scholarly journals Deep Neural Networks for Detection and Location of Microseismic Events and Velocity Model Inversion from Microseismic Data Acquired by Distributed Acoustic Sensing Array

Sensors ◽  
2021 ◽  
Vol 21 (19) ◽  
pp. 6627
Author(s):  
Daniel Wamriew ◽  
Roman Pevzner ◽  
Evgenii Maltsev ◽  
Dimitri Pissarenko
Keyword(s):  
Deep Learning ◽  
Real Time ◽  
Wave Velocity ◽  
Velocity Model ◽  
P Wave ◽  
Model Parameters ◽  
Acoustic Sensing ◽  
Microseismic Data ◽  
Microseismic Events ◽  
Distributed Acoustic Sensing

Fiber-optic cables have recently gained popularity for use as Distributed Acoustic Sensing (DAS) arrays for borehole microseismic monitoring due to their physical robustness as well as high spatial and temporal resolutions. As a result, the sensors record large amounts of data, making it very difficult to process in real-/semi-real-time using the conventional processing routines. We present a novel approach, based on deep learning, for handling the large amounts of DAS data in real-/semi-real-time. The proposed neural network was trained on synthetic microseismic data contaminated with real-ambient noise from field data and was validated using field DAS microseismic data obtained from a hydraulic fracturing operation. The results indicate that the trained network is capable of detecting and locating microseismic events from DAS data and simultaneously update the velocity model to a high degree of precision. The mean absolute errors in the event locations and the velocity model parameters are 2.04, 0.72, 2.76, 4.19 and 0.97 percent for distance (x), depth (z), P-wave velocity, S-wave velocity and density, respectively. In addition to automation and computational efficiency, deep learning reduces human expert data handling during processing, thus preserving data integrity leading to more accurate and reproducible results.

Reconnaissance seismic refraction-reflection surveys in northwestern New Mexico

1984 ◽  
Vol 74 (4) ◽  
pp. 1263-1274
Author(s):  
Lawrence H. Jaksha ◽  
David H. Evans
Keyword(s):  
New Mexico ◽  
Wave Velocity ◽  
Lower Crust ◽  
Seismic Waves ◽  
Seismic Refraction ◽  
Velocity Model ◽  
P Wave ◽  
Uppermost Mantle ◽  
P Wave Velocity ◽  
Crustal Thinning

Abstract A velocity model of the crust in northwestern New Mexico has been constructed from an interpretation of direct, refracted, and reflected seismic waves. The model suggests a sedimentary section about 3 km thick with an average P-wave velocity of 3.6 km/sec. The crystalline upper crust is 28 km thick and has a P-wave velocity of 6.1 km/sec. The lower crust below the Conrad discontinuity has an average P-wave velocity of about 7.0 km/sec and a thickness near 17 km. Some evidence suggests that velocity in both the upper and lower crust increases with depth. The P-wave velocity in the uppermost mantle is 7.95 ± 0.15 km/sec. The total crustal thickness near Farmington, New Mexico, is about 48 km (datum = 1.6 km above sea level), and there is evidence for crustal thinning to the southeast.

scholarly journals Sensitivities of surface wave velocities to the medium parameters in a radially anisotropic spherical Earth and inversion strategies

2015 ◽  
Vol 58 (5) ◽  
Author(s):  
Sankar N. Bhattacharya
Keyword(s):  
Surface Wave ◽  
Phase Velocity ◽  
Group Velocity ◽  
Wave Velocity ◽  
Love Waves ◽  
P Wave ◽  
Model Parameters ◽  
Nonlinear Inversion ◽  
Wave Velocities ◽  
Spherical Earth

<p>Sensitivity kernels or partial derivatives of phase velocity (<em>c</em>) and group velocity (<em>U</em>) with respect to medium parameters are useful to interpret a given set of observed surface wave velocity data. In addition to phase velocities, group velocities are also being observed to find the radial anisotropy of the crust and mantle. However, sensitivities of group velocity for a radially anisotropic Earth have rarely been studied. Here we show sensitivities of group velocity along with those of phase velocity to the medium parameters <em>V<sub>SV</sub>, V<sub>SH </sub>, V<sub>PV</sub>, V<sub>PH , </sub></em><em>h</em><em> </em>and density in a radially anisotropic spherical Earth. The peak sensitivities for <em>U</em> are generally twice of those for <em>c</em>; thus <em>U</em> is more efficient than <em>c</em> to explore anisotropic nature of the medium. Love waves mainly depends on <em>V<sub>SH</sub></em> while Rayleigh waves is nearly independent of <em>V<sub>SH</sub></em> . The sensitivities show that there are trade-offs among these parameters during inversion and there is a need to reduce the number of parameters to be evaluated independently. It is suggested to use a nonlinear inversion jointly for Rayleigh and Love waves; in such a nonlinear inversion best solutions are obtained among the model parameters within prescribed limits for each parameter. We first choose <em>V<sub>SH</sub></em>, <em>V<sub>SV </sub></em>and <em>V<sub>PH</sub></em> within their corresponding limits; <em>V<sub>PV</sub></em> and <em>h</em> can be evaluated from empirical relations among the parameters. The density has small effect on surface wave velocities and it can be considered from other studies or from empirical relation of density to average P-wave velocity.</p>

scholarly journals Research on prediction method of S-wave velocity based on deep learning

2021 ◽  
Vol 2083 (4) ◽  
pp. 042065
Author(s):  
Guojie Yang ◽  
Shuhua Wang
Keyword(s):  
Deep Learning ◽  
Wave Velocity ◽  
Prediction Method ◽  
P Wave ◽  
Learning Technology ◽  
S Wave ◽  
Advantages And Disadvantages ◽  
Wave Prediction ◽  
Wide Range ◽  
Velocity Prediction

Abstract Aiming at the s-wave velocity prediction problem, based on the analysis of the advantages and disadvantages of the empirical formula method and the rock physics modeling method, combined with the s-wave velocity prediction principle, the deep learning method is introduced, and a deep learning-based logging s-wave velocity prediction method is proposed. This method uses a deep neural network algorithm to establish a nonlinear mapping relationship between reservoir parameters (acoustic time difference, density, neutron porosity, shale content, porosity) and s-wave velocity, and then applies it to the s-wave velocity prediction at the well point. Starting from the relationship between p-wave and s-wave velocity, the study explained the feasibility of applying deep learning technology to s-wave prediction and the principle of sample selection, and finally established a reliable s-wave prediction model. The model was applied to s-wave velocity prediction in different research areas, and the results show that the s-wave velocity prediction technology based on deep learning can effectively improve the accuracy and efficiency of s-wave velocity prediction, and has the characteristics of a wide range of applications. It can provide reliable s-wave data for pre-stack AVO analysis and pre-stack inversion, so it has high practical application value and certain promotion significance.

scholarly journals Distributed acoustic sensing as a tool for exploration and monitoring: a proof-of-concept

2021 ◽  
Author(s):  
Nicola Piana Agostinetti ◽  
Alberto Villa ◽  
Gilberto Saccorotti
Keyword(s):  
High Resolution ◽  
Plane Wave ◽  
Geothermal Field ◽  
P Wave ◽  
Apparent Velocity ◽  
Near Surface ◽  
Acoustic Sensing ◽  
Geothermal Activity ◽  
Distributed Acoustic Sensing ◽  
Microseismic Event

Abstract. We use PoroTOMO experimental data to compare the performance of Distributed Acoustic Sensing (DAS) and geophone data in executing standard exploration and monitoring activities. The PoroTOMO experiment consists of two "seismic systems": (a) a 8.6 km long optical fibre cable deployed across the Brady geothermal field and covering an area of 1.5 x 0.5 km with 100 m long segments, and (b) an array of 238 co-located geophones with an average spacing of 60 m. The PoroTOMO experiment recorded continuous seismic data between March 10th and March 25th 2016. During such period, a ML 4.3 regional event occurred in the southwest, about 150 km away from the geothermal field, together with several microseismic local events related to the geothermal activity. The seismic waves generated from such seismic events have been used as input data in this study. For the exploration tasks, we compare the propagation of the ML 4.3 event across the geothermal field in both seismic systems in term of relative time-delay, for a number of configurations and segments. Defined the propagation, we analyse and compare the amplitude and the signal-to-noise ratio (SNR) of the P-wave in the two systems at high resolution. For testing the potential in monitoring local seismicity, we first perform an analysis of the geophone data for locating a microseismic event, based on expert opinion. Then, we a adopt different workflow for the automatic location of the same microseismic event using DAS data. For testing the potential in monitoring distant event, data from the regional earthquake are used for retrieving both the propagation direction and apparent velocity of the wavefield, using a standard plane-wave-fitting approach. Our results indicate that: (1) at a local scale, the seismic P-waves propagation and their characteristics (i.e. SNR and amplitude) along a single cable segment are robustly consistent with recordings from co-located geophones (delay-times δt ∼ 0.3 over 400 m for both seismic systems) ; (2) the interpretation of seismic wave propagation across multiple separated segments is less clear, due to the heavy contamination of scattering sources and local velocity heterogeneities; nonetheless, results from the plane-wave fitting still indicate the possibility for a consistent detection and location of the event; (3) at high-resolution (10 m), large amplitude variations along the fibre cable seem to robustly correlate with near surface geology; (4) automatic monitoring of microseismicity can be performed with DAS recordings with results comparable to manual analysis of geophone recordings (i.e. maximum horizontal error on event location around 70 m for both geophones and DAS data) ; and (5) DAS data pre-conditioning (e.g., temporal sub-sampling and channel-stacking) and dedicated processing techniques are strictly necessary for making any real-time monitoring procedure feasible and trustable.

Multi-parameter model building for the Qiuyue structure using four-component ocean-bottom seismometer data

Geophysics ◽  
2021 ◽  
pp. 1-52
Author(s):  
Yuzhu Liu ◽  
Xinquan Huang ◽  
Jizhong Yang ◽  
Xueyi Liu ◽  
Bin Li ◽  
...  
Keyword(s):  
Wave Velocity ◽  
Waveform Inversion ◽  
Velocity Model ◽  
P Wave ◽  
Velocity Analysis ◽  
Full Waveform Inversion ◽  
Ocean Bottom ◽  
Ocean Bottom Seismometer ◽  
Full Waveform ◽  
P Wave Velocity

Thin sand-mud-coal interbedded layers and multiples caused by shallow water pose great challenges to conventional 3D multi-channel seismic techniques used to detect the deeply buried reservoirs in the Qiuyue field. In 2017, a dense ocean-bottom seismometer (OBS) acquisition program acquired a four-component dataset in East China Sea. To delineate the deep reservoir structures in the Qiuyue field, we applied a full-waveform inversion (FWI) workflow to this dense four-component OBS dataset. After preprocessing, including receiver geometry correction, moveout correction, component rotation, and energy transformation from 3D to 2D, a preconditioned first-arrival traveltime tomography based on an improved scattering integral algorithm is applied to construct an initial P-wave velocity model. To eliminate the influence of the wavelet estimation process, a convolutional-wavefield-based objective function for the preprocessed hydrophone component is used during acoustic FWI. By inverting the waveforms associated with early arrivals, a relatively high-resolution underground P-wave velocity model is obtained, with updates at 2.0 km and 4.7 km depth. Initial S-wave velocity and density models are then constructed based on their prior relationships to the P-wave velocity, accompanied by a reciprocal source-independent elastic full-waveform inversion to refine both velocity models. Compared to a traditional workflow, guided by stacking velocity analysis or migration velocity analysis, and using only the pressure component or other single-component, the workflow presented in this study represents a good approach for inverting the four-component OBS dataset to characterize sub-seafloor velocity structures.

scholarly journals Application of machine learning to microseismic event detection in distributed acoustic sensing data

Geophysics ◽  
2020 ◽  
Vol 85 (5) ◽  
pp. KS149-KS160 ◽  
Author(s):  
Anna L. Stork ◽  
Alan F. Baird ◽  
Steve A. Horne ◽  
Garth Naldrett ◽  
Sacha Lapins ◽  
...  
Keyword(s):  
Detection Algorithm ◽  
Data Sets ◽  
Data Set ◽  
Acoustic Sensing ◽  
Detection Techniques ◽  
Real Noise ◽  
Temporal Sampling ◽  
Manual Inspection ◽  
Microseismic Events ◽  
Distributed Acoustic Sensing

This study presents the first demonstration of the transferability of a convolutional neural network (CNN) trained to detect microseismic events in one fiber-optic distributed acoustic sensing (DAS) data set to other data sets. DAS increasingly is being used for microseismic monitoring in industrial settings, and the dense spatial and temporal sampling provided by these systems produces large data volumes (approximately 650 GB/day for a 2 km long cable sampling at 2000 Hz with a spatial sampling of 1 m), requiring new processing techniques for near-real-time microseismic analysis. We have trained the CNN known as YOLOv3, an object detection algorithm, to detect microseismic events using synthetically generated waveforms with real noise superimposed. The performance of the CNN network is compared to the number of events detected using filtering and amplitude threshold (short-term average/long-term average) detection techniques. In the data set from which the real noise is taken, the network is able to detect >80% of the events identified by manual inspection and 14% more than detected by standard frequency-wavenumber filtering techniques. The false detection rate is approximately 2% or one event every 20 s. In other data sets, with monitoring geometries and conditions previously unseen by the network, >50% of events identified by manual inspection are detected by the CNN.

scholarly journals The potential of seismic cross-hole tomography for geotechnical site investigation

2019 ◽  
Vol 92 ◽  
pp. 18006
Author(s):  
Yannick Choy Hing Ng ◽  
William Danovan ◽  
Taeseo Ku
Keyword(s):  
Wave Velocity ◽  
Site Investigation ◽  
Velocity Model ◽  
P Wave ◽  
Reconstruction Technique ◽  
Sufficient Information ◽  
Eastern Region ◽  
S Wave ◽  
Near Surface ◽  
Soil Layers

Seismic cross-hole tomography has been commonly used in oil and gas exploration and the mining industry for the detection of precious resources. For near-surface geotechnical site investigation, this geophysical method is relatively new and can be used to supplement traditional methods such as the standard penetration test, coring and sampling, thus improving the effectiveness of site characterization. This paper presents a case study which was carried out on a reclaimed land in the Eastern region of Singapore. A seismic cross-hole test was performed by generating both compressional waves and shear waves into the ground. The signals were interpreted by using first-arrival travel time wave tomography and the arrival times were subsequently inverted using Simultaneous Iterative Reconstruction Technique (SIRT). A comparison with the borehole logging data indicated that P-wave velocity model cannot provide sufficient information about the soil layers, especially when the ground water table is near the surface. The S-wave velocity model seemed to agree quite well with the variation in the SPT-N value and could identify to a certain extent the interface between the different soil layers. Finally, P-wave and S-wave velocities are used to compute the Poisson's ratio distribution which gave a good indication of the degree of saturation of the soil.

scholarly journals Reference P-wave velocity in coal seams at great depths in Jastrzebie coal mine

2019 ◽  
Vol 133 ◽  
pp. 01011
Author(s):  
Jakub Kokowski ◽  
Zbigniew Szreder ◽  
Elżbieta Pilecka
Keyword(s):  
Wave Velocity ◽  
Coal Mine ◽  
Velocity Model ◽  
P Wave ◽  
Coal Seams ◽  
Data Set ◽  
Seismic Profiling ◽  
Reference Velocity ◽  
P Wave Velocity ◽  
Geological Conditions

In the study, the determining of the reference velocity of the P-wave in coal seams used in seismic profiling to assess increases and decreases in relative stresses at large depths has been presented. The seismic profiling method proposed by Dubinski in 1989 covers a range of depth up to 970 m. At present, coal seams exploitation in Polish coal mines is conducted at greater depths, even exceeding 1200 m, which creates the necessity for a new reference velocity model. The study presents an empirical mathematical model of the change of the P-wave velocity in coal seams in the geological conditions of the Jastrzebie coal mine. A power model analogous to the Dubinski’s one was elaborated with new constants. The calculations included the results from 35 measurements of seismic profiling carried out in various coal seams of the Jastrzebie mine at depths from 640 to 1200 m. The results obtained cause changes in the result of calculations of seismic anomalies. Future validation of the proposed model with larger data set will be required.

Guidelines for building a detailed elastic depth model

Geophysics ◽  
2000 ◽  
Vol 65 (1) ◽  
pp. 35-45
Author(s):  
Jarrod C. Dunne ◽  
Greg Beresford ◽  
Brian L. N Kennett
Keyword(s):  
Wave Velocity ◽  
Mode Conversion ◽  
Poisson Ratio ◽  
Velocity Model ◽  
P Wave ◽  
Time Interval ◽  
S Wave ◽  
Sea Floor ◽  
P Wave Velocity ◽  
Inelastic Attenuation

We developed guidelines for building a detailed elastic depth model by using an elastic synthetic seismogram that matched both prestack and stacked marine seismic data from the Gippsland Basin (Australia). Recomputing this synthetic for systematic variations upon the depth model provided insight into how each part of the model affected the synthetic. This led to the identification of parameters in the depth model that have only a minor influence upon the synthetic and suggested methods for estimating the parameters that are important. The depth coverage of the logging run is of prime importance because highly reflective layering in the overburden can generate noise events that interfere with deeper events. A depth sampling interval of 1 m for the P-wave velocity model is a useful lower limit for modeling the transmission response and thus maintaining accuracy in the tie over a large time interval. The sea‐floor model has a strong influence on mode conversion and surface multiples and can be built using a checkshot survey or by testing different trend curves. When an S-wave velocity log is unavailable, it can be replaced using the P-wave velocity model and estimates of the Poisson ratio for each significant geological formation. Missing densities can be replaced using Gardner’s equation, although separate substitutions are required for layers known to have exceptionally high or low densities. Linear events in the elastic synthetic are sensitive to the choice of inelastic attenuation values in the water layer and sea‐floor sediments, while a simple inelastic attenuation model for the consolidated sediments is often adequate. The usefulness of a 1-D depth model is limited by misties resulting from complex 3-D structures and the validity of the measurements obtained in the logging run. The importance of such mis‐ties can be judged, and allowed for in an interpretation, by recomputing the elastic synthetic after perturbing the depth model to simulate the key uncertainties. Taking the next step beyond using simplistic modeling techniques requires extra effort to achieve a satisfactory tie to each part of a prestack seismic record. This is rewarded by the greater confidence that can then be held in the stacked synthetic tie and applications such as noise identification, data processing benchmarking, AVO analysis, and inversion.

Inversion of azimuthally dependent NMO velocity in transversely isotropic media with a tilted axis of symmetry

Geophysics ◽  
2000 ◽  
Vol 65 (1) ◽  
pp. 232-246 ◽  
Author(s):  
Vladimir Grechka ◽  
Ilya Tsvankin
Keyword(s):  
Wave Velocity ◽  
Symmetry Axis ◽  
Transverse Isotropy ◽  
Transversely Isotropic ◽  
P Wave ◽  
Model Parameters ◽  
Symmetry Direction ◽  
Time Migration ◽  
Inversion Procedure ◽  
Tti Media

Just as the transversely isotropic model with a vertical symmetry axis (VTI media) is typical for describing horizontally layered sediments, transverse isotropy with a tilted symmetry axis (TTI) describes dipping TI layers (such as tilted shale beds near salt domes) or crack systems. P-wave kinematic signatures in TTI media are controlled by the velocity [Formula: see text] in the symmetry direction, Thomsen’s anisotropic coefficients ε and δ, and the orientation (tilt ν and azimuth β) of the symmetry axis. Here, we show that all five parameters can be obtained from azimuthally varying P-wave NMO velocities measured for two reflectors with different dips and/or azimuths (one of the reflectors can be horizontal). The shear‐wave velocity [Formula: see text] in the symmetry direction, which has negligible influence on P-wave kinematic signatures, can be found only from the moveout of shear waves. Using the exact NMO equation, we examine the propagation of errors in observed moveout velocities into estimated values of the anisotropic parameters and establish the necessary conditions for a stable inversion procedure. Since the azimuthal variation of the NMO velocity is elliptical, each reflection event provides us with up to three constraints on the model parameters. Generally, the five parameters responsible for P-wave velocity can be obtained from two P-wave NMO ellipses, but the feasibility of the moveout inversion strongly depends on the tilt ν. If the symmetry axis is close to vertical (small ν), the P-wave NMO ellipse is largely governed by the NMO velocity from a horizontal reflector Vnmo(0) and the anellipticity coefficient η. Although for mild tilts the medium parameters cannot be determined separately, the NMO-velocity inversion provides enough information for building TTI models suitable for time processing (NMO, DMO, time migration). If the tilt of the symmetry axis exceeds 30°–40° (e.g., the symmetry axis can be horizontal), it is possible to find all P-wave kinematic parameters and construct the anisotropic model in depth. Another condition required for a stable parameter estimate is that the medium be sufficiently different from elliptical (i.e., ε cannot be close to δ). This limitation, however, can be overcome by including the SV-wave NMO ellipse from a horizontal reflector in the inversion procedure. While most of the analysis is carried out for a single layer, we also extend the inversion algorithm to vertically heterogeneous TTI media above a dipping reflector using the generalized Dix equation. A synthetic example for a strongly anisotropic, stratified TTI medium demonstrates a high accuracy of the inversion (subject to the above limitations).

Sign in / Sign up

Export Citation Format

Share Document