Date of Award


Degree Type


Degree Name

Master of Science


School of Engineering


Computing, Health and Science

First Advisor

Associate Professor Ute Mueller


Many iron ore deposits have a weathered zone (Hardcap) near the surface which is highly variable in grades. Estimating the amount of ore grade material (HG) in this zone is difficult as a result of this variability. The Section Seven Deposit at Tom Price is largely mined out and has production data available in the form of grade blocks that were marked out during mining as HG and non- HG. Hardcap domains and a block model representing them were created and estimates were made from original exploration data using Ordinary Kriging, Global Change of Support, Indicator Kriging and Median Indicator Kriging techniques. The estimates were compared to the production data The production data total HG blocked out was 6.4 Mt and the best central estimator of ore was Ordinary Kriging (2.0 Mt). Indicator and Median Indicator Kriging E-type estimates of ore were very similar at ~ 1.6 Mt. The Global Change of Support estimate was 4.0 Mt. An effective way of seeing the excessive smoothing in the central estimates was to compare the grade tonnage curves. All the central estimate of grades (OK, IK and Median IK E-type) were inaccurate and over smoothed. Given good quality samples and assays as well as sound estimation parameters the accuracy of these methods fundamentally comes down to the amount of data available to estimate from. There is insufficient data to get accurate estimates using these techniques. The main information that Indicator Kriging provides is not the E-type estimate but an estimate of the distribution of grades for each block from which a pseudo-probability that the block is HG can be derived. The pseudo probability was used to create maps of HG at different probability levels and there was a good match visually and between the production data HG blocks and blocks that had a greater than 0 chance of being HG. In comparison to the maps of HG generated from Ordinary Kriging which feature very few HG blocks and many sub-HG blocks these are a great improvement. Median Indicator Kriging was just as effective as Indicator Kriging in this regard, which is an important point as the former is less work than the later. Quantitative reconciliation of the Median Indicator Kriging results against the production data showed that blocks with a probability of 0.3 of being HG totalled 6.7 Mt and 49% of this matched HG production data. This gives rise to a methodology as follows: If OK has been used in estimating hardcap and if the Global Change of Support estimate indicates that there is a risk of oversmoothing with regard to the HG cut-off then Median IK should be used to identify areas which have a chance of being HG and then deciding on the best way to take advantage of this information. Some possibilities would be to:

  • target these areas for closer spaced drilling in order to generate an improved OK estimate;
  • use the area defined above to sub-domain the hardcap and re-estimate using OK;
  • target these areas for mining first as they have a good chance of being HG.


Paper Location