Numerous cases of induced/triggered seismicity resulting either directly or indirectly from injection/extraction associated with anthropogenic activity related to geo-resources exploration have been reported in the last decades. Induced earthquakes felt by the general public can often negatively affect public perception of geo-energies and may lead to the cancellation of important projects. Furthermore, large earthquakes may jeopardize wellbore stability and damage surface infrastructure. Thus, monitoring and modeling processes leading to fault slip, either seismic or aseismic, are critical to developing effective and reliable forecasting methodologies during deep underground exploitation. The complex interaction between injected fluids, subsurface geology, stress interactions, and resulting fault slip requires an interdisciplinary approach to understand the triggering mechanisms, and may require taking coupled thermo-hydro-mechanical-chemical processes into account.
In this session, we invite contributions from research aimed at investigating the interaction of the above processes during exploitation of underground resources, including hydrocarbon extraction, wastewater disposal, geothermal energy exploitation, hydraulic fracturing, gas storage and production, mining, and reservoir impoundment for hydro-energy. We particularly encourage novel contributions based on laboratory and underground near-fault experiments, numerical modeling, the spatio-temporal relationship between seismic properties, injection/extraction parameters, and/or geology, and fieldwork. Contributions covering both theoretical and experimental aspects of induced and triggered seismicity at multiple spatial and temporal scales are welcome.
vPICO presentations: Mon, 26 Apr
In the past decades, induced seismicity has become a major concern due to its correlation with oil and gas production and wastewater disposal. Unlike the induced seismicity observed in the United States that is associated with massive saltwater disposal, the induced seismicity observed in the Duvernay formation, a shale target in Alberta, Western Canada, is associated with hydraulic fracturing operations. In this work, we explore the possible mechanisms and the hydro-geological factors responsible for the seismic events that occurred between 2014 and 2015 in the Duvernay formation. By a two-dimensional finite element poroelastic model, using COMSOL Multiphysics, we couple fluid flow and solid deformation to estimate the change in the Coulomb Failure Stress (CFS) along two critically stressed faults existing near the hydraulic fracturing operations. One fault (Fault 1) is 1.01 km away from the location of hydraulic fractures while the second fault (Fault 2) is 0.425 km below the location of hydraulic fractures. The variations of the CFS along the two pre-existing faults are analyzed and compared to the seismic events obtained from the observational data in the Duvernay formation from December 2014 to March 2015 (Bao & Eaton, 2016). Our results show that most of the seismic events correlate spatially and temporally with positive CFS values that imply a risk of failure. During the early stages of hydraulic fracturing, the triggering failure mechanism of “Fault 1” is the increase in the shear stress on portions of the fault that are under extension and that of “Fault 2” is the pore pressure diffusion. Moreover, the distance between the centers of the two faults must range between 1.5 km and 2 km for the CFS results to agree with the observed seismic events. Under this condition, the shallower sections of “Fault 1” are under compression and show a stabilizing behavior (i.e., negative CFS) that is confirmed by the lack of seismic events from observational data, and the deeper sections of “Fault 1” are under extension and show a destabilizing behavior (i.e., positive CFS), which correlates with the measured seismic events. If the distance between “Fault 1” and “Fault 2” is less than 1.5 km, the shallower section of “Fault 1” would be destabilized by the effect of pore pressure, which does not agree with the observed seismic data. Moreover, if the distance between “Fault 1” and “Fault 2” is greater than 2 km, “Fault 1” would be entirely stabilized. Hence, the position of the faults with respect to the location of the hydraulic fracturing operations played an important role in the induced earthquakes triggering mechanisms and in the spatiotemporal distribution of the seismic events.
How to cite: Yassine, D., Yehya, A., and Maalouf, E.: Numerical Analysis of the Induced Seismicity Triggered by Hydraulic Fracturing in the Duvernay Formation in Alberta, CA, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-4740, https://doi.org/10.5194/egusphere-egu21-4740, 2021.
An increasing number of hydraulic fracturing (HF) operations in low-permeable tight shales in the Kiskatinaw area, northeastern British Columbia, have been associated with M3+ earthquakes in the last decade, including a ML 4.5 on 11/30/2018 near Dawson Creek. Here, we use a catalog of 8285 events ranging from magnitude ML -0.5 to 4.5 between July 2017 and July 2020 to investigate their source parameters. We identify event families using waveform cross-correlation and event temporal correlation, and estimate the focal mechanism solutions (FMS) of the highest-magnitude event within each family using the probabilistic earthquake source inversion framework Grond. We also estimate FMS for events with a magnitude larger than ML 2.5 that do not belong to a family (independent events). We compile a FMS catalog using the robustly constrained solutions for the largest events, and associate all smaller earthquakes with a cross-correlation coefficient (CCC) > 0.8 with the corresponding FMS. In addition, we estimate seismic moment and static stress drop values using spectral fitting methods on both single spectra and spectral ratios, and investigate their scaling relations.
In total, we constrain 65 FMS, of which 53 are clustered events, and the remaining 12 are independent events. An additional 4255 events have a CCC > 0.8 with one of the constrained FMS and are listed accordingly in the catalog. Of the total 4320 FMS, 93% are strike-slip events with one nodal plane at low angles to SH, 3% are dominantly strike-slip with thrust-faulting components, and the remaining 4% have a dominantly thrust-faulting mechanisms perpendicular to SH. The thrust-style events comprise the relatively larger magnitudes contained in the catalog, and may indicate slip on pre-existing faults. Most strike-slip events are part of an event family with multiple matching waveforms, while most thrust-faulting events are isolated with a low number of matching waveforms.
We fit the spectral corner frequency of 2360 P-phases and 1981 S-phases using single spectra estimates, and 1031 P-phases and 919 S-phases using the spectral ratios. While results from spectral ratios suggest a roughly constant stress drop of ~1 MPa for all magnitudes, the constant stress drop trend from single spectrum fitting breaks down at magnitudes smaller ~ ML 2.0, as has commonly been observed for events recorded by surface stations. We do not observe significant dependence of stress-drop values with the faulting style, nor with event depth.
How to cite: Roth, M. P., Kemna, K. B., Harrington, R. M., and Liu, Y.: Dominant strike-slip faulting and near-constant stress drop of induced earthquakes in the Kiskatinaw area, northeastern British Columbia, Canada, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-2437, https://doi.org/10.5194/egusphere-egu21-2437, 2021.
Over the last decade, low-permeability tight shale formations in the Western Canada Sedimentary Basin (WCSB) have been extensively developed using hydraulic fracturing (HF) techniques for oil- and gas exploration. In the meantime, an increasing number of M3+ earthquakes (e.g., ML 4.5 11/30/2018 near Dawson Creek, and an Mw 4.6 08/17/2016 near Ft. St. John) has been associated with HF operations. By increasing the seismicity in areas of low historical seismicity, the relationship between operational parameters and the rate of fault activation needs to be fully understood to avoid economic losses due to operation shutdowns or damages caused by ground shaking.
As earthquakes follow a well-known power-law magnitude-frequency relation, standard earthquake catalogs are typically dominated by microearthquakes in quantity. However, they usually still miss a large number of earthquakes due to insufficient station coverage and/or limited duration of observation. The latter could also lead to an inadequate time window for detecting larger earthquakes, which results in uncertainties of the power-law relation parameters for a particular area.
Here, we enhance a local seismic catalog derived from a dense seismic network in the Kiskatinaw (Montney Formation) area in British Columbia, Canada, using a multi-station matched filter technique. The existing automated STA/LTA catalog > 8000 earthquakes contains earthquakes from July 2017 - July 2020 with manually revised phase arrivals from up to 25 broad-band stations, ranging between ML -0.6 to 4.5. Using all the ~8000 events from the initial catalog as templates, we detected > 40,000 additional earthquakes, lowering the magnitude of completeness Mc from ~1.3 to ~0.2. We observe a b-value of approximately 1, and the majority of events occurred between 1.0 - 3.0 km depth, where injection depths range from 1.5 - 2.5 km.
In addition to the previously observed clustering of earthquakes around specific HF wells, we also observe ~8000 earthquakes with no apparent spatial (up to 5 km) or temporal (within two weeks of the reported HF stimulation) connection to HF operations. We estimate the contribution of the uncorrelated events to background seismicity rates. Furthermore, we detect earthquakes with templates related to HF operations, with spatial, but lacking temporal correlation to HF stimulation. Spatially correlated earthquakes with a no temporal correlation could highlight either areas with delayed induced seismicity (if following well stimulation) or areas with previous background seismicity (if preceding it). We will also show the correlation between background seismicity rates and cumulative injection rates and volumes of all wells within the study area.
How to cite: Kemna, K. B., Roth, M. P., Wache, R. M., Harrington, R. M., and Liu, Y.: The relevance of small-magnitude earthquakes in detailing the spatiotemporal correlation between hydraulic fracturing related injection and seismicity, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-5244, https://doi.org/10.5194/egusphere-egu21-5244, 2021.
Source parameters can help constrain the causes and mechanics of induced earthquakes. In particular, systematic variations of stress drops of fluid-injection induced seismicity have been interpreted in terms of the role of fluids, differences between tectonic and induced events, and self-similarity. The empirical basis for the variations, however, remains controversial. Here, we test three hypotheses about stress drops with observations of seismicity induced by hydraulic fracturing in the Horn River basin (Canada). First, stress drop is self-similar and independent of magnitude. Second, stress drop increases with distance from the point of fluid injection, which might be expected if in-situ effective stresses increase away from the point of fluid injection. Third, stress drops estimated with empirical Green’s functions (EGFs) are systematically larger than those estimated from direct fits to source models, which is expected if seismic waves attenuate in a frequency-dependent manner or experience site effects.
We probe the hypotheses with a large microseismic dataset collected during hydraulic fracturing operations in the Horn River shale gas play in British Columbia. 90,000+ seismic events were recorded by three borehole geophone arrays with a moment magnitude range of -3 < Mw < 0.5. To calculate corner frequencies, we assume small, co-located seismic events can be approximated as EGFs, which effectively remove propagation and site effects from a larger target event. We target 34 Mw > 0 events and search for EGFs over a 100 m radius for each event, choosing only those EGFs that satisfy multiple quality criteria. This study builds on previous work that estimated stress drops from direct fitting of standard Brune source models and found systematic high frequency resonances recorded by the geophones.
Of the 34 target events, we retrieve corner frequency and stress drop estimates for 22 events to test the three hypotheses. We observe that stress drop appears relatively constant over Mw , but the magnitude range (0 < Mw < 0.5) is currently too limited to draw strong conclusions. Second, stress drop appears to decrease, rather than increase, with distance from the point of injection (with a moderate Pearson’s correlation co-efficient of -0.5 ± 0.2); this could be caused by a direct hydraulic connection causing a reduction of in-situ effective normal stresses distal to the point of injection. Third, we observe no systematic difference between stress drops from direct source fits and EGF-based estimates, although stress drop uncertainties are large compared to standard earthquake source studies because of limited azimuthal coverage and high-frequency instrument resonances. These initial results do not support the systematic variations of stress drop for fluid-injection induced seismicity that have been observed in other datasets.
How to cite: Klinger, A., Holmgren, J., and Werner, M.: Testing hypotheses of stress drop variations with hydraulic fracturing induced seismicity in the Horn River basin, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-14795, https://doi.org/10.5194/egusphere-egu21-14795, 2021.
In the scope to investigate the possible interactions between injected fluids, subsurface geology, stress field and triggering earthquakes, we investigate seismic source parameters related to the seismicity in West Texas (USA). The analysis of seismic moment tensor is an excellent tool to understand earthquake source process kinematics; moreover, changes in the fluid volume during faulting leads to existence of non-double-couple (NDC) components (Frohlich, 1994; Julian et al., 1998; Miller et al., 1998). The NDC percentage in the source constitutes the sum of absolute ISO and CLVD components so that %NDC= % ISO + %CLVD and %ISO+%CLVD+%DC=100%. It is currently known that the presence of NDC implies more complex sources (mixed shear-tensile earthquakes) correlated to fluid injections, geothermal systems and volcano-seismology where induced and triggered seismicity is observed.
With this hypothesis, we analyze the micro-earthquakes (M <2 .7) recorded by the Texas Seismological Network (TexNet) and a temporary network constituted by 40 seismic stations (equipped by either broadband or 3 component geophones). Our study area is characterized by Northwest-Southeast faults that follow the local stress/field (SHmax) and the geological characteristic of the shallow basin structure of the study area. After a selection based on signal-to-noise ratio, we filter (1-50 Hz) the seismograms and estimate P-wave pulse polarities and the first P-wave ground displacement pulse in time domain. Then, we perform the full moment tensor analysis by using hybridMT technique (Andersen, 2001; Kwiatek et al., 2016) with a detailed 1D velocity model. The key parameter is the polarity/area of the first P-wave ground displacement pulse in time domain. Uncertainties of estimated moment tensors are expressed by normalized root-mean-square (RMS errors) between theoretical and estimated amplitudes (Vavricuk et al., 2014). We also evaluate the quality of the seismic moment tensors by bootstrap and resampling. In our preliminary results we obtain NDC percentage (in terms of %ISO and %CLVD components), Mw, seismic moment, P, T and B axes orientation for each source inverted.
How to cite: Alexandros, S. and Pamela, R.: Analysis of non-double-couple source mechanisms in an area of induced seismicity, West Texas (USA)., EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-13438, https://doi.org/10.5194/egusphere-egu21-13438, 2021.
On October 27th, 2017, a Mw 4 earthquake occurred close to the municipality of Montesano sulla Marcellana, less than 10 km external to the concession of the largest European on-shore hydrocarbon reservoir - the Val d’Agri oilfield (Southern Italy). Being a weak event located outside the extended monitoring domain of the industrial concession, the relevance of this earthquake and possible links with the hydrocarbon exploitation were not deepened. The study of weak to moderate earthquakes can improve the characterization of the potentially destructive seismic hazard of this particular area, already struck by M>6.5 episodes in the past. Taking advantage of a wide coverage of seismic stations deployed in the VA region, we analyze the source parameters of this Mw 4 earthquake applying advanced seismological techniques to estimate the uncertainties derived from the moment tensor inversion and identify plausible directivity effects. The moment tensor is dominated by a NW-SE oriented normal faulting with a centroid depth of 14 km. A single ML 2.1 aftershock was recorded and used as empirical Green function to calculate the apparent source time function for the mainshock. Apparent durations (in the range 0.11 - 0.21 s, obtained from S-waves) define an azimuthal pattern which reveals an asymmetric bilateral rupture with the 70% of the rupture propagation in the N310°W direction, suggesting a rupture plane dipping to the SW. Our results conclude that the Montesano earthquake activated a deeper fault segment associated to the Eastern Agri Fault System close to the basement. The relative low trigger potential below 10% based on depletion-induced stress changes discards an induced or triggered event due to the long-term hydrocarbon extraction in the Val d’Agri oilfield, and it rather suggests a natural cause due to the local tectonic stress.
How to cite: López-Comino, J. Á., Braun, T., Dahm, T., Cesca, S., and Danesi, S.: Deciphering the source parameters and genesis of the 2017, Mw 4 Montesano earthquake close to the Val d’Agri Oilfield (Italy), EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-1076, https://doi.org/10.5194/egusphere-egu21-1076, 2021.
The United Downs enhanced geothermal system in Cornwall, UK, has induced several microseismic events since flow testing began in August 2020, targeting a granitic intrusion at 5 km depth. As of January 2021, two events exceeding local magnitudes (ML) 1.5 have occurred, highlighting the associated seismic hazard and providing initial data for a preliminary assessment of the region’s ground motion response. However, with only one national seismic station publicly available within 90 km of the site, public data are scarce. In an effort to involve the surrounding communities in the geothermal project, United Downs provided Raspberry Shake 1D or 4D (one vertical geophone, with 4D containing an additional three accelerometers) seismographs to nearby schools, increasing the number of publicly available seismic stations to ten within 15 km of the site. In this study, we assess the ground motions recorded by the Raspberry Shake stations and evaluate their utility for probing ground motions models (GMMs) and the effects of the local geology.
171 earthquakes between ML -1.3 to 1.7 originating at United Downs have been recorded to date, with 37 events above ML 0.0. Unfortunately, the accelerometer components of the Raspberry Shake instruments contained too high background noise levels to be useable, leaving only the vertical geophone component to be analysed for each of the instruments. We find that while the peak ground velocity (PGV) values are in line with those predicted from the Douglas et al. (2013) geothermal GMM, the area experiences higher peak ground acceleration (PGA) than expected. We also find that the observed PGVs and PGAs match the region’s geological features, consisting of a combination of igneous intrusions and sedimentary sandstones and mudstones. For sparse national seismic networks, Raspberry Shake stations can provide a quick initial evaluation of seismic events and their ground motions before industry releases private data for more detailed analyses.
How to cite: Holmgren, J. and Werner, M.: Raspberry Shakes provide initial ground motion assessment of the geothermally induced seismicity at United Downs in Cornwall, UK, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-14448, https://doi.org/10.5194/egusphere-egu21-14448, 2021.
Possible geothermal water resources in the eastern part of Vienna could ideally provide hundreds of thousands of Viennese households with environmentally friendly heat and hot water. Currently, the potential of low-lying hot water resources in this area is being explored geologically and geophysically within the scope of the multi-disciplinary project GeoTief Explore 3D.
For the development of geothermal projects, studies of historical seismicity and monitoring of current seismicity are basic requirements. Therefore, we deployed a seismic network consisting of four stations in the area of interest in 2019. After 2 years of recording, we search the data for previously undetected small magnitude local earthquakes. Also we evaluate the local background noise and its variability with time. Based on these results, we estimate site-specific detection thresholds within in the project area.
How to cite: Apoloner, M.-T., Rodler, F.-A., and Lenhardt, W.: Evaluation of 2 years seismic monitoring for geothermal development in Vienna, Austria , EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-7124, https://doi.org/10.5194/egusphere-egu21-7124, 2021.
The Ruhr district meets the necessary elements to carry out geothermal projects due to its geothermal potential and demand, as it is a densely populated industrial area. Currently, there are projects for direct use, whereas projects for electricity generation are planned. The latter, due to greater depths, reservoir enhancement techniques are required in some cases. This may increase the associated seismic risk which should be elaborated in detail.
With available data, a three-dimensional geological and structural model was created. The shallower parts have been widely studied and documented by mining activity in the Ruhr region during the last century. Below a depth of 1 km, data are scarce, and uncertainties increase. The full elastic wavefield emitted by a realistic seismic source has been simulated using a finite differences scheme and the derived geological model. The elastic properties were estimated with well data. The source has common characteristics of real seismic events in the area.
The wave propagation simulations let us analyze the seismic response with different sources and velocities models. Three cases are considered, two seismic events with distinct depths based on real events. The third case is based on the proposed location of a deep geothermal project.
Especially for the case with the deeper source, the areas with relatively high amplitudes of displacement correlated with structural features of the model. Applying the imaging condition of maximum energy density allows us to define zones with a potentially increased seismic risk that should be monitored more closely.
How to cite: Gonzalez de Lucio, G. D. L. A., Finger, C., and Saenger, E.: Wave propagation simulation on a 3D model of the Ruhr district for locations of seismic monitoring., EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-12774, https://doi.org/10.5194/egusphere-egu21-12774, 2021.
Since Nov 2019, a series of seismic of events were felt by the population of the city of Strasbourg, France. The first main event (Ml3.0) on Nov 12, 2019 was part of a seismic swarm that has been initiated a few days before, lasted four month and was located by the BCSF-RéNaSS (EOST) in the northwestern part of the town (Robertsau area) at a depth of 5 km. Its location in the vicinity of the deep geothermal wells (GEOVEN), the temporal correlation with the injection activity on site, the similarity of the depth between the bottom of the wells and the hypocenter of the event, the lack of local seismicity before the event occurrence, the known geological structures including crustal faults in the area, all strongly support the possible triggering of the events by the deep geothermal activities despite the relatively large distance (4-5km). Template matching has been applied and allowed for a significant improvement of the detections. Double-difference relocations evidenced a complex fault zone in the swarm area that extends over 800m. Focal mechanisms of the two main events are consistent with the known orientation of the fault zone. The regional stress field in combination with the fault orientation and a Coulomb failure criterion, shows that the swarm location is in an unstable domain if the cohesion of the fault is weak, particularly sensitive to stress perturbations. Since Oct 2020, hydraulic tests were initiated and a second cluster of seismic events with more felt earthquakes developed closer to the geothermal wells. It includes the largest event (Ml3.6) that was induced on Dec 4, 2020 and caused the definitive arrest of the project. A preliminary analysis shows that most of the largest events happened along the same fault zone as in Nov 2019 but very close to the injection well, where a significant over-pressure has been maintained over time.
This presentation is dedicated to the memory of Prof. François Cornet.
How to cite: Schmittbuhl, J., Lengline, O., Lambotte, S., Grunberg, M., Doubre, C., Vergne, J., Cornet, F., and Masson, F.: Induced and triggered seismicity from Nov 2019 to Dec 2020 below the city of Strasbourg, France, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-8374, https://doi.org/10.5194/egusphere-egu21-8374, 2021.
The research site of Soultz-sous-Forêts (Alsace, France) was a pioneer pilot geothermal site in Europe. In this study, we use the available data from 2000 and 2003 hydraulic stimulation tests to analyze the seismicity evolution. We apply the ETAS (Epidemic-Type Aftershock Sequence) model to extract the background seismicity rate during the two stimulation periods.
For the 2003 sequence, to retrieve the nonstationary seismicity component, we use a moving window of 400 events for the whole catalog. The evolution of the background seismicity rate μ is successfully retrieved with an evolution in two peaks coherent with the wellhead pressure evolution, while the triggering parameter Κ is stable. At the end of the stimulation μ decrease significantly. Then we look at the evolution of ETAS parameter by selecting five clusters of seismicity. The evolution of μ for each cluster is in agreement with a propagation of the pressure away from the well with the cluster closer to the well showing one early peak only, the middle clusters showing two peaks and the far cluster showing a later peak. All clusters show a decrease of μ at the end of stimulation.
For the 2000 sequence, the background seismicity rate is less well constrained but it stays globally constant during the stimulation with some decrease after its end. We see no clear peak in μ as was present during 2003 and K is relatively low. However, μ also decreases at the end of the stimulation. The selection of clusters does not change this global behavior and all clusters present grossly the same characteristics.
Our results are in agreement with the different characteristics observed by several authors (e.g. Calo and Dorbath, 2013; Dorbath et al, 2009) between these two stimulations. On one hand, the 2003 stimulation consists in an activation of several existing structures that yields a seismicity well explained by the ETAS model with a combined effect of Coulomb stress transfer and perturbation induced by the stimulation (e.g. pore pressure variation). The evolution in space is also coherent with the finding of Calo and Dorbath (2013) that the injected water goes far from the well avoiding increase in effective stress near the well. In this case, background seismicity rate can be related to the measured pressure. On the other hand, the 2000 stimulation developed a 3D reservoir with the creation of a fresh shear zone (Cornet et al, 2015) and so the direct effects of the stimulation are dominants. However, no clear relation between the background seismicity rate and the operational parameters can be observed. At the end of stimulation, we observe a decrease of background rate corresponding to a progressive return to a natural background rate, similar to what is observed in other settings (Oklahoma, Rousse).
How to cite: Maury, J. and Aochi, H.: Comparison of the seismicity evolution during the 2000 and 2003 stimulations at Soultz-sous-Forêts., EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-8864, https://doi.org/10.5194/egusphere-egu21-8864, 2021.
In addition to stable and accurate hypocenters of seismic events, the characterisation of events is crucial for the investigation of seismicity in the context of geothermal reservoirs, CO2-sequestration and other geotechnical applications. Since the origin and nature of the seismicity in such cases is still under investigation, tools should rely on as few a priori assumptions about the sources as possible. Here, an approach is presented to determine the time-dependent moment tensor and origin time in addition to commonly derived hypocenter locations of seismic events using time-reverse imaging (TRI). The full six component moment tensor is derived and may be used to display for example focal mechanisms. The workflow consists of determining the location of potential sources, discriminating artificial and true source locations and obtaining the time-dependent moment tensors by recording the stress components at the derived source locations. Since TRI does not rely on the identification of seismic phases but on the simulation of the time-reversed wavefield through an adequate velocity model, no assumptions about the source location or the type of source mechanism is made. TRI is less affected by low signal-to-noise ratios and is thus promising for noisier sites and quasi-simultaneous events. However, a sufficient number of seismic stations are needed to accurately sample the wavefield spatially. The proposed workflow is demonstrated by locating and characterising microseismic events in the geothermal field of Los Humeros, Mexico. Although higher levels of noise are present and only a one-dimensional velocity model is available at this time, selected events could be located and characterised.
How to cite: Finger, C. and Saenger, E. H.: Characterisation of seismic events using time-reverse imaging , EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-7206, https://doi.org/10.5194/egusphere-egu21-7206, 2021.
In this study we investigate the statistical spatio-temporal characteristics induced seismicity associated with two stimulation campaigns performed in 2018 and 2020 in a 6.1 km deep geothermal well near Helsinki, Finland as part of the St1 Deep Heat project. We aim to find out whether the seismic activity is passively responding to injection operations, or whether we observe signatures of significant stress transfer and strong interactions between events. The former suggests stable relaxation of seismic energy proportional to hydraulic energy input, while the latter includes stress transfer as an additional source of stress perturbation, hence implying larger seismic hazard.
The selected catalogs from 2018 and 2020 stimulation contained in total 60,814 and 4,368 seismic events, respectively, recorded during and after stimulation campaigns and above the local magnitude of M -1.5. The analyzed parameters include magnitude-frequency b-value, correlation integral (c-value), fractal dimension (D-value), interevent time statistics, magnitude correlation, interevent time ratio and generalized spatio-temporal distance between earthquakes. The initial observations suggest significant time-invariance of the magnitude-frequency b-value, and increased D and c-values only at high injection rates, the latter also guiding the rate of seismicity. The seismicity covering the stimulation period neither provide signatures of magnitude correlations, nor temporal clustering or anticlustering. The interevent time statistics are generally characterized with Gamma distribution (close to Poissonian distribution), and the generalized spatio-temporal distance suggest very limited triggering (90% of the catalog was classified as background seismicity). The observable parameters suggest the seismicity passively respond to hydraulic energy input rate with little to no time delay, and the total seismic moment is proportional to total hydraulic energy input. The performed study provides the base for implementation of time-dependent probabilistic seismic hazard assessment for the site.
How to cite: Kwiatek, G., Leonhardt, M., Martínez-Garzón, P., Pentti, M., Bohnhoff, M., and Dresen, G.: Seismicity associated with 2018 and 2020 hydraulic stimulations at EGS in Helsinki, Finland, shows limited earthquake interaction: Implication for seismic hazard assessment, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-12217, https://doi.org/10.5194/egusphere-egu21-12217, 2021.
The evolution and characteristics of induced seismicity in geothermal stimulations can shed light on water pathways and fracture network development. However, these seismic sources are usually difficult to characterize due to their small magnitudes and the low signal-to-noise ratio (SNR) of observational recordings. Heterogeneous and ill-constrained 3D subsurface structure further restricts the local-scale application of array based methods, such as the back-projection method. The 2018 st1 Deep Heat geothermal stimulation experiment in Espoo, Finland, induced thousands of seismic events in the 5-6 km depth range with magnitudes smaller or equal to ML 1.8 (Hillers, et al., 2020). The competent bedrock and absence of a dissipating sedimentary layer results in high SNR seismograms collected by three 4-station arrays, three 25-station arrays and tens of standalone stations located within 5 km distance around the wellhead. These high-quality data facilitate the application of multi-array beamforming and the back-projection methods, to image small-magnitude induced seismicity sources and characterize their properties at reservoir scales.
The beamforming results demonstrate array, frequency and phase (P or S) dependent slowness biases of catalog locations, which are obtained using standard location procedures with manually picked P- and S-wave arrivals. This indicates multi-scale heterogeneity in the study region. Specifically, we find that the back azimuth of the slowness at each array points to inconsistent locations and leads to poorly constrained epicenters. We show that the systematic slowness variability can be reduced and multi-array location estimates can be greatly improved by calibration using well-constrained catalog events.
To perform the back-projection, we select unclustered stations from narrow epicentral distance ranges to avoid unfavorably large variations in the duration of the body phases, and we set the azimuth gap threshold to less than 40 degrees. The locations determined by the back-projection are close to the catalog locations, with the majority of them within 150 m, suggesting a successful application of the back-projection technique using local stations to study small events. We repeatedly observe “swimming” artifacts (Ishii et al., 2007; Walker and Shearer, 2009), i.e. the back-projection locations migrate in a certain direction with time. This is typically attributed to array-source directivity effects in teleseismic applications, but in our case the stations are well-distributed around the source. We next use numerical wave propagation simulations, with receivers homogeneously azimuthal distributed at constant epicentral distance to a point-source. We apply the back-projection using synthetic seismograms. The results confirm the consistent appearance of “swimming” patterns and the apparent migration direction which changes in dependence on the focal mechanism of the point source. We conclude that the back-projection method may provide useful proxies for source mechanisms to help track and link the evolving fracture network.
How to cite: Li, B., Gabriel, A.-A., Rintamäki, A., and Hillers, G.: Array based analysis of induced earthquake characteristics using beamforming and back-projection methods in Helsinki, Finland, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-12888, https://doi.org/10.5194/egusphere-egu21-12888, 2021.
Enhanced Geothermal Systems apply the pressurized fluid injection to fracture impermeable rocks to form pathways in which water circulates. The cold water under high pressure is pumped into the hot subsoil, where it heats up and returns to the surface. However, the induced fractures may coalesce into unwanted paths that allow the fluids to reach pre-existing faults, triggering major seismic events.
This work investigates the relationship between injection and a degree of disordering of sources, ZZ, at Cooper Basin geothermal field in Australia, following the methodology developed and applied to study The Geysers geothermal field case (Lasocki & Orlecka-Sikora, 2020). The parameter ZZ quantifies the potential of seismicity to build pathways for fluid migration. It is the average distance between the seismic events in the eight-dimensional parameter space consisting of three hypocentral coordinates, T- and P-axis plunges, T-axis trend, and polar and azimuthal angles in the spherical system of coordinates beginning at the open hole of an injection well. A decrease of ZZ indicates an increasing hazard of forming far-reaching migration pathways. In The Geysers case, ZZ turned out to be highly correlated with the injection rate.
Here we focus on the case of Habanero 4 well stimulation from 17 - November 30, 2012 (data access, see: IS EPOS, 2020). We processed 489 seismic events with known focal mechanisms. The events moment magnitude varies between 0.8 and 3.1.
Our analysis shows that ZZ is significantly correlated with both the injection rate and the wellhead pressure. The higher the injection rate / the wellhead pressure was, the less probable was the creation of undesired fluid migration pathways. The Cooper Basin’s and The Geyser’s reservoir rocks are vastly different, the former – granite, the latter – greywacke sandstone, likewise the stimulation techniques applied in these two reservoirs. However, in both cases, ZZ was positively correlated with injection rate; thus, the potential to build unwanted paths for fluids was negatively correlated. These results suggest that such correlation may be a global feature of rock fracturing caused by pressurized fluid injections.
This work has been supported by S4CE (Science for Clean Energy) project, funded from the European Union’s Horizon 2020 - Framework Programme, under grant agreement No 764810 and by PRIN-MATISSE (20177EPPN2) project funded by Italian Ministry of Education and Research.
IS EPOS (2020), Episode: COOPER BASIN, https://tcs.ah-epos.eu/#episode:COOPER_BASIN, doi:10.25171/InstGeoph_PAS_ISEPOS-2020-001
Lasocki, S., & Orlecka-Sikora, B. (2020). High injection rates counteract formation of far-reaching fluid migration pathways at The Geysers geothermal field. Geophysical Research Letters, 47, e2019GL086212. https://doi.org/10.1029/2019GL086212
How to cite: Battimelli, E., Lasocki, S., and Capuano, P.: High injection rates decrease the probability of creating undesired, far-reaching fluid migration pathways at Cooper Basin geothermal field, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-4902, https://doi.org/10.5194/egusphere-egu21-4902, 2021.
Fluid-injections under high pressures into deep “hot” rock formations are routinely performed during the development of Enhanced Geothermal Systems (EGS). Such fluid-injections, which aim to enhance the permeability in the targeted rock formation, can induce intense microseismicity and in some cases even larger magnitude earthquakes. A characteristic of injection-induced seismicity is its spatial migration with time, which is considered indicative of pore-pressure diffusion and the geometry of the stimulated volume in which permeability is enhanced. Understanding the details of earthquake migration during stimulation operations is particularly important for the design of EGS, the management of operations, as well as for the mitigation of hazardous induced earthquakes. Herein, we develop a stochastic model to map the spatiotemporal evolution of injection-induced seismicity. The model is based on the well-established Continuous Time Random Walk (CTRW) theory that has widely been applied in nonlinear transport phenomena in complex heterogeneous media. Within this context, we describe the spatiotemporal evolution of injection-induced seismicity with an appropriate master equation and the time-fractional diffusion equation. Application of the model to two stimulation experiments in the Cooper Basin (Australia) EGS shows that induced seismicity migrates slowly with time away from the injection points according to a subdiffusive process, with waiting times between the successive earthquakes drawn from a broad probability density function with asymptotic power-law behavior. Moreover, we show that the solution of the time-fractional diffusion equation adequately describes the propagation of induced seismicity in time and space, showing a peak of earthquake concentration close to the injection point and a stretched exponential decay for the concentration of distant events. The results demonstrate that the CTRW model can efficiently describe nonlinear diffusion of injection-induced seismicity during stimulation operations in EGS.
The research project was supported by the Hellenic Foundation for Research and Innovation (H.F.R.I.) under the “2nd Call for H.F.R.I. Research Projects to support Post-Doctoral Researchers” (Project Number: 00256).
How to cite: Michas, G. and Vallianatos, F.: Stochastic modelling of injection-induced seismicity in the Cooper Basin enhanced geothermal system, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-5138, https://doi.org/10.5194/egusphere-egu21-5138, 2021.
Hydraulic stimulation for the creation of an Enhanced Geothermal System (EGS) reservoir could potentially reactivate a nearby fault and result in man-made earthquakes. In November 15, 2017, an Mw 5.5 earthquake, the second largest after the initiation of the South Korean national instrumental monitoring system, occurred near an EGS project in Pohang, South Korea. The earthquake occurred on a previously unmapped fault, that is here denoted the Mw 5.5 Fault. A number of previous studies to model the hydraulic stimulation in the Pohang EGS project have been carried out to identify the mechanism of seismic events. Those previous studies focused on coupled hydro-mechanical processes without the consideration of pre-existing fractures and thermal effects. This study presents an investigation of the mechanisms of induced and triggered seismicity in the Pohang EGS project through three-dimensional coupled thermo-hydro-mechanical numerical simulations. Fractures intersecting the open-hole sections of two deep boreholes, PX-1 and PX-2, clearly indicated by field observations are modeled along with the Mw 5.5 Fault. Models of stress-dependent permeability models are calibrated based on the numerical reproduction of the pressure-time evolution during the field hydraulic stimulations. The Coulomb failure stress change at the Mw 5.5 Fault is calculated to quantify the impact of five hydraulic stimulations. In the case of PX-2 stimulations, the pore pressure buildup results in a volumetric expansion of the reservoir and thereby the perturbation of stresses is transferred to the Mw 5.5 Fault. The volumetric contraction of the reservoir by the temperature reduction could slightly perturb the stress distribution at the Mw 5.5 Fault. In the case of PX-1 stimulations, shear slip of the PX-1 fracture is explicitly modeled. The modeling shows that transfer of the shear stress drop by the shear slip stabilizes the Mw 5.5 Fault, which is consistent with the field observation that the seismicity was not induced at the Mw 5.5 Fault by the PX-1 stimulations. The cooling-induced thermal stress additionally reduces the effective normal stress of PX-1 fracture. Thus, some additional shear slip of the PX-1 fracture is induced by the thermal effect. However, the modeling shows that for both PX-1 and PX-2 stimulations, thermally-induced stress perturbations are very small compared to pressure-induced stress perturbations.
How to cite: Kim, K.-I., Yoo, H., Park, S., Yim, J., Xie, L., Min, K.-B., and Rutqvist, J.: A numerical study of the mechanism of injection-induced and triggered seismicity at the Pohang Enhanced Geothermal Systems project, South Korea, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-6912, https://doi.org/10.5194/egusphere-egu21-6912, 2021.
As many industrial activities impacting the underground, deep geothermal projects can be associated with the occurrence of induced seismic events. This seismicity is sometimes a direct consequence of stimulation operations needed to enhance the permeability of geothermal reservoirs, but, in other cases, it can also occur in different phases of geothermal projects, as during wells shut-in, after injection operations, or during the production phase, which generally implies lower flow rates and injection pressures. The intensity of this seismicity, in terms of magnitudes of seismic events, can be extremely variable, from microseismic events (M < 2), not felt at the surface, to large earthquakes (M > 5) that pose a serious risk to neighboring populations and may lead to the abandon of geothermal projects. In this context, it is of paramount importance to: i) better characterize and understand the interactions between natural and anthropogenic factors which may lead to geothermal-induced seismicity and ii) evaluate currently applied approaches to handle and minimize associated risks.
The objective of this work is to establish a state of the art about deep geothermal-induced seismicity, by describing factors that have a bearing on the generation of seismic events, as well as by discussing existing means to handle their occurrence. Based on a worldwide review of geothermal projects, we created a large database describing each selected case study in terms of geological properties and tectonic setting, operational parameters and type of geothermal systems, as well as spatio-temporal characteristics of the observed induced seismicity. Collected data are analyzed in order to better understand possible cause-effect relationships between induced seismicity and geothermal operations with the aim of identifying the most important preexisting and anthropogenic factors, as well as their interactions, which may have a key role on the occurrence of seismic activity.
How to cite: De Santis, F., Klein, E., and Thoraval, A.: Deep geothermal-induced seismicity: controlling factors and hazard mitigation measures, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-14518, https://doi.org/10.5194/egusphere-egu21-14518, 2021.
Estimation of hypocenter location errors is not a simple task. These errors are influenced by many factors. The most important are: the quality of velocity model, the configuration of stations in the observation network and the noise level recorded at stations. While the network configuration affects the error distribution in a deterministic manner, the noise level is largely random. It means that the uncertainties cannot be determined in a deterministic way and only statistical approach can be used. There are several methods for estimating location errors for particular seismic network. Some techniques use synthetic seismograms to calculate the detection range related to each station. However, this approach requires very precise knowledge of the geological model, which is not always possible. Instead, in this work we present a different approach, which uses only phase data for events included in the catalog. In this method, the detection range for each station is estimated using the detection probability (Schorlemmer & Woessner, 2008) used for both P- and S- waves first arrivals. The usefulness of this approach is discussed assuming the shape of LUMINEOS seismic network which operates in the Legnica-Głogów Copper District (LGCD), Poland. In the LGCD region seismic activity is related to three deep underground copper mines. Every year thousand of seismic events with magnitudes up to M4.0 are registered here. Some of them are followed by tragic mining collapses and are widely felt by local residents.
How to cite: Kokowski, J. and Rudziński, Ł.: Estimation of earthquakes location errors distribution for LUMINEOS local seismic network in Poland. , EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-7475, https://doi.org/10.5194/egusphere-egu21-7475, 2021.
Mining exploitation is associated with the occurrence of adverse environmental effects. The most serious of such effects is land subsidence. Although land subsidence can be well predicted and mitigated by several methods, nevertheless, the extraction of mineral deposits is also associated with induced seismicity. The occurrence of seismic events causes ground surface vibrations, land surface displacements and, in many cases, has a negative impact on the safety of surface infrastructure and the inhabitants of endangered areas. Despite this, the issue of induced seismicity is much less recognized and often ignored in the assessment of the negative impacts of mining exploitation.
Induced seismicity is related to stress changes in the reservoir and surrounding rock mass that may be caused by a variety of mechanisms. Consequently, the patterns of induced seismicity vary greatly over time and space for different fields or events within the same field. It is often difficult to determine the correlation between seismicity and mining precisely because of the lack of data detailing the pattern of exploitation at the various wells. As a result, the source mechanism of mining-induced tremor remains a subject of active research.
The research aimed to better identify the phenomenon of induced seismicity caused by mining operations. Research has been conducted in the area of underground copper ore mining in Poland. Firstly, we investigate the pre-and post-seismic land-surface movements following 8 mining-induced Mw 3.6-4.8 earthquakes that occurred between 2016 and 2018. We use Sentinel 1 data to derive these movements 2 weeks before and 4 weeks after the mainshock. The results of these studies show that no substantial pre-seismic surface movements are indicating the possibility of a seismic event occurring. However, the co-seismic deformation fields are quite symmetrical, the maximum land subsidence is almost 10 cm and occurs within a few days after the mainshock. In addition, the time series of post-seismic deformation shows a gradual decay and a good correspondence with the post-shock distribution.
Secondly, we use the Mogi model, assuming the elastic half-space, to invert co-seismic deformation fields and to obtain the source parameters of the mine-induced earthquakes. The spatial distribution of the epicenters indicates a correlation with the fields of mining exploitation. The results also show that the average depth of the hypocenter tremor is approx. 650 m. This corresponds to the depth of the stiff sandstone layers adjacent to the exploration. These layers accumulate the stress of post-exploitation voids. In addition, the modeling results indicate an approx. the volume of the displaced rock layers of 1.2 x 105 m3. This value shows a high correlation with the volume of post-shock troughs determined based on InSAR data.
The results of this study contribute to research into activities related to mining operations resulting in an induced-earthquake occurrence. This demonstrates InSAR's potential for quasi-constant monitoring of large-scale areas against seismic hazards caused by ongoing mining operations.
How to cite: Witkowski, W., Łukosz, M., Guzy, A., and Hejmanowski, R.: Mining-Induced Tremors Source Modelling Applying InSAR, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-7269, https://doi.org/10.5194/egusphere-egu21-7269, 2021.
FloodRisk is an interdisciplinary project focusing on the effects of mine water level rise in abandoned coal mine regions in Germany. Such effects are heterogeneous ground uplift, stress changes due to the change in pore pressure and the reactivation of potential faults. One of the most directly measurable effects is certainly the induced micro seismicity. It is known from previous studies that the flooding of old mines can lead to a renewed increase level in induced micro seismicity in these regions.
In this study the relationship between mine water rise, fluid-induced stress changes and induced seismicity in the Haus Aden dewatering area in the eastern Ruhr area (Germany) will be investigated in more detail.
For this purpose, we operate a network of currently 21 short period seismic stations in the region of the former "Bergwerk Ost" colliery, which had the highest seismicity rate in the Ruhr area during active underground coal mining. This network is still to be expanded to cover the entire water drainage area, about 30 Raspberry Shake sensors are waiting for the possibility of installation.
Nevertheless, the existing network registered almost 1000 induced micro seismic events in a magnitude range from -0.7 up to 2.6 MLv. Many of these events are spatially clustered and some show quite high waveform similarity. This allows relative localisation and can increase the accuracy of the location. The depth location of the earthquakes, within the limits of localisation accuracy, agrees very well with the distribution of seismicity at the time of active mining. The spatial distribution so far seems to be limited by a large inactive transverse fault in the west. It needs to be clarified what influence this fault has on the propagation of mine water in the underground.
The measured temporal trend of the mine water level, after pumps were shut down in mid-2019, shows a strong correlation with the temporal evolution of the observed micro seismicity. In the first months after the pumps are switched off, the water levels at the observation points rise only slowly and isolated microseismic events occur again. In November 2019, the rise in water levels doubled and at the same time, the strongest induced event in the measurement period was recorded with a magnitude of 2.6 MLv. In the following months, the seismicity rate ranged from 8 to 34 events above 0.5 MLv per month, some of which were felt. A structural geological 3D subsurface model is developed to help to understand the distribution of induced seismicity and the role of the raising mine water level.
How to cite: Rische, M., Fischer, K. D., Allgaier, F., and Friederich, W.: Induced micro seismicity due to raising mine water level in former coal mines in the eastern Ruhr area (Germany), EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-13040, https://doi.org/10.5194/egusphere-egu21-13040, 2021.
Reservoir-triggered seismicity (RTS) is the longest known anthropogenic seismicity type. It has the potential to generate seismic events of M6 and bigger. Previous studies of this phenomenon have proved that major events are triggered on preexisting major discontinuities, forced to slip by stress changes induced by water level fluctuations and/or pore-pressure changes in the rock mass in the vicinity of reservoirs. Song Tranh 2 is an artificial water reservoir located in Central Vietnam. Its main goal is back up the water for hydropower plant. High seismic activity has been observed in this area since the reservoir was first filled in 2011. The relation between water level and seismic activity in the Song Tranh area is complex, and the lack of clear correlation between water level and seismic activity has led to the conclusion that ongoing STR2 seismic activity is an example of the delayed response type of RTS. However, the first phase of the activity observed after impoundment has been deemed a rapid response type. In this work, we proved that the seismicity recorded between 2013 and 2016 manifested seasonal trends related to water level changes during wet and dry seasons. The response of activity and its delay with respect to water level changes suggest that the main triggering factor is pore pressure change due to the significant water level changes observed. A stress orientation difference between low and high water periods is also revealed. The findings indicate that water load and related pore pressure changes influence seismic activity and stress orientation in this area.
This work was partially supported by research project no. 2017/27/B/ST10/01267, funded by the National Science Centre, Poland, under agreement no. UMO-2017/27/B/ST10/01267.
How to cite: Lizurek, G., Leptokaropoulos, K., Wiszniowski, J., Nowaczyńska, I., Van Giang, N., Plesiewicz, B., and Quoc Van, D.: Seasonal trends of reservoir-triggered seismicity in Song Tranh 2 reservoir, Vietnam, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-5495, https://doi.org/10.5194/egusphere-egu21-5495, 2021.
The Song Tranh 2 hydropower construction is located in the Quang Nam province (central Vietnam), it has a reservoir volume of 740 million cubic meters of water and a dam height of 96 m. The reservoir was filled to capacity for the first time in February 2011. The seismicity in the vicinity of reservoir is example of reservoir triggered seismicity(RTS). The natural seismic activity of the Song Tranh 2 reservoir is very low. After the reservoir was filled, the seismic activity increased, and the number and frequency of the tremors also changed as the water level changed. Water level changes are accelerating the tectonic process leading the critically stressed faults to slip. Data suggest that reservoir exploitation stress field changes as triggering origin of this seismicity. The stress inversion method was used to check if there were any seasonal trends. The inverted stress tensor and, in particular, the stress ratio, which is very sensitive to data quality and scope and difficult to accurately retrieve, can be influenced by porous pressure changes. Has been checked, how the average annual seismic activity is related to the change of the water level and if it implies the orientation of the principal stress during high and low water levels in the reservoir. The pore pressure changes and the stress ratio changes were also estimated in relation to the high and low water level periods.
How to cite: Nowaczyńska, I., Lizurek, G., Wiszniowski, J., and Plesiewicz, B.: Seasonal stress inversion trends of RTS in Song Tranh2 reservoir, Vietnam, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-13027, https://doi.org/10.5194/egusphere-egu21-13027, 2021.
It is important to distinguish between natural earthquakes and those induced by CO2 injection at carbon capture and storage sites. For example, the 2004 Mw 6.8 Chuetsu earthquake occurred close to the Nagaoka CO2 storage site during gas injection, but we could not quantify whether the earthquake was due to CO2 injection or not. Here, changes in pore pressure during CO2 injection at the Nagaoka site were simulated and compared with estimated natural seasonal fluctuations in pore pressure due to rainfall and snowmelt, as well as estimated pore pressure increases related to remote earthquakes. Changes in pore pressure due to CO2 injection were clearly distinguished from those due to rainfall and snowmelt. The simulated local increase in pore pressure at the seismogenic fault area was much less than the seasonal fluctuations related to precipitation and increases caused by remote earthquakes, and the lateral extent of pore pressure increase was insufficient to influence seismogenic faults. We also demonstrated that pore pressure changes due to distant earthquakes are capable of triggering slip on seismogenic faults. The approach we developed could be used to distinguish natural from injection-induced earthquakes and will be useful for that purpose at other CO2 sequestration sites.
This research was published in “Sustainability”, https://doi.org/10.3390/su12229723.
Keywords: pore pressure; CO2 injection; induced earthquakes; seasonal earthquakes; remote earthquakes; seismogenic faults.
How to cite: Chhun, C. and Tsuji, T.: Natural and artificial pore pressure variation for distinguishing earthquakes induced by CO2 injection from natural earthquakes, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-1725, https://doi.org/10.5194/egusphere-egu21-1725, 2021.
Modern geophysics highlights that the slip behaviour response of faults is variable in space and time and can result in slow or fast ruptures. Despite geodetical, seismological, experimental and field observations, the origin of this variation of the rupture velocity in nature, as well as the physics behind it, is still debated. Here, we first discuss the scaling relationships existing for the different types of fault slip observed in nature and we highlight how they appear to stem from the same physical mechanism. Second, we reproduce at the scale of the laboratory the complete spectrum of rupture velocities observed in nature. Our results show that when the nucleation length is within the fault length, the rupture velocity can range from a few millimetres to kilometres per second, depending on the available energy at the onset of slip. Our results are analysed in the framework of linear elastic fracture mechanics and highlight that the nature of seismicity is governed mostly by the initial stress level along the faults. Our results reveal that faults presenting similar frictional properties can rupture at both slow and fast rupture velocities. This combined set of field and experimental observations bring a new explanation of the dominance of slow rupture fronts in the shallow part of the crust and in areas presenting large fluid pressure, where initial stresses are expected to remain relatively low during the seismic cycle.
How to cite: Passelegue, F., Almakari, M., Dublanchet, P., Barras, F., Fortin, J., and Violay, M.: Initial effective stress controls the nature of earthquakes , EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-9737, https://doi.org/10.5194/egusphere-egu21-9737, 2021.
Fluid injection causes fault slip that is partitioned in aseismic and seismic moment release. EGS stimulation campaigns have shown that in addition to total fluid volume injected also the rates of injection and fluid pressure increase affect seismic moment release. We investigate the effect of injection rate on slip characteristics, strain partitioning and energy budget in laboratory fluid injection experiments on reservoir sandstone samples in a triaxial deformation apparatus equipped with a 16-channel acoustic emission (AE) recording system. We injected fluid in sawcut samples containing a critically stressed fault at different pressurization rates. In general, fluid-induced fault deformation is dominantly aseismic. We find slow stick-slip events are induced at high fluid pressurization rate while steady fault creep occurs in response to low fluid pressurization rate. The released total seismic moment is found to be related to total injected volume, independent of fault slip behavior. Seismic moment release rate of AE is related to measured fault slip velocity. Total potential energy change and fracture energy release rate are defined by fault stiffness and largely independent of injection rate. Breakdown power density scales with slip rate and is significantly higher for fast injection and pressurization rates. The relation between moment release and injected volume is affected by fault slip behavior, characterized by a linear relation for slip at constant rate and fault creep while a cubic relation is evident for unstable and dynamic slip. Our experimental results allow separating a stable pressure-controlled injection phase with low rate of energy dissipation from a run-away phase, where breakdown power is high and cumulative moment release with injected volume is non-linear.
How to cite: Dresen, G., Wang, L., Kwiatek, G., Rybacki, E., Bonnelye, A., and Bohnhoff, M.: Fluid-induced Seismicity Depends on Injection Rate, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-5275, https://doi.org/10.5194/egusphere-egu21-5275, 2021.
Reactivation of pre-existing faults/fractures in the reservoir due to the deep injection is a key concern in designing and running geothermal and water/CO2 injection projects. Therefore, we investigate potential methods to manage injection-induced seismicity. Recent laboratory and field studies recommend that changes in injection pattern (e.g., cyclic injection) might trigger less seismicity than monotonic injection. This study presents results from uniaxial compressive laboratory experiments performed on high porosity Red Felser sandstone that provide new information about the effect of loading pattern and rate on injection‐driven seismicity. Red Felser sandstone samples with identical porosity and dimensions were subjected to three different loading patterns, including cyclic recursive (CR), cyclic progressive (CP), and monotonic loading. Besides, three different loading rates (displacement control) were applied for each loading pattern: low, medium, and high rates that are 10-4 mm/s, 5×10-4 mm/s, and 5×10-3 mm/s, respectively. Microseismicity analysis shows that (i) the maximum magnitude of seismic events and seismic radiated energy at failure decrease for lower loading rates and during the cyclic loading scenario, (ii) the b-value (magnitude-frequency distribution of events) increases on average 40% for a low-rate cyclic recursive loading in comparison with high-rate cyclic recursive and monotonic loading at different rates. The largest b-value resulted from a low-rate cyclic recursive (LCR) loading pattern. The b-value was estimated and compared using different methods, including a least-square regression on either an incremental frequency distribution or a cumulative frequency distribution, and with the maximum likelihood method (MLM) to provide a reliable b-value estimation. The analyses indicate that by considering the accurate magnitude of completeness, MLM, and, with a least-square regression, the incremental frequency distribution, both result in a reliable b-value. From a mechanical perspective, a low loading rate reduces the sample's final strength by 19%. Moreover, samples subjected to cyclic loading display more complex fracture patterns and more disintegration. In our laboratory study, a combination of low-rate loading and a recursive cyclic loading pattern resulted in reduced seismicity through decreasing the maximum seismicity magnitude and increasing the b-value.
How to cite: Naderloo, M., Barnhoorn, A., Veltmeijer, A., and Jansen, J. D.: Laboratory study on seismicity mitigation: The role of loading pattern and rate, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-13656, https://doi.org/10.5194/egusphere-egu21-13656, 2021.
In 2018-2019, the STIMTEC hydraulic stimulation experiment was conducted at the Reiche Zeche underground laboratory in Freiberg, Saxony/Germany, to investigate the role of hydro-mechanical processes for the often required enhancement of hydraulic properties in deep geothermal projects. We applied the same injection protocol to each of the ten stimulated intervals in the 63 m-long injection borehole. Yet, we observed significant small-scale variability in the seismic and hydraulic responses to stimulation and in parallel stress field heterogeneity on the meter scale. While acoustic emission (AE) activity was high in the upper part of the injection borehole, no AE events were detected in its deepest part, ending in a high-permeability damage zone.
To investigate the stress field and seismic variability throughout the experimental volume and their interrelation further, we started the follow-on experiment STIMTEC-X. The initial phase involved eleven local stress measurements performed in October 2020 in three existing boreholes, previously used for monitoring purposes, with varying orientations and lengths. This phase of the experiment was seismically monitored in real-time using an adaptive, high-resolution seismic monitoring network comprising six AE-type hydrophones, six regular AE sensors and four accelerometers. The hydrophones were installed in combination with hydraulic gauges or the double packer probe used for localized injection to make best use of the existing infrastructure. Hydrophones were optimally placed for each measurement configuration anew with at least one deployed in the direct vicinity (~3-4 m) of the injection interval. We detected low-magnitude AE activity (M<-3.5) at high resolution, spatially distributed between distinct clusters identified previously during the STIMTEC experiment. Overall, these records indicate a doubling of the seismically active volume. We also performed eight dilatometer tests to determine deformation characteristics of induced hydrofracs and pre-existing fractures. A circulation experiment between the injection borehole and two newly drilled boreholes of 23 m and 30 m depth is pending. Here, we present the seismicity associated with the STIMTEC and STIMTEC-X hydraulic stimulation campaigns and focal mechanism solutions. We focus on how they contribute to 3-D volumetric stress field characterisation between local stress measurement points.
Figure 1: Borehole layout (cyan - injection borehole, yellow: seismic monitoring boreholes, green: hydraulic monitoring borehole, red: mine-back validation boreholes) and acoustic emission (AE) events during the STIMTEC (yellow and orange circles) and STIMTEC-X (purple circle) experiments at the Reiche Zeche underground laboratory in Freiberg, Germany. Damage zones (transparent red) and hydraulically stimulated (dark blue rings) and/or hydraulically tested intervals (light blue rings) are shown. Stimulation of the intervals resulted in >11000 AE events with most events occurring during the periodic pumping sequences following the hydrofracturing. The seismic clouds extend about 5 m radially around the boreholes.
How to cite: Boese, C., Renner, J., and Dresen, G. and the STIMTEC-X Team: The STIMTEC-X hydraulic stimulation campaign at the URL Reiche Zeche Mine, Germany, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-11025, https://doi.org/10.5194/egusphere-egu21-11025, 2021.
Anomalies in seismic ambient noise, defined as strong spectral amplification of the vertical components at frequencies of several Hertz, are currently observed on sites located above hydrocarbon reservoirs. If properly understood, these anomalies could have a potential for applications such as geothermal reservoir exploration or underground gas storage monitoring. Under purely elastic modeling, the nature of these anomalies was mainly explained by the geological structure more than the fluid reservoir itself. The main objective of the present work is to explain the exact origin of the anomalies by numerical simulations of the 3D wave propagation using specfem3D code. The simulated spectral anomalies are essentially static and determined by the typical geological reservoir environments. The effect of an anticline structure, which is a common characteristic of hydrocarbon reservoirs, is investigated using different types of sources. Results show that the spectral anomalies caused by the presence of the anticline structure have similarities with the anomalies observed in real data. More work is needed to extract laws linking geometrical characteristics of the anticline to spectral properties. Future works will also include analysis on real gas storage sites, followed by a transposition to the geothermal field applications, for which more complicated parameters appear to participate to the phenomenon.
How to cite: El Khoury, C., Chauris, H., and Kazantsev, A.: Analysis of anomalies in seismic ambient noise above fluid reservoirs, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-8709, https://doi.org/10.5194/egusphere-egu21-8709, 2021.
Introduction. Previous studies (e.g., Harrington and Brodsky, 2006) documented a relative scarcity of remote triggering in Japan, compared to other seismic regions. For example, in California, dynamic triggering is reported to occur at levels of stress as small as 0.1 kPa, while in Japan it was reported that levels of 30 kPa or more are required to trigger detectable events (van der Elst and Brodsky, 2010). However, the threshold dynamic triggering level following the 2016 M7.3 Kumamoto earthquake was of just a few kPa (Enescu et al., 2016). Enescu et al. (2016) proposed that one of the possibilities to explain this observation is a change of stress triggering threshold that may have taken place after the 2011 M9.0 Tohoku-Oki earthquake.
Motivation. Given the above observations, this study investigates 1) the occurrence of dynamically triggered earthquakes in Japan after some large earthquakes from 2004, and 2) whether the threshold of dynamic triggering may have changed due to the 2011 Tohoku-Oki earthquake and why this threshold might have changed.
Analysis and Results. First, we investigated dynamic triggering throughout Japan, following some large earthquakes occurred after 2004. As a result, the threshold appears to decrease following the 2011 Tohoku-Oki earthquake, however the number of earthquakes we have investigated was relatively small, so we could not draw statistically significant conclusions. In the second part of the study, we have focused on a few specific areas within Japan to systematically investigate dynamic triggering, which reduced significantly the computational costs. Thus, we focused on some areas in Tohoku and Hida, where swarm earthquakes occurred after the 2011 Tohoku-Oki earthquake. As a result, the change of the triggering level in an area close to the Yamagata-Fukushima border is considered to be statically significant at a 5% significance level. In other regions, the significance at a 5% level could not be established, however a decrease of this threshold is apparent, except for one region. We speculate that changes in the stress triggering threshold levels might be related to pore pressure changes in the crust following the 2011 Tohoku-Oki earthquake.
How to cite: Enescu, B. and Takeda, Y.: Systematic Investigation of Dynamic Earthquake Triggering in Japan, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-2548, https://doi.org/10.5194/egusphere-egu21-2548, 2021.
Permeability changes induced by earthquakes have been widely studied. The question remains of how multiple large earthquakes influence permeability at different depths in the far-field and permeability changes could possibly be employed for hydraulic characterization of the aquifers has not yet been investigated. We study the change in permeability in fractured aquifers of the North China Paraplatform based on 11 years of groundwater hydrographs of 7 wells and 62 earthquakes. From 2008 to 2018, the permeability changes varied from well to well, all aquifers showed a consistent and distinct magnitude of change in permeability (decrease, increase and no change) following each earthquake. From the perspective of a single well to multiple shocks, the permeability variation of the JN well is the most sensitive to seismic events. From the perspective of multiple wells to one single earthquake, there were no cases of simultaneous permeability changes in all 7 wells induced by a single earthquake. Permeabilities varying within a wide range at a given depth implies that it could be considered as a dynamically self-regulating value, while permeability changes indicate great differences at varying depths. We found that the correlations between permeability changes and seismic energy density or depth are weak, however, the azimuths of seismic waves could greatly impact the changes in permeability, i.e., from 25° to 295°, and the most significant span is 250° to 295°, and fault distribution around the monitoring wells may also contribute to this result. Employing a seismic waves-pressure amplitude model, the mobilization of colloids driven by the oscillation of pressure head as a possible mechanism of permeability change. Distant, large magnitude earthquakes can alter the permeability, also can accelerate or slow down the rate of permeability change of the aquifer material.
How to cite: Gu, H., Xu, Y., Wang, M., Su, Z., Lan, S., Woo, N. C., and Sauter, M.: Spatial permeability variations of aquifers in North China Plain derived from large magnitude earthquake signals, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-8670, https://doi.org/10.5194/egusphere-egu21-8670, 2021.
Injection-induced seismicity has become a central issue in the development of subsurface energy technologies such as enhanced geothermal energy, unconventional hydrocarbon production, wastewater injection, geologic carbon sequestration, or underground gas storage. The effect of the hydraulic properties of faults on the nucleation of earthquakes is a key aspect poorly understood. Our research question is how these properties may alter the onset of slip, the nucleation pattern, the nucleation length, and the time to nucleation.
We simulate earthquakes by means of sophisticated 2-dimensional computational models where earthquakes are triggered by fluid injection. The fault frictional contact is described by the Dieterich–Ruina rate-and-state law. Rock is simulated as a poroelastic solid and we couple fluid flow and rock mechanics.
Our model allows us to explain the impact of longitudinal and transverse fault permeability on the mechanisms that control the evolution of fault strength and shear stress during the nucleation. We find that the nucleation is controlled by the pressure and shear stress profiles along the fault, which in turn are driven by the fault hydraulic properties. Therefore, fault permeability exerts a fundamental control on the scaling of the nucleation length, the nucleation pattern, and the time to nucleation.
Acknowledgements: Project supported by a 2019 Leonardo Grant for Researchers and Cultural Creators, BBVA Foundation. The BBVA Foundation accepts no responsibility for the opinions, statements and contents included in the project and/or the results thereof, which are entirely the responsibility of the authors.
How to cite: Santillán Sanchez, D., Mosquera Feijoo, J. C., and Cueto-Felgueroso Landeira, L.: The impact of fault permeability on the nucleation of injection-induced earthquakes, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-3329, https://doi.org/10.5194/egusphere-egu21-3329, 2021.
Injection-induced seismicity is usually observed as an enlarging cloud of seismic events that grows in a diffusive manner around the injection zone. These observations are commonly interpreted as the triggering of instabilities in pre-existing fractures and faults due to the direct effect of pore pressure increase (Shapiro, 2015), whereas poroelastic stressing is usually associated with the occurrence of seismic events beyond the plausible zone affected by pore pressure diffusion (Segall and Lu, 2015). However, an alternative triggering mechanism based on the elastic transfer of stress due to injection- induced aseismic slip has been recently proposed (Viesca, 2015; Guglielmi et al, 2015). Previous studies have shown that in critically stressed faults, the aseismic rupture front can outpace fluid diffusion (Garagash and Germanovich, 2012; Bhattacharya and Viesca, 2019), and in turn be the primary cause that controls the evolution of seismicity as it has been recently inferred from in-situ experiments of fluid injection (Duboeuf et al., 2017) and recent cases of injection-induced earthquakes (Eyre et al, 2019).
Despite the great relevance of aseismic slip on injection-induced seismicity, the conditions that control the three-dimensional propagation of aseismic ruptures are still poorly constrained. This is in part due to the challenge of solving such a 3D moving boundary problem in which both fault slip and rupture shape are unknown. Here, we study the mechanics of injection-induced aseismic ruptures on a planar fault characterized by a strength equal to the product of a constant friction coefficient and the effective normal stress. We systematically track the temporal evolution of the rupture area relative to the evolution of the pressurized zone and focus on the effect of the initial stress state and injection scenario. For injection at constant flux, we derive a semi-analytical solution for circular ruptures (for a Poisson’s ratio equal to zero), which gives the ratio between the rupture radius and a nominal pore pressure front location, which we named as amplification factor λ. This amplification factor is a function of a unique dimensionless parameter that depends on the initial fault stress criticality and the fluid-induced overpressure. Then, we generalize the semi-analytical solution to the case of non-circular ruptures (for any value of the Poisson’s ratio) by solving numerically for the spatiotemporal evolution of fault slip using a fully implicit boundary-element-based solver with quadratic triangular elements. We show that the rupture front is nearly elliptical and the rupture area Ar evolves in a self-similar diffusive manner such that Ar(t) = 4παλ2t, where α is the fault hydraulic diffusivity and λ is the amplification factor for circular ruptures. The rupture area is greater than the nominal pressurized area if λ > 1. The semi-analytical solution for the rupture area provides a unique opportunity for verifying numerical hydro-mechanical solvers. After, we investigate numerically the case of circular and non-circular ruptures driven by injection at constant pressure instead of constant flux. We show that the self-similar property of the rupture growth is lost under this injection scenario.
How to cite: Sáez, A., Lecampion, B., Bhattacharya, P., and Viesca, R. C.: Three-dimensional aseismic ruptures driven by fluid injection, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-13394, https://doi.org/10.5194/egusphere-egu21-13394, 2021.
Fluid injections at depth can trigger seismic swarms and aseismic deformations. Similarly, some natural sequences of seismicity occur clustered in time and space, without a distinguishable mainshock. They are usually interpreted as driven by fluid and/or aseismic processes. Those seismic swarms, natural or injection-induced, present similarities in their behavior, such as a seismic front migration. The effective stress drop, defined as a ratio between seismic moment and cluster size, is also weak for all swarms, when compared to usual earthquakes values. However, the physical processes that drive both types of swarms, and that can explain such similarities are still poorly understood. Here, we propose a mechanical model in which the fluid primarily induces an aseismic slip, which then triggers and drives seismicity within and on the edges of the active zone. This model is validated using a global and precise dataset of 16 swarms, from natural or induced origins, in different geological contexts. Consequently, our measurements of the migration velocity of the seismicity front, and of the effective stress drop for our swarms can be related to the seismic-to-aseismic moment. Using our model, we are then able to compute an estimate of the volume of fluids circulating during natural earthquake swarms, assuming the total moment is related to the volume of fluids. Our study highlights common characteristics and novel insights into the physical processes at play during seismic swarms.
How to cite: Danre, P., De Barros, L., and Cappa, F.: A common model to explain similarities between injection-induced and natural earthquake swarms, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-1085, https://doi.org/10.5194/egusphere-egu21-1085, 2021.
Anthropogenic injection of fluid into tight fractured reservoirs is known to alter the stress state of the Earth`s crust, inducing micro-seismicity and eventually significant earthquakes. The injection scenario, in terms of injection pressure or injection rate, is one of the key controlling parameters for injection-induced seismicity. Although a number of studies have been carried out on understanding the effects of injection strategy on seismicity rates, less is known about its effect on the nucleation of dynamic slip on a pressurized fault, especially for non-stationary injection protocols.
In this contribution we study the effects of injection rate variation on the transition between aseismic and seismic slip along a frictional weakenig fault. Notably, we parametrize the injection strategy by assuming an initial linear increase of injection rate in time, up to a value after which it remains constant. We perform a scalying analysis and identify the governing parameters that control the fault response. We solve numerically the coupled hydro-mechanical problem using a fast boundary element solver for localized inelastic deformations . Upon benchmarking the numerical results with the semi-analytical ones of Garagash and Germanovich  for the specific case of constant injection rate, we investigate the effect of injection rate variation on critically stressed and marginally pressurized faults. We derive analytical expressions for nucleation time and we confirm them via numerical results. Furthermore, we present a small scale yielding solution for marginallly pressurized faults and investigate the influence of injection scenario on shear crack run-out distances (when occuring).
 Ciardo, F., Lecampion, B., Fayard, F., and Chaillat, S. (2020), A fast boundary element based solver for localized inelastic deformations, Int J Numer Methods Eng. 2020; 1–23.
 Garagash, D., and L. N. Germanovich (2012), Nucleation and arrest of dynamic slip on a pressurized fault, J. Geophys. Res., 117, B10310.
How to cite: Ciardo, F., Rinaldi, A. P., and Wiemer, S.: Effects of varying injection rate on dynamic slip nucleation along a frictional weakening fault, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-8704, https://doi.org/10.5194/egusphere-egu21-8704, 2021.
The increasing rate of induced seismicity in subsurface reservoirs, exceeding occasionally moment magnitude 5, has generated significant attention among earthquake scientists and regulators over the last decade. Fluid injection activity during the operation stage often produces a significant, sometimes even destructive, earthquake. Many approaches have been proposed to monitor, model, and predict the injection-related seismicity to avoid an earthquake larger than a threshold set by the regulator (e.g., Mw 2.0). However, unexpected higher magnitude events occur exceeding what is predicted by empirical models, theoretical relations, or computer simulations.
Current models do not consider that subsurface reservoirs consist of complex fracture networks characterized by connected and unconnected individual fracture planes, often comprising a larger but inactive fault (unfavorably oriented with respect to regional stress). Fluid injection may then perturb stress conditions and trigger an initial rupture on fractures close to the injection well; this initial event may then dynamically trigger other fractures and potentially generate a large earthquake.
We inspect conditions leading to induced earthquakes taking into account the complex fracture network intersected to an inactive fault using dynamic earthquake rupture simulations. We generate the fracture network using a nearest-neighbor method following statistical parameters (power-law distribution of fracture length and fracture density) based on field data. There are 134 fractures consisting of 95 connected fractures, 3 fractures connected with at least one fracture, and 38 unconnected fractures. We focus on two fracture populations oriented in strike N110E ± 10° and N210E ± 10°, respectively. The main fault has a depth-dependent dip orientation which results in a listric fault geometry.
For our dynamic rupture simulations, we use the open-source software SeisSol (https://github.com/SeisSol/SeisSol), apply a laboratory-based rate-and-state with rapid velocity weakening friction law, and assign source radius-dependent characteristic length (L parameter) to the fractures. We vary stress conditions (maximum horizontal orientation, static-pore pressure, and prestress ratio) and conduct an initial static Mohr-Coulomb analysis before running the expensive dynamic rupture simulation. We choose conditions that lead to cascading rupture with (case 1) and without (case 2) the involvement of the main fault. Case 1 has higher artificial overstress within the nucleation area than case 2. Our simulation shows intricate rupture progression over small fractures via rupture branching with the parallel and orthogonal connected fractures. The rupture can also transfer to the unconnected fractures through dynamic triggering from the closest neighboring fracture. Case 1 produces a moment magnitude of Mw 6.36 that is equivalent to case 2. Our preliminary result reveals that connected fractures can generate a significant and potentially large induced earthquake if all fractures are favorable to the stress condition.
How to cite: Palgunadi, K. H., Gabriel, A.-A., Garagash, D., and Mai, P. M.: Cascading earthquakes on a fracture network in a geo-energy reservoir, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-16385, https://doi.org/10.5194/egusphere-egu21-16385, 2021.
As a result of the world-wide interest in carbon storage and geothermal energy production, increased emphasis is nowadays placed on the development of reliable microseismic monitoring techniques for hazard monitoring related to fluid movement and reactivation of faults. In the process of developing and benchmarking these techniques, the incorporation of realistic noise into synthetic datasets is of vital importance to predict their effectiveness once deployed in the real world. Similarly, the recent widespread use of Machine Learning in seismological applications calls for the creation of synthetic seismic datasets that are indistinguishable from the field data to which they will be applied.
Noise generation procedures can be split into two categories: model-based and data-driven. The distributed surface sources approach is the most common method in the first category: however, it is well-known that this fails to capture the complexity of recorded noise (Dean et al., 2015). Pearce and Barley (1977)’s convolutional approach offers a data-driven procedure that has the ability to accurately capture the frequency content of noise however imposes that noise must be stationary. Birnie et al. (2016)’s covariance-based approach removes the stationarity requirement accurately capturing spatio-temporal characterisations of noise, however, like all other data-driven approaches it is constrained to the survey geometry in which the noise data has been collected.
In this work, we propose an extension of the covariance-based noise modelling workflow that aims to generate a noise model over a user-defined geometry. The extended workflow comprises of two steps: the first step is responsible for the characterisation of the recorded noise field and the generation of multiple realisations with the same statistical properties, constrained to the original acquisition geometry. Gaussian Process Regression (GPR) is subsequently applied over each time slice of the noise model transforming the model into the desired geometry.
The workflow is initially validated on synthetically generated noise with a user-defined input covariance matrix. This allows us to prove that the noise statistics (i.e., covariance and variogram) can be kept almost identical between the noise extracted from the synthetic dataset and the various steps of the noise model procedure. The workflow is further applied to the openly available ToC2ME passive dataset from Alberta, Canada consisting of 69 geophones arranged in a pseudo-random pattern. The noise is modelled and transformed into a 56-sensor, gridded array, which is shown to a very close resemblance to the recorded noise field.
Whilst the importance of using realistic noise in synthetic datasets for benchmarking algorithms or training ML solutions cannot be overstated, the ability to transform such noise models into arbitrary receiver geometries opens up a host of new opportunities in the area of survey design. We argue that by coupling the noise generation and monitoring algorithms, the placement of sensors can be optimized based on the expected microseismic signatures as well as the surrounding noise behaviour. This could be of particular interest for geothermal and CO2 storage sites where processing plants are likely to be in close proximity to the permanent monitoring stations.
How to cite: Birnie, C. and Ravasi, M.: On the generation of geometry-independent noise models for microseismic monitoring purposes, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-8524, https://doi.org/10.5194/egusphere-egu21-8524, 2021.