scholarly journals Prediction of seismic p-wave velocity using machine learning

2019 ◽  
Author(s):  
Ines Dumke ◽  
Christian Berndt
Keyword(s):  
Machine Learning ◽  
Wave Velocity ◽  
Seismic Velocity ◽  
P Wave ◽  
Training Dataset ◽  
Seismic Velocities ◽  
P Wave Velocity ◽  
Machine Learning Approach ◽  
Global Datasets ◽  
Spatial Predictors

Abstract. Measurements of seismic velocity as a function of depth are generally restricted to borehole locations and are therefore sparse in the world's oceans. Consequently, in the absence of measurements or suitable seismic data, studies requiring knowledge of seismic velocities often obtain these from simple empirical relationships. However, empirically derived velocities may be inaccurate, as they are typically limited to certain geological settings, and other parameters potentially influencing seismic velocities, such as depth to basement, crustal age, or heatflow, are not taken into account. Here, we present a machine learning approach to predict seismic p-wave velocity (vp) as a function of depth (z) for any marine location. Based on a training dataset consisting of vp(z) data from 333 boreholes and 38 geological and spatial predictors obtained from publically available global datasets, a prediction model was created using the Random Forests method. In 60 % of the tested locations, the predicted seismic velocities were superior to those calculated empirically. The results indicate a promising potential for global prediction of vp(z) data, which will allow improving geophysical models in areas lacking first-hand velocity data.

scholarly journals Prediction of seismic P-wave velocity using machine learning

Solid Earth ◽  
2019 ◽  
Vol 10 (6) ◽  
pp. 1989-2000 ◽  
Author(s):  
Ines Dumke ◽  
Christian Berndt
Keyword(s):  
Machine Learning ◽  
Wave Velocity ◽  
Seismic Velocity ◽  
P Wave ◽  
Training Dataset ◽  
Seismic Velocities ◽  
P Wave Velocity ◽  
Machine Learning Approach ◽  
Global Datasets ◽  
Spatial Predictors

Abstract. Measurements of seismic velocity as a function of depth are generally restricted to borehole locations and are therefore sparse in the world's oceans. Consequently, in the absence of measurements or suitable seismic data, studies requiring knowledge of seismic velocities often obtain these from simple empirical relationships. However, empirically derived velocities may be inaccurate, as they are typically limited to certain geological settings, and other parameters potentially influencing seismic velocities, such as depth to basement, crustal age, or heat flow, are not taken into account. Here, we present a machine learning approach to predict the overall trend of seismic P-wave velocity (vp) as a function of depth (z) for any marine location. Based on a training dataset consisting of vp(z) data from 333 boreholes and 38 geological and spatial predictors obtained from publicly available global datasets, a prediction model was created using the random forests method. In 60 % of the tested locations, the predicted seismic velocities were superior to those calculated empirically. The results indicate a promising potential for global prediction of vp(z) data, which will allow the improvement of geophysical models in areas lacking first-hand velocity data.

scholarly journals A P-wave velocity model of the upper crust of the Sannio region (Southern Apennines, Italy)

1998 ◽  
Vol 41 (4) ◽  
Author(s):  
G. Iannaccone ◽  
L. Improta ◽  
P. Capuano ◽  
A. Zollo ◽  
G. Biella ◽  
...  
Keyword(s):  
Wave Velocity ◽  
Seismic Velocity ◽  
Carbonate Platform ◽  
Velocity Model ◽  
P Wave ◽  
Seismic Velocities ◽  
Upper Crust ◽  
Near Surface ◽  
P Wave Velocity ◽  
Apulia Platform

This paper describes the results of a seismic refraction profile conducted in October 1992 in the Sannio region, Southern Italy, to obtain a detailed P-wave velocity model of the upper crust. The profile, 75 km long, extended parallel to the Apenninic chain in a region frequently damaged in historical time by strong earthquakes. Six shots were fired at five sites and recorded by a number of seismic stations ranging from 41 to 71 with a spacing of 1-2 km along the recording line. We used a two-dimensional raytracing technique to model travel times and amplitudes of first and second arrivals. The obtained P-wave velocity model has a shallow structure with strong lateral variations in the southern portion of the profile. Near surface sediments of the Tertiary age are characterized by seismic velocities in the 3.0-4.1 km/s range. In the northern part of the profile these deposits overlie a layer with a velocity of 4.8 km/s that has been interpreted as a Mesozoic sedimentary succession. A high velocity body, corresponding to the limestones of the Western Carbonate Platform with a velocity of 6 km/s, characterizes the southernmost part of the profile at shallow depths. At a depth of about 4 km the model becomes laterally homogeneous showing a continuous layer with a thickness in the 3-4 km range and a velocity of 6 km/s corresponding to the Meso-Cenozoic limestone succession of the Apulia Carbonate Platform. This platform appears to be layered, as indicated by an increase in seismic velocity from 6 to 6.7 km/s at depths in the 6-8 km range, that has been interpreted as a lithological transition from limestones to Triassic dolomites and anhydrites of the Burano formation. A lower P-wave velocity of about 5.0-5.5 km/s is hypothesized at the bottom of the Apulia Platform at depths ranging from 10 km down to 12.5 km; these low velocities could be related to Permo-Triassic siliciclastic deposits of the Verrucano sequence drilled at the bottom of the Apulia Platform in the Apulia Foreland.

scholarly journals The Local Earthquake Tomography of Erzurum (Turkey) Geothermal Area

2019 ◽  
Vol 23 (3) ◽  
pp. 209-223 ◽  
Author(s):  
Caglar Ozer ◽  
Mehmet Ozyazicioglu
Keyword(s):  
Wave Velocity ◽  
Seismic Velocity ◽  
Three Dimensional ◽  
P Wave ◽  
Geothermal Area ◽  
S Wave ◽  
Velocity Models ◽  
Local Earthquake Tomography ◽  
P Wave Velocity ◽  
Local Earthquake

Erzurum and its surroundings are one of the seismically active and hydrothermal areas in the Eastern part of Turkey. This study is the first approach to characterize the crust by seismic features by using the local earthquake tomography method. The earthquake source location and the three dimensional seismic velocity structures are solved simultaneously by an iterative tomographic algorithm, LOTOS-12. Data from a combined permanent network comprising comprises of 59 seismometers which was installed by Ataturk University-Earthquake Research Center and Earthquake Department of the Disaster and Emergency Management Authority  to monitor the seismic activity in the Eastern Anatolia, In this paper, three-dimensional Vp and Vp/Vs characteristics of Erzurum geothermal area were investigated down to 30 km by using 1685 well-located earthquakes with 29.894 arrival times, consisting of 17.298 P- wave and 12.596 S- wave arrivals. We develop new high-resolution depth-cross sections through Erzurum and its surroundings to provide the subsurface geological structure of seismogenic layers and geothermal areas. We applied various size horizontal and vertical checkerboard resolution tests to determine the quality of our inversion process. The basin models are traceable down to 3 km depth, in terms of P-wave velocity models. The higher P-wave velocity areas in surface layers are related to the metamorphic and magmatic compact materials. We report that the low Vp and high Vp/Vs values are observed in Yedisu, Kaynarpinar, Askale, Cimenozu, Kaplica, Ovacik, Yigitler, E part of Icmeler, Koprukoy, Uzunahmet, Budakli, Soylemez, Koprukoy, Gunduzu, Karayazi, Icmesu, E part of Horasan and Kaynak regions indicated geothermal reservoir.

scholarly journals Seismic velocity change in sandstone during CO2 injection

2020 ◽  
Vol 205 ◽  
pp. 02001
Author(s):  
Marte Gutierrez ◽  
Daisuke Katsuki ◽  
Abdulhadi Almrabat
Keyword(s):  
Wave Velocity ◽  
Supercritical Co2 ◽  
Seismic Velocity ◽  
Saline Water ◽  
Pore Fluid ◽  
P Wave ◽  
Co2 Injection ◽  
Porous Rocks ◽  
Deep Saline Aquifer ◽  
P Wave Velocity

This paper presents analytical and experimental studies of the effects of supercritical CO2 injection on the seismic velocity of sandstone initially saturated with saline water. The analytical model is based on poroelasticity theory, particularly the application of the Biot-Gassmann substitution theory in the modeling of the acoustic velocity of porous rocks containing two-phase immiscible fluids. The experimental study used a high pressure and high temperature triaxial cell to clarify the seismic response of samples of Berea sandstone to supercritical CO2 injection under deep saline aquifer conditions. Measured ultrasonic wave velocity changes during CO2 injection in the sandstone sample showed the effects of pore fluid distribution in the seismic velocity of porous rocks. CO2 injection was shown to decrease the P-wave velocity with increasing CO2 saturation whereas the S-wave velocity was almost constant. The results confirm that the Biot-Gassmann theory can be used to model the changes in the acoustic P-wave velocity of sandstone containing different mixtures of supercritical CO2 and saline water provided the distribution of the two fluids in the sandstone pore space is accounted for in the calculation of the pore fluid bulk modulus.

scholarly journals Applying Statistical Analysis and Machine Learning for Modeling the UCS from P-Wave Velocity, Density and Porosity on Dry Travertine

2020 ◽  
Vol 10 (13) ◽  
pp. 4565
Author(s):  
Manuel Saldaña ◽  
Javier González ◽  
Ignacio Pérez-Rey ◽  
Matías Jeldres ◽  
Norman Toro
Keyword(s):  
Machine Learning ◽  
Compressive Strength ◽  
Statistical Analysis ◽  
Uniaxial Compressive Strength ◽  
Wave Velocity ◽  
Accurate Determination ◽  
Effective Porosity ◽  
P Wave ◽  
Strength Parameter ◽  
P Wave Velocity

In the rock mechanics and rock engineering field, the strength parameter considered to characterize the rock is the uniaxial compressive strength (UCS). It is usually determined in the laboratory through a few statistically representative numbers of specimens, with a recommended minimum of five. The UCS can also be estimated from rock index properties, such as the effective porosity, density, and P-wave velocity. In the case of a porous rock such as travertine, the random distribution of voids inside the test specimen (not detectable in the density-porosity test, but in the compressive strength test) causes large variations on the UCS value, which were found in the range of 62 MPa for this rock. This fact complicates a sufficiently accurate determination of experimental results, also affecting the estimations based on regression analyses. Aiming to solve this problem, statistical analysis, and machine learning models (artificial neural network) was developed to generate a reliable predictive model, through which the best results for a multiple regression model between uniaxial compressive strength (UCS), P-wave velocity and porosity were obtained.

P-wave velocity anisotropy in an active methane venting pockmark: The Scanner Pockmark, northern North Sea

2020 ◽  
Author(s):  
Gaye Bayrakci ◽  
Timothy A. Minshull ◽  
Jonathan M. Bull ◽  
Timothy J. Henstock ◽  
Giuseppe Provenzano ◽  
...  
Keyword(s):  
Wave Velocity ◽  
North Sea ◽  
Image Resolution ◽  
P Wave ◽  
Bright Spot ◽  
Seismic Velocities ◽  
Clay Layer ◽  
Travel Times ◽  
P Wave Velocity ◽  
Northern North Sea

<p>Scanner pockmark is an active and continuous methane venting seafloor depression of ~ 900 x 450 m wide and 22 m deep. It is located in the northern North Sea, within the Witch Ground basin where the seafloor and shallow sediments are heavily affected by pockmarks and paleo-pockmarks of various sizes. A seismic chimney structure is present below the Scanner pockmark. It is expressed as a near-vertical column of acoustic blanking below a bright zone of gas-bearing sediments. Seismic chimneys are thought to host connected vertical fractures which may be concentric within the chimney and align parallel to maximum compression outside it. The crack geometry modifies the seismic velocities, and hence, the anisotropy measured inside and outside of the chimney is expected to be different.</p><p> </p><p>We carried out anisotropic P-wave tomography with a GI-gun wide-angle dataset recorded by the 25 Ocean Bottom Seismometers (OBSs) of the CHIMNEY experiment (2017). Travel times of more than 60,000 refracted phases propagating within a volume of 4 x 4 x 2 km were inverted for P-wave velocity and the direction and degree of P-wave anisotropy. The grid is centred on the Scanner Pockmark and has a y-axis parallel to -34<sup>o</sup> N. The horizontal node interval is denser in the zone covered by the OBSs and the vertical node interval is denser near the seabed. A 3 iteration inversion leads to a chi<sup>2</sup> misfit value of 1 and a root-mean-square misfit of <10 ms. The results show a maximum P-wave anisotropy of 5%, and higher degrees of anisotropy correlates well with higher velocities. The fast P-wave velocity orientation, a proxy for fracture orientations, is 46<sup>o</sup> N. The top of the chimney possibly links a bright spot mapped at 270 ms in two way travel time using RMS amplitudes of MCS data, to the surface gas emission. The bright spot corresponds to low tomographic P-wave velocity and anisotropy, suggesting that gas is located in a zone with unaligned fractures or porosity. This observation is in good agreement with early multi-channel seismic data interpretations which suggested that the gas is trapped within a sandy clay layer, the Ling Bank Formation, capped by an upper clay layer, the Coal Pit Formation. In the next step, we will invert the travel-times of reflected phases in order to increase the image resolution.  </p>

Constraining crustal structure in the presence of sediment: a multiple converted wave approach

2019 ◽  
Vol 219 (1) ◽  
pp. 313-327 ◽  
Author(s):  
Erin Cunningham ◽  
Vedran Lekic
Keyword(s):  
Wave Velocity ◽  
Seismic Velocity ◽  
Seismic Station ◽  
Sedimentary Basins ◽  
Crustal Thickness ◽  
Receiver Functions ◽  
P Wave ◽  
Converted Wave ◽  
P Wave Velocity ◽  
Multiple Data Sets

SUMMARY Receiver functions are sensitive to sharp seismic velocity variations with depth and are commonly used to constrain crustal thickness. The H–κ stacking method of Zhu & Kanamori is often used to constrain both the crustal thickness (H) and ${V_P}$/${V_S}$ ratio ($\kappa $) beneath a seismic station using P-to-s converted waves (Ps). However, traditional H–κ stacks require an assumption of average crustal velocity (usually ${V_P}$). Additionally, large amplitude reverberations from low velocity shallow layers, such as sedimentary basins, can overprint sought-after crustal signals, rendering traditional H–$\ \kappa $ stacking uninterpretable. We overcome these difficulties in two ways. When S-wave reverberations from sediment are present, they are removed by applying a resonance removal filter allowing crustal signals to be clarified and interpreted. We also combine complementary Ps receiver functions, Sp receiver functions, and the post-critical P-wave reflection from the Moho (SPmp) to remove the dependence on an assumed average crustal ${V_P}$. By correcting for sediment and combining multiple data sets, the crustal thickness, average crustal P-wave velocity and crustal ${V_P}$/${V_S}$ ratio is constrained in geological regions where traditional H–$\ \kappa $ stacking fails, without making an initial P-wave velocity assumption or suffering from contamination by sedimentary reverberations.

Experimental study of fracture size effect on elastic-wave velocity dispersion and anisotropy

Geophysics ◽  
2018 ◽  
Vol 83 (1) ◽  
pp. C49-C59 ◽  
Author(s):  
Da Shuai ◽  
Jianxin Wei ◽  
Bangrang Di ◽  
Sanyi Yuan ◽  
Jianyong Xie ◽  
...  
Keyword(s):  
Wave Velocity ◽  
Seismic Velocity ◽  
Effective Medium Theory ◽  
Transversely Isotropic ◽  
Elastic Wave Velocity ◽  
P Wave ◽  
S Wave ◽  
P Wave Velocity ◽  
Fracture Size ◽  
Good Agreement

We have designed transversely isotropic models containing penny-shaped rubber inclusions, with the crack diameters ranging from 2.5 to 6.2 mm to study the influence of fracture size on seismic velocity under controlled conditions. Three pairs of transducers with different frequencies (0.5, 0.25, and 0.1 MHz) are used for P- and S-wave ultrasonic sounding, respectively. The P-wave measurements indicate that the scattering effect is dominant when the waves propagate perpendicular to the fractures. Our experimental results demonstrate that when the wavelength-to-crack-diameter ratio ([Formula: see text]) is larger than 14, the P-wave velocity can be described predominantly by the effective medium theory. Although the ratio is larger than four, the S-wave velocity is close to the equivalent medium results. When [Formula: see text] < 14 or [Formula: see text] is < 4, the elastic velocity is dominated by scattering. The magnitudes of the Thomsen anisotropic parameters [Formula: see text] and [Formula: see text] are scale and frequency dependent on the assumption that the transversely isotropic models are vertical transversely isotropic medium. Furthermore, we compare the experimental velocities with the Hudson theory. The results illustrate that there is a good agreement between the observed P-wave velocity and the Hudson theory when [Formula: see text] > 7 in the directions parallel and perpendicular to the fractures. For small fracture diameters, however, the P-wave velocity perpendicular to the fractures predicted from the Hudson theory is not accurate. When [Formula: see text] < 4, there is good agreement between the experimental fast S-wave velocity and the Hudson theory, whereas the experimental slow S-wave velocity diverges with the Hudson theory. When [Formula: see text] > 4, the deviation of fast and slow S-wave velocities with the Hudson prediction is stable.

Wavelet distortion correction due to domain conversion

Geophysics ◽  
2010 ◽  
Vol 75 (6) ◽  
pp. V77-V87 ◽  
Author(s):  
Rishi Bansal ◽  
Mike Matheney
Keyword(s):  
Wave Velocity ◽  
Seismic Data ◽  
Seismic Velocity ◽  
Distortion Correction ◽  
P Wave ◽  
Converted Wave ◽  
Shaping Filter ◽  
Seismic Wavelet ◽  
Attribute Analysis ◽  
P Wave Velocity

Converted-wave (PS) data, when converted to PP time, develop time- and location-varying compression of the seismic wavelet due to a variable subsurface [Formula: see text] [Formula: see text]. The time-dependent compression distorts the wavelet in a seismic trace. The lack of a consistent seismic wavelet in a domain-converted PS volume can eventually lead to an erroneous joint PP/PS inversion result. Depth-converted seismic data also have wavelet distortion due to velocity-dependent wavelet stretch. A high value of seismic velocity produces more stretch in a seismic wavelet than a low value. Variable wavelet stretch renders the depth data unsuitable for attribute analysis. A filtering scheme is proposed that corrects for distortion in seismic wavelets due to domain conversions (PS to PP time and time-to-depth) of seismic data in an amplitude-preserving manner. The method uses a Fourier scaling theorem to predict the seismic wavelet in the converted domain and calculates a shaping filter for each time/depth sample that corrects for the distortion in the wavelet. The filter is applied to the domain-converted data using the method of nonstationary filtering. We provide analytical expressions for the squeeze factor [Formula: see text] that is used to predict the wavelet in the converted domain. The squeeze factor [Formula: see text] for PS to PP time conversion is a function of the subsurface [Formula: see text] whereas for PP time-to-depth conversion [Formula: see text] is dependent on subsurface P-wave velocity. After filtering, the squeezed wavelets in domain-converted PS data appear to have resulted from a constant subsurface [Formula: see text], which we denote as [Formula: see text]. Similarly, the filtered depth-converted data appear to have resulted from a constant subsurface P-wave velocity [Formula: see text].

Sign in / Sign up

Export Citation Format

Share Document