Probabilistic Magnetotelluric Inversion with Adaptive Regularisation Using the No-U-Turns Sampler

A Bayesian Inference Approach to Reduce Uncertainty in Magnetotelluric Inversion: A Synthetic Case Study

“…The generated frequency bandwidth for the magnetotelluric impedance data ranged from 0.01 to 100 Hz. A 5% Gaussian noise was added at each frequency in order to simulate errors to our synthetic results according to (Conway et al, 2018;Grandis et al, 1999;Mandolesi et al, 2018). We set the number of iterations to 30,000 times for each model.…”

“…In addition, the uncertainty of the earth model parameters can be included in the form of prior distributions and quantified by examining the posterior distribution model. Normally, this task is achieved by utilizing Markov Chain Monte Carlo (MCMC) methods (Conway, Simpson, Didana, Rugari, & Heinson, 2018). Therefore, in this paper, we present the use of stochastic method for inverting magnetotelluric data.…”

“…More recently, Conway et al (2018) successfully applied 1D MCMC magnetotelluric inversion using a No-U-Turns Sampler. However, Mandolesi, Ogaya, Campany and Piana (2018) attest that magnetotelluric data are sensitive to the depth-distribution of rock resistivity.…”

The Geosciences DeVL Experiment: new information generated from old magnetotelluric data of The University of Adelaide on the NCI High Performance Computing Platform

“…HPC can allow us to do more interesting modelling with our MT data. For example, there is growing interest in probabilistic modelling of MT data which can provide interesting ways of modelling time-lapse data (Rosas-Carbajal et al, 2015), exploration of the solution space (Rosas-Carbajal et al, 2013;Conway et al, 2018;Mandolesi et al, 2018), and the ability to more accurately model the errors in the MT response (Sielle and Visser, 2018). However, such techniques are computationally burdensome and could greatly benefit from HPC.…”

Reliable 1D magnetotelluric probabilistic inversion considering modelling assumption violations

“…In order to tackle the problem of the uncertainty assessment, the geophysical community started to make more use of probabilistic inversion methods. In particular, the reversible jump Markov-chain Monte Carlo (MCMC) sampling algorithm was recently applied to the 1D MT problem (Mandolesi et al, 2018, Xiang et al, 2018, Conway et al, 2018, Brodie and Jiang, 2018. These trans-dimensional samplers treat the model parameterisation dimension as an unknown during the inversion.…”