MTGeophysics.jl

Magnetotelluric modelling, inversion, and visualization workflows in Julia.

Documentation | Issues | Releases

---

Features

• 1D and 2D MT forward solvers (analytical and finite-difference)
• 2D/3D VFSA inversion with ensemble uncertainty quantification
• ModEM 3D model and data I/O
• Interactive 3D slice viewers (GLMakie) with GIS overlays and coordinate reprojection
• Shapefile export for GIS integration

Installation

julia --project=. -e 'using Pkg; Pkg.instantiate()'

Quick start: 2D benchmark + VFSA inversion

Note: This is a quick demonstration workflow. For production-quality inversion, tune key settings such as the number of VFSA iterations, number of chains, cooling schedule, regularization choices, and uncertainty/ensemble controls for your survey and model size.

1. Generate the COMMEMI 2D benchmarks. Writes the true model, halfspace starting model, reference data template, and noisy observed data into examples/0COMEMI2D-I/, -II/, and -III/.

julia --project=. helpers/benchmarks_2d.jl

2. (Optional) Inspect the 2D forward response of the benchmark model.

julia --project=. examples/response_2d.jl

3. Run the 2D VFSA inversion (multi-chain ensemble, RBF parameterization) on the COMMEMI-I benchmark. Results are written to a timestamped examples/run_VFSA2DMT_<timestamp>/ directory containing per-chain logs, best models, and ensemble mean/median/std.

julia --project=. examples/run_vfsa2dmt.jl

4. Compute ensemble statistics (posterior mean/median/std and misfit summaries).

julia --project=. helpers/run_statistics_2d.jl examples/run_VFSA2DMT_<timestamp>

5. Create a convergence GIF to visualize model evolution across iterations (requires keep_models = true).

julia --project=. helpers/make_gif_2d.jl examples/run_VFSA2DMT_<timestamp>

From Julia, the inversion can also be driven directly:

using MTGeophysics

result = VFSA2DMT(
    VFSA2DMTParams(
        script_path      = @__FILE__,
        start_model_path = "examples/0COMEMI2D-I/Comemi2D1.ini",
        data_path        = "examples/0COMEMI2D-I/Comemi2D1.obs",
        config = VFSA2DMTConfig(
            n_chains    = 2,
            n_ctrl      = 400,
            max_iter    = 3000,
            n_trials    = 4,
            log_bounds  = (0.0, 4.0),
            seed        = 20260308,
            keep_models = true,
        ),
    ),
)

See the documentation for the full 2D/3D workflow, configuration options, and visualization examples.

Research using this code

• Mishra, P. K., Kamm, J., Patzer, C., Autio, U., and Sen, M. K.: Building uncertainty-aware subsurface models with 3D magnetotelluric inversion, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4367, https://doi.org/10.5194/egusphere-egu26-4367, 2026.

• Mishra, P. K.: MTGeophysics.jl: A software repository for magnetotelluric research and application, 27th Electromagnetic Induction Workshop (EMIW 2026), St. John's, Newfoundland and Labrador, Canada, 2026.

• Mishra, P. K., Kamm, J., Patzer, C., Autio, U., Xiao, L., and Sen, M. K.: Stochastic model exploration in three-dimensional inversion of magnetotelluric data, 27th Electromagnetic Induction Workshop (EMIW 2026), St. John's, Newfoundland and Labrador, Canada, 2026.

• Mishra, P. K.: MTGeophysics.jl: A software repository for magnetotelluric research and application, JuliaCon 2026.

• Patzer, C., Mishra, P. K., and Kamm, J.: Studying the Wiborg Rapakivi Batholith in SE Finland, 27th Electromagnetic Induction Workshop (EMIW 2026), St. John's, Newfoundland and Labrador, Canada, 2026.