Induced Seismicity Monitoring Research Articles

Induced Seismicity Forecasting with Uncertainty Quantification: Application to the Groningen Gas Field

Abstract Reservoir operations for gas extraction, fluid disposal, carbon dioxide storage, or geothermal energy production are capable of inducing seismicity. Modeling tools exist for seismicity forecasting using operational data, but the computational costs and uncertainty quantification (UQ) pose challenges. We address this issue in the context of seismicity induced by gas production from the Groningen gas field using an integrated modeling framework, which combines reservoir modeling, geomechanical modeling, and stress-based earthquake forecasting. The framework is computationally efficient thanks to a 2D finite-element reservoir model, which assumes vertical flow equilibrium, and the use of semianalytical solutions to calculate poroelastic stress changes and predict seismicity rate. The earthquake nucleation model is based on rate-and-state friction and allows for an initial strength excess so that the faults are not assumed initially critically stressed. We estimate uncertainties in the predicted number of earthquakes and magnitudes. To reduce the computational costs, we assume that the stress model is true, but our UQ algorithm is general enough that the uncertainties in reservoir and stress models could be incorporated. We explore how the selection of either a Poisson or a Gaussian likelihood influences the forecast. We also use a synthetic catalog to estimate the improved forecasting performance that would have resulted from a better seismicity detection threshold. Finally, we use tapered and nontapered Gutenberg–Richter distributions to evaluate the most probable maximum magnitude over time and account for uncertainties in its estimation. Although we did not formally account for uncertainties in the stress model, we tested several alternative stress models, and found negligible impact on the predicted temporal evolution of seismicity and forecast uncertainties. Our study shows that the proposed approach yields realistic estimates of the uncertainties of temporal seismicity and is applicable for operational forecasting or induced seismicity monitoring. It can also be used in probabilistic traffic light systems.

Estimation of location errors for local seismic network in an area with intense and weak seismicity

SUMMARYWe have modified a method for estimation of location errors distribution for local seismic networks in area with intense shallow seismicity. We have used the variance-covariance matrices in a regular grid of points. The traveltime residuals variance is calculated as a function of source–receiver distance, while the new idea is to estimate the detection range of seismic stations on the basis of the probability of detection plots. In consequence, the method enables the evaluation of the applied velocity models during the analysis of traveltime residuals, as well as evaluation of the usefulness of individual stations with the use of probability plots. Ultimately, it provides assessment of the station configuration by analysing the distribution of location errors. The data needed for the analysis are a seismic catalogue containing earthquakes locations together with traveltime residuals and the velocity model, as well as stations metadata. The method is tested on using the real data which come from the LUMINEOS local seismic network, the network used for induced seismicity monitoring in the Legnica–Głogow Copper District in Poland.

Monitoring the Microseismicity through a Dense Seismic Array and a Similarity Search Detection Technique: Application to the Seismic Monitoring of Collalto Gas-Storage, North Italy

Seismic monitoring in areas where induced earthquakes could occur is a challenging topic for seismologists due to the generally very low signal to noise ratio. Therefore, the seismological community is devoting several efforts to the development of high-quality networks around the areas where fluid injection and storage and geothermal activities take place, also following the national induced seismicity monitoring guidelines. The use of advanced data mining strategies, such as template matching filters, auto-similarity search, and deep-learning approaches, has recently further fostered such monitoring, enhancing the seismic catalogs and lowering the magnitude of completeness of these areas. In this framework, we carried out an experiment where a small-aperture seismic array was installed within the dense seismic network used for monitoring the gas reservoir of Collalto, in North Italy. The continuous velocimetric data, acquired for 25 days, were analysed through the application of the optimized auto-similarity search technique FAST. The array was conceived as a cost-effective network, aimed at integrating, right above the gas storage site, the permanent high-resolution Collalto Seismic Network. The analysis allowed to detect micro-events down to magnitude Ml = −0.4 within a distance of ~15 km from the array. Our results confirmed that the system based on the array installation and the FAST data analysis might contribute to lowering the magnitude of completeness around the site of about 0.7 units.

A multibranch, multitarget neural network for rapid point-source inversion in a microseismic environment: examples from the Hengill Geothermal Field, Iceland

SUMMARYDespite advanced seismological techniques, automatic source characterization for microseismic earthquakes remains difficult and challenging since current inversion and modelling of high-frequency signals are complex and time consuming. For real-time applications such as induced seismicity monitoring, the application of standard methods is often not fast enough for true complete real-time information on seismic sources. In this paper, we present an alternative approach based on recent advances in deep learning for rapid source-parameter estimation of microseismic earthquakes. The seismic inversion is represented in compact form by two convolutional neural networks, with individual feature extraction, and a fully connected neural network, for feature aggregation, to simultaneously obtain full moment tensor and spatial location of microseismic sources. Specifically, a multibranch neural network algorithm is trained to encapsulate the information about the relationship between seismic waveforms and underlying point-source mechanisms and locations. The learning-based model allows rapid inversion (within a fraction of second) once input data are available. A key advantage of the algorithm is that it can be trained using synthetic seismic data only, so it is directly applicable to scenarios where there are insufficient real data for training. Moreover, we find that the method is robust with respect to perturbations such as observational noise and data incompleteness (missing stations). We apply the new approach on synthesized and example recorded small magnitude (M ≤ 1.6) earthquakes at the Hellisheiði geothermal field in the Hengill area, Iceland. For the examined events, the model achieves excellent performance and shows very good agreement with the inverted solutions determined through standard methodology. In this study, we seek to demonstrate that this approach is viable for microseismicity real-time estimation of source parameters and can be integrated into advanced decision-support tools for controlling induced seismicity.

Well-Log-Based Velocity and Density Models for the Montney Unconventional Resource Play in Northeast British Columbia, Canada, Applicable to Induced Seismicity Monitoring and Research

AbstractWe present detailed velocity and density models for the Montney unconventional resource play in northeast British Columbia, Canada. The new models are specifically essential for robust hypocenter determination in the areas undergoing multistage hydraulic-fracturing operations and for detailed analysis of induced seismicity processes in the region. For the upper 4 km of the sedimentary structure, we review hundreds of well logs and select sonic and density logs from 19 locations to build the representative models. For depths below 4 km, we extend our models using data from the southern Alberta refraction experiment (Clowes et al., 2002). We provide one set of models for the entire Montney play along with two separated sets for the southern and northern areas. Specifically, the models for the southern and northern Montney play are based on logs located in and around the Kiskatinaw Seismic Monitoring and Mitigation Area and the North Peace Ground Motion Monitoring area, respectively. To demonstrate the usefulness of our detailed velocity model, we compare the hypocenter location of earthquakes that occurred within the Montney play as determined with our model and the simple two-layered model (CN01) routinely used by Natural Resources Canada. Locations obtained by our velocity model cluster more tightly with the majority of events having root mean square residual of <0.2 s compared with that of <0.4 s when the CN01 model is used. Cross sections of seismicity versus depth across the area also show significant improvements in the determination of focal depths. Our model results in a reasonable median focal depth of ∼2 km for events in this area, which is consistent with the completion depths of hydraulic-fracturing operations. In comparison, most solutions determined with the CN01 model have fixed focal depths (0 km) due to the lack of depth resolution.

Micro-Seismic Monitoring Data Analysis System Based on Sequentially Discounting Autoregressive and its Application to Offshore CO2 Storage Safety Operation

Micro-seismic monitoring is widely used as an effective way in detecting the seismic events during injection to ensure the safety during the injection of CO2 into geologic formations and is beneficial for increasing the public acceptance. At the offshore site, the challenges in monitoring involve processing high noise record and distinguishing the induced seismic events from the background natural earthquakes. We introduce the Sequentially Discounting AutoRegressive (SDAR) learning method to identify seismic events against the background noise. The SDAR method has a higher detection rate than traditional methods (e.g., STA/LTA), even for low signal to noise (< 1.0) events. We apply the SDAR method to synthetic record and the Tomakomai Ocean Bottom Cable (OBC) baseline data to test the method effectiveness. The detected background natural earthquakes will then be used to construct the seismic model and identify the induced seismic events during the injection. Based on the SDAR method, an Advanced Traffic Light System (ATLS) is being built for induced seismicity monitoring and injection safety management at the Tomakomai storage project. Our results suggest that the SDAR-based ATLS will be a promising tool for safety management, especially for injection projects in offshore storage project.

A review of the current status of induced seismicity monitoring for hydraulic fracturing in unconventional tight oil and gas reservoirs

During the past decades, significant progress has been made in the development of induced seismicity monitoring for related human activities. Hydraulic fracturing and induced seismicity monitoring are operating procedures for safe and effective production of oil and gas from unconventional resources, particularly shales. Hydraulic fracturing can induce seismicity through fluid injection and disturbance of subsurface stress in tight reservoirs. Most seismic events associated with hydraulic fracturing exhibit magnitude of Mw ≤ 3 and are referred to as microseismicity, while a few larger-magnitude earthquakes (e.g. Mw > 3) could also be induced by reactivating pre-existing faults. Here, we review the current status of research concerning induced seismicity monitoring for shale hydraulic fracturing. Induced seismicity contains information relating to important subsurface characteristics, e.g. rock failure potential and seismogenic zones. Microseismic monitoring is essential for reservoir characterization, e.g. fracture geometry delineation and reservoir geomechanical analysis. It is carried out with advanced acquisition, processing, and interpretation techniques, while larger-magnitude earthquakes are mainly exploited for potential geohazard management and mitigation. Challenges and prospects associated with multi-disciplines for future research and applications of induced seismicity monitoring are identified, and it contributes to achieve safe and efficient unconventional (tight) oil and gas resource exploitation.

Five key lessons gained from induced seismicity monitoring in western Canada

In response to induced seismicity observed in western Canada, existing public networks have been densified and a number of private networks have been deployed to closely monitor induced earthquakes associated with oil and gas operations in the region. Over the past three years, we generated an unprecedented volume of seismic data from monitoring induced seismicity for some of the most active operators in western Canada. This rich data set can be used to understand preexisting geologic structures, the activation mechanisms and probabilities, and seismological attributes of the resultant ground motions. Acknowledging that the primary goal of private networks is assisting operators in making operational decisions, these insights can play key roles in improving the accuracy of event magnitudes, ground-motion predictions, and hazard estimates, which successively can be used for developing effective risk management strategies.

Combining network and array waveform coherence for automatic location: examples from induced seismicity monitoring

Combining network and array waveform coherence for automatic location: examples from induced seismicity monitoring

Ambient Seismic Imaging Throughout the Unconventional Field’s Life Cycle

Technology Update Newly developed ambient seismic imaging methods provide valuable information throughout the life cycle of an unconventional field. Methods developed by Global Geophysical Services enable new insights for exploration, development planning, fracture and refracture design, reservoir management, production forecasting, and reserves estimation. Ambient seismic imaging has also been used to monitor conventional reservoirs, waterfloods, CO2 sequestration, mine stability, and water influx into mines. Additional applications include geothermal systems and induced seismicity monitoring. Many assume that the term “ambient seismic” (imaging artificial or natural seismic emissions) is synonymous with “microseismic” (imaging earthquakes with a magnitude less than 0, i.e., microearthquakes or MEQs). However, advances in the understanding of natural and induced seismic emissions and new imaging and analytical methods have differentiated these terms. Microseismic methods typically locate MEQs by use of classical seismological methods: polarization angles and arrival times of P and S waves. Thus, the term “microseismic” implies methods relying only on MEQs, the “dots-in-a-box” long familiar to fracturing engineers and development geologists. In recent years, research has shown that the majority of the seismic energy radiated during a fracture treatment is in seismic phenomena other than MEQs. Other emission sources include longperiod, long-duration (LPLD) activity (Das and Zoback 2013a and 2013b), long-duration signals (LDS) (Sicking et al. 2014), and fluid resonance in the fracture system (Tary and van der Baan 2012). These types of emissions cannot be analyzed with classical methods because they do not have distinct, noticeable first arrivals or recognizable P/S wave separation. Such emissions can persist or fade in and out for minutes to hours, rather than milliseconds, and therefore cannot be classified as distinct events such as MEQs. Furthermore, there may be other seismic phenomena that have yet to be identified and categorized. Seismic emission tomography (SET) is an alternative to locating earthquakes with P and S wave arrivals and polarization angles. It is in wide use for surface monitoring and most recently for downhole monitoring. SET focuses total trace energy by means of ray tracing through the velocity model in short time steps (on the order of 100 ms), thus imaging energy from many types of seismic emissions, including MEQs, LPLD, LDS, fluid resonance, and possibly unknown phenomena. Each computed SET time step produces a voxel-by-voxel depth image of seismic emissions in the same way that a 3D surface reflection seismic survey produces a 3D time image. MEQs located in individual time steps are termed “proxy hypocenters” because they represent brief, high-energy events. However, they are not identified uniquely as MEQs by classical methods.

Induced Seismicity Monitoring of an Underground Salt Cavern Prone to Collapse

Within the framework of a large research project launched to assess the feasibility of microseismic monitoring of growing underground caverns, this specific work focuses on the analysis of the induced seismicity recorded in a salt mine environment. A local seismic network has been installed over an underground salt cavern located in the Lorraine basin (Northeast of France). The microseismic network includes four 3-components and three single component geophones deployed at depths between 30 and 125 m in cemented boreholes drilled in the vicinity of the study area. The underground cavern under monitoring is located within a salt layer at 180 m depth and it presents a rather irregular shape that can be approximated by a cylindrical volume of 50 m height and 180 m diameter. Presently, the cavern is full of saturated brine inducing a significant pressure on its walls (~2.0 MPa) to keep the overburden mechanically stable. Nevertheless some small microseismic events were recorded by the network and analyzed (approximately 2,000 events in 2 years of recording). In October 2005 and April 2007, two controlled pressure transient experiments were carried out in the cavern, in order to analyze the mechanical response of the overburden by tracking the induced microseismicity. The recorded events were mainly grouped in clusters of 3–30 s of signal duration with emergent first arrivals and rather low frequency content (between 20 and 120 Hz). Some of these events have been spatially located by travel-time picking close to the actual cavern and its immediate roof. Preliminary spectral analysis of isolated microearthquakes suggests sources with non-negligible tensile components possibly related to fluid-filled cracks. Rock-debris falling into the cavern from delamination of clay marls in the immediate roof is probably another source of seismic excitation. This was later confirmed when the most important seismic swarms occurred at the site during May 2007, accompanied by the detachment of more than 8 × 104 m3 of marly material on top of the cavern roof. In any case, no clear evidence of classical brittle ruptures in the most competent layers of the overburden has been observed during the analyzed period. Current work is focused on the discrimination of all these possible mechanisms to better understand the damage processes in the cavern overburden and to assess its final collapse hazard.

Induced Seismicity Monitoring and Procedure of Rock-Burst Focus Localization

On the basis of concentration criterion of solid body failure, an approach is proposed for identifying and determining the location of failure foci in the rock mass. The efficiency, stability, and possibility of using the approach in question are considered for predicting the rock bursts.

Seismological Studies of a Fluid Injection in Sedimentary Rocks, East Texas

Induced microseismicity data from a large volume fluid injection into sedimentary rock was analyzed to study the fracture system, fluid pathways, and state of stress in the lower Frio formation in east Texas. Seismicity data are from two arrays of 25 3-component geophone packages sited in two monitoring boreholes. From a total of 2,894 event triggers, a subset of 54 microearthquakes was chosen for their high quality seismograms and clear P and S arrivals. Arrival times were picked with a precision of 0.5 to 1.0 ms, and microearthquakes were located with hypocentral uncertainties estimated as less than 10–20 m. Hypocenters farthest from the injection well define a nearly horizontal tube of seismicity approximately aligned in the direction of the injection well. A simultaneous inversion of arrival times for transverse isotropic velocity structure and hypocenters yielded P-wave anisotropy of –14% and S-wave anisotropy of –2%. Thus, velocities along vertical ray paths are higher than those along horizontal paths, probably because of lithologic differences. Single-event focal mechanisms were determined for 47 events, and many of them are normal fault type. The minimum principal stress derived from the focal mechanisms is nearly horizontal and trends approximately north-south, consistent with the regional stress state. An imaging analysis of the seismograms shows the presence of strong seismic scatterers at positions that correlate with boundaries seen in the hypocenters; both features probably result from a similar set of heterogeneities. This study demonstrates the abundance of information that can be extracted from induced seismicity data and underscores the value of induced seismicity monitoring for studying the fluid and fracture systems created by fluid injections.