Mineral prospectivity mapping integrating multi-source geology spatial data sets and logistic regression modelling

A method of integrating multi-source geology spatial data sets and logistic regression modelling for mineral prospectivity mapping is described in this paper. Logistic regression model describes the relationship between a dependent variable, which is a binary variable representing the presence or absence of the mineral deposits, and k independent variables represent ore-controlled geological features such as faults, lithology, geochemical anomaly, which may be continuous or discrete or any combination of both types. A case study was selected located in East Kunlun region of Qinghai Province, China. Multi-source geospatial data contain geological data, geophysical data, geochemical data and remotely sensed data. The potential prospectivity map was produced by logistic regression on the resulting revised binary map patterns, same as in weights of evidence modelling, two logistic probability thresholds of 0.52 and 0.58 were used to classify the study area into three classes of high potential, moderate potential, and low potential areas, which high potential areas contain 54% of the total gold deposits, covering 7.3% of the total area. Moderate potential area contains 8% of the gold deposits, covering 8% of the total area. Low potential areas contain 38% of the gold deposits, covering 84.7% of the total area.