Grzegorz Kwiatek

Spatiotemporal changes, faulting regimes, and source parameters of induced seismicity: A case study from The Geysers geothermal field

The spatiotemporal, kinematic, and source characteristics of induced seismicity occurring at different fluid injection rates are investigated to determine the predominant physical mechanisms responsible for induced seismicity at the northwestern part of The Geysers geothermal field, California. We analyze a relocated hypocenter catalog from a seismicity cluster where significant variations of the stress tensor orientation were previously observed to correlate with injection rates. We find that these stress tensor orientation changes may be related to increased pore pressure and the corresponding changes in poroelastic stresses at reservoir depth. Seismic events during peak injections tend to occur at greater distances from the injection well, preferentially trending parallel to the maximum horizontal stress direction. In contrast, at lower injection rates the seismicity tends to align in a different direction which suggests the presence of a local fault. During peak injection intervals, the relative contribution of strike-slip faulting mechanisms increases. Furthermore, increases in fluid injection rates also coincide with a decrease in b values. Our observations suggest that regardless of the injection stage, most of the induced seismicity results from thermal fracturing of the reservoir rock. However, during peak injection intervals, the increase in pore pressure may likewise be responsible for the induced seismicity. By estimating the thermal and hydraulic diffusivities of the reservoir, we confirm that the characteristic diffusion length for pore pressure is much greater than the corresponding length scale for temperature and also more consistent with the spatial extent of seismicity observed during different injection rates.

Effects of long‐term fluid injection on induced seismicity parameters and maximum magnitude in northwestern part of The Geysers geothermal field

The long-term temporal and spatial changes in statistical, source, and stress characteristics of one cluster of induced seismicity recorded at The Geysers geothermal field (U.S.) are analyzed in relation to the field operations, fluid migration, and constraints on the maximum likely magnitude. Two injection wells, Prati-9 and Prati-29, located in the northwestern part of the field and their associated seismicity composed of 1776 events recorded throughout a 7 year period were analyzed. The seismicity catalog was relocated, and the source characteristics including focal mechanisms and static source parameters were refined using first-motion polarity, spectral fitting, and mesh spectral ratio analysis techniques. The source characteristics together with statistical parameters (b value) and cluster dynamics were used to investigate and understand the details of fluid migration scheme in the vicinity of injection wells. The observed temporal, spatial, and source characteristics were clearly attributed to fluid injection and fluid migration toward greater depths, involving increasing pore pressure in the reservoir. The seasonal changes of injection rates were found to directly impact the shape and spatial extent of the seismic cloud. A tendency of larger seismic events to occur closer to injection wells and a correlation between the spatial extent of the seismic cloud and source sizes of the largest events was observed suggesting geometrical constraints on the maximum likely magnitude and its correlation to the average injection rate and volume of fluids present in the reservoir.

Twenty Thousand Aftershocks of a Very Small (M 2) Earthquake and Their Relation to the Mainshock Rupture and Geological Structures

We have determined the locations of more than 20,000 aftershocks (as small as moment magnitude Mw 4:4 or even smaller) following an M 2 event in a South African gold mine, using manually picked arrival times. Spatial clustering into fivegroupswasclearlydiscerned.Amajorityoftheaftershocksformedaplanarcluster (∼4 m in apparent thickness, ∼100 × 80 m in areal extent). This cluster is thought to delineate the rupture area of the mainshock because its orientation and spatial extent wereconsistent with thenodalplaneofthecentroid moment tensor (CMT)solution and withthecornerfrequencyofthemainshock,respectively.Theclustersattitudesuggests that the mainshock was a Mohr-Coulomb failure (or formation of a shear rupture sur- faceinintactrockatananglethatobeystheCoulombfailurecriterion)thattookplacein a vertical compression stress field that is indicated by borehole breakout patterns. The aftershock distribution also shows that the mainshock rupture was largely confined to the interior of a 25-m-thick vertical dike, although there are indications of interactions taking place between the rupture and the dikes material boundary with the host rock.

Volumetric components in the earthquake source related to fluid injection and stress state

We investigate source processes of fluid-induced seismicity from The Geysers geothermal reservoir in California to determine their relation with hydraulic operations and improve the corresponding seismic hazard estimates. Analysis of 869 well-constrained full moment tensors (Mw 0.8–3.5) reveals significant non-double-couple components (>25%) for about 65% of the events. Volumetric deformation is governed by cumulative injection rates with larger non-double-couple components observed near the wells and during high injection periods. Source mechanisms are magnitude dependent and vary significantly between faulting regimes. Normal faulting events (Mw   2.5. Our results imply that source processes and magnitudes of fluid-induced seismic events are strongly affected by the hydraulic operations, the reservoir stress state, and the faulting regime.

HybridMT: A MATLAB/Shell Environment Package for Seismic Moment Tensor Inversion and Refinement

We present the software package hybridMT which allows performing seismic moment tensor inversion and refinement, optimized for earthquake data recorded by regional‐to‐local seismic networks as well as for acoustic emission activity. The provided software package is designed predominantly for use in MATLAB (see [Data and Resources][1])/shell environments. The algorithm uses first P ‐wave amplitudes to invert for unconstrained full, deviatoric, and double‐couple constrained moment tensors. Uncertainty assessment is performed by bootstrap resampling. The moment tensor inversion may be performed directly in the shell environment (by a dedicated command‐line tool) or conveniently through the MATLAB interface (m‐functions). In addition to moment tensor inversion, we also provide the MATLAB implementation of the hybrid moment tensor technique. This methodology increases the quality of calculated seismic moment tensors from events forming a spatial cluster by assessing and correcting for poorly known path and site effects. We tested hybridMT on synthetic datasets, acoustic emission data recorded during laboratory rock deformation experiments, and induced seismicity data from a geothermal reservoir. The package is supplemented with extensive documentation, tutorials, and a dedicated website. HybridMT is freely available and distributed under General Public License. [1]: #sec-10

Theoretical limits on detection and analysis of small earthquakes

We investigate theoretical limits on detection and reliable estimates of source characteristics of small earthquakes using synthetic seismograms for shear/tensile dislocations on kinematic circular ruptures and observed seismic noise and properties of several acquisition systems (instrument response, sampling rate). Simulated source time functions for shear/tensile dislocation events with different magnitudes, static stress drops, and rupture velocities provide estimates for the amplitude and frequency content of P and S phases at various observation angles. The source time functions are convolved with a Greens function for a homogenous solid assuming given P, S wave velocities and attenuation coefficients and a given instrument response. The synthetic waveforms are superposed with average levels of the observed ambient seismic noise up to 1 kHz. The combined seismograms are used to calculate signal-to-noise ratios and expected frequency content of P and S phases at various locations. The synthetic simulations of signal-to-noise ratio reproduce observed ratios extracted from several well-recorded data sets. The results provide guidelines on detection of small events in various geological environments, along with information relevant to reliable analyses of earthquake source properties.

Seismic moment tensor and b value variations over successive seismic cycles in laboratory stick‐slip experiments

The formation of fault damage due to slip under high normal stresses can rarely be monitored under in situ conditions. To advance our understanding of microfracture processes, we investigated stick-slip events on Westerly granite samples containing the following: (1) a planar saw cut fault and (2) a fault developed from a fresh fracture surface. We examined temporal changes of seismic moment tensors and b values of acoustic emission (AE) events. During experiment on the saw cut surface, small AEs exhibiting non-double-couple components were observed continuously and strong AEs displaying double-couple components were visible only when approaching the slip onsets. Sliding on naturally fractured surfaces showed, in addition to double-couple components, significant volumetric contributions, especially during the interslip periods and immediately after stick-slip events indicating substantial shear-enhanced compaction within a relatively broad damage zone. The obtained results shed light on how differences in fault structure control the kinematics of microseismicity during different periods of the seismic cycle.

A refined methodology for stress inversions of earthquake focal mechanisms

We develop an improved methodology for reliable high-resolution inversions of focal mechanisms to background stress field orientation and stress ratio R in two or three dimensions. The earthquake catalog is declustered to remove events likely affected strongly by local stress interactions rather than reflecting the large-scale background stress field. The declustered data are discretized with the k-means algorithm into groups containing a number of focal mechanisms between a minimum number Nmin and 2Nmin. Synthetic tests indicate that Nmin ≈ 30 provides stable inversion results under different stress regimes and noise conditions when R ≈ 0.5, while Nmin ≈ 45 is needed for R near 0 or 1. Additional synthetic tests compare the performance of selecting the fault plane of each focal mechanism using (a) the plane with lowest misfit angle between the slip vector from the focal mechanism and shear traction from the stress tensor and (b) the plane with highest instability coefficient representing proximity to the optimally oriented fault for given stress field and friction coefficient. The instability criterion is found to provide more accurate inversion results under all tested stress regimes, stress ratios, and noise conditions. The refined inversion methodology combines selecting fault planes using the instability criterion iteratively with a damped simultaneous inversion of different focal mechanism groups. Results characterizing neighborhoods of discretized domains merged during the damped inversion provide high-resolution information independent of the discretization. Some aspects of the methodology are illustrated with focal mechanism data from the San Jacinto Fault Zone in Southern California.

Self-similar rupture implied by scaling properties of volcanic earthquakes occurring during the 2004-2008 eruption of Mount St. Helens, Washington

We analyze a group of 6073 low-frequency earthquakes recorded during a week-long temporary deployment of broadband seismometers at distances of less than 3 km from the crater at Mount St. Helens in September of 2006. We estimate the seismic moment (M0) and spectral corner frequency (f0) using a spectral ratio approach for events with a high signal-to-noise (SNR) ratio that have a cross-correlation coefficient of 0.8 or greater with at least five other events. A cluster analysis of cross-correlation values indicates that the group of 421 events meeting the SNR and cross-correlation criteria forms eight event families that exhibit largely self-similar scaling. We estimate the M0 and f0 values of the 421 events and calculate their static stress drop and scaled energy (ER/M0) values. The estimated values suggest self-similar scaling within families, as well as between five of eight families (i.e., M0∝f0−3 and ER/M0∝ constant). We speculate that differences in scaled energy values for the two families with variable scaling may result from a lack of resolution in the velocity model. The observation of self-similar scaling is the first of its kind for such a large group of low-frequency volcanic tectonic events occurring during a single active dome extrusion eruption.

Impact of fluid injection on fracture reactivation at The Geysers geothermal field

We analyze the spatiotemporal distribution of fault geometries from seismicity induced by fluid injection at The Geysers geothermal field. The consistency of these faults with the local stress field is investigated using (1) the fault instability coefficient I comparing the orientation of a fault with the optimal orientation for failure in the assumed stress field and (2) the misfit angle β between slip vectors observed from focal mechanisms and predicted from stress tensor. A statistical approach is applied to calculate the most likely fault instabilities considering the uncertainties from focal mechanisms and stress inversion. We find that faults activated by fluid injection may display a broad range in orientations. About 72% of the analyzed seismicity occurs on faults with favorable orientation for failure with respect to the stress field. However, a number of events are observed either to occur on severely misoriented faults or to slip in a different orientation than predicted from stress field. These events mostly occur during periods of high injection rates and are located in proximity to the injection wells. From the stress inversion, the friction coefficient providing the largest overall instability is μ = 0.5. About 91% of the events are activated with an estimated excess pore pressure <10 MPa, in agreement with previous models considering the combined effect of thermal and poroelastic stress changes from fluid injection. Furthermore, high seismic activity and largest magnitudes occur on favorably oriented faults with large instability coefficients and low slip misfit angles.