Logging stratigraphy on a drill rig to decide to continue or end the hole is a skill that every geologist should have. Ideally logging happens at the rig in realtime as we are often wanting to determine end of hole critera. Let’s consider how well hyperspectral and image data perform in classifing samples into prospective and non-prospective units. As usual we are going to use this dataset from from the C3DMM project.

Let’s talk about the performance of using chips vs hyperspectral to classify ore and waste in an Iron ore deposit. Before we go onto the technical work let’s discuss the practical uses of this information. Common use cases would be deciding if a sample is ore/waste, screening samples for further analysis or simply selecting the right sensor for you application and trading off cost/speed/performance/technical difficulty considerations. So with all that in mind I hope that you will find a smart application in your workplace.

Good Afternoon Spectroscopists, I though someone might be interested in using pytsg to process hyperspectral data. For this example we are going to use this dataset provided by CSIRO under a CC4 licence collected for the C3DMM project In this post we will use of PLS to predict Fe grade from the spectra. Let us start by installing the libraries that we are going to need. You can use the tsg file reader that I developed here or you can write your own…

Hi All, Let’s talk about improving these estimates or how to do a site specific adjustment. Revisiting the literature I found this paper Álvaro Corral, Álvaro González (2019). Power Law Size Distributions in Geoscience Revisited one could argue (but you won’t argue) that I should use a power law distribution for the cavities instead of an exponential, but that would require two parameters instead of one and it is easier to reason with less parameters at least initially.

Hi All, In this article we are going to integrate our estimate of background rate of cavities from caliper data we modelled in Part 1 with the likelyhood of detecting a cavity by drillhole spacing we generated in Part 2 . One of the issues with the simulation modelling estimating the background rates in Part 1 was the estimate of cavity distribution was done for the entire data set. As this is an Iron Ore data set from open file WAMEX data I know from experience that there are more cavities in certain stratigraphic units and alteration zones (hard cap).

Wecome back, Before integrating the data we generated in part 1 we are going to estimate the chance of intersecting a cavity on a blast pattern for a series of cavity sizes and blast pattern designs. To start the demonstration we build a baseline drill pattern of 5x7m with a fixed 0.27m hole diameter that we will keep fixed for all simulations. Now that we have a grid, we add a cavity of 1m diameter.

Hi All, Here is the first in a series of posts covering a bit of ground on cavity detection. From estimating the proportion of cavities in exploration drill holes to their chance of detection with MWD on a blast pattern. First off we define a cavity, for the sake of an exploration drill hole a simple definition might be a caliper measurement that flatlines at the maximum value. Looking over the histogram we can see a few points of interest: