MineSight, made by Mintec Inc, was actually the first program to bring out a true implicit modelling module to compete against Leapfrog software. Like Micromine this module uses a true RBF with the ability to apply various adjustments and manipulations to create the final output. Being a true RBF based system puts it one step ahead of Maptek's Vulcan, GEOVIA's GEMS or Surpac and Datamine's Studio3 all of which are simple block model estimation and contouring systems. Mintec have allowed me to step through their program and have a play with the tool to see how it stacks up against the competition. Having never seen nor used MineSight in the past it has taken me a while to get inside the product in order to assess the IM tool. This is not a review of MineSight, rather the capability of its IM module, which is in short is impressive
. The team at Mintec have put a significant amount of capability into the module with a significant number of settings that allow the user to sculpt the final product.
The module is accessed through the following utilities menu;
This fires up the Implicit modelling window that floats in the foreground, allowing you to work whilst reviewing the data in the back ground (Figure 1). Figure 1. The IM form pops up in the foreground allowing you to work on the modelling whilst reviewing the results in the main MineSight window.
You have the ability to model both grade and lithology from a drillhole database, if you have set up composite files or point files in the Torque database these can be used also. The workflow is well thought out, broken out into input, process control and output areas (Figure 2). Once the connection to the Torque database has been made you step through each level where there are a significant amount options available to the user.
At the input level there is the ability to determine how the downhole survey runs, apply filters to the data, select how you want the input data to be treated (top and bottom surfaces, grade shells, various volume selections etc) and depending on the process selection method you can run a true thickness calculation. Also under the input option is a geometry selection which allows you to add additional point, lines or surfaces to influence or control the interpolation. You may for instance add some additional surface points to pull the interpolation beyond the limits of the drilling, it also gives you the option of selecting off-surface point distances to be calculated on these. The option does allow you to generate a IM mesh from polygonal inputs like explicitly digitised sections. 3D points may be modelled either selected through the Torque database, or as a point file selected through the Geometry node. Any geometry items (points, polylines, polygons or surfaces) that you wish to use in the process can be selected directly from screen. There are numerous methods available for determining the off-surface points for these inputs to ensure correct modelling.
The control level is where you set up a lot of the processing options. These include an option to select the processing Kernel - either thin plate spline (best for surface options) or Biharmonic (better for volumes). You can opt for an anisotropic search where you can fine tune your search parameters, as part of this process you have the option of previewing the search in the main screen, this is a great way of assessing the likely output prior to processing, sometimes it is hard to visualise the search parameters so being able to assess the ellipse is a great help. You also have the ability to decluster on the fly if required.
The output level is for defining the final product. There are options to control how the contouring is completed and an option to determine the marching cube size - this option directly affects the resolution of the final product so must be selected with care. Options to write the true thickness to file and where to save the output are all selected in this area, as are the extents of the project.
Figure 2. Various options for setting up the estimate.
As well as saving the surface files you have the option to save the parameters used and the RBF itself so that it is easy to generate new surfaces or regenerate the interpolant at later stage (Figure 3). Figure 3. You have the ability to save both the RBF and the parameters used for future reference.
The processing is quick and the processing is shown in the information window on the IM dialogue. Once the processing is complete it is displayed on screen in the main viewer.
So what about the results? As I have done with the previous comparisons I have taken the tutorial datasets I used for the Leapfrog-Micromine comparison run them through MineSight. Below I present some comparisons between Leapfrog and MineSight.
Figure 4 shows the results of a basic search of the copper variable from Leapfrog’s Marvin dataset. MineSight does a very good job replicating the result. The MineSight IM method appears to produce more volume although differences would likely have been easily fixed through more familiarity with the program, but from an un-educated start the initial output is very close to the Leapfrog output.
Figure 4. Isotropic interpolation on copper from the Leapfrog Marvin dataset, Leapfrog on the top, MineSight on the bottom
Figure 5 shows the results of using the Lithology modelling. Here I have modelled the QzP rocktype from the Leapfrog Marvin Dataset. The final output shows a very close similarity between the two.
Figure 6 shows the total lithology modelling output when all 3 rocktypes have been modelled. Given the simplicity of this model the resulting wireframes are quite good and very comparable. Figure 5. QzP wireframe of the Leapfrog Marvin dataset, Leapfrog on the top, MineSight on the bottom Figure 6. All the rock types in the Marvin project modelled in MineSight to build a solid model of the geology, the output is quite acceptable and with the ability to add geometry controls you could manage quite complex models.
Figure 7 shows the results of modelling the 5gpt gold grade shell from the Micromine NVG dataset using the same anisotropy. As can be seen here the increased volume generated in MineSight creates a significantly larger shell. Again with some fine tuning this could likely be improved somewhat.
Figure 7. 5gpt shell from the Micromine NVG dataset, Leapfrog on the top, MineSight on the bottom
Figure 8 shows the results of a largely unguided lithology interpolation of an NVG vein, the results from both programs are unacceptable as it has been for all the comparisons I have run. Whilst I have used an anisotropy for the search both programs have failed to adequately model the vein. This is the trap in believing IM can make modelling simple and easy. IM is a tool to help in the generation of a model. Knowing how to use the tool makes all the difference.
The ability to model a vein volume (Leapfrog), a single surface (in the case of Micromine) or the footwall and hanging wall (top and bottom) in MineSight allows you to then use the standard Boolean options in the programs with the IM outputs to generate a robust vein model.
Figure 8. Unguided vein interpolant of a single vein from the Micromine NVG dataset, Leapfrog on the top, MineSight on the bottom
Figure 9 shows the results using a vein modelling approach. The Leapfrog vein uses the interval selection and vein modelling method as discussed in the Micromine comparison. With MineSight I have modelled the hanging wall and footwall sections as separate volumes and combined them using Boolean processes. The results are slightly different based on the different generation methods but the results are definitely comparable
Figure 9. The same vein from the Micromine NVG dataset, Leapfrog using vein modelling on the top, MineSight using top and bottom surfaces and Boolean methods and polygons on the bottom
The MineSight implementation is certainly a powerful and comprehensive modelling tool which when combined with the full suite of tools available in MineSight gives the user the ability to integrate IM into their geological workflow with ease. The ability to fine tune the outputs through numerous options and additional data means that the program is not a black box, rather a powerful tool for the modelling geologist.
As a first step there are a couple of things where I think the program could be improved. There is no way I could determine for including known dip and strike (or dip direction) information into the generation of a surface or volume. Being able to include this data either as separate measurements or a combined structural trend is important when modelling faults or folded stratigraphy. Also the Help files are very limited with respect to details, telling you what the module does rather than how to do it. Training courses are rare and a working knowledge of how to achieve realistic results uncommon amongst the general geological community, many users out there would use the Help files as a first point of call. Given IM is such a very new method of modelling it would be nice to see more information around what changes to various options will do and how to generate the models.
The MineSight IM module is certainly powerful and will present a challenge to Leapfrog for MineSight users. Does MineSight do all Leapfrog can do – no, but it certainly does a lot and it might be all many need to achieve their desired model output. As time goes on the module is sure to increase in capability and functionality, certainly a module any geology modeller with a MineSight licence might like to add to their toolkit.
Not being a MineSight user my comparison is somewhat limited but what I could achieve is in no small part due to the help of Mark Gabbitus from Mintec's Perth office, from setting up the software for the trial, setting up the drill hole databases though to procedural workflows, his help was instrumental in bringing you this comparison, and I thank Mark for his input.