GeoModeller is able to run forward and inverse modelling for gravity and magnetics in 2D and 3D.

That is, it is possible to either:

- Forward from a 3D grid (voxet) or 2D grid (section) to a modeled 2D grid or 1D line, respectively.
- Invert from a 2D grid (observed) or 1D line to a 3D grid or 2D grid, respectively.

### Forward modeling

In Geophysics, Forward modeling refers to the computation of the theoretical signal or response about a remotely measurable physical property for a specified body of rock.

A signal is obtained using passive survey methods while a response is obtained using active survey methods. The difference lies in the presence of a controlled human-made source used to force a response out of the surveyed area in the case of active methods. Physical properties may be measured passively or actively depending on the setting and equipment available.

For example:

- Gravity is passive
- Magnetics is passive
- Electrical is active (IP)
- Electromagnetics may be passive (MT) or active (airborne EM)
- Seismics may be passive or active (reflected)

Therefore, forward modelling requires as input,

- a 3D geological model and,
- a set of measured or estimated values of the physical properties of interest for each formation

The 3D geological model and physical properties are then combined into an indexed 3D grid with physical properties.

The physical properties 3D grid is the basis for the Forward calculation of the theoretical signal/response along with the simulated survey parameters.

The forward modelling algorithm in GeoModeller will calculate the theoretical response of the modeled area based on all aforementioned data and settings.

In that sense, forward modeling can be seen as a geophysical survey simulation.

GeoModeller offers a range of forward methods:

- 2D/3D Gravity (Bouguer, Free Air, Full Tensor, Gradiometry)
- 2D/3D Magnetics (TMI, Gradients, Full Tensor)
- 2D Seismic
- 3D Geothermal (Temperature, Heat flow)

### Inverse modeling

Geophysical inverse modeling refers to the computation of the physical properties of a specified body of rock based on observed data and, optionally, a geological model/constraints. That is, inversion is...the inverse of forward modelling.

Note that any type of inversion may only be run if an observed dataset for the corresponding property is provided

- For 2D inversions, a 1D profile will constitute the observed data.
- For 3D inversions a 2D grid will constitute the observed data.

Unlike forward modelling, inversion is non-deterministic because of mathematical ambiguities. Except for very specific cases, there is usually no analytical solutions for inversions and the process is solved numerically.

More specifically,

- Starting from a "blank" physical properties grid, many forward models are built in successive iterations and compared to the observed data and geological model/constraints
- The process measures misfit (how much a given forwarded model fits the data and constraints) and favors low misfit models over high misfit ones
- Every iteration the inversion algorithm attempts to produce a model with a lower misfit until a stable state is attained. A stable state is reached when the misfit does not decrease anymore with successive iterations

GeoModeller offers two inverse methods:

- 3D Gravity (Bouguer, Free Air, Full Tensor, Gradiometry)
- 3D Magnetics (TMI, Gradients, Full Tensor)