SUMMARY We test a fully non-linear method to solve Bayesian seismic tomographic problems using data consisting of observed traveltimes first-arriving waves. Rather than Monte Carlo methods sample the posterior probability distribution that embodies solution inverse problem, we use variational inference. Variational inference problem under an optimization framework by seeking best approximation from family distributions, while still providing probabilistic results. introduce new for...

SUMMARY Seismic full-waveform inversion (FWI) provides high resolution images of the subsurface by exploiting information in recorded seismic waveforms. This is achieved solving a highly non-linear and non-unique inverse problem. Bayesian inference therefore used to quantify uncertainties solution. Variational method that probabilistic, solutions efficiently using optimization. The has been applied 2-D FWI problems produce full posterior distributions. However, due higher dimensionality more...

Seismic surface wave tomography is a tried and tested method to reveal the subsurface structure of Earth. However, conventional 2-step scheme inverting first for 2-D maps phase or group velocity then 3-D spatial preserves little information about lateral correlations, introduces additional uncertainties errors into result. We introduce 1-step non-linear that removes these effects by directly from frequency-dependent traveltime measurements. achieve this using reversible jump Markov chain...

SUMMARY Seismic full-waveform inversion (FWI) can produce high-resolution images of the Earth’s subsurface. Since modelling is significantly nonlinear with respect to velocities, Monte Carlo methods have been used assess image uncertainties. However, because high computational cost sampling methods, uncertainty assessment remains intractable for larger data sets and 3-D applications. In this study, we propose a new method called variational FWI, which uses Stein gradient descent solve FWI...

Abstract Constraining geophysical models with observed data usually involves solving nonlinear and nonunique inverse problems. Neural mixture density networks (MDNs) provide an efficient way to estimate Bayesian posterior marginal probability functions (pdf's) that represent the solution. However, it is difficult infer correlations between parameters using MDNs, in turn draw samples from pdf. We introduce alternative resolve these issues: invertible neural (INNs). These are simultaneously...

Summary Seismic tomography is used to image subsurface structures at various scales, accomplished by solving a nonlinear and nonunique inverse problem. It therefore important quantify velocity model uncertainties for accurate earthquake locations geological interpretations. Monte Carlo sampling techniques are usually this purpose, but those methods computationally intensive, especially large datasets or high-dimensional parameter spaces. In comparison, Bayesian variational inference provides...

A wide range of academic and practical applications require that we interrogate the Earth’s subsurface for answers to scientific questions. common approach is image properties using data recorded at or above surface, interpret those images address questions interest. Seismic tomograph one such method which has been used widely generate images. In order obtain robust well-justified answers, it important assess uncertainties in property estimates.To solve seismic tomographic problems...

Seismic full-waveform inversion (FWI) uses full seismic records to estimate the subsurface velocity structure. This requires a highly nonlinear and nonunique inverse problem be solved; therefore, Bayesian methods have been used quantify uncertainties in solution. Variational inference optimization efficiently provide solutions. However, previously method has only applied transmission FWI with strong prior information imposed on such as is never available practice. We found that works well...

Research Article| May 01, 2014 Surface Microseismic Monitoring of Hydraulic Fracturing a Shale‐Gas Reservoir Using Short‐Period and Broadband Seismic Sensors Xiangfang Zeng; Zeng aWantai‐BMT Lab School Earth Space Sciences, University Science Technology China, Hefei 230026, Chinazhang11@ustc.edu.cn Search for other works by this author on: GSW Google Scholar Haijiang Zhang; Zhang Xin Hua Wang; Wang Yingsheng bBeijing Miseis Technologies, Beijing 100096, China Qiang Liu Author Article...

Surface wave tomography uses measured dispersion properties of surface waves to infer the spatial distribution subsurface such as shear-wave velocities. These can be estimated vertically below any geographical location at which data are available. As inversion is significantly non-linear, Monte Carlo methods often used invert curves for velocity profiles with depth give a probabilistic solution. Such provide uncertainty information but computationally expensive. Neural network based provides...

Abstract In a variety of geoscientific applications we require 3‐D maps properties the Earth's interior and corresponding map uncertainties to assess their reliability. On seabed it is common use Scholte wave dispersion data infer these using inversion‐based imaging theory. Previously introduced fully nonlinear Monte Carlo tomography method that inverts for shear velocities directly from frequency‐dependent travel time measurements which improves accuracy results better estimates...

Seismic tomography is a methodology to image the interior of solid or fluid media, and often used map properties in subsurface Earth. In order better interpret resulting images it important assess imaging uncertainties. Since significantly nonlinear, Monte Carlo sampling methods are for this purpose, but they generally computationally intractable large datasets high-dimensional parameter spaces. To extend uncertainty analysis larger systems we use variational inference conduct seismic...

SUMMARY Seismic body wave traveltime tomography and surface dispersion have been used widely to characterize earthquakes study the subsurface structure of Earth. Since these types problem are often significantly non-linear non-unique solutions, Markov chain Monte Carlo methods find probabilistic solutions. Body data usually inverted separately produce independent velocity models. However, is generally sensitive around subvolume in which occur produces limited resolution shallower Earth,...

Abstract We present a 1D shear-velocity model for Los Humeros geothermal field (Mexico) obtained from three-component beamforming of ambient seismic noise, imaging the first time bottom sedimentary basement ∼5 km below volcanic caldera, as well brittle-ductile transition at ∼10 depth. Rayleigh-wave dispersion curves are extracted noise measurements and inverted using Markov chain Monte Carlo scheme. The resulting probability density function provides distribution down to 15 depth, hence,...

We test a fully non-linear method to solve seismic tomographic problems using data consisting of observed travel times first-arriving waves. use variational inference calculate the posterior probability distribution which describes solution Bayesian inverse problem. The is an efficient alternate Monte Carlo methods, seeks best approximation distribution. This found optimization framework, and provides probabilistic results. apply new for geophysics -- normalizing flows. models by employing...

Abstract The ultimate goal of a scientific investigation is usually to find answers specific, often low‐dimensional questions: what the size subsurface body? Does hypothesized feature exist? Existing information reviewed, an experiment designed and performed acquire new data, most likely answer estimated. Typically interpreted from geological geophysical data or models, but biased because only one particular forward function considered, inversion method applied, human interpretation process....

Nano zero-valent iron (nZVI) is a promising remediation material for Cd-contaminated soil, but questions remain regarding the effects of nZVI-induced Fe oxides on Cd availability with different soil types and moisture conditions. To identify changes in mineral phases resulting from application nZVI, three Cd-spiked soils 0.1% nZVI amendment were incubated under conditions water-holding capacities (WHCs) 30%, 60%, 180%. The was significantly decreased yellow black amended fewer being observed...

Abstract It has been a challenge to image velocity changes in real time by seismic travel tomography. If more events are included the tomographic system, inverted models do not have necessary resolution resolve changes. But if fewer used for real‐time tomography, system is less stable and model may contain some artifacts, thus, resolved be real. To mitigate these issues, we propose wavelet‐based time‐dependent double‐difference (DD) tomography method. The new method combines multiscale...