Magnetic anomaly orders of magnitude greater than observed

[gabrahamastro]

Overview of Problem

I've been creating a 3D magnetic model based off those created by Bott & Tantrigoda 1987 (Figure 8). I've modelled them as cylindrical (layering cylinders 100m thick) before calculating the anomaly for each cylinder, summing these anomalies up.

However, I've found that, when computing the magnetic anomaly produced, though it has the same morphology as that observed (using data from the Great Britain Aeromagnetic Survey; in blue dashed) it is over 10 times greater in magnitude.

I believe the source of error could be from the following:

• Incorrect units for magnetisation (assumed to be $\mathrm{Am^{-1}}$)
• Incorrect units for dimensions (assumed to be metres)
• Incorrect method for summing the anomaly up (assumed to be simply summing up the values outputted by a forward modelled anomaly, using the 'TFA' forward operator)

Figures of interest

Below is a figure I produced which shows a 2D slice of the 3D model (based on Figure 8c from Bott & Tantrigoda 1987), along with the computed anomaly and observed anomaly (where the computed anomaly uses the same IGRF as that which was used to calculate the anomaly reduced to the pole, in green).

Also, here is the 3D model itself which I used, showing the xz plane and cylinders (100m thickness) stacked on top of each other (each with the anomaly individually computed).

Method

• I assigned the magnetic anomaly of $6.1 \mathrm{Am^{-1}}$ to each cylinder using pg.solver.parseMapToCellArray.
• I then computed the response using MagneticsModelling.response and summing the outcome using a simple np.sum.
• The 3D model was constructed using units in meters, quality of 34 and rotated using pg.Mesh().rotate.

The following is how the cylinders were created:

cylinder = mt.createCylinder(radius=r_x,
                                             height=spacing,
                                             pos=[centre_x,0,z_true-(spacing/2)],
                                             nSegments=25,
                                             marker=1, # defining as a different magnetisation
                                             markerPosition=[centre_x,0,z_true-(spacing/2)]) 
                # only gets up to surface (cylinder top defined)

                cylinder_mesh = mt.createMesh(cylinder,
                                              quality = 34)

where r_x is the radius, spacing is 100m, centre_x is the central x coordinate of the system, z_true is the depth of the cylinder, cylinder_mesh is what was used to compute the anomaly and marker=1 ensured that the magnetic anomaly could be parsed to the mesh.

Also, the method for noise addition follows that of this example.

• The anomaly was computed over a single line, defined in Bott & Tantrigoda 1987 as PP' (which aligned with the modelled body once this body was rotated).

Concluding Comments

I also realise this may simply be the size of anomaly which would be expected by the values I've inputted, though I'm not experienced enough to identify if this is the case.