Trans-dimensional inversion of microtremor array dispersion data with hierarchical autoregressive error models

Abstract

S U M M A R YThis paper applies a general trans-dimensional Bayesian inference methodology and hierarchical autoregressive data-error models to the inversion of microtremor array dispersion data for shear wave velocity (v s ) structure. This approach accounts for the limited knowledge of the optimal earth model parametrization (e.g. the number of layers in the v s profile) and of the dataerror statistics in the resulting v s parameter uncertainty estimates. The assumed earth model parametrization influences estimates of parameter values and uncertainties due to different parametrizations leading to different ranges of data predictions. The support of the data for a particular model is often non-unique and several parametrizations may be supported. A transdimensional formulation accounts for this non-uniqueness by including a model-indexing parameter as an unknown so that groups of models (identified by the indexing parameter) are considered in the results. The earth model is parametrized in terms of a partition model with interfaces given over a depth-range of interest. In this work, the number of interfaces (layers) in the partition model represents the trans-dimensional model indexing.In addition, serial data-error correlations are addressed by augmenting the geophysical forward model with a hierarchical autoregressive error model that can account for a wide range of error processes with a small number of parameters. Hence, the limited knowledge about the true statistical distribution of data errors is also accounted for in the earth model parameter estimates, resulting in more realistic uncertainties and parameter values. Hierarchical autoregressive error models do not rely on point estimates of the model vector to estimate data-error statistics, and have no requirement for computing the inverse or determinant of a data-error covariance matrix. This approach is particularly useful for trans-dimensional inverse problems, as point estimates may not be representative of the state space that spans multiple subspaces of different dimensionalities. The order of the autoregressive process required to fit the data is determined here by posterior residual-sample examination and statistical tests. Inference for earth model parameters is carried out on the trans-dimensional posterior probability distribution by considering ensembles of parameter vectors. In particular, v s uncertainty estimates are obtained by marginalizing the trans-dimensional posterior distribution in terms of v s -profile marginal distributions. The methodology is applied to microtremor array dispersion data collected at two sites with significantly different geology in British Columbia, Canada. At both sites, results show excellent agreement with estimates from invasive measurements.

Bayesian inversion of marine controlled source electromagnetic data offshore Vancouver Island, Canada

“…Correlated errors can also be addressed using an hierarchical scheme based on sampling over first-order autoregressive parameters as additional unknowns (see e.g. Dettmer et al 2012;Steininger et al 2013). However, additional inversion parameters may not be constrained by the data and introduce spurious structure, and fixed data covariance matrices were found to produce more stable inversion results in this work.…”

Seabed roughness parameters from joint backscatter and reflection inversion at the Malta Plateau

“…6,17 The priors for the roughness and geoacoustic parameters consist of bounded uniform distributions constraining the parameters to physically meaningful values, as listed in Table I. The relationship between sound velocities and densities for the sediments and basement are also constrained by 2D bounds 6 derived from a large set of measurements.…”

“…Bayesian inversion has been applied widely to other geoacoustic inverse problems. [7][8][9][10][11][12][13] The inversion is conducted transdimensionally, a relatively new method that has been applied recently to several problems in geophysics [14][15][16][17][18] and geoacoustics. 19 Trans-D inversion samples over model dimension (number of unknowns) and intrinsically account for model selection uncertainty in parameter estimate uncertainty.…”

A trans-dimensional polynomial-spline parameterization for gradient-based geoacoustic inversion

“…[2][3][4][5][6][7][8] In particular, trans-D Bayesian inversion provides an effective, automated approach to model selection (e.g., determining the number of sediment layers consistent with the resolving power of the data), such that the parameter estimate uncertainties account for model selection uncertainties. [9][10][11][12][13][14][15][16] As the concept of model parsimony is linked to the choice of model parameterization type, the posterior uncertainty of parameter estimates are still affected by a priori parameterization selection decisions. This work highlights the importance of model parameterization in Bayesian inversion a) Author to whom correspondence should be addressed.…”