Overview

QtomoMCMC is a Julia-based inversion tool developed for 3D body-wave attenuation tomography, supporting two inversion modes: relative t* and absolute t*. It uses a transdimensional Markov Chain Monte Carlo (MCMC) method within a Bayesian framework.

Unlike traditional linearized tomography, QtomoMCMC provides robust uncertainty estimates and adaptive model complexity, making it a powerful tool for exploring non-unique geophysical problems. Below are key motivations for adopting this MCMC-based approach:

Why Transdimensional MCMC?

1. Uncertainty Quantification

MCMC naturally yields model uncertainty maps that highlight well-resolved versus poorly constrained regions, in contrast to traditional linear methods that often mask this variability.

2. Less Sensitivity to Human Interefence

Linearized methods are sensitive to manually chosen regularization parameters (e.g., damping, smoothing), whereas MCMC relies on prior distributions that have a milder impact on final results.

3. Flexible Model Complexity through Posterior Sampling

MCMC samples the posterior distribution, allowing representation of multiple plausible models — a significant advantage over single-solution linear inversions in underdetermined problems.

4. Data-Driven Transdimensional Parameterization

QtomoMCMC employs Voronoi cells whose number and geometry are determined by the data. This adaptive parameterization increases resolution in data-dense regions and avoids arbitrary gridding assumptions.

Files You Need to Care About

If you're just trying to run an inversion, here are the essential files to pay attention to:

File	Purpose
inp.yml	Main config file (paths, priors, iteration settings)
Data/	Directory for travel time and event/station input files
scripts/load.jl	Optional: customize loading for non-standard input formats