moombarriga-geoscience

The Power of Properly Done MT

Wed, 17 Dec 2025 00:59:22 GMT

The inversion of magnetotelluric (MT) data is as much an art form as it is a science. Not only does a full 3D MT inversion (which is what everyone should be doing these days!) take a lot of computational resources (we're talking supercomputers here), but it also demands a high degree of expertise from the professional performing the inversion. To get good inversion results, you really need someone who has access to good inversion tools and who really knows how to use them. The black-box approach doesn't work here!

When MT is done well, the results can be spectacularly valuable to a project. When MT is done poorly, you can end up feeling like you dropped a big chunk of cash for nothing. The difference between success and failure here depends highly on the expertise of your contractor.

So, to illustrate this point, let's take a look at an example of well-done MT.

Geologic cross section through the Nifty Syncline and the Nifty deposit. From Maidment et al., 2017.

The Nifty Deposit

For this example, we're going to look at an MT dataset that was collected adjacent to the Nifty deposit, in Western Australia.

Nifty is a sediment-hosted copper deposit within the limb of a southeast-plunging syncline. The mine has largely tapped supergene and hypogene enrichment zones, but primary copper mineralization is stratabound chalcopyrite. The host rock for this mineralization is the Nifty member of the Broadhurst Formation, part of a sequence of Neoproterozoic low-grade metasedimentary rocks. The figure below shows a cross section through the Nifty Syncline and the Nifty deposit.

The Broadhurst Formation can be generally divided into three broad sub-units: the lower and upper portions largely consist of carbonaceous and sulfidic pelitic rocks with interbedded carbonates, and the middle portion broadly consists of psammitic rocks (Porter, 2017). The Nifty member, which hosts the Nifty deposit, is in the lower part of the upper pelitic sub-unit.

Due to their carbonaceous and sulfidic nature, these upper and lower sub-units are highly electrically conductive. Geophysically, they contrast strongly with the electrically resistive psammitic rocks that constitute the middle sub-unit of the Broadhurst Formation and the underlying Coolbro Formation.

So, let's say you want to look for additional mineralized zones within the Nifty Syncline, away from the Nifty deposit. To do this, you would most likely want to map the structure and extent of the syncline. Fortunately, the electrically conductive units within the Nifty Syncline make excellent geophysical marker horizons for accomplishing these purposes. The Syncline, however, is covered by conductive overburden, and it also extends deep (>1 km). Due to these factors, MT is an excellent choice for geophysical imaging of the Nifty Syncline.

The Nifty Dataset

The map below shows the MT dataset that we collected over the Nifty Syncline, to the northwest of the Nifty deposit itself. These data were collected in 2014 with Phoenix MT instruments.

Inversions Done Well

We've performed 3D inversion of the Nifty dataset using CGG's Geotools platform and their RLM3D inversion code. Images of our inversion results are shown below. As you can see, the Nifty Syncline is imaged exceptionally well. The conductivity images show the textbook map pattern of a southeast-plunging syncline.

Clearly, there is exploration value here: we can see the structure of the Nifty Syncline, as defined by the conductive marker beds, and we can trace out those key markers. The ore-bearing Nifty member occurs in the lower part of the upper pelitic sub-unit of the Broadhurst Formation, and we can see spatially where that is located in these conductivity images.

AMT/MT sites (white inverted triangles) collected by Moombarriga Geoscience to the northwest of the Nifty deposit, over the Nifty Syncline.

Slice at an elevation of 50 m above sea level (~250 m below the surface) through the electrical conductivity model developed for the Nifty dataset. Note that warm colors denote low resistivity (high conductivity) values.

Cross section through the electrical conductivity model developed for the Nifty dataset. Cross section location is shown with the white "XS" line in the previous figure.

For those of you curious about data misfit, we can assure you that this conductivity models fits and explains the data quite well!

Normalized root-mean-square (nRMSE) site-by-site misfit for the conductivity images shown above.

Now, the value delivered with these images was unlocked because we know how to get good inversion results from the tools we have on-hand. Improperly setting even one inversion parameter can result in complete destruction of that value. This is demonstrated by the image below, which was derived by slightly changing one set of regularization parameters in the inversion.

Depth slice (at 50 m above sea level / ~250 m below the surface) through a "bad" conductivity model derived from the Nifty dataset; the only difference from the depth slice above is a change in regularization.

About Us

Moombarriga Geoscience is a full-service MT and DC/IP services company. We collect and process data, and we perform inversions and interpretations. Contact us today to discuss your geophysical needs!

References

D.W. Maidment, D.L. Huston, and T. Beardsmore (2017). Paterson Orogen geology and metallogeny. In G.N. Phillips (ed), Australian Ore Deposits, The Australasian Institute of Mining and Metallurgy, Monograph 32.

T.M. Porter (2017). Nifty and Maroochydore copper deposits. In G.N. Phillips (ed), Australian Ore Deposits, The Australasian Institute of Mining and Metallurgy, Monograph 32.

2D versus 3D Magnetotelluric Imaging

Fri, 16 May 2025 01:07:50 GMT

Geology & Geophysics

Magnetotelluric (MT) imaging has come a long way in the last 10 years. A decade ago, you'd have to leave a 2D inversion running on your desktop computer over night; now, it'll run on modern laptops in just 15 minutes. Similarly, a decade ago, 3D inversions were barely possible due to computational resource limitations; now, they're the industry standard.

Until somewhat recently, most MT studies (academic as well as industrial) were performed in 2D, due to the aforementioned computational limitations. 2D inversion of MT data is certainly a useful approximation in some situations; however, 3D MT inversion now makes it possible to extract more information from the data and to gain much more robust insights into what the data are telling us about subsurface conductivity structure.

In this article, we're going to take a look at some comparisons between 2D and 3D inversion of the same data. As you'll see, sometimes the 2D and 3D inversion results are similar, but sometimes they're very different, with implications for the questions you're trying to answer.

Dataset and Methodology. The dataset we're using here is shown in Figure 1 below. We inverted the full dataset in a fully 3D inversion, and we also selected four profiles of sites with which we performed 2D profile inversions and individual 3D inversions just of the profile sites. For each of the four profiles, this gives us three images to compare: the corresponding slice through the full 3D model, the slice through the 3D inversion using only the profile data, and the 2D inversion for the profile.

We won't go into the details of our inversion methodology here, but suffice to say that our approach for 2D inversions is similar to what has often been done (and still is being done to some extent) in industrial MT applications.

2D vs 3D Comparisons

Figure 1 shows a depth slice through the full 3D conductivity model derived from the full dataset. As you can see, there are several major conductivity lineaments, and we chose our four data profiles to cross these lineaments.

Figure 1: Depth slice through the full 3D conductivity model. Inverted black triangles denote MT sites used in the inversion. White lines denote the four profiles shown below. White halos denote sites used along the profiles.

Figures 2 through 5 below show the comparisons of the three different inversion images (full 3D, profile 3D, and 2D) along each profile.

Figure 2: Comparison of three inversion images along profile A-A'.

The images along profile A-A' (Fig. 2, above) show major differences between the 2D and 3D imaging results. The full 3D and 3D profile inversions show broadly similar structures, with a resistive upper crust (<5 km depth) overlying a conductive mid crust (>5 km depth) and an isolated shallow (~2 km depth) east-dipping conductor at the right (east) end of the profile. In contrast, the 2D image shows moderately conductive channels in the upper crust (<5 km depth), and the isolated shallow conductor becomes west-dipping. These differences demonstrate the complexity of structure along this profile.

Figure 3: Comparisons of three inversion images along profile B-B'.

The images along profile B-B' (Fig. 3, above) are broadly similar between different inverse solutions, but neither the 2D inversion nor the 3D profile inversion captures the details of conductors imaged in the full 3D inversion. The 2D inversion also generally makes the shallow crust (<5 km depth) more conductive than the two 3D inversions, with an extra weakly conductive feature midway along the profile.

Figure 4: Comparisons of three inversion images along profile C-C'.

The images along profile C-C' (Fig. 4, above) are again broadly similar between inversions, but structural details are often markedly different in each solution. For example, all three solutions capture the shallow (~2 km) conductor on the right (northeast) side of the profile, but each solution shows a different structural attitude of that conductor to the southwest. The highly conductive mid crustal conductor (>10 km depth) is also imaged differently in the 2D solution compared to two 3D solutions.

Figure 5: Comparisons of three inversion images along profile D-D'.

Finally, the images along profile D-D' are yet again broadly similar, but in detail they differ markedly. Notably, the full 3D and 3D profile inversions resolve the two separate deep (>5 km depth) conductors, whereas the 2D inversion only resolves a laterally continuous dipping conductive zone at the corresponding depths. The 3D inversions are revealing more structural information here.

Conclusions

As you can see, sometimes the 2D and 3D inversion results would lead you to similar interpretations. However, sometimes the 2D inversion results show features that are not supported by the 3D inversions (e.g., Fig. 2), and the 3D images reveal important information about our view of the subsurface.

Although 2D inversions are very useful in certain situations, we consider the 3D images to be a more robust, information-rich depiction of subsurface conductivity structure. The 2D inversions are not necessarily "wrong", but they are constrained by assumptions of 2D structure. In contrast, 3D inversions permit more degrees of freedom to, for example, place localized structures off-axis of a profile of sites, rather than directly along the profile as with 2D inversions. The 3D images are consequently able to more readily honor the requirements of the data and thereby yield different, potentially more reliable insights into subsurface structure.

Ideally, for resource exploration work, you would have access to an array of sites across a project area that could be used for a fully 3D inversion of the data. However, if you've only got MT sites along a profile, then even just running a 3D inversion on that data profile can provide new insights compared to a 2D inversion (e.g., Figs. 2, 5). The 3D profile inversion is not necessarily the most "correct" (e.g., Figs. 3, 4), but it nevertheless yields more information than 2D inversions alone. Comparing the 3D profile image to the 2D image can show you, for example, which structures are reliable and which structures you'd want to constrain with more data before attempting to drill.

If you're working with legacy MT data, these will most likely have only been inverted in 2D (or even just 1D), since that's all that could be done when they were originally collected. Today, it's trivial to re-invert legacy MT data using 3D inversion tools, and doing so can reveal much more robust insights into subsurface structure than the original 2D (or 1D) images alone.

About Us

Moombarriga Geoscience is a full-service MT and DC/IP services company. We collect and process data, and we perform inversions and interpretations -- including 3D re-inversions of legacy MT data that may have previously only been analyzed under 2D assumptions. We also provide off-the-shelf 3D re-inversion data products of legacy government geophysical reports, ready for incorporation into modern exploration programs.