TEST RESULTS COMPARING THE NEW US GEOLOGICAL SURVEY MONTE CARLO QUANTITATIVE RESOURCE ESTIMATION SIMULATION SOFTWARE APPLICATION TO MARK3

The U.S. Geological Survey (USGS) is currently improving how mineral resource assessments are performed. Steps include updating and designating new grade and tonnage models, determining appropriate economic filters for mine development feasibility analysis, and developing new software for quantifying the probable amounts of contained commodities in undiscovered deposits. This poster presentation focuses on testing the quantitative resource estimations of the new USGS Monte Carlo simulation software against those of the old MARK3 software.

Alpha tests of the new software show that it is able to perform thousands of simulations in only a few seconds’ time with results comparable to MARK3. Deposit forecasts by the two simulators using identical distributions show that the algorithms behave similarly. Discrepancies between simulators in the probable range of estimated metal amounts are attributed to differences in philosophy regarding the truncation of forecast results. In contrast to the new software, the MARK3 algorithm truncates the range of output forecast values. This results in large differences (> 90%) between the minimum and maximum forecast amounts of metals when comparing unbounded forecasts using the new simulator to those of MARK3.

For bounded predictions, MARK3 uses a weighted piecewise linear approximation of the lognormal probability density functions and assumes a dependency between grade and tonnage. The new simulator uses a lognormal distribution truncated at the 5 and 95 percentiles with no dependency between variables. For this case, the minimum and maximum extremes predicted by both simulators are closer (approximately 10% difference) and could be made identical depending on the bounding percentiles used when running the new simulator. Additional differences in the predictions of both simulators are attributed to how both simulators aggregate the results. MARK3 assumes 100 percent correlation between the input variables, whereas the new simulator allows correlation coefficients be specified by the user. Reproducible outcomes combined with ease of use, increased functionality and integrated graphical and tabulated reporting of results encourage future development of this new software application.