Statistically Optimal Technique of Simultaneous Event Location and Focal Mechanism Determination of Weak Microseismicity Using Surface Arrays

Ilya Dricker*, Paul Friberg, Earth Imaging Inc., Alexey Epiphansky, Alexander Kushnir, Mikhail Rozhkov, Alexander Varypaev, SYNAPSE Science Center


In this paper we show that complex radiation patterns of microseismic events appear to cause false detections and degrade event location accuracy of conventional surface array microseismic processing techniques. A computational procedure for microseismic monitoring with 1 and 3-C surface arrays is proposed. The procedure is based on maximum likelihood statistic approach considering noise affecting array sensors as correlated in time and between sensor components. Along with enhancement of reliability and accuracy of the source location the procedure allows for simultaneously estimating the parameters of the source mechanism. All available information about medium structure and source mechanism can be accounted for in this procedure. Its efficiency was tested using simulated mixtures of synthetic array seismograms with real array noise recordings. Significant improvements of location accuracy and detection capability threshold were established when compared with the conventional Seismic Emission Tomography (SET) processing.

Statistical Analysis of Microseismic Noise during Hydraulic Fracturing

M.V. Rozhkov* (SYNAPSE Science Center), A.F. Kushnir (SYNAPSE Science Center), N.M. Rojkov (SYNAPSE Science Center), I.G. Dricker (Earth Imaging Inc.) & S. Hellman (Earth Imaging Inc.)


Spatial and temporal spectral analysis of the background noise before and during hydraulic fracturing shows that surface noise is generally uncorrelated before and after well injections. Noise becomes sufficiently coherent (correlated) during fluid injection and even during a break between fracturing stages. Stacking of seismic array records with proper move-out corrections of Seismic Emission Tomography (SET) helps to suppress non-correlated noise component, but enhances both correlated technogenic noise and signals from microseismic events. Presence of correlated noise can be a factor increasing number of false alarms in microseismic bulletins which are created by semblance-based SET techniques.

Evaluation of Location Capabilities of Statistically Optimal Algorithms for Microseismic Monitoring

A.F. Kushnir (SYNAPSE Science Center), M. V. Rozhkov* (SYNAPSE Science Center), A. Varypaev (SYNAPSE Science Center) & I.G. Dricker (Earth Imaging Inc.)


The benefits of data processing approach based on the Statistically Optimal (SO) Algorithms for processing of microseismic data observed during hydraulic fracturing jobs is presented in the paper. The presence of correlated noise is a factor that increases the number of false detections in microseismic bulletins created by semblance-based SET techniques. Significant suppression of correlated noise can be achieved by application of the SO algorithms for surface array data processing if statistical characteristics of noise are taken into account; specifically, if spatial and/or temporal correlations of noise are strong enough. The two SO-methods outlined in the paper: Adaptive Microseismic Location Algorithm and Frequency-Phase Microseismic Location Algorithm. Each SO-algorithm has its advantages and limitations, so the best results of detection/location are reachable by a combination of SO-algorithms. Results of SO-algorithms benchmarking for microseismic synthetic events using a double-couple source mixed with the recorded noise during hydraulic fracturing demonstrate that for 141-sensor surface array SO-algorithms allow to accurately locate microseismic events with SNR ~ 0.05, while traditional SET processing fails to even detect events with SNR lower than 0.3.

Enhancements of Microseismic Surface Array Monitoring of Hydraulic Fracturing Based on Statistically Optimal Algorithms

Alexey Epiphansky, Alexander Kushnir, Mikhail Rozhkov, Alexander Varypaev, Nikita Rojkov, SYNAPSE Science Center; Ilya Dricker, Sidney Hellman, Earth Imaging Inc.


Surface array microseismic monitoring during induced hydraulic fracturing (fracking) will be the subject of our presentation. Patterns of microseismicity associated with fracking of oil and shale gas reservoirs reveal the nature of fracture network geometry, connectivity, and density and can be used to improve oil and gas production. Surface array monitoring is positioned as a cheaper alternative to a borehole monitoring procedure, providing, in addition, better ray coverage of the study volume. The disadvantage of the surface monitoring is that arrivals of events caused by fracking activities (M=-1 to -3) are buried in surface noise on individual seismic records, prohibiting traditional location procedures (signal-to-noise ratio, or SNR << 1). Therefore, in surface microseismic monitoring one seeks for a functional over seismic records which would unmask microseismic signal. The most known surface array technique is called Seismic Emission Tomography (SET). SET associates a maximum of semblance functional over space, time, and sensors with the event hypocenter. Starting with the SET as a benchmark we discuss potential improvements of surface microseismic monitoring. First, we show that seismic noise during fracking is coherent, and, therefore, forms semblance maxima causing false event locations in SET processing. Next, we present a statistically optimal functional (SO) and demonstrate that a semblance operator is just a subset of a generic SO functional suitable for microseismic event location. We then show how different SObased operators can significantly increase signal-to-noise ratio. In particular, we demonstrate that statistically optimal algorithms are able to suppress man-made correlated noise of hydraulic fracturing. Finally, we discuss the effects of event source mechanisms on the microseismic locations, showing that nonexplosive sources appear as ambiguous, erroneous or multiple maxima of semblance volumes and produce redundant events or misallocations in final catalogs. We demonstrate that our Statistically Optimal (SO) technique applied is capable to properly locate non-explosive events with complex radiation patterns and – what is even more exciting – compute microseismic event focal mechanisms even if the original signal completely hidden in the background noise. In our presentation we provide both theoretical background and numerical modeling using noise patterns from the actual hydrofracturing site.