عنوان مقاله [English]
Nowadays the use of Geostatistics to estimate mineral reserves is prevalent because this method uses a powerful tool for detecting changes in regional variable which is variogram. One of the most important prerequisites to use Geostatistics is that the variable of interest (grade, thickness) should bear an appropriate structure in area. Otherwise, Geostatistics will be unable for providing accurate estimates. In such cases (for instance vein gold deposit) where a high ratio between nugget effect and sill of variograms can be found, data aren’t adequate to cover the area, or uncertainty factors play major role in the overall variability of dataset, methods based on fuzzy logic can be applied as one of the options. Fuzzy estimation method operates through using Fuzzy C-Means (FCM) clustering algorithm. Since this algorithm uses random functions to create the initial cluster centers, and because of specific types (which bears data points which mainly belong to the background population and low anomalous data) of grade data in gold deposit, FCM technique is unable to reproduce valid cluster centers. In this study, the objective was to determine the initial cluster centers properly by separation of the data set and get rid of the abovementioned prerequisites such as random space. By doing this, estimation algorithm was able to provide more accurate results. Finally, the validation results showed that the modified method is more accurate based on the error functions we used. Root Mean Square Error values were 0.205, 0.162 and 0.107 for Simple FCM, Kriging and Modified FCM respectively which clearly shows better outputs for Modified FCM. Moreover, this method had also been able to reproduce the statistical and geostatistical parameters of the real data set with more accurate than other techniques used in this study.