Influence of models on seismic-event localization

Abstract

Source localization is a fundamental problem in seismology. Current localization techniques often rely on homogeneous isotropic models, even if a survey region is known to be geologically complex or anisotropic. We investigated a model's influence on localization, using data from a hydraulic injection experiment at the Continental Deep Drilling ͑Kontinentale Tiefbohrung, or KTB͒ site in Germany. We performed the localization with a grid-search algorithm and two additional methods for verification. From previous work, a homogeneous isotropic model based on a borehole check shot and a heterogeneous 3D isotropic model were available. Vertical seismic profiling ͑VSP͒ and borehole data as well as laboratory data from rock samples provided homogeneous anisotropic models. Although localization with the isotropic and anisotropic models led to the same magnitude of residuals and therefore to a comparable quality of fit, the distribution of the event cloud differed significantly for isotropic and anisotropic media. The isotropic model leads to a southward lateral shift of the event cloud of about 800 m from the injection point; the cloud appears almost centered at the injection point for the anisotropic models. This implies the classical diffusion concept for event generation, whereas isotropic models suggest the existence of a fault system providing a migration path for injected fluid, generating seismicity at an offset from the injection point. These results indicate that meticulous model building is essential for seismic-event localization and subsequent interpretation.

Recent Advances and Challenges of Waveform‐Based Seismic Location Methods at Multiple Scales

“…As opposed to weak elastic anisotropy of the earth at the global scale, the anisotropy of tight reservoirs (e.g., kerogen‐rich shales) at the exploration‐scale may be strong. The high magnitude of observed anisotropy is another challenge that impedes reliable field microseismic application, and the small‐scale natural and hydraulic fractures in these formations make the anisotropy even more complicated (Gajewski et al, ; Grechka, ; Grechka et al, ). Moreover, industrial activities (e.g., hydraulic stimulation) may produce or extend fractures that change the local seismic velocity (Fehler et al, ), which in turn affects seismic location.…”

Inversion of microseismic data for triclinic velocity models

“…Fortunately, its quality typically suffices for microseismic data that can be processed in homogeneous azimuthally anisotropic models (Gajewski et al . ; Grechka et al . ) because angular deviations of the P‐wave polarization vectors from the corresponding ray directions r P , amounting to just a few degrees in such media (Crampin, Stephen and McGonigle ), are smaller than uncertainties in estimating β U even for strong microseismic events (Drew, White and Wolfe ; Eisner, Fischer and Rutledge ).…”

“…The quality of approximation (2) is critical for the accuracy of event locations. Fortunately, its quality typically suffices for microseismic data that can be processed in homogeneous azimuthally anisotropic models (Gajewski et al 2009;) because angular deviations of the P-wave polarization vectors U P,e from the corresponding ray directions r P , amounting to just a few degrees in such media (Crampin, Stephen and McGonigle 1982), are smaller than uncertainties in estimating β U even for strong microseismic events (Drew, White and Wolfe 2008;Eisner, Fischer and Rutledge 2009). When approximation (2) is satisfactory, the number of Inversion of microseismic data for triclinic velocity models 1161 unknown coordinates of ξ e reduces from three to two: ξ e ≡ ζ e,1 cos β e , ζ e,1 sin β e , ζ e,2 ≈ ζ e,1 cos β U , ζ e,1 sin β U , ζ e,2 ,…”

Regional noise source location based on the time delays between station pairs from ambient noise interferometry

“…Since the 1990s, nonlinear methods have been applied to search for the source location in the regular or stochastic model space to minimize the misfit between the theoretical and observed traveltimes, such as the genetic algorithm 5 , and the Monte Carlo technique 6 . Significant improvements have been introduced to enhance the performance of the traveltime-based methods, including the relative location method 7 , double-difference relocation method 2 , 8 , and cluster-based relocation methods 9 , 10 . The traveltime-based location methods generally require phase-picking of the first arrival body wave, which brings high measuring error for the low signal-to-noise ratio seismic waveforms, leading to unreliable location results from insufficient spatial network coverage.…”