0% found this document useful (0 votes)
49 views5 pages

GMSYS-3D Magnetic Inversion Tutorial

This tutorial document describes how to perform a 1-step lateral susceptibility inversion in GMSYS-3D to estimate susceptibility variations in the subsurface from magnetic anomaly data. It involves loading a starting model with a laterally uniform susceptibility of 0, inverting to minimize misfit between calculated and observed magnetic fields, and examining the resulting inverted susceptibility distribution. Profiles through the calculated, observed, and inverted grids show that while the calculated field matches the general shape, it lacks some of the short-wavelength details present in the observed data.

Uploaded by

Raden Tuna
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
49 views5 pages

GMSYS-3D Magnetic Inversion Tutorial

This tutorial document describes how to perform a 1-step lateral susceptibility inversion in GMSYS-3D to estimate susceptibility variations in the subsurface from magnetic anomaly data. It involves loading a starting model with a laterally uniform susceptibility of 0, inverting to minimize misfit between calculated and observed magnetic fields, and examining the resulting inverted susceptibility distribution. Profiles through the calculated, observed, and inverted grids show that while the calculated field matches the general shape, it lacks some of the short-wavelength details present in the observed data.

Uploaded by

Raden Tuna
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

GMSYS-3D TUTORIAL II

Magnetic 1-Step Susceptibility Inversion

Introduction
This tutorial illustrates the use of 1-Step Lateral Susceptibility Inversion using a single
layer model.
The GMS3OSSI GX used in the 1-Step Susceptibility Inversion applies the Oasis
MAGMAP Apparent Susceptibility Filter (SUSC) to arrive at the lateral susceptibility
variation, rather then utilizing the normal inversion routines of GMSYS-3D. This is a
somewhat faster process than the “normal” GMSYS-3D routines which are used by the
GMS3SUSCINV GX (Menu item, “Calculate | Magnetic Susceptibility Inversion”).
Using the “Magnetic Susceptibility Inversion” and GMS3SUSCINV would follow an
analogous process, which could also be applied to the model in this tutorial. The
parameters in the GMS3SUSCINV dialog box are somewhat more involved and are
documented in the help file for that GX.

Tutorial Files
Create a new directory for this exercise (e.g., c:\temp\3d_mag_tut\). Copy all of the
tutorial files into this directory.
There are four grids (and associated .gi files):
• mag_tut_obs3.grd observed magnetic anomalies
• mag_tut_calc3.grd calculated magnetic anomalies
• mag_tut_err.grd error (mis-fit) grid
• topo_s4.grd; topography grid

three maps:
• mag_tut_obs3.map
• mag_tut_err.map
• mag_tut_calc3.map;

and a GMSYS-3D model file:


• mag_tutorial_1.g3D;

Getting Started
We will start with an initial Lateral Susceptibility Distribution (LSD) grid that is all zero
and invert for a lateral susceptibility distribution that minimizes the misfit between the
calculated and observed grids.

Magnetic Inversion Tutorial Page 1


Rev 1: 3/28/2006 12:50 AM
Remember that GMSYS-3D only uses micro-cgs units for susceptibility (i.e. cgs*106)!
For example, the average susceptibility of basic igneous rocks as reported by Dobrin is
2596 in micro-cgs units.
Create a new workspace in the new directory containing the tutorial files named
“mag_tut”. Load the GMSYS-3D menu (GX | Load menu…). Open the
“mag_tutorial_1.g3D” model (GMSYS-3D | Model Definition | Open model) and save it.
Display the observed and calculated grids (GMSYS-3D | Display | Display Mag Anomaly
Grids as 2D Maps). Note that the calculated anomaly grid is all zeros.
Open the “Model Explorer tool to view a summary of
the model. Select GMSYS-3D | Model Definition
|Model Explorer (or click on the icon). Expand the
“Layer 1” branch (click on the “+” in the Model
Explorer) to check the source of the susceptibility info
for Layer 1 (as shown in the “before” and “after”
images on the right).
Create a starting Lateral Susceptibility Distribution
(LSD) grid that is all zero. Select GMSYS-3D | New
Constant Grid (or click on the icon); name the grid
“tut_susc_inv.grd” and enter “0.0” for the Constant Grid Value.
Edit the model layers so that the first layer uses “tut_susc_inv.grd” as a LSD grid. Select
GMSYS-3D | Model Definition | Edit Model and click on the Edit Layers button (or just
click on the icon). Select the “topo_s4.grd” layer and click the Edit button. Select the
Magnetics tab and then the Lateral Susceptibility Grid radio button. Browse for the Dist
Grid and select “tut_susc_inv.grd”.
Do a forward calculation to bring the model up to date (GMSYS-3D | Calculate |
Magnetic Forward Calculation or click on the icon). The forward calculation step also
generates a spreadsheet listing the statistics of the “mag_tut_calc3.grd” (i.e., the “error
grid”). You may want to adjust the window size so that it only displays a few lines and
covers a smaller portion of the screen as shown below.

Lateral Susceptibility Inversion


Select GMSYS-3D | Calculate | 1-Step Suscept Inversion. There is no icon in the tool bar
for this function. (Note that there is another “Susceptibility Inversion” option in the
menu. It uses a different approach that works better in some situations. For this exercise
we will use the 1-Step option.)

Magnetic Inversion Tutorial Page 2


Rev 1: 3/28/2006 12:50 AM
Choose “topo_s4.grd” for the “Inversion Surface”. This selects the LSD grid below this
surface for inversion.
The inversion scheme does a downward continuation of the magnetic anomaly grid to the
“Z0” level, which in theory, is the top of the magnetic layer. If you leave this parameter
blank, the GX sets it to the maximum elevation in the structure grid. If you set this level
too deep into the layer, the inversion process tends to “blow up”. Enter “845” (the
maximum of the topo_s4.grd) for “Z0”.
Normally you will leave the Thickness Factor at 1.0. Calculated susceptibilities will be
multiplied by the thickness factor after inversion
Click on “OK” to start the inversion. If the Calculated and Error grids are currently
displayed, they will be updated at the end of the inversion process. A new entry will also
be added to the error statistics database.
(Each time you run the susceptibility inversion, it saves a backup copy of the LSD grid.
If you select the [Undo] button”, it will restore the backup LSD, Calculated, and Error
grids and refresh the maps.)
Display the LSD grid resulting from the inversion. Select Grid | Display grid | Single
grid; choose Grid name “tut_susc_inv.grd”; then click on New Map . Let’s try more
iterations of inversion to see if we can improve the fit. Note that the Standard Deviation
is less than 10 so that the convergence criteria will have to be adjusted. Select GMSYS-
3D | Calculate | 1-Step Suscept Inversion. Change the Max. Std. Error (nT) to 1.0 and
click on “OK”.
Refresh the LSD map (make it the active map and click on the “Redraw” icon. You may
be able to see a small amount of low-amplitude “ringing” in the LSD.
If you look at the statistics of the susceptibility grid, you will note that the values are both
positive and negative with a minimum value of approximately –2000 and a mean value of
approximately –800 micro-cgs. You may want to add 800 to the susceptibility grid to
make most of the susceptibility values positive. Adding a constant will not affect the
calculated magnetic anomaly from a single layer.
Select “Grid | Expressions | Remove base…” Enter “tut_susc_inv.grd” for the “Input
Grid”, “tut_susc_inv_pos.grd” for the “Output Grid”, “-800” for the “Base value to
remove”, and click on “OK”.
Now change the lateral susceptibility to “tut_susc_inv_pos.grd” in the model and do a
forward calculation to demonstrate that it does not change the calculated anomaly.

Magnetic Inversion Tutorial Page 3


Rev 1: 3/28/2006 12:50 AM
Discussion:

You can see that the calculated map looks very similar to the observed map but the error
map and the error statistics show that there are differences in detail. To illustrate those
differences, we will extract some profiles through both the calculated and observed grids.

Create a new database (Data | New


database…) named “[Link]”. Make sure
the “tut_susc_inv_pos.grd” Map is selected.
Then select Grid | Utilities | Grid
profile…Enter “mag_tut_calc3.grd”,
mag_tut_obs3.grd”, and
“tut_susc_inv_pos.grd” as grid names, 500
for the sample interval, and “digitize from
map” as the Method to set profile
coordinates. Then click “OK”. Draw a
profile that crosses all of the major features
of the map, similar to the one shown the left.
Plot profiles of the calculated magnetic
response (“G_mag_tut_cal3”) and observed magnetic anomaly (“G_mag_tut_obs3”) in
the same profile window (right click on the channel header and select Show profile).
Make sure that you set “Same scale for all profiles in panel” under Panel Options. Then
plot the inverted susceptibility (“G_tut_susc_inv_pos”) in another panel as shown in the
figure below.

Magnetic Inversion Tutorial Page 4


Rev 1: 3/28/2006 12:50 AM
Although the observed and calculated curves fit well, the calculated curve does not have
as much short-wavelength information as the observed. The feature on the right of the
profile above corresponds to the feature in the middle of the map. The general shape of
the calculated curve is correct but it lacks the short wavelength power of the observed
curve. The susceptibility distribution shown in the bottom panel demonstrates the type of
boundaries produced by the inversion. The inversion cannot generate sharp boundaries
without ringing.

Magnetic Inversion Tutorial Page 5


Rev 1: 3/28/2006 12:50 AM

Common questions

Powered by AI

When performing a 1-Step Susceptibility Inversion, several factors related to grid setup should be considered: the correct initial setup of the Lateral Susceptibility Distribution (LSD) grid, ensuring that parameters like "Z0" are appropriately set to avoid computational blow-ups, and using grids such as "topo_s4.grd" that match the inversion's requirements. Also, it's essential to correctly apply any constant values to the grid if needed to balance the susceptibility statistics, such as adding 800 to ensure positive values without affecting calculated anomalies.

The advantages of using observed magnetic anomalies for model calibration include providing a real-world benchmark against which calculated models can be compared and adjusted, allowing for a data-driven approach to model refinement. However, limitations exist in terms of capturing only certain wavelength information, which may lead to inaccuracies in reproducing shorter wavelength features that are part of the observed data but not adequately represented in the model due to limitations in the inversion processes and grid resolution. This disparity can affect the interpretation of fine-scale geological features.

Susceptibility grids generated by inversion processes struggle to produce sharp boundaries without "ringing" due to the inherent limitation of the inversion algorithms in resolving detailed features. The inversion seeks to minimize differences in the fit between calculated and observed magnetic anomalies, often leading to smoothed susceptibility distributions. Attempts to create sharp changes in susceptibility can introduce artifacts like "ringing," which are oscillations or ripples caused by the algorithm's inability to abruptly change values without mathematical overshoot.

The error grid in the susceptibility inversion process acts as a measure of the misfit between the observed and calculated magnetic anomalies. It is crucial for assessing the inversion's accuracy and guides further model adjustments by highlighting areas where prediction discrepancies exist. After each inversion iteration, the error grid gets updated, allowing the user to visualize the difference between observed data and model output and to evaluate the improvements made by adjustments or additional calculations.

The primary advantage of using the 1-Step Lateral Susceptibility Inversion in GMSYS-3D is that it is a faster process compared to the normal inversion routines used by GMSYS-3D. The 1-Step approach employs the Oasis MAGMAP Apparent Susceptibility Filter instead of the traditional inversion routines, which accelerates the process of determining lateral susceptibility variations.

Adjusting the standard deviation criteria during susceptibility inversion is meant to allow for more iterations and improve the fit between calculated and observed grids. This is achieved by lowering the standard deviation threshold, for instance, changing the Max. Std. Error to a smaller value like 1.0 nT from a higher threshold. This modification forces the algorithm to continue iterating until the error map meets the new stringent criteria, leading to a potentially better solution convergence.

The forward calculation process contributes to updating the geophysical model by recalculating the magnetic anomalies based on current susceptibility grids and model parameters. This step ensures that any changes made to the model, such as adding new grid values or adjusting layer settings, are reflected in the resulting calculated magnetic anomaly grids. Additionally, it generates a report of error statistics that helps in evaluating the accuracy of the model update.

Adding a constant to the susceptibility grid does not change the calculated magnetic anomaly from a single layer. This operation is primarily a mathematical adjustment to ensure that most of the susceptibility values are positive, but it does not influence the physical magnetic properties or the anomaly calculations since it uniformly changes the grid values without altering their relative differences.

In the 1-Step Susceptibility Inversion process, the "Z0" parameter represents the theoretical top of the magnetic layer and is determined by setting it to the maximum elevation in the structure grid ("topo_s4.grd"). This parameter is crucial because if set too deep into the layer, it can lead to divergence in the inversion process. It is significant because it affects the downward continuation of the magnetic anomaly grid, which is crucial for accurately approximating subsurface susceptibility distribution.

The tutorial suggests plotting the inverted susceptibility separately from the calculated and observed magnetic responses to distinctly analyze the susceptibility distribution and evaluate how well it matches the magnetic response profiles. This separation allows the user to specifically focus on the success and limitations of the inversion without conflating it with the uncertainties present in the observed data and calculated model. It helps in identifying areas where the inversion is producing unrealistic boundaries or artefacts, like "ringing," and where further refinement might be needed.

You might also like