Geology Differentiation

What is geology differentiation?

Geology differentiation takes (usually multiple) physical property models recovered from geophysical inversions as input and classifies the physical property values into distinct classes, each of which is characterized by a unique distribution of physical property values, and can be regarded as a representation or approximation of the geology in a study area. The output of geology differentiation is a 3D quasi-geology model that shows the spatial distribution of different geological units.

The following figure from Li et al. (2019, Geology differentiation: A new frontier in quantitative geophysical interpretation in mineral exploration) summarizes the general workflow of geology differentiation.

A general workflow for geology differentiation.

Why is it important?

As can be seen from the above figure, geology differentiation takes geophysics one step further toward our ultimate goal of characterizing and understanding the geology. Traditionally, geophysicists would typically call it a day and be happy when they obtain a density model, a susceptibility model, or a resistivity model from their inversions. These models would then be delivered to geologists for subsequent interpretations. It is, therefore, up to geologists to make some geological sense of the anomalous features in the physical property models. It is also geologists' job to integrate other prior information (such as prior geological knowledge, petrophysical data, etc) into their interpretation and the final decision making.

There are several problems with this approach. First, some of practical insights into the quality of the inverted models might not be effectively communicated from geophysicists to geologists. It is well known to geophysicists that every inverted model has Uncertainties and not every feature present in an inverted model is trustworthy. Some of them are simply artefacts from regularization, noise, lack of data, lack of illumination, inherent ambiguity from physics, etc. However, this is less well known to geologists who might treat all the anomalous features equally valid and interpret them that way. Geophysicists need to work more closely with geologists, after inversions are completed, to make sure that the geophysical models are properly interpreted. Some of the geological inputs might also be useful to further constrain the geophysical inversions.

Secondly, if geophysicists simply define their job as to produce some physical property models from inversions, "our profession (i.e., geophysics) may risk becoming relegated to the equivalent of imaging technologists and geophysicists becoming the ultimate technicians", as pointed out by Li et al. (2019, Geology differentiation: A new frontier in quantitative geophysical interpretation in mineral exploration),

Therefore, geology differentiation is important because:

It serves as a platform for promoting and supporting greater levels of interaction between geophysicists and geologists. Inputs from geophysicists ensure that geophysical models are properly interpreted with their uncertainties taken into account. Inputs from geologists would allow geophysicists to further refine their models that are more consistent with existing geological data and expert geologists' perceptions.
It maximizes the value of geophysics and geophysicists! Geophysicists could be replaced by AI if their job is only to produce models from data. In the foreseeable future, AI would produce better models than geophysicists. Geophysicists need to think about how their work can be made more relevant to, say deposit discoveries, now! Geology differentiation is one way of elevating the value and relevance of geophysicists!

Some of my work on geology differentiation

Geology differentiation has been an important component of my research work. Below I provide an example of my recent work on geology differentiation. This work has been published in Interpretation as an open-access article (so anyone can download it!). You can access this article from this webpage.

Figure 1: (a) The airborne gravity gradiometry data (Gzz) after terrain correction using a density value of 2.4 g/cm3. (b) The airborne magnetic TMI data after regional field removal. Data were collected over the northeast Iowa and southeast Minnesota border, and are publicly available at this USGS webpage. (Sun et al., 2020)

We have performed a separate 3D inversion of Gzz data and a separate 3D inversion of magnetic data and ended with a density model and a susceptibility model. We have also carried out a Joint inversion of both data sets and ended up with another set of jointly inverted density and susceptibility models. Therefore, we have obtained two pairs of density and susceptibility models, one from separate inversions and the other from joint inversion. Figures 2 and 3 below show the separately inverted physical property models in 3D.

Figure 2: The recovered density contrast model from a separate inversion of Gzz data shown in Figure 1(a).

Figure 3: The recovered susceptibility model from a separate inversion of magnetic data shown in Figure 1(b).

Based on the recovered physical property models obtained above, we have cross-plotted the inverted values and performed a classification based on (1) the natural groupings observed in the crossplot, (2) available geology information and (3) prior work by Drenth et al. (2015).

Figure 4: Classification results on top of the scatterplot of separately inverted density contrast and susceptibility values. Each black point corresponds to one pair of density contrast and susceptibility values at one model cell overlain by the classification (indicated by the polygons of different colors).

Figure 5: Differentiation results based on jointly inverted density contrast and susceptibility values.

With the above classification results in crossplots, we can now map them back to 3D spatial domain and visualize them in 3D. The third column in Figure 6 shows our final quasi-geology model in 3D (based on separate inversions). The different colors represent different geological units that we have identified based on geophysical inversions, prior geology information and previous researchers' work. The 3D quasi-geology model shows the spatial distribution of various geological units and is a better representation of the subsurface geology than the geophysical models shown in the first and second columns in Figure 6.

Figure 6: A 3D separately inverted susceptibility, density contrast, and quasi-geology model of the study area at (a) 180 m, (b) 1250 m, showing the high-susceptibility anomaly of class 3 by the white ellipse, (c) 4500 m, showing the high-susceptibility anomaly of class 7 by the pink ellipse, and (d) 6700 depths.

In our work (Sun et al., 2020), we have also investigated what additional information joint inversion can provide. Figure 7 summarizes the new geological units that we have managed to identify based on the jointly inverted density and susceptibility models.

Figure 7: A 3D view of the newly identified subclasses based on joint inversion results.

In his Master's thesis, my student Jae Deok Kim took a similar approach but extended it to a much larger regional scale mineral exploration project. He inverted airborne gravity and magnetic data over an area of 46,500 square km in British Columbia, built a 3D quasi-geology model and identified a few prospective exploration regions. His thesis can be accessed here via Texas Digital Library.

Page updated

Google Sites

Report abuse