Microseismic Monitoring and Earthquakes I
Presiding: S Shapiro, Freie Universitaet Berlin; J Kendall, University of Leeds
S34A-01 INVITED 15:35h
Graph Theory for Seismic Analyses
The interpretation of seismic events is often as challenging as their acquisition. To visualize and effectively analyze the seismic datasets, we propose to represent the data on a graph in which the individual events are represented as nodes and the physical relations between the events are represented as edges. For example, one physical relationship could be a measure of the waveform similarity; i.e., doublets. This representation allows us to use the numerous graph theory algorithms already existing to quickly quantify complex relationships. To illustrate, doublets are grouped into multiplets using a simple search for the connected parts of a graph. Depending on which physical relationship will be analyzed, the weighting function can be generalized to represent other attributes, such as distance between hypocenters or relative source parameters. The graphs may also be oriented (e.g., a delay between origin times may determine the orientation of the edges). As an example of an attribute quantifying the relationship between microseismic events, we use the similarity of the waveforms representing doublets. The graphic representation facilitates the interpretation because the datasets often include large numbers of similar events originating on the same fault and producing similar waveforms. The grouping of doublets into multiplets illustrates the mutual interactions among different fault planes and simplifies the complexity of the analysis. For example, if separate multiplets with a high correlation threshold become part of a single multiplet with a lower correlation threshold, we can conclude that these multiplets are part of one large fault system.
S34A-02 INVITED 15:50h
Fluid Induced Microseismicity: A Possible Triggering Mechanism And Analysis For Reservoir Characterization
The attention to the microseismic monitoring during operation of geothermal or hydrocarbon reservoirs has grown considerably over the last several years. The observation of microseismicity occurring during borehole fluid injections or extractions has a large potential in characterizing reservoirs in terms of their hydraulic properties as far as several kilometers from boreholes. An approach for the interpretation of microseismic data was proposed to provide in-situ estimates of hydraulic diffusivity or permeability characterizing a geothermal or hydrocarbon reservoirs on the large spatial scale (of the order of 103 m). We call this approach "Seismicity Based Reservoir Characterization" (SBRC). It considers microseismicity triggered by a pore pressure relaxation process at critically stressed locations in rocks. The approach uses a spatio-temporal analysis of fluid-injection induced microseismicity to reconstruct scalar values as well as the tensor of hydraulic diffusivity and to estimate the tensor of permeability for the seismically active volume. The method provides a possibility to invert for hydraulic diffusivity distributions in fluid-saturated rocks. Estimates of hydraulic diffusivity tensors on large spatial scales as well as imaging of its distributions in space resulting from this concept can be of significant importance for industrial applications and understanding of physical properties of geological structures. Here we introduce the fundamental concepts for the interpretation of microseismic data and show numerical verifications of the method. We propose an approach for the numerical modeling of microseismicity. We focus on the verification of the SBRC inversion algorithms using synthetic data. The results of numerical modeling show that in spite of the apparent simplicity of this approach it reproduces significant features of microseismicity observed in reality. The structure and temporal evolution of microseismicity clouds depends on hydraulic properties of rocks as well as on the statistics and spatial distribution of their criticality. The inversion algorithms have been successfully validated. We consider this as an indication that our description of the main physical features of the triggering phenomenon are adequate. The pore pressure relaxation process seems to be an important mechanism for triggering microearthquakes in fluid-saturated rocks. We also show the application of the SBRC approach to various data sets. Data examples of Hot Dry Rock experiments in crystalline rocks (Fenton Hill and Soultz sous Forets) are presented as well as the analysis of microseismicity obtained during fluid injections in sedimentary environment. We show that it seems to be promising to apply the diffusion-process-based approach even for hydraulic fracturing experiments.
S34A-03 16:05h
Four Dimensional Velocity Structure of the Soultz-sous-Forets Geothermal Reservoir During the 2003 Stimulation
The stimulation of 2003 carried out on the HDR site of Soultz-sous-Forets (Alsace, France) has generated a continuous, small magnitude earthquake activity. During this experiment, a seismological network composed of about 25 stations was installed by the Ecole et Observatoire des Sciences de la Terre (EOST). More than 3500 seismic events have been pointed and located. In order to estimate the variation of the physical properties of the reservoir due to the fluid circulation, we have performed a four dimensional tomography: the 3D velocity structure for different time windows. The whole data set has been apportioned into different temporal sets with a fixed number of events. We use a new, more accurate tomographic algorithm based on a double difference tomography (Zhang & Thurber, 2003). This method gives a good resolution on the reservoir volume and allows a relatively good resolution in the vicinity of this volume. It enlights the velocity structure variation of the reservoir through the stimulation period. Simultaneously, this 3D velocity model is used to relocate the events. The final uncertainty in the location is therefore lowered and ranges from 15 to 30 meters in each direction. In this study, we propose a dicussion about the variation of the velocity in time during the stimulation based on the correlation with the hydraulic parameters and other known geological features, for example the fracture zones near the injection well. This analysis allows us to point out the effect of the water circulation on the reservoir properties. The finality of this study aims to appreciate the mechanism which controls the velocity change: fracturation, overpressure, saturation, etc.
S34A-04 INVITED 16:20h
Triggering of Great Earthquakes
There have been many studies showing that of great earthquakes appear to be triggered by slow stress changes. In particular, stress changes due to strain diffusion from other large events in the region have been shown to be effective. Four centuries of earthquake occurrence data in Northeast Japan allowed a ~36 year lag time of subduction events after on-land earthquakes to be determined (Rydelek and Sacks, 1990). The elastic-viscoelastic model of the crust-lithosphere derived from geodetic observations over about 50 years can explain the time lag. The stress diffusion from the 1940's Nankai trough earthquakes, M ~8, slowly unloaded the normal stress clamping the Nojima fault over a 50 year period, resulting in the 1995 Kobe earthquake, m=6.9 (Pollitz and Sacks, 1997). From the earthquake record spanning about 12 centuries, the 1940's Nankai trough earthquakes were themselves advanced in time, the interval since the previous event of 1854 being clearly the shortest on record. Strain diffusion from the on-land great Nobi earthquake of 1891 explains not only the advance, but also the two year delay between the eastern (Tonankai) and western (Nankaido) events. The failure mechanism was modeled by Rydelek and Sacks, 2003. In the above examples, the strain diffusion has created stress changes of a fraction to a few bars at the fault so as to increase the Coulomb failure and modify the occurrence time by a significant amount, i.e. many years on a fault with recurrence interval of more than a century. Continuous GPS observations enable insight into much lower stress triggering of great events. It has long been recognized that most Nankai trough events (since 684) occur in the winter. A seasonal shortening of the continental plate (Heki, 2004) overlying the subducting Philippine sea plate, causes a reduction of stress on the thrust fault. Even though this stress change is less than 0.1 bar, and the yearly stress loading of the fault may be 0.5 to 1 bar, it seems to be sufficient to influence the failure time.
S34A-05 16:35h
Case Study of Injection and Induced Seismicity at Paradox Valley, Colorado
The US Bureau of Reclamation's Paradox Valley Unit (PVU) extracts aquifer brine from 9 shallow wells along the Dolores River, Paradox Valley, southwestern Colorado and, after treating, high-pressure injects the brine 4.3 to 4.8 km below the surface. PVU injects at rates between ~800 and ~1,300 l/min. Since 1991, PVU has emplaced over 4.4x106 m3 of fluid and induced more than 4,000 surface-recorded seismic events. The events are recorded on the local, 15-station, Paradox Valley Seismic Network. The induced seismicity at Paradox separates into two distinct, source zones: a principle zone (>95% of the events) asymmetrically surrounding the injection well to a maximum radial distance of ~3+ km and a secondary, ellipsoidal zone, ~2.5 km long and centered ~8 km northwest of the injection well. The expansion of these zones has stabilized since mid-1999, ~3 years after the onset of continuous injection. Within the principal zone, hypocenters align in distinct, linear patterns showing at-depth stratigraphy and the local Wray Mesa fracture and fault system. The primary faults of the Wray Mesa system are aseismic, striking sub-parallel to the inferred maximum principal stress direction, with one or more faults, probably, acting as fluid conduits to the secondary seismic zone. Individual seismic events, in both zones, do not discernibly correlate with short-term injection parameters; however, a 0.5 km2 region immediately north west of the injection well responds to long-term, large-scale changes in injection rate and the surpassing of a threshold injection pressure. Focal mechanisms of the induced events are consistent with simple double-couple, strike-slip moments and sub-horizontal extension to the northeast. In addition, the fault planes are consistent with principal stress directions determined from borehole breakouts. More than 99.9% of the PVU seismicity is below human detection (~M 2.5). However, approximately fifteen events have been felt locally with the largest being a magnitude M 4.3. Because of the M 4.3 and earlier, felt M 3.5 and M 3.6 events and injection economics, PVU changed injection strategies 3 times, since 1996. These changes reduced seismicity from ~1,100 events/year to as low as ~60 events/year. Finally, it is important to recognize that PVU is not a research operation, but an economically-governed project to remove and dispose brine. Hence, the project has two goals: (1) maximize brine disposal while minimizing objectionable (i.e., felt) seismicity and (2) characterize the reservoir (e.g., lifetime, injectability, etc.) within the parameters dictated by (1). Analyzing and interpreting the induced seismicity helps meet these goals.
S34A-06 INVITED 16:50h
Microseismic Monitoring and Seismic Anisotropy in Oilfields
Passive seismic monitoring in oil fields provides insights into spatial and temporal variations in the stress field. Microseismic activity can be induced by production, injection and regional tectonic processes. Such earthquake activity can delineate faults, identify reservoir compartmentalization, and monitor the progress of injection fronts. Microseismic data can be also used to estimate seismic anisotropy in hydrocarbon settings. Anisotropy is an indicator of order within a medium (e.g., aligned cracks, crystals or layering) and as such offers information about the stress field and dynamic processes within the reservoir. Here we present evidence of seismic anisotropy based on observations of shear-wave splitting in two quite different oilfields, one in the North Sea and one in Oman. In each experiment data were acquired by downhole arrays of three-component sensors. The datasets are very large as 10-20 events are normally recorded each day and the experiments can last many weeks (North Sea) to many months (Oman). We have therefore developed automated techniques to analyze shear-wave splitting in large datasets. The magnitude of the anisotropy varies considerably throughout the reservoirs, ranging from near isotropy to almost 10% anisotropy. Variations in anisotropy magnitude are often sharp in transition, being controlled by faulting or lithology. In both fields we find the dominant orientation of the fast shear-wave parallels known fracture trends and the maximum compressive stress directions. More detailed investigation of spatial variations in the anisotropy reveal that it is also controlled by the intrinsic crystalline anisotropy of the rock. Collectively, our results suggest that the anisotropy is controlled by both rock-type and fault-related fracturing. We also consider the frequency-dependent nature of shear-wave splitting in an effort to better constrain the length-scales of fractures or microcracks responsible for the anisotropy. We estimate the fractures in the competent carbonate reservoir rock of the Oman field to be on length scales of meters in size, whilst in the less competent siltstones on the North Sea field, we estimate the cracks to be cm or smaller in length scale. These results are consistent with geological information for these reservoirs and suggest that passive seismic monitoring can be used to study the fracture character of reservoirs.