We have produced a high-resolution microseismic image of a hydraulic fracture stimulation in the Carthage Cotton Valley gas field of east Texas. Gas is produced from multiple, low-permeability sands within an interbedded sand-shale sequence. We improved the precision of microseismic event locations 4-fold over initial locations by manually repicking the waveforms in a spatial sequence, allowing us to visually correlate waveforms of adjacent sources. The new locations show vertical containment within individual, targeted sands, suggesting little or no hydraulic communication between the discrete perforation intervals simultaneously treated within an 80-m section. Treatment lengths inferred from event locations are about 200 m greater at the shallow perforation intervals than at the deeper intervals. The highest quality locations indicate fracture zone widths as narrow as 6 m. Similarity of adjacent-source waveforms, along with systematic changes of phase amplitude ratios and polarities, indicate fairly uniform focal mechanisms (fracture plane orientation and sense of slip) over the treatment length. Composite focal mechanisms indicate both left-and right-lateral strike-slip faulting along near-vertical fractures that strike subparallel to maximum horizontal stress (S Hmax). The focal mechanisms and event locations are consistent with activation of the reservoir's prevalent natural fractures, fractures that are isolated within individual sands and trend subparallel to expected hydraulic fracture orientation (S Hmax direction). Shear activation of these fractures indicates a stronger correlation of induced seismicity with low-impedance flow paths than is normally found or assumed during injection stimulation.

We analyzed precisely located microearthquake data detected during five hydraulic fracture treatments in the Carthage gas field of east Texas. The treatments were conducted in two adjacent boreholes within interbedded sands and shales of the Upper Cotton Valley formation. The microearthquakes were induced within narrow horizontal bands that correspond to the targeted sandstone layers. Events throughout all the treatments show strike-slip faulting occurring uniformly along vertical fractures trending close to maximum horizontal stress direction. These events are consistent with the reservoir's prevalent natural fractures, known to be isolated within the sands and trending subparallel to the expected hydraulic fracture orientation. When this uniform fracture system was activated exclusively, the detected shear deformation, measured as the moment release per unit volume of fluid injected, was constant, independent of various fluid viscosities and flow rates used. Within the base of the Upper Cotton Valley formation, anomalous event counts and moment release occurred within dense clusters that delineate bends or jogs in the fracture zones. The mechanisms are also strike-slip, but the fault planes are more favorably oriented for failure. The dense clusters show location patterns diverging in time, suggesting the expulsion of fluids from compressive fault jogs. Fluid flow forced by the adjacent slip-induced loading appears to initially extend the treatments, but the loading also tends to lock up and concentrate stress at the jogs, as evident by fewer but larger events populating the structures as treatments progress. As a result, effective drainage lengths from the boreholes may be shorter than would be inferred from the seismicity extending past the jogs. These high-moment asperities are similar to dense patches of seismicity observed along creeping sections of the San Andreas fault, where they have been attributed to localized zones of strength or stress concentration, surrounded by larger regions undergoing stable, aseismic slip. This similarity, plus large moment deficits in terms of volume injected, suggests a large component of aseismic slip is induced by the Cotton Valley treatments.

Imaging of microseismic data is the process by which we use information about the source locations, timing, and mechanisms of the induced seismic events to make inferences about the structure of a petroleum reservoir or the changes that accompany injections into or production from the reservoir. A few key projects were instrumental in the development of downhole microseismic imaging. Most recent microseismic projects involve imaging hydraulic-fracture stimulations, which has grown into a widespread fracture diagnostic technology. This growth in the application of the technology is attributed to the success of imaging the fracture complexity of the Barnett Shale in the Fort Worth basin, Texas, and the commercial value of the information obtained to improvecompletions and ultimately production in the field. The use of commercial imaging in the Barnett is traced back to earlier investigations to prove the technology with the Cotton Valley imaging project and earlier experiments at the M-Site in the Piceance basin, Colorado. Perhaps the earliest example of microseismic imaging using data from downhole recording was a hydraulic fracture monitored in 1974, also in the Piceance basin. However, early work is also documented where investigators focused on identifying microseismic trace characteristics without attempting to locate the microseismic sources. Applications of microseismic reservoir monitoring can be tracked from current steam-injection imaging, deformation associated with reservoir compaction in the Yibal field in Oman and the Ekofisk and Valhall fields in the North Sea, and production-induced activity in Kentucky, U.S.A.

The injection or production of fluids can induce microseismic events in hydrocarbon and geothermal reservoirs. By deploying sensors downhole, data sets have been collected that consist of a few hundred to well over 10,000 induced events. We find that most induced events cluster into well-defined geometrical patterns. In many cases, we must apply highprecision, relative location techniques to observe these patterns. At three sedimentary sites, thin horizontal strands of activity are commonly found within the location patterns. We believe this reflects fracture containment between stratigraphic layers of differing mechanical properties or states of stress. At a massive carbonate and two crystalline sites, combinations of linear and planar features indicate networks of intersecting fractures and allow us to infer positions of aseismic fractures through their influence on the location patterns. In addition, the fine-scale seismicity patterns often evolve systematically with time.At sedimentary sites, migration of seismicity toward the injection point has been observed and may result from slip-induced stress along fractures that initially have little resolved shear. In such cases, triggering events may be critical to generate high levels of seismic activity. At one crystalline site, the early occurrence of linear features that traverse planes of activity indicate permeable zones and possible flow paths within fractures. We hope the continued development of microseismic techniques and refinement of conceptual models will further increase our understanding of fluid behavior and lead to improved resource management in fractured reservoirs.Running Head: Microearthquake Patterns in Reservoirs

We have analyzed seismic anisotropy using shear-wavesplitting measurements made on microseismic events recorded during a hydraulic fracture experiment in a tight gas reservoir in Carthage, east Texas. Microseismic events were recorded on two downhole arrays of three-component sensors, the geometry of which provided good ray coverage for anisotropy analysis. A total of 16,633 seismograms from 888 located events yielded 1545 well-constrained shear-wave-splitting measurements. Manual analysis of splitting from a subset of this data set reveals temporal changes in splitting during fracturing. Inversion using the full data set allows the identification of fracture strike and density, which is observed to vary during fracturing. The recovered fracture strike in the rock mass is parallel to directions of regional borehole breakout, but oblique to the hydraulic fracture corridor as mapped by the microseismic event. We relate this to en-echelon fracturing of preexisting cracks. The magnitude of shear-wave splitting shows a clear temporal increase during each pumping stage, indicating the generation of cracks and fissures in a halo around the fracture corridor, which thus increase the overall permeability of the rock mass. Our results show that shear-wave-splitting analysis can provide a useful tool for monitoring spatial and temporal variations in fracture networks generated by hydraulic stimulation.