Toward a digital twin for seismic waves of induced events in a geothermal reservoir

Ermert, Laura A., Lanza, Federica, Shi, Peidong ORCID: https://orcid.org/0000-0001-5782-245X, Ritz, Vanille A., Tuinstra, Katinka, Finger, Claudia, Muff, Aaron, Obermann, Anne, Pio Rinaldi, Antonio, Meier, Men-Andrin and Wiemer, Stefan 2025. Toward a digital twin for seismic waves of induced events in a geothermal reservoir. Bulletin of the Seismological Society of America 10.1785/0120250125 Item availability restricted.

PDF - Accepted Post-Print Version Restricted to Repository staff only until 17 September 2026 due to copyright restrictions. Download (5MB)

Abstract

Monitoring induced seismicity is a key component of risk management for enhanced geothermal systems (EGSs). Induced seismicity presents particular monitoring requirements related to high event rates, small magnitudes, and difficulties of workflow testing due to the scarcity of manually labeled and ground‐truth data sets. Digital replicas of geothermal reservoirs could help address these requirements. Here, we present a seismic wave propagation digital replica inspired by the Utah Frontier Observatory for Research in Geothermal Energy (FORGE). We simulate over 20,000 induced event waveforms with a spectral‐element solver and source–receiver reciprocity, assemble them by origin time, and superimpose either low‐amplitude Gaussian noise or site‐specific correlated noise. The resulting continuous synthetic waveforms are used to test a real‐time and a postprocessing monitoring workflow based on SeisComP and MALMI, respectively. Both workflows perform well at retrieving input frequency–magnitude distributions. The synthetic waveform data set is made available online to facilitate future testing. Completeness strongly varies from the injection to the postinjection phase. The machine learning‐enabled processing tool performs well at identifying S phases, quantifying event locations and magnitudes, dealing with site noise, and reliably retrieving all input events at magnitudes >−0.25. The real‐time setup rapidly provides accurate estimates of the b value, robust characterization of large events, and approximate source information. Our study also provides insights into the feasibility of a digital “twin” component of wave propagation in an EGS. For time‐invariant setups like the one presented, the generation of synthetic data is computationally feasible, whereas frequent or operational changes of the structure model would likely require surrogate modeling. Near‐real‐time source characterization of observed induced events, including focal mechanisms, could greatly enhance the value of geothermal reservoir digital twin components.

Item Type:	Article
Date Type:	Published Online
Status:	In Press
Schools:	Schools > Earth and Environmental Sciences
Publisher:	Seismological Society of America
ISSN:	0037-1106
Date of First Compliant Deposit:	21 September 2025
Date of Acceptance:	6 August 2025
Last Modified:	22 Sep 2025 08:30
URI:	https://orca.cardiff.ac.uk/id/eprint/181216

Actions (repository staff only)

Edit Item