GIS-based mineral prospectivity modeling


The great leader Comrade Kim Jong Il said as follows.

"Scientists and technicians should work to overcome by their own efforts the problems which require an urgent solution for the development of the national economy of our country, and to introduce the scientific and technical successes of developed countries in accordance with its specific reality." ("KIM JONG IL SELECTED WORKS" Vol. P. 195~196)

GIS-based mineral prospectivity modeling plays a significant role to decide where to drill in new deposits by integrating various kinds of data which is obtained in the exploration.

GIS-based mineral prospectivity modeling which uses data-driven, knowledge-driven and hybrid predictive models has already been studied in many countries thanks to rapid development of GIS techniques. To summarize the previous studies, we can conclude that we have no single best model that works effectively in all situations thus it is reasonable to use at least two predictive models and to compare their results with each other. Recently, some new models such as geometric average, restricted Boltzmann machine and expected value have been suggested to be applied to mineral prospectivity modeling, but the possibility of their application and effectiveness haven't been studied yet.

The Faculty of Geology, Kim Il Sung University, have achieved some successes in the study of introducing both geometric average and fuzzy logic models for GIS-based mineral prospectivity modeling.

We applied fuzzy logic and geometric average models for mineral prospectivity modeling, verified the effectiveness of their application and targeted new prospecting areas. First, we suggested new fuzzy membership functions to simultaneously introduce fuzzy logic and geometric average models for GIS-based prospectivity modeling and modified two models. We selected and used fuzzy sigmoid and fuzzy Gaussian functions by considering the nature of data used for mineral prospectivity modeling. Then we implemented GIS-based fuzzy logic modeling for 5 factors of fault, Cu, Pb, Zn geochemical anomaly maps and magnetic anomaly map in the study area and produced mineral prospectivity maps by simultaneously applying two knowledge-driven models.

The target areas classified by the fuzzy logic occupy 15% of the study area and contain 78% of mineral occurrences which are known. It also means that only 11 out of 14 known mineral occurrences are present in the most favorable area with high prospectivity values. Meanwhile, the resulting areas obtained by the geometric average model occupy 13% of the study area, but contain 93% of known mineral occurrences. In other words, 13 out of 14 known mineral occurrences are present in the most favorable area with high prospectivity values and only one is out of its site. The AUC values of two predictive models based on the fuzzy logic (0.90) and geometric average (0.93) are both above 0.5 and it indicates that both results are definitely better than a result based on a random selection of target areas and two models are useful for mineral prospectivity modeling.

The results of our study have been published in the title of "Application of fuzzy logic and geometric average: A Cu sulfide deposits potential mapping case study from Kapsan Basin, DPR Korea"( in "Ore Geology Reviews" of Elsevier.