Maximum entropy modeling for orogenic gold prospectivity mapping in the Tangbale-Hatu belt, western Junggar, China

Abstract

The Tangbale-Hatu belt (western Junggar region), located in the Central Asian Orogenic Belt (CAOB), has undergone complicated accretion and collision processes during the evolution of the Paleo-Asian Ocean. The geological events contributed to orogenic gold mineral systems in the region. In the present study, mineral systems approach was employed to evaluate critical ore-forming processes such as fluid migration pathways, the formation of trap zones, and deposition of metals. By means of translating these critical processes into mappable ore-controlling variables, we attempt to establish a process-based quantitative evaluation model. A maximum entropy (MaxEnt) model was proposed to predict the potential distribution of orogenic gold deposits based on known gold deposits/occurrences, and ore-controlling variables. Nine ore-controlling variables including fault intersection density, fault linear density, the first factor score map, Au singularity indices, Au regional anomaly, Au local anomaly, proximity to NE-trending faults, proximity to intrusion contacts and proximity to stratigraphic contacts, were selected to identify the most potential zones endowed with gold deposits. The spatial association of individual ore-controlling variable with the incidence of gold deposit occurrences was investigated by response curves, and the relative importance of each ore-controlling variable was determined by jackknife analysis in the MaxEnt model. The results indicate that Au regional anomaly derived from spectrum–area (S–A) fractal model is the most important variable, followed by the first factor score map and proximity to stratigraphic contacts. The model accuracy was evaluated by ROC curve giving a high predictive ability (AUC=0.915). Favorable zones for gold mineralization illustrated by the prospectivity map are spatially coincident with known gold deposits/occurrences and the main mineralized trends, showing that the MaxEnt model can be efficiently employed for spatial fusion analysis of multisource geospatial data to support conceptual mineral systems modeling of geologic controls on mineral deposit occurrences.