Erik Ronald, PG
Mining Geology HQ
This installment of the Rules of Thumb (RoT) series is focused on geological modeling. I’d like to share a list of what I feel are fundamental considerations for the successful creation of a robust and useful geological model. There is a wonderful quote from the statistician George Box: “For such a model there is no need to ask the question ‘Is the model true?’ If ‘truth’ is to be the ‘whole truth’ the answer must be ‘No’. The only question of interest is ‘Is the model illuminating and useful?’” (Box, 1979). It is sometimes shortened to “All models are wrong, some are useful”. This insightful quote should be stamped onto the cover of all geological model reports as a reminder that when interpreting and creating a geological model, it is not about getting it “right”, it is about generating a fit-for-purpose representation which respects the data and is useful to whomever uses it.
This article is not meant to be an exhaustive list of hard rules on the subject. It should be noted that these rules of thumb hold true regardless of whether you use implicit, explicit, or other modeling techniques available. The list is compiled from my personal experiences modeling a variety of commodities in the western U.S., central Mexico, Canada, and Australia.
Let us start with a basic definition of a geological model. For the purposes of this article, it is a computer based three-dimensional (sometimes 2D) wireframe model that is the culmination of interpreted geoscientific data for a particular area of interest or deposit. The article is only focused on the geological model and not the Resource block model that includes estimated grade, tonnage, and other attributes. There are many types of geological models but this article is concerned with those of economic geology consequence.
1. Answer the question: What will the model be used for?
Surprisingly, many geologists can’t answer this question. A geological model is created for a specific purpose and it is critical to have a clear understanding of what that purpose is prior to interpreting and modeling geology. For example, a model used for early exploration drill planning may only require basic outlines of alteration, mineralization, and structure. However, a geological model to be used for metallurgical feasibility and plant/mill development will require greater detail and data. That will include domained units based on petrography, comminution tests, chemistry, predicted recovery, and deleterious attributes, and may not resemble geologic units in the traditional sense.
2. Understand the deposit/area geology
This may sound like a no-brainer but it has to be high on this list. The person or team modeling the deposit should be an experienced and competent geologist who is knowledgeable about site geology. Unfortunately, there are occasions when companies employ an individual who doesn’t understand or have experience with a particular deposit style or site geology. These instances rarely end well.
A model must adhere to the geochronologic sequence of ore body genesis and make geological sense. The relationships of units, structures, alteration, and other features must be modeled in the same fashion while respecting observed cross-cutting relationships.
3. Incorporate all (trusted) data
There is a disturbing trend in the industry where geologists rely solely on drilling data to perform geological modeling. With the advent of implicit modeling, a novice can rapidly take any drill database and generate shiny 3D shapes that are rubbish. If one recalls their university training, creating a cross-section is fundamentally the same as performing 3D geological modeling except this time you’re likely using complex and expensive software instead of your trusty colored pencils and India ink.
A geologic model works best when it is well-informed. This statement means the model respects all existing data sets including surface mapping, geomorphic analyses, geophysics, drilling data (resource, blast hole, exploration, or even water wells), open cut or underground information, road cuts, trenches, and anything else you trust. Historic datasets can be fraught with inconsistencies or inaccuracies, so it is important to have confidence in the data, but even untrustworthy data can be insightful to the experienced geologist.
4. Get to know your fundamental data
Whether you have the luxury of a large, multi-sourced data set or a small number of surface grab samples, be sure you have a fundamental understanding of the data. This involves performing exploratory data analysis (EDA) on all categorical and numerical data to understand and test appropriate groupings, relationships, and domains of data. Knowing the data in combination with understanding the business decisions that will be made from the model (point #1) will allow the modeler to best determine the appropriate amount of data grouping or splitting required. These can be rock types, physical characteristics, chemical assays, mineralogical groups, or a myriad of other data traits.
5. Think regionally first, then model down to deposit scale
Geological models should always be constructed from a wide regional perspective first, then modified and interpreted down to smaller areas of interest or deposit-scale. All too often, geologists (especially mine geologists) interpret features or structures within an ore body incorrectly due to local-scale complexity while not realizing the regional tectonics or macro-scale setting. For example, I was in a highly altered and deformed skarn ore body that contained high concentrations of talc and chlorite that resulted in extensive soft-sediment deformation from a large regional structure. Pit mapping was highly complex, with structures discontinuous from one bench to the next. The whole thing would do one’s head in trying to figure out what was going on. The structural controls on mineralization could actually be understood easily once you got out of the pit and grasped the regional structure. As the regional structure entered the softer lithology within the deposit, the ore body behaved like faulted toothpaste and there was little hope of determining what was happening. Fortunately, the micro-scale structure was irrelevant to the deposit economics and therefore irrelevant to the model.
6. Start with a structural “skeleton”
Keeping with the structural theme, it is advised to start any geological model by understanding the structural regime and modeling a structural skeleton of the major faults and fold hinges prior to working out the lithology, stratigraphy, alteration, and the rest. The reality of lacking quality structural data may make this step easier said than done, but constructing the structure skeleton first will help with interpretations and produce a more fundamentally sound model. Additionally, relationships of alteration and mineralization usually become quite clear and even predictable once you have a grasp of the structure.
7. Keep it as simple as it needs to be
There is a quote attributed to Einstein that goes “If you can’t explain it simply, you don’t understand it well enough”. A similar philosophy should be taken when undertaking geological modeling. The simplest explanation of geologic complexity is usually more correct than an overly sophisticated and complex interpretation. Additionally, I’d refer back to point #1 that a model must be fit-for-purpose. For example, there may be second order parasitic folding present in an ore body that is geologically interesting but irrelevant for bench-scale mining. In cases like this, keep it simple and unless it affects the mine plan or the business, it should not be modeled.
8. Don’t forget the waste!
Often times in exploration and mining, geologists are overly focused on the ore body and fail to dedicate sufficient time to characterize or even understand major waste units. A wise mining engineer once told me “we mine a lot more waste than ore so you should probably understand that s#*t as well as the ore”. Compounding this problem is the fact that drilling and evaluation work is usually focused on the ore body which can result in little waste characterization or overly clustered data. In these cases, sometimes those “dry” holes are extremely valuable in working out stratigraphy and lithology.
9. Reconcile your shapes in 3D
The last rule of thumb I’ll present here is to simply ensure all geological wireframes make sense regardless of which direction they are cut for cross-sections. All too often in elongated ore bodies, the modeler will use a sectional interpretation but fail to go back and check the geological wireframes in long-section and plan view. The whole point of a geologic model is to create a 3D representation of a deposit or area thus a good test is to pick random orientations for cross-sections and see if they still make geological sense. Thankfully, some software packages such as Leapfrog and GoCAD have come a long way in ensuring wireframes are truly 3D and not just 2D polygons extended in a third dimension.
That’s it for now but I believe that following these simple rules of thumb will greatly help geological modelers ensure their 3D models are useful, make geologic sense, and are applicable in exploration or mining projects. I’d enjoy hearing from practicing geologists on their thoughts, what I may have missed, or additional “rules”. Be sure to check out the other articles in the ROT series including Mineral Exploration success and Geological field mapping.
Box, G. E. P. (1979), Robustness in the strategy of scientific model building, in Launer, R. L. & Wilkinson, G. N., Robustness in Statistics, Academic Press, pp. 201–236.