Comparing new drillhole assay data with an old block model
I am working on a project which includes a block model and a drillhole database; the drillhole database includes holes (with assays data) that were drilled after the block model was created.
I would like to compare the new drillholes' assay data with the grades of the block model where the drillhole samples intercept the block model. Can a set of 'pseudo assay' results (based on the block model data and with the same sample lengths as in the new drillholes' assays) be created for holes that did not contribute to the creation of the block model?
The purpose of this exercise would be to see how well the block model has predicted the ore grade and width. Can someone please let me know what tool/function/procedure, if any, I should be using?
Customer support service by UserEcho
I think you want to use either Modelling>3d Block Estimate>Statistical or use IDW as a nearest neighbor back into the assay interval.
Statistical averages xyz points within a define block extents. So no data that is physically outside the block will be averaged. This is your typical route for blast holes to block comparison. One annoyance is you will have to merge based on the xyz centroid from the previous model, I don't believe you can use an existing model for this function.
Using Modelling>3d Block Estimate>Inverse distance weighting you could do a nearest neighbor (set max samples to 1, not explicit nearest neighbor option).
As usual, I imagine there are several other ways that other suggest doing this exact thing.
One note, this is my favorite aspect of micromine. Using block models as comps or assays as blocks. In my opinion no other software comes close the flexibility and do what you want capabilities of MicroMine.
May I suggest a fast approach that I used recently, under Modelling/Block Model Tools/Assign. It is enough to have X Y Z for your assay file and create a new BlockModel_Grade Column to be populated with the Block Model grades. BM grades will be assigned straight to your interval centroids and you can compare them easily with the assayed grades.
I think that is a simpler way!
Thanks for the reply. I did try this at first, to a get rough version of the Block Model 'assays'. However, I was told by colleagues that we needed a more accurate procedure that takes into account assays that overlap blocks; so I've ended up creating a wireframe for each assay instead.
In the first approach above, using IDW to power 1 and a search that approximates the size of blocks in the original model will give you (approximately) average grades from the new drilling in each of the blocks to compare against previous block grade estimates. The second approach will give you a single block model grade for a series of intervals in the new assay file (I think). You'll then have to calculate grades for composited intervals that correspond to the groups of assay intervals that have received assigned grades from the original model in order to compare.
If you have the wireframing module, then try the following steps:
Thank you. This is what I have done in the end (but set the circular profile diameter to 0.1 m instead). It seems like a very fiddly way of getting the information though. Fortunately I only needed the pseudo-grades from 831 assays and not all 17,000 assays in the database. Trying to create 17,000 wireframes might have melted my laptop.
Thanks for the replies everyone. I didn't realise this would be so tricky. I think Datamine has an 'Interrogate the wireframe/block model' tool, or something, that does the same thing; maybe something for MM to add in the future?
Andrew, this is already logged as an issue. I will add your name, so that it will have a higher priority the next time we review improvements. You are right to decrease the diameter of the profile. The methodology gives an accurate result.
Another (simpler) way to handle this is to use the Implicit Attribute Modelling function, with the Block Model as an input and an Implicit model as the output. Now you can use the Output Implicit Model to Point function to populate the psuedo-grade field. The results are very similar, but:
I am interested in your opinion of the value to this process. Are you validating the block model? The assay results? Using it for planned holes?
I know that writing psuedo-grades back to a hole that WAS used in the modelling process does not appear to generate any meaningful information. Of course this will depend somewhat on the relationship between block size and interval length.
Thanks for the tip.
Yes, I wanted pseudo-assay grades for validating the block model. We are evaluating another party's resource estimate and want an idea of how reliable their block modelling technique is. I only started using MM a few months ago and have no real prior experience with similar modelling software but some of my senior colleagues consider the ability to 'interrogate' a block model this way as very important. I think it is a useful tool for planning future holes as well though.