Prediction of Au grade in Carlin type using pathfinder elements by GMDH and MCMC in Zarshuran deposit

Pathfinder elements play a crucial role in the exploration of concealed and deep-seated mineral deposits. Their significance is particularly pronounced in the context of epithermal gold (Au) deposits, where their presence may serve as an indicator of nearby gold mineralization. Among these pathfinder elements, arsenic (As) and antimony (Sb) are considered the most critical for the exploration of epithermal Au systems. This study investigated Au Carlin type in the Zarshuran to highlight the utility of pathfinder elements in gold estimation. The analysis was conducted using the concentrations of ۳۵ elements measured across ۱۰۸ samples. The mineralization characteristics of the Zarshuran deposit exhibit notable similarities to those of epithermal gold deposits hosted in sedimentary rocks (Carlin-type), thus presenting a suitable exploration model for the northern Takab region. Selection of pathfinder elements was carried out through factor analysis, which revealed a strong positive correlation among Au, As, Cd, Pb, Sb, and Zn. Two predictive approaches were employed to estimate gold content: the Group Method of Data Handling (GMDH) neural network, and the Monte Carlo Markov Chain (MCMC) simulation. Neural network techniques, such as GMDH, are particularly well-suited for modeling datasets with both linear and nonlinear characteristics. In these models, As, Cd, Pb, Sb, and Zn were used as input variables. The predictive performance of the models was assessed using the coefficient of determination (R²). The GMDH neural network achieved a superior performance with an R² value of ۰.۹۴۸۳, outperforming the MCMC simulation. Based on these findings, the GMDH neural network is recommended as a robust and reliable method for predicting Au mineralization in other prospective exploration areas.

کلیدواژه ها:

Monte Carlo Markov chain simulation ، GMDH Neural Network ، Au prediction ، pathfinder element ، Au-Zarshuran