal Field (Japan) data. The deep learning model could provide P- and S- arrival times that can lead to qualitatively satisfiable hypocenter distributions in Basel Geothermal Field (Switzerland). Transfer learning model with the Basel Geothermal Field data improved the accuracy of picking, especially for S-waves, and more precise hypocenter distributions were obtained for the test data set from of Basel field. Further transfer learning using seismic data in many geothermal fields will lead to a more robust deep learning model for global microseismic monitoring.