Backward modeling approach of Trenches in epithermal Au deposit of Glojeh, Zanjan, Iran

This paper proposes a statistical Backward Modeling approach that quantitatively predicts Au concentration from the known concentrations of the other associated elements (Ag, Cu, Pb and Zn) from the Glojeh deposit in Zanjan province. A geologically constrained data set is 288 samples and collected from 631.9 meter of trenches. These samples divided for training model and validation model by the ratio sample of 2:1, respectively. A full Cubic Model (CM) including independent variables (X), order 2 terms (X2) and interaction between variables (Xi×Xj) created. By removing Zn, Ag×Pb, Zn^2, Ag×Zn, Cu^2, Pb, Ag, Cu×Zn and Cu×Pb predictors, respectively in different steps from CM, this process lead to create Reduced Cubic Model (RCM). RCM is calculated with R2 (pred) equal to 74.90% and best convergence for different R's and the PRESS index show the minimum value of 47.312 while F-value confirm improvement of the modeling. RCM consist of Cu, Ag×Cu, Pb×Zn, Ag^2 and Pb^2 as a main predictor in the equation and validation group confirm RCM by R2=73.90%.

Keywords:

Backward Modeling Approach , Glojeh deposit , training model , RCM , R2(pred)