Hossein Rahimi - School of Mining Engineering, College of Engineering, University of Tehran, Tehran, Iran
Maysam Abedi - School of Mining Engineering, College of Engineering, University of Tehran, Tehran, Iran
Abbas Bahroudi - School of Mining Engineering, College of Engineering, University of Tehran, Tehran, Iran
Mohammad Javad Rezapour - Geo-Exploration Targeting Lab (GET-Lab), School of Mining Engineering, College of Engineering, University of Tehran, Tehran, Iran
Gholam-Reza Elyasi - Geo-Exploration Targeting Lab (GET-Lab), School of Mining Engineering, College of Engineering, University of Tehran, Tehran, Iran
Soheila Aslani - School of Mining Engineering, College of Engineering, University of Tehran, Tehran, Iran

This work presents a hybrid-based clustering approach for mineral potential mapping (MPM) of porphyry-type Cu mineralization at Kerman province in the SE of Iran. Whereby a multidisciplinary geospatial data set was processed and integrated in the Chahargonbad district. Data-driven prediction-area (P-A) plots were drawn for each evidence layer derived from geological, geochemical, geophysical and satellite imagery data. The P-A plots provide insight into the weight of evidence for synthesizing all geospatial layers. Out of many knowledge-driven methods which biasing from experts' opinions, index overlay and fuzzy operators were employed to find out an optimum Cu favorability map through calculating an efficiency index representing the performance of each MPM. A concentration-area (C-A) fractal model was implemented to separate the mineral favorability map into some populations to ensure correct determining the cluster numbers. Clusters number is a prerequisite which must be defined correctly to increase the performance of clustering analysis for generating reliable results in MPM. Such an appropriate number of clusters can be incorporated in running three prevalent groups of clustering methodologies as data-driven approaches in MPM. They are self-organizing map, fuzzy c-means, and k-means algorithms. One of the reasons for this tendency to consider a hybrid-based method is that it overcomes the shortcomings of the both methods (bias of experts’ opinions and unknown clusters number) in mineral favorability mapping. The unknown number of clusters was determined through a knowledge-driven method, and then it was passed to an unsupervised data-driven method, i.e. clustering algorithm. This hybrid method produces synthesized maps in close association with known porphyry-Cu mineralization in the Chahargonbad area.

Clustering, hybrid method, Mineral Potential Mapping, Porphyry copper, Chahargonbad