A hybrid residual neural network–Monte Carlo approach to invert surface wave dispersion data

Aleardi, Mattia; Stucchi, E.

doi:10.1002/nsg.12163

Cited by 14 publications

(19 citation statements)

References 65 publications

(83 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As an alternative, Cao et al (2020) proposed to split the velocity model into several depth intervals and determine the averaged V s values for each interval from DCs data by mixed density neural networks (MDN). Aleardi and Stucchi (2021) train a residual neural network (ResNet) to map the spectral dispersion image of the surface wave into S-wave velocity and layer thicknesses. The advantages of using artificial neural networks (ANN) are higher computational efficiency without a need to adjust optimization parameters and the lack of necessity to include any model constraint into the error function, unlike global optimization methods (Yablokov and Serdyukov, 2020;Aleardi and Stucchi, 2021;Yablokov et al, 2021).…”

Section: Introductionmentioning

confidence: 99%

“…Aleardi et al (2020) performed a rigorous study and comparison of transdimensional and reversible-jump MCMC inversions of Rayleigh-wave DCs. Aleardi and Stucchi (2021) state MCMC methods are computationally expensive due to the huge number of samples needed to attain stable PPDs. For this reason, the MCMC approach becomes computationally impractical for inverting big data.…”

Section: Introductionmentioning

confidence: 99%

“…For this reason, the MCMC approach becomes computationally impractical for inverting big data. Aleardi and Stucchi (2021) considered the machine learning based inversion of a full dispersion images and suggested a Monte Carlo simulation strategy that propagates onto the model space the uncertainties related to noise in the data and the modelling error introduced by the network approximation.…”

Section: Introductionmentioning

confidence: 99%

“…Zhang and Chan (2003) state that mode-misidentification errors on the experimental DCs are often greater than errors resulting from inaccurate data for a given mode. An inversion of the full dispersion image of the surface wave (Aleardi and Stucchi, 2021) seems a reasonable alternative to the multiple-mode inversion. This approach is based on a full surface wavefield modelling, which is challenging due to the need to take into account attenuation, scattering of the surface waves, the presence of other types of waves and unknown source wavelet as well.…”

Section: Introductionmentioning

confidence: 99%

“…The pre-trained ANN enables one to project dispersion data noise and inverse operator errors onto a space of the velocity model parameters. Thus, we adapt the Monte-Carlo simulation strategy applied by Aleardi and Stucchi (2021) to estimate uncertainties of our ANN multiple DCs inversion method.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Uncertainty quantification of multi-modal surface wave inversion using artificial neural networks

Yablokov¹,

Lugovtsova²,

Serdyukov³

2022

Preprint

View full text Add to dashboard Cite

An inversion of surface waves dispersion curves is a non-unique and ill-conditioned problem. The inversion result has a probabilistic nature, what becomes apparent when simultaneously restoring the shear wave (S-wave) velocity and layer thickness. Therefore, the problem of uncertainty quantification is relevant. Existing methods through deterministic or global optimization approaches of uncertainty quantification via posterior probability density (PPD) of the model parameters are not computationally efficient since they demand multiple solutions of the inverse problem. We present an alternative method based on a multi-layer fully connected artificial neural network (ANN). We improve the previous approach from uni-modal to multi-modal inversion and use the existing algorithm to determine optimal parametrization (number of layers) and the ranges of possible model parameters. We uniformly draw training data sets within estimated ranges and train the ANN. Saved ANN's weights map the phase velocity dispersion curves to values of the S-wave velocity and layers thickness. To estimate the uncertainties, we adapt the Monte-Carlo simulation strategy and project onto the resulting velocity model both frequency-dependent data noise and inverse operator errors, which are evaluated by the prediction of the training data set. We first test our approach on synthetic experiments for various velocity models: a positive velocity gradient, a low-velocity layer and a high-velocity layer. This is done considering uni-modal inversion at first and then compared to the multi-modal inversion. Afterwards, we apply our approach to field data and compare resulting models with the body S-wave processing by the generalized reciprocal method. The experiments show high-potential results - using ANN yields the possibility to accurately estimate PPD of restored model parameters without a significant computational effort. The PPD-based comparison demonstrates advantages of a multi-modal inversion over uni-modal inversion. The trained ANN provides reasonable model parameters predictions and related uncertainties in real-time.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Uncertainty quantification of multi-modal surface wave inversion using artificial neural networks

Yablokov¹,

Lugovtsova²,

Serdyukov³

2022

Preprint

View full text Add to dashboard Cite

show abstract

Application of a global–local full‐waveform inversion of Rayleigh wave to estimate the near‐surface shear wave velocity model

Lamuraglia

Stucchi

Aleardi

2022

Near Surface Geophysics

Self Cite

View full text Add to dashboard Cite

We present a global–local full‐waveform inversion (FWI) of surface waves to estimate a high‐resolution shear wave velocity model of the near‐surface at the site of Grenoble (France). The seismic data we use have been acquired in the framework of the InterPACIFIC project. A first attempt is made by employing the multichannel analysis of surface waves (MASW) to invert the fundamental modes of Rayleigh waves. This inversion is solved through a global optimization driven by the neighbourhood algorithm. However, the limiting 1D assumption, together with the severely ill‐posedness of the MASW inversion motivate us to exploit all the information content of the recorded Rayleigh waves (i.e., travel‐times and amplitudes) by implementing a global–local FWI approach. We first perform a global FWI in which a genetic algorithm (GA) is used to minimize the L2 norm difference between observed and simulated seismic waveforms up to 30 Hz. This inversion employs a two‐grid scheme in which the subsurface is described by a fine grid input to the forward modelling, whereas the unknowns are defined on a coarse grid. A bilinear interpolation brings the coarse‐grid model into the fine grid. To accelerate the convergence, two strategies are considered; offset marching and frequency marching. Then, the long‐wavelength velocity profile provided by the GA inversion is used as the starting point for a local FWI for further refinement of the predicted model. In this case, we employ a frequency‐marching strategy up to a maximum frequency of 60 Hz. Both the global and local approaches use a finite difference code for the forward computation. Despite the limited a priori information infused into the inversion framework, the presented global–local FWI scheme provides reliable results as demonstrated by the good match between recorded and predicted seismograms and by the agreement of the estimated velocity with both nearby borehole data and stratigraphic information on the study site. Our global–local strategy also achieves better data fitting and improved matching with the available borehole data with respect to the velocity profile estimated when the local FWI is started from the MASW predictions.

show abstract

Prediction of S‐wave velocity models from surface waves using deep learning

Cho,

Pyun,

Choi

et al. 2023

Near Surface Geophysics

View full text Add to dashboard Cite

Surface wave (SW) methods extract dispersion properties of wavefields propagating through a seismic array (1D or 2D). This is achieved by analysing the phase velocity versus frequency (or wavelength) data. Afterwards, an inversion process is performed to construct near‐surface S‐wave velocity models. Among the SW methods, multichannel analysis of SWs (MASW) is commonly used for engineering applications, analysing dispersion characteristics by generating a dispersion image. However, classical MASW depends on the manual picking of dispersion curves, which can lead to subjective outcomes and require time and effort to obtain precise results. To avoid these pitfalls, many studies, including deep‐learning techniques, have focused on automating the process. Similarly, we propose a deep‐learning‐based algorithm that estimates the S‐wave velocity directly from the dispersion image of SWs. This algorithm consists of a convolutional neural network (CNN) designed to directly yield S‐wave velocity profiles and a fully connected network (multi‐layer perceptron) added to regularize predictions. Unlike typical SW techniques, the proposed approach does not incorporate prior information such as layer count and thickness. To ensure successful training, we modified the loss function to exploit the normalized mean squared error. The training dataset was generated by modelling synthetic shot gathers and transforming them into dispersion images for various 1D stratified velocity structures. After a sample is fed to the CNN network for inversion, the inversion network's output subsequently goes through an additional simple neural network (NN) to regularize the predicted S‐wave velocity model (which is the post‐processing step). The combined usage of deep‐learning‐based SW inversion with NN‐based post‐processing was assessed using test data. The proposed algorithm achieved an average relative error of approximately 7.49% in predicting the S‐wave velocity and was successfully applied to the field data. Additionally, we discuss its performance on noisy data as well as its applicability to out‐of‐training data. Numerical examples demonstrated that the proposed method is robust to noise, whereas it requires additional training to handle data beyond the distribution of the training data.

show abstract

A hybrid residual neural network–Monte Carlo approach to invert surface wave dispersion data

Cited by 14 publications

References 65 publications

Uncertainty quantification of multi-modal surface wave inversion using artificial neural networks

Uncertainty quantification of multi-modal surface wave inversion using artificial neural networks

Application of a global–local full‐waveform inversion of Rayleigh wave to estimate the near‐surface shear wave velocity model

Prediction of S‐wave velocity models from surface waves using deep learning

Contact Info

Product

Resources

About