Effective anisotropy parameters from microseismic reservoir monitoring

Many formations in sedimentary basins are seismically anisotropic, i.e., they are characterized by seismic velocities varying with the direction of propagation. Standard seismic data processing methods generally consider only isotropic media. However, processing and estimation of subsurface velocities become inaccurate when anisotropic data are treated under the general assumption of isotropy. Ignoring anisotropy can lead to inaccuracies in imaging and significant misties in time-to-depth conversion (e.g. Alkhalifah et al., 1996; Sarkar and Tsvankin, 2006). Consequently it is important to define the model and strength of anisotropy for a study area and use this information in data processing. Anisotropy significantly affects the normal-moveout velocity (NMO) in shale reservoir. A simple methodology for estimating effective anisotropy parameters from long-offset seismic data can be based on the inversion of seismic waves arrival times. Vertical Transverse Isotropic (VTI) media are characterized by nonhyperbolic moveouts, which are more significant in large-offset arrivals for qP- and SH-waves. The deviation of arrival times vs offset depends on the strength in anisotropy and can be used in an inversion procedure to compute the Thomsen parameters for weak VTI media. In this study we obtain effective VTI parameters of a gas shale play from qP- and SH-wave traveltime inversion of microseismic monitoring data recorded at the Earth surface during a hydraulic fracturing experiment. The reservoir is located in North America and seismic data are recorded with a ten arms star-like array of 1C geophones with a high maximum-offset-to-source-depth ratio and a cross-like array of 3C accelerometers. Long offset arrays and small receivers spacing increase the accuracy of the computed anisotropy parameters (Gei et al., 2011). The inversion procedure for qP-waves is based on a truncated Taylor series-type characterization of moveout in transversely isotropic media with a vertical symmetry axis obtained by Alkhalifah and Tsvankin (1995). Traveltimes for qP-waves are computed as a function of offset and the anisotropy parameters δ (Thomsen, 1986) and η (Alkhalifah and Tsvankin, 1995). Similarly synthetic traveltimes for SH-waves can be computed as a function of offset and the anisotropy parameter γ (Thomsen, 1986). The inversion algorithm consists in a nonlinear iterative minimization of the residual traveltimes given by the difference between the computed and experimental traveltimes for both qP- and SH-waves, independently. The methodology and inversion setting was tested by a synthetic dataset obtained with a pseudospectral full-wave modeling code. We inverted picked arrival times from four perforation shots and ten microseismic events. Only qP-waves are generated from perforation shots and we obtain consistent results from the four independent inversions, resulting in approximately δ=0.12 and η=0.27. Furthermore, we invert the synthetic arrival times computed with an isotropic layered model suitable for this reservoir and obtain an effective anisotropy approximately 50% in strength. Thus, we conclude that the effective anisotropy observed in the field data is caused partially by the intrinsic anisotropic properties of the formations from the reservoir up to the surface. Inversions of qP-waves from the 10 microseismic events resulted in average values of δ=0.14 and η=0.22. Consistency between anisotropy parameters of the different events is smaller relatively to the results of perforation shots inversions. This is most probably caused by the lower accuracy in the source depth of microseismic events with respect to the perforations and the reliability of results depends on the correctness of source location (Gei et al., 2011). SH-wave traveltime inversions of the microseismic events results on average γ=0.63, a value outside the range of the weak anisotropy approximation. The SH traveltimes are much worse fitted than qP-wave traveltimes and show significant residuals. The results of the inversion of the 10 microseismic events show high values of γ, but both the Thomsen parameters and the vertical S-wave velocities are quite consistent. It is important to remark that these high values of γ are expression of the effective and not intrinsic anisotropy of the gas shale and overburden.