Using migration-based methods, we developed a workflow to obtain high-quality time picks for microseismic event data. The workflow crosscorrelates manually picked waveforms of template microseismic events with waveforms from the remaining events to refine the first arrival time picks of P and S phases. We applied our workflow to 50 minutes of microseismic data from a hydraulic fracturing monitoring campaign. The workflow provides high-quality picks for 689 out of a total of 805 detected events by using three representative template events. We validate the time picks by comparing event locations using the crosscorrelation picks with locations from the migration based method. The workflow improves time-pick accuracy, which subsequently may be used to improve associated source parameters and other attributes that require a precise time-pick of the signal onset. Presentation Date: Wednesday, October 14, 2020 Session Start Time: 9:20 AM Presentation Time: 9:45 AM Location: Poster Station 11 Presentation Type: Poster

An accurate estimation of microseismic event locations plays an important role in the success of fluid-injection programs. Wavefield-based elastic time-reverse imaging (TRI) offers a robust approach to locate microseismic events that occur due to induced seismicity. The event location accuracy, though, is greatly dependent on the veracity of the elastic velocity models. In this study, we propose a methodology for microseismic image-domain wavefield tomography using the elastic wave equation and extended source images. Gradients are computed using the adjoint-state method and build on the extended image residuals that satisfy the differential semblance optimization (DSO) criterion. The objective function is designed to optimize the focusing of timereversed microseismic energy at zero lag in extended source images. The function applies a penalty operator to the extended images to annihilate the energy at zero lag and highlight residual energy at non-zero lags caused by backpropagation through erroneous velocity models. Minimizing the objective function leads to a model optimization problem aimed at improving the image-focusing quality. Realistic synthetic experiments demonstrate that one can compute accurate velocity model gradients using the proposed method, which can significantly improve the focusing of imaged events, leading to enhanced fluid-injection programs. Presentation Date: Monday, October 12, 2020 Session Start Time: 1:50 PM Presentation Time: 2:40 PM Location: Poster Station 8 Presentation Type: Poster

Noise in surface microseismic data usually has complicated features, signal reconstruction from various types of noise is very challenging for existing reconstruction techniques. In this paper, we propose to use a trilateral weighted sparse coding scheme within the block matching framework for microseismic signal reconstruction. Block matching requires identification of similar patches in the data. This becomes more challenging for low quality data. We overcome this obstacle by using Zernike moments whose amplitudes are rotation invariant and less sensitive to noise. This in turn benefits the subsequent filtering stage by trilateral weighted sparse coding. We introduce two weight matrices to characterize the complex noise properties, and a third weight matrix to characterize the sparsity priors of signal. The improved sparse coding model is solved by the alternating direction method of multipliers. Tests on synthetic and field surface microseismic datasets show that the proposed strategy is very effective for microseismic signal recovery from the data contain complex noise interference. Presentation Date: Wednesday, October 14, 2020 Session Start Time: 1:50 PM Presentation Time: 3:30 PM Location: 360C Presentation Type: Oral

Event detection is one of the time-consuming parts in microseismic processing. Different automated event detection algorithms have been proposed, such as the short-time average over the long-time average (STA/LTA), power spectral density (PSD), and subspace detection (SD). However, these approaches are not convenient for big data sets. In this study, we introduce a fast matched filter (MF) algorithm, which can solve the efficiency challenge for these detectors. The proposed fast MF is built based on a fast normalized cross-correlation (NCC) technique. This method detects events in the data based on their similarity with template events by comparing the NCC coefficients between the template events and the data with a specific user-defined threshold. The detection workflow consists of six steps, namely data preconditioning, selecting template waveforms, multiplexing, fast NCC computation, extracting potential events, and quality control of the detection results. We have implemented the MF algorithm on a microseismic data set with over 19,000 events detected on two monitoring wells. The MF detection results are compared with the results from STA/LTA. The MF algorithm is more efficient in event detection with fewer false alarms, higher detection probability, and shorter processing time than STA/LTA, especially when dealing with big, noisy data sets. Presentation Date: Monday, October 12, 2020 Session Start Time: 1:50 PM Presentation Time: 4:20 PM Location: 360A Presentation Type: Oral

Earthquakes induced by hydraulic fracturing are typically believed to be caused by elevated pore pressure or increased shear stress. However, according to a recent study, these mechanisms are incompatible with experiments and rate-state frictional models that predict stable sliding (aseismic slip) for faults with high clay content or total organic carbon, as well as observations of the timings and locations of the seismicity. An alternative model was therefore proposed, in which distal, unstable regions of a fault are loaded by aseismic slip on stable regions of the fault stimulated by hydraulic fracturing. This model has significant implications in terms of mitigating induced seismicity, as it suggests that there may be a potentially measurable deformation signal tens of hours before earthquake nucleation. However, the conclusions of that study were based on a relatively small number of events from a small local broadband network. In this study we integrate a high-resolution microseismic dataset from the same treatment with that previous work, and demonstrate that the microseismic data provides an even more compelling case that aseismic slip plays a role in induced seismicity. Presentation Date: Monday, October 12, 2020 Session Start Time: 1:50 PM Presentation Time: 2:15 PM Location: 360A Presentation Type: Oral

We present a case study from the West Texas Permian Basin of constraining anisotropic velocity and improving microseismic event locations by using microseismic shear-wave splitting data (SH-wave arrival is earlier than Sv-wave arrival). There are 2 downhole arrays with 12 geophones in each array for monitoring the zipper hydraulic fracturing in 3 horizontal wells in the target formation. We built an anisotropic velocity model by joint inversion of an anisotropic velocity model with about 100 evenly distributed microseismic events as well as about 20 perforation shots in one of the horizontal wells. We started from a perforation-shot-calibrated vertical transversely isotropic (VTI) model provided by a vendor. Perforation shots cannot usually constrain anisotropic parameters, especially Thomsen’s (1986) parameter gamma. We found that the vendor’s calibrated VTI model could not fit the large shear-wave splitting in the raw microseismic data, so we manually picked Sv-wave arrivals of selected microseismic events and used them to constrain the anisotropic model. The average gamma parameter in the new VTI model increased to 0.2-0.25 in the target formation. The new VTI model reduced the average Sv-wave arrival time residual to 0.0 ms from 4.8 ms in the model supplied by the vendor. We used this new VTI model to locate 85 large microseismic events (Mb magnitude > -1.8) in one completion stage of this well. We picked the Sv-wave arrivals of these 85 large events, which the vendor did not use for locating microseismic events. We used the layered linear inversion (LLI) algorithm (Li et al., 2015, 2018) to locate the events. The new microseismic event cloud is much thinner in the vertical direction, indicating that hydraulic fractures are confined within the thin, low-impedance target shale layer. Event locations were shifted closer to the perforations, moving on average about 50 ft laterally and 25 ft vertically. Presentation Date: Wednesday, October 14, 2020 Session Start Time: 1:50 PM Presentation Time: 2:15 PM Location: 360C Presentation Type: Oral

ABSTRACT Two primary classes of seismic events are typically associated with hydraulic fracturing of unconventional reservoirs. Small microseismic dip- or oblique-slip events are interpreted as a direct indicator of the opening of hydraulic fractures. A second class of events that is sometimes observed are strike-slip events commonly related to fault activation. These strike-slip events are usually very small; however, there are cases where they are large enough that they are detected on surface seismometers. These cases of fault-related seismicity, in the absence of microseismic monitoring, are commonly attributed to basement related faults. Recent cases of relatively low-level induced seismicity in the Anadarko basin have been captured on a number of high-resolution microseismic and regional seismometer arrays. Analysis reveals that these fault related events are only slightly deeper than the treated interval, well above the basement. Activation occurs with direct fluid or proximal fluid communication. Once activated by a stage, activity remains for up to 10 days, compared to hydraulic fracture events where seismicity ends immediately with cessation of pumping. There appears to be a limit on the maximum moment release on any fault, related to the length of the feature that is illuminated. Subsequent activation of the same fault generally does not result in a lower seismicity rate, although the largest events tend to occur on primary activation of the fault. We also note that a second type of strike-slip event is often observed independent of fault activation. These events occur in the completion interval, strike parallel to the hydraulic fracture growth and trend towards pressurized offsetting wellbores. The mechanism of these events is possibly due to a stress rotation and shear on existing hydraulic fractures. Presentation Date: Wednesday, September 18, 2019 Session Start Time: 8:30 AM Presentation Start Time: 8:55 AM Location: 217B Presentation Type: Oral

ABSTRACT As a major field for unconventional oil and gas exploration and development in China, Sichuan Basin hosts a great quantity of marine shale successions. A typical example is Longmaxi Formation shale, which contributes to 30% of shale gas production in China over the past decade. However, further shale gas exploitation by hydraulic fracturing in horizontal wells has been greatly hindered, due to complex network of regional structural stresses and heterogeneity of fault distribution, which are basically attributable to the multiple stages of tectonic movements in the geological history. According to incomplete statistics over the period from 2009 to 2015, 31.7% of the horizontal wells in the south of Sichuan Basin suffer casing deformation issues, more than half of which are related to the faults beside the wells or natural fracture zones (Chen et al., 2016). Hence, it is critical to effectively assess stability of faults beside wells for more efficient oil and gas development in the study area. In this study, a comprehensive analysis method based on microseismic results and sliding trend seismic attributes is proposed. Its application succeeds to reduce the overall casing deformation probability in horizontal well groups based on approaches such as advanced judgment and on-site adjustment of construction parameters. Therefore, faults beside wells with different fracture features are well controlled during large-scale hydraulic fracturing in the strata with complicated structures. Presentation Date: Wednesday, September 18, 2019 Session Start Time: 8:30 AM Presentation Time: 9:20 AM Location: 221B Presentation Type: Oral

ABSTRACT The accuracy of velocity models built from microseismic data can be increased by employing 3D anisotropic waveform inversion (WI). Here, we apply our previously developed elastic WI algorithm to velocity analysis of synthetic and field microseismic data from VTI (transversely isotropic with a vertical symmetry axis) media. A synthetic test demonstrates that WI may be able to delineate hydraulic fractures if they are properly illuminated for the employed source-receiver geometry. We evaluate the sensitivity of the inversion results to errors in the source locations and to band-limited noise in the input displacements. The WI algorithm is also applied to a microseismic data set acquired during hydraulic fracturing of a shale reservoir. The inversion of the P-wave direct arrivals helps refine the 3D fields of the horizontal velocity and anellipticity coefficient . Further improvement is expected from joint inversion of the P and S data. Presentation Date: Wednesday, September 18, 2019 Session Start Time: 8:30 AM Presentation Time: 11:25 AM Location: 221C Presentation Type: Oral

ABSTRACT Seismicity with moment magnitudes () of up to 4.1 has been observed since late 2013 close to the town of Fox Creek, Alberta (Canada). This induced seismicity has been correlated with hydraulic-fracturing (HF) operations in the area. This study analyzes regional seismic data recorded from November 2015 to March 2016. During this period, seismic activity occurred within five main clusters located near distinct HF operations. Three of these clusters were short-lived, diminishing in intensity quickly after the end of HF operations; however, the remaining two exhibited unusual long-lived seismicity that continued beyond the treatment period and persisted for at least several months. Cluster 1 contained a 4.1 induced event along a north-south fault that is well oriented with respect to the regional stress field. The persistence of this cluster could be explained as a long-lived aftershock sequence. However, cluster 2 was more swarm-like in character, lacking any mainshock event and containing no events with =2.0. The swarm-like behavior of cluster 2 resumed and intensified after the 4.1 event in cluster 1. Downhole microseismic data at cluster 2 reveal a series of lineaments with a NW-SE strike direction that is mis-oriented for dynamic rupture in the regional stress field. Moreover, prior to swarm intensification, microseismicity started to penetrate into the formation below the reservoir along inferred, pre-existing faults and/or fractures in the underlying formation. Based on previous interpretations, both of the persistent clusters could represent fault systems that were pressurized by HF operations. Differences in response of clusters 1 and 2 (i.e. mainshock-aftershock versus swarm-like) may be linked to whether the activated fault is optimally oriented for slip in the present stress field. Presentation Date: Thursday, October 18, 2018 Start Time: 8:30:00 AM Location: 210C (Anaheim Convention Center) Presentation Type: Oral

Microseismic data is often used to locate fracture locations and their origin in time created by fracking. Although surface microseismic data can have large apertures and is easier to acquire than the borehole data, it often suffers from poor signal to noise ratio (S/R). Poor S/R poses a challenge in terms of estimating the correct location and source-time function of a microseismic source. In this work, we propose a denoising step in combination with a computationally cheap debiasing based approach to locate microseismic sources and to estimate their source-time functions with correct amplitude from extremely noisy data. Through numerical experiments, we demonstrate that our method can work with closely spaced microseismic sources with source-time functions of different peak amplitudes and frequencies. We have also shown the ability of our method with the smooth velocity model. Presentation Date: Tuesday, October 16, 2018 Start Time: 8:30:00 AM Location: 208A (Anaheim Convention Center) Presentation Type: Oral

3D anisotropic waveform inversion (WI) could provide highresolution velocity models and improved event locations for microseismic surveys. Here we extend our previously developed 2D WI methodology for microseismic borehole data to 3D anisotropic models. This extension allows us to invert multicomponent data recorded in multiple boreholes and properly account for vertical and lateral heterogeneity. Synthetic examples illustrate the performance of the inversion for layer-cake and “hydraulically fractured” VTI (transversely isotropic with a vertical symmetry axis) models. In both cases, the algorithm is able to reconstruct parts of the model which are sufficiently illuminated for the employed source-receiver geometry. We also present preliminary results including moment-tensor estimation for a field data set recorded during hydraulic stimulation of a shale reservoir Presentation Date: Wednesday, October 17, 2018 Start Time: 1:50:00 PM Location: 205A (Anaheim Convention Center) Presentation Type: Oral

Microseismic data is being routinely collected as part of large pad scale hydraulic fracturing developments. The large lateral and sometimes, vertical spread of the pad wells allow the possibility of correlating observations made from the microseismicity during treatment phase with known properties of the reservoir from 3D seismic as well as well log data. This study from the Permian Basin looks at the microseismic data from the treatment of 11 well laterals in both the upper and the middle Wolfcamp formations (UWC & MWC) and proposes the use of frequency magnitude or "b-value" distributions to understand fracturing behavior within the reservoir. Based on analysis of fractures from image logs and through fracture cores from target reservoir, we correlate the direct observations with the indirect measurements made though microseismic data analysis. Our work provides a valid, reproducible approach towards improved understanding of presence of natural fractures in the subsurface and their interaction with hydraulic fracturing operations using microseismic measurements. Presentation Date: Wednesday, October 17, 2018 Start Time: 1:50:00 PM Location: Poster Station 12 Presentation Type: Poster

ABSTRACT Cross-well tomography as a technology is the bridge between spatially detailed wellbore measurements and regional 3D reflection seismic surveys. Repeat cross-well tomography, if applied properly, is able make high resolution measurements of rock property changes, and similar to 4D surface seismic analysis, data differencing can enhance subtle variations not immediately apparent from a single acquisition. In this study, cross-well tomography measurements were made before and after hydraulic fracturing of a Marcellus reservoir in Clearfield County, Pennsylvania. Though imaging of the cross-well data was problematic, the change in Pwave and Swave velocities show remarkable patterns that shed light on fractures generated and possible residual fluid and pressure remaining in the fractured zone (Figure 1). Evidence from microseismic recording during hydraulic fracturing supports the possible interpretation that significant induced fracturing occurred above the Marcellus reservoir interval (here referred to as the Lower and Upper Hamilton interval), and that residual pressure was left behind post-fracturing as evidenced by the significant drop in Pwave velocities. A corresponding lack of shear wave velocity change suggests that proppant did not reach the Upper Hamilton and Tully formations, and therefore these fractures did not remain connected to the borehole. Significant rock property changes were introduced in the Lower Hamilton, evidenced by the change in the shear velocities through this interval, but the minor Pwave velocity changes suggest that the zone is in communication with the borehole, and therefore pressure was able to dissipate. Comparison with the microseismic data recorded between cross-well surveys shows interesting but debatable relationships with the tomography in that the density of microseismic events agrees with the Pwave velocity changes, but the rock property changes interpreted from the shear velocities is inferred. Invoking a bedding-plane slip model (Tan et al, 2016), it is possible that shear-slip events were infrequent through the Hamilton, and tensile fractures dominated until reaching the upper Hamilton and Tully carbonate upper bound on fracturing. The paper concludes with some reflections on tomographic processing and applications for further analysis. Presentation Date: Tuesday, October 16, 2018 Start Time: 1:50:00 PM Location: 212A (Anaheim Convention Center) Presentation Type: Oral

ABSTRACT A challenge in microseismic monitoring is quantification of survey acquisition and processing error, and how these errors jointly effect estimated locations. Quantifying acquisition and processing error, and uncertainty has multiple benefits, such as more accurate and precise estimation of locations, anisotropy, moment tensor inversion, and, potentially, allowing for detection of 4D reservoir changes. Here we quantify uncertainty due to acquisition, receiver orientation error, and hodogram analysis. We use a probabilistic location approach to identify the optimal bottom well location based upon known source locations. Probability density functions (PDF) are utilized to quantify uncertainty and propagate it through processing, including in source location inversion to describe the 3D event location likelihood. Changes in the early processing steps have allowed for understanding of location uncertainty and improved the mapping of the microseismic events. Presentation Date: Tuesday, October 16, 2018 Start Time: 8:30:00 AM Location: 208A (Anaheim Convention Center) Presentation Type: Oral

ABSTRACT When assessing the effectiveness of hydraulic stimulations in unconventional reservoirs, the extension of the microseismic event distribution is often taken as an important proxy for the underlying fracture geometry. Almost always the process ignores the inherent location uncertainty and its apparent inflation of the event distribution. In addition, translating the event distribution into fracture dimensions is generally a highly subjective process with often arbitrarily set rules. Both deficiencies can be overcome by using a statistical approach that uses Bayes theorem to incorporate all available information, including the location uncertainty, to quantify the probability of different plausible fracture heights. The output is a probability curve with values for the P10-P50-P90 exceedance probabilities for the different fracture heights which can be further used to assess the likelihood of vertical fracture barriers etc. Especially for datasets with considerable location uncertainty, the resulting fracture heights are generally much less than obtained from traditional interpretations. The methodology is not limited to microseismic data but can be used to assess and incorporate any pertinent objective and subjective information. Presentation Date: Wednesday, October 17, 2018 Start Time: 1:50:00 PM Location: 208A (Anaheim Convention Center) Presentation Type: Oral

ABSTRACT Microseismic methods are crucial for real-time monitoring of the hydraulic fracturing dynamic status. However, unlike the active-source seismic events, the microseismic events usually have very low signal-to-noise ratio (SNR), which makes its processing challenging. To overcome the noise issue of the weak microseismic events, we propose a novel method for microseismic event detection based on the support vector machine (SVM) classification with a Gaussian kernel. For the preprocessing, fix-sized segmentation with a length of 2* is used to divide the data into segments. 123 features in total, which are used as input data to train the SVM model, have been extracted. These features include 63 1D time/spectral-domain features, and 60 2D texture features. Afterwards, we use a combination of univariate feature selection and random-forest-based recursive feature elimination for feature selection. This feature selection strategy not only finds the best features but also decides the number of features that are needed for the best accuracy. Regarding the essential training process, a C-SVM model, where coefficient C is used to control the tolerance of error item, is considered and a cross-validation (CV) process is implemented for automatic parameter setting. In the end, a group of synthetic and real microseismic data with different levels of complexity show the effectiveness of the proposed method. Presentation Date: Wednesday, October 17, 2018 Start Time: 1:50:00 PM Location: Poster Station 2 Presentation Type: Poster

ABSTRACT Time-lapse, multicomponent seismic data were acquired in conjunction with microseismic monitoring during hydraulic fracture stimulation of eleven horizontal wells at Wattenberg Field, Colorado. By integrating microseismic data with surface multicomponent seismic monitoring, we can develop a broader understanding of how hydraulic fracture treatments can be monitored in the subsurface. Presentation Date: Thursday, September 28, 2017 Start Time: 9:45 AM Location: 351F Presentation Type: ORAL

ABSTRACT Least-squares elastic reverse-time migration (LSERTM) using microseismic sources is a low-cost and efficient method to image subsurface structures below and above microseismic event locations. However, explosive or vector sources conventionally used in elastic wavefield simulation are incapable of generating the realistic radiation patterns from microseismic sources. This limitation leads to disagreement between simulated and observed seismic phases. We develop an LSERTM method using a moment-tensor source for forward elastic-wavefield simulation. We validate the method using microseismic data acquired using a buried geophone string at the Aneth CO-EOR field. We first locate the microseismic events and invert for their moment tensors. We then use these information to perform LSERTM, and obtain high-resolution images of interfaces below and above these microseismic events. These structural information can in turn help improve microseismic event location and moment-tensor inversion. Presentation Date: Thursday, September 28, 2017 Start Time: 8:55 AM Location: 362D Presentation Type: ORAL

ABSTRACT This paper presents a random noise suppression method for multicomponent microseismic data by orthogonal matching pursuit of multiple measurement vectors (OMPMMV). Unlike the conventional orthogonal matching pursuit for single measurement vector (OMPSMV), this method assumes all different components data have the same sparsity structure and provides an additional informative coupling between multiple components. So it can process the multicomponent data simultaneously and protect the weak energy on one component by the strong energy constraint on the other components. The examples of synthetic and real data demonstrate the potential of this method for multicomponent data denoising and weak signal protection. Presentation Date: Wednesday, September 27, 2017 Start Time: 2:15 PM Location: Exhibit Hall C/D Presentation Type: POSTER