Volker Oye

Automated microearthquake location using envelope stacking and robust global optimization

Most earthquake location methods require phase identification and arrival-time measurements. These methods are generally fast and efficient but not always applicable to microearthquake data with low signal-to-noise ratios because the phase identification might be very difficult. The migration-based source location methods, which do not require an explicit phase identification, are often more suitable for such noisy data. Whereas some existing migration-based methods are computationally intensive, others are limited to a certain type of data or make use of only a particular phase of the signal. We have developed a migration-based source location method especially applicable to data with relatively low signal-to-noise ratios. We projected seismograms onto the ray coordinate system for each potential source-receiver configuration and subsequently computed their envelopes. The envelopes were time shifted according to synthetic P- and S-wavearrival times (computed using an eikonal solver) and stacked for a pre...

Automated seismic event location for hydrocarbon reservoirs

An automatic monitoring system has been developed to process continuously recorded microseismic data and locate the associated events. To this end we use P- and S-wave travel times and the direction of the incoming wave field. The processing is organised in four modules: (i) a multi-channel detection algorithm based on signal-to-noise ratios, (ii) a P-wave onset determination based on error prediction filtering with an auto-regressive model, (iii) a P-wave polarisation analysis providing the direction of the incoming wave field, and (iv) a rotation of seismic traces into the ray coordinate system with subsequent S-wave onset determination. For a homogeneous velocity model the event hypocenter is determined by a linearised inversion technique, and for a three-dimensional (3D) velocity model a directed grid search method or the neighbourhood algorithm is applied. We applied the monitoring system to a microseismic data set from the Ekofisk oil field in the North Sea and located about 2000 microseismic events close to the borehole receiver string. Most of the microseismic events occurred in clusters and in a depth range from 2800 to about 3050 m depth. Location errors were estimated by bootstrapping, and a comparison between results from directed grid search and neighbourhood algorithm revealed a high level of consistency.

Orientation of Three-Component Geophones in the San Andreas Fault Observatory at Depth Pilot Hole, Parkfield, California

To identify and constrain the target zone for the planned safod Main Hole through the San Andreas Fault (saf) near Parkfield, California, a 32-level three-component (3C) geophone string was installed in the Pilot Hole (ph) to monitor and improve the locations of nearby earthquakes. The orientation of the 3C geophones is essential for this purpose, because ray directions from sources may be determined directly from the 3D particle motion for both P and S waves. Due to the complex local velocity structure, rays traced from explosions and earthquakes to the ph show strong ray bending. Observed azimuths are obtained from P -wave polarization analysis, and ray tracing provides theoretical estimates of the incoming wave field. The differences between the theoretical and the observed angles define the calibration azimuths. To investigate the process of orientation with respect to the assumed velocity model, we compare calibration azimuths derived from both a homogeneous and 3D velocity model. Uncertainties in the relative orientation between the geophone levels were also estimated for a cluster of 36 earthquakes that was not used in the orientation process. The comparison between the homogeneous and the 3D velocity model shows that there are only minor changes in these relative orientations. In contrast, the absolute orientations, with respect to global North, were significantly improved by application of the 3D model. The average data residual decreased from 13° to 7°, supporting the importance of an accurate velocity model. We explain the remaining residuals by methodological uncertainties and noise and with errors in the velocity model.

Simultaneous Microearthquake Location And Moment-tensor Estimation Using Time-reversal Imaging

We use time-reversal imaging for microearthquake location and investigate the possibility of a simultaneous qualitative moment-tensor estimation. We cross-correlate the data with the synthetic strain Green’s tensor and stack individually for each moment-tensor component. The objective function for the source location is then formulated as the squared sum of those stacked components. The maximum value of the objective function corresponds to the estimated source location and origin time. Similarly, the corresponding stacked components at the estimated source location give the entire time history of the qualitative estimation of the moment tensor. We apply the method to synthetic data of various types of moment-tensor sources computed for a complex and heterogeneous model, namely the Pyhasalmi ore mine in Finland. We also test the method with the same data adding white noise up to 40% of the absolute maximum. Although the method is computationally intensive, it is fully automatic and can easily be adapted to parallel processing. The preliminary results show that the method is robust and reliable.

Travel times and waveforms of microseismic data in heterogeneous media

We compute Pand S-wave first arrival times for a heterogeneous model of the Pyhasalmi ore mine using a finite-difference Eikonal code (Podvin and Lecomte, 1991). For the very same model we are computing the complete wavefield using a viscoelastic finite difference scheme (Larsen and Schultz, 1995). We compare the synthetic results amongst each other and with real microseismic data recorded with an in-mine seismic network. The resulting first arrival times generally agree with the onset of the synthetic wavefield, whereas the amplitude distribution is strongly affected by the heterogeneities in the model. Such amplitude variations at the different in-mine geophones as well as Pand S-wave coda lengths generally agree with the observed microseismic data.

Characterization of induced seismicity patterns derived from internal structure in event clusters

Seismicity induced by CO2 injection at Decatur, Illinois, occurs in distinct clusters and shows no obvious correlation with the proceeding pressure front. We analyse some of these clusters in more depth by using a waveform cross-correlation approach. With this approach we can associate about 1400 events from two clusters, with moment magnitudes between 1.1 and -1.7, with specific formations of much smaller vertical dimensions (10th of meters) than the depth resolution of traveltime-based event locations. The differentiation of reservoir and basement events, and the definition of sub-clusters by waveform correlation, rather than by location, helps to better analyse the spatio-temporal evolution of the events within a cluster. In the Decatur case, this is characterized by event migration from the reservoir into the adjacent basement. The spatial variation of Brune stress drop and Gutenberg b-value exhibit signs of a fluid-driven triggering mechanism at the cluster level, revealing a punctual hydraulic connection between reservoir and basement, most likely associated with basement faults cutting into the reservoir. The observed clustering of seismicity can thus be explained by the lateral heterogeneity of permeability and crustal strength, and is overall consistent with a pressure-induced triggering mechanism. Hence, proper long-term risk mitigation for large-scale fluid injection close to the basement requires prior mapping of small sub-seismic basement-connected faults.

Source parameters determined from microearthquakes in an underground ore mine

To bridge the gap between source parameters estimated from tectonic earthquakes and from laboratory experiments, we investigate source-scaling relations for 42 microearthquakes (M W -1.3 to 0.1) from the Pyhasalmi ore mine in Finland. Source parameters such as seismic moment (M 0 ), radiated seismic energy (E R ), corner frequency (f c ) and stress drop are estimated by fitting a Brune-shaped spectrum to the observed source spectra. We use both, a constant Q model and a frequency dependent Q model determined by the multiple empirical Greens function (MEGF) method, which eliminates local site and path effects. Assuming constant Q for the spectral fit, the energy-to-moment ratio e seems to decrease slightly with decreasing moment, while for frequency dependent Q the ratio e seems to be constant. Both methods support a modified M 0 ∼f c -(3+e) (e between 0.5 and 1) scaling relation that implies a slight decrease of e with seismic moment. Within this studys small range of seismic moments it is difficult to estimate a clear trend in e, however, the absolute values estimated for e as well as the estimated static stress drops are roughly a factor of 10 to 100 smaller than the ones reported for larger earthquakes from other studies. We estimated source mechanisms from P-wave polarities and found indications for dilatational source character and spalling, verified by inmine inspections. To investigate the impact of the mines structural complexity on the seismic waveforms, we carried out 3D finite-difference computations with an explosion source in the mine model and illustrate the effects of near-source inhomogeneities.

The Importance of Coupling Passive and Active Seismic Methods in Geothermal Fields - A Case Study at Paralana, Australia

The Paralana Enhanced Geothermal System is situated in South Australia and is a promising project aiming to exploit natural heating from the radiogenic basement. A 5-day stimulation of the well, in July 2011, induced more than 7000 microearthquakes. Seismic event waveforms are complex, exhibiting a high amplitude seismic phase arrival between direct-P and direct-S waves, a phase which complicates event analyses and needs to be explained. In such projects, site characterization should use all available data, a key issue being a proper knowledge of both P- and S-velocity models. In the present case, a combination of active and passive seismic data, acquired before and/or during hydraulic fracturing, was paramount to the definition of a proper model. Using existing 2D seismic profile interpretations and a seismic log, and applying ray tracing to constrain interval velocities, a 3D velocity model was built in order to improve the event locations. Furthermore, to define the origin of the unknown phase, 3D ray-tracing modeling was systematically applied to test different scenarios. Actual results show that in term of arrival time, these phases most likely correspond to converted waves on interfaces between 1.6- and 2.5-km depth rather than on nearby steep faults.

A new processing package for microseismic monitoring of hydrocarbon reservoirs

Summary We are developing an automatic processing system for the analysis of induced microseismicity. In the first stage of our ongoing project we built up the basic signal processing modules and focused on the rapid and accurate location of microseismic events related to the production and stimulation of hydrocarbon reservoirs. The present system is adapted (but not restricted) to seismic data recorded with three-component geophones deployed in a VSP-configuration in a borehole. Two localisation modules have been developed and are implemented, a linear inversion and a directed grid search algorithm allowing for complex 3D velocity structures. One example using hydrofracturing data and another one implying evaluation of a complete passive monitoring data set from the Ekofisk field are documented here.

Integrating Active with Passive Seismic Data to Best Constrain CO2 Injection Monitoring

The Illinois Basin - Decatur Project (IBDP) is to date one of the largest CO2 sequestration projects in the United States. So far, 1 Mio tonnes of CO2 have been injected over 3 years into the Mt. Simon sandstone formation at about 2 km depth. A suite of various active and passive seismic monitoring techniques have been applied at the site, providing a rich monitoring dataset. Time-lapse 3D surface seismic and VSP measurements were carried out to delineate the progression of the CO2 front. In addition, passive seismic monitoring revealed over 10’000 microseismic events. As a novel method, we attempt to combine the active and passive seismic data for seismic tomographic inversion for the 4D velocity- and attenuation structure in the reservoir. The combined aperture and higher resolution focuses on the reservoir and may allow a more precise mapping of the injected fluid over time. To investigate 4D changes of velocities and attenuation a similar source and receiver distribution is required. This is a particular challenge for microseismic events. High microseismic event location accuracy is essential, which we intend to improve by near surface material characterization, both from downhole petrophysical logging and seismic velocity logging within newly drilled shallow wells.