Kamran Mostafaei

In mineral exploration projects, reducing uncertainty and increasing the success rate of exploration has always been a significant challenge. To reduce this uncertainty and increase the accuracy of mineral exploration, techniques such as information layer integration and mineral potential modeling are used. Mineral potential modeling is performed to identify target areas using various methods, including data-driven, knowledge-driven, and hybrid approaches. Therefore, the objective of this thesis is to model the mineral potential of gold mineralization in the 1:100,000 Alut sheet in Kurdistan. To achieve this goal, information layers such as lithology, alteration, faults, and stream sediment geochemistry were created for the region based on the genetic model of gold deposits and previous studies. These layers were first prepared and then processed for integration. Due to the different basis of the layers used for integration, all layers were fuzzified. The generated layers were then integrated using three methods: Fuzzy Gamma, Artificial Neural Networks (ANN), and Support Vector Machines (SVM), and areas with potential for gold mineralization were identified. The locations of points with evidence of gold mineralization were used for validation. The results of the three methods were compared. The comparison of the results shows that the models obtained from the artificial neural network and support vector machine methods perform better than the fuzzy gamma method. Based on the results, the use of support vector machines and artificial neural network models is more suitable and reliable for conducting more detailed exploration in the study area. In addition, several areas of potential mineralization were identified that can be considered as exploration targets in future studies.