No abstract

SUMMARY We investigate the relation between geothermal field production and fracture density and orientation in the Ngatamariki and Rotokawa geothermal fields, located in the Taupo Volcanic Zone, New Zealand using shear wave splitting (SWS). We determine the SWS parameters for 17 702 microseismic events across 38 stations spanning close to 4 yr from 2012 to 2015. We compare the strength of anisotropy to changes in field production and injection. We also compare the orientation of the anisotropy to in situ and regional measurements of maximum horizontal stress orientation. ($S_{\mathrm{ H}_{\mathrm{ max}}}$). Due to the volume of unique events (approximately 160 000), shear wave phases are picked automatically. We carry out automatic SWS measurements using the Multiple Filter Automatic Splitting Technique (MFAST). The SWS measurements are interpreted in the context of stress aligned microcracks. Outside both fields and within Ngatamariki, fast polarizations align with the NE–SW regional orientation of $S_{\mathrm{ H}_{\max}}$. Within Rotokawa a greater complexity is observed, with polarizations tending toward N–S. We observe increases in per cent anisotropy coinciding with the start of production/injection in Ngatamariki and then a later correlated drop in per cent anisotropy and vP/vS ratios in southern Ngatamariki as injection is shifted to the north. This relationship is consistent with pore fluid pressure within the reservoir being affected by local changes in production and injection of geothermal fluids causing cracks to open and close in response.

Fluid injection into the Earth's crust can induce seismic events that cause damage to local infrastructure but also offer valuable insight into seismogenesis. The factors that influence the magnitude, location, and number of induced events remain poorly understood but include injection flow rate and pressure as well as reservoir temperature and permeability. The relationship between injection parameters and injection‐induced seismicity in high‐temperature, high‐permeability reservoirs has not been extensively studied. Here we focus on the Ngatamariki geothermal field in the central Taupō Volcanic Zone, New Zealand, where three stimulation/injection tests have occurred since 2012. We present a catalog of seismicity from 2012 to 2015 created using a matched‐filter detection technique. We analyze the stress state in the reservoir during the injection tests from first motion‐derived focal mechanisms, yielding an average direction of maximum horizontal compressive stress (SHmax) consistent with the regional NE‐SW trend. However, there is significant variation in the direction of maximum compressive stress (σ1), which may reflect geological differences between wells. We use the ratio of injection flow rate to overpressure, referred to as injectivity index, as a proxy for near‐well permeability and compare changes in injectivity index to spatiotemporal characteristics of seismicity accompanying each test. Observed increases in injectivity index are generally poorly correlated with seismicity, suggesting that the locations of microearthquakes are not coincident with the zone of stimulation (i.e., increased permeability). Our findings augment a growing body of work suggesting that aseismic opening or slip, rather than seismic shear, is the active process driving well stimulation in many environments.

<p>In order to investigate the cracks/fractures in the geothermal fields of Rotokawa and Ngatamariki, we measure seismic anisotropy across both fields and interpret the results in the context of stress aligned microcracks. Cracks aligned perpendicular to the direction of maximum horizontal stress close and their fluid is forced into cracks aligned with maximum horizontal stress (SHmax). Seismic anisotropy is the directional dependence of a seismic wave's velocity and provides a measure of crack orientation and density.  To measure seismic anisotropy we conduct shear wave splitting measurements on 52,000 station-earthquake pairs across both Rotokawa and Ngatamariki from earthquakes recorded during 2015. Both fields are the subject of other geophysical and geological studies. Thus they are excellent subjects for studying seismic anisotropy. We cluster our measurements by their station-event path and fit the parameters from these clusters to those from theoretical crack planes. We also apply 2-D tomography to shear wave splitting time delays (𝛿t) and spatial averaging to shear wave splitting fast polarisations (∅). In addition, we compare time delays with P-wave to S-wave velocity ratios (νP / vS).  Local measurements of stress within Rotokawa and regional measures of stress within the Taupo Volcanic Zone provide a comparison for the shear wave splitting measurements. We measure ∅ which agrees with the NE-SW regional direction of SHmax across Ngatamariki and parts of Rotokawa. Within Rotokawa, we observe a rotation of ∅ away from NE-SW toward N-S that agrees with borehole measurements of direction of SHmax of 023° and 030°. Spatial averaging of ∅ reveals mean orientations close to the strike of nearby active faults.  The theoretical crack planes, that fit best to the shear wave splitting measurements, correspond to aligned cracks striking 045° outside of both fields, 035° within Ngatamariki, and 035° through to 0° within Rotokawa.  The average percent anisotropy for the full dataset, approximately 4%, is close to the upper bound for an intact rock. Delay time tomography shows regions of higher delay time per kilometre of path length (s=km) within both fields and possibly associated with the production field fault in Rotokawa.  vP =vS shows a wide range of normally distributed values, from 1.1 through to 2.4 with a mean of 1.6, indicating a mixture of gas filled and saturated cracks. A positive correlation between delay time per kilometre (𝛿tpkm) and νP /νS indicates that the majority of the cracks are saturated.</p>

Catalogs of microseismicity are routinely compiled at geothermal reservoirs and provide valuable insights into reservoir structure and fluid movement. Hypocentral locations are typically used to infer the orientations of structures and constrain the extent of the permeable reservoir. However, frequency-magnitude distributions may contain additional, and underused, information about the distribution of pressure. Here, we present a four-year catalog of seismicity for the Rotokawa geothermal field in the central Taupō Volcanic Zone, New Zealand starting two years after the commissioning of the 140 MWe Nga Awa Purua power station. Using waveform-correlation-based signal detection we double the size of the previous earthquake catalog, refine the location and orientation of two reservoir faults and identify a new structure. We find the rate of seismicity to be insensitive to major changes in injection strategy during the study period, including the injectivity decline and shift of injection away from the dominant injector, RK24. We also map the spatial distribution of the earthquake frequency-magnitude distribution, or b-value, and show that it increases from ∼1.0 to ∼1.5 with increasing depth below the reservoir. As has been proposed at other reservoirs, we infer that these spatial variations reflect the distribution of pressure in the reservoir, where areas of high b-value correspond to areas of high pore-fluid pressure and a broad distribution of activated fractures. This analysis is not routinely conducted by geothermal operators but shows promise for using earthquake b-value as an additional tool for reservoir monitoring and management.

No abstract

Summary In the field of seismic interferometry, cross-correlations are used to extract Green’s function from ambient noise data. By applying a single station variation of the method, using auto-correlations, we are in principle able to retrieve zero-offset reflections in a stratified Earth. These reflections are valuable as they do not require an active seismic source and, being zero-offset, are better constrained in space than passive earthquake based measurements. However, studies that target Moho signals with ambient noise auto-correlations often give ambiguous results with unclear Moho reflections. Using a modified processing scheme and phase-weighted stacking, we determine the Moho P wave reflection time from vertical auto-correlation traces for a test station with a known simple crustal structure (HYB in Hyderabad, India). However, in spite of the simplicity of the structure, the auto-correlation traces show several phases not related to direct reflections. Although we are able to match some of these additional phases in a qualitative way with synthetic modelling, their presence makes it hard to identify the reflection phases without prior knowledge. This prior knowledge can be provided by receiver functions. Receiver functions (arising from mode conversions) are sensitive to the same boundaries as auto-correlations, so should have a high degree of comparability and opportunity for combined analysis but in themselves are not able to independently resolve VP, VS, and Moho depth. Using the timing suggested by the receiver functions as a guide, we observe the Moho S wave reflection on the horizontal auto-correlation of the north component but not on the east component. The timing of the S reflection is consistent with the timing of the PpSs-PsPs receiver function multiple, which also depends only on the S velocity and Moho depth. Finally, we combine P receiver functions and auto-correlations from HYB in a depth-velocity stacking scheme that gives us independent estimates for VP, VS, and Moho depth. These are found to be in good agreement with several studies that also supplement receiver functions to obtain unique crustal parameters. By applying the auto-correlation method to a portion of the EASI transect crossing the Bohemian Massif in central Europe, we find approximate consistency with Moho depths determined from receiver functions and spatial coherence between stations, thereby demonstrating that the method is also applicable for temporary deployments. Although application of the auto-correlation method requires great care in phase identification, it has the potential to resolve both average crustal P and S velocities alongside Moho depth in conjunction with receiver functions.

Abstract. We use seismic waveform data from the AlpArray Seismic Network and three other temporary seismic networks, to perform receiver function (RF) calculations and time-to-depth migration to update the knowledge of the Moho discontinuity beneath the broader European Alps. In particular, we set up a homogeneous processing scheme to compute RFs using the time-domain iterative deconvolution method and apply consistent quality control to yield 112 205 high-quality RFs. We then perform time-to-depth migration in a newly implemented 3D spherical coordinate system using a European-scale reference P and S wave velocity model. This approach, together with the dense data coverage, provide us with a 3D migrated volume, from which we present migrated profiles that reflect the first-order crustal thickness structure. We create a detailed Moho map by manually picking the discontinuity in a set of orthogonal profiles covering the entire area. We make the RF dataset, the software for the entire processing workflow, as well as the Moho map, openly available; these open-access datasets and results will allow other researchers to build on the current study.