Three-dimensional Metallogenic Prediction Of The Deep And Peripheral Parts Of The Lujing Uranium Deposit

Lujing ore field,in the Hunan-Jiangxi border,is an important uranium resource base in China.It is in the Wugong-Zhuguang fault-uplift area of South China active belt(Huaxia block)and the middle part of Zhuguang complex rock mass.The Lujing deposit in the middle of Lujing ore field is a large granite-type uranium deposit in which there is a potential to discover concealed ore bodies in deep and peripheral areas.However,the overall exploration depth is small in this deposit,the exploration investment in peripheral is insufficient,and there is still a technical bottleneck in terms of prospecting direction.Therefore,this paper evaluates and predicts the deep and peripheral uranium potential of the deposit through 3-D metallogenic modelling.In this paper,the geological prospecting model of the Lujing uranium deposit is established based on a metallogenic model and the summary of ore-controlling factors and metallogenic regularities.After the construction of a 3-D geological model and the quantitative analysis of favorability of various ore-controlling factors,a quantitative prediction model has been established,and a variety of mineral prediction methods,such as 3-D weights of evidence,3-D information quantity and machine learning are used for the quantitative prediction in order to expand uranium resources and discover blind ore bodies in the deep and periphery of Lujing deposit.This paper summarizes the metallogenic geological characteristics and metallogenic regularities of the Lujing uranium deposit,analyzes the control of strata,lithologies,structures and contact zones to uranium mineralization,determines that strata,faults and rock contact zones are the most important ore controlling factors of the deposit,and establishes a prospecting model.The carbonaceous slate of the Cambrian in this area is rich in uranium.The uranium content of the Indosinian and early Yanshanian granites is higher than the average of South China granites.Besides,the uranium-bearing hydrothermal fluid is also as a good uranium provenance.All these uranium sources founded a good basis for uranium metallogenic in this area.Frequent tectonic movements activated the channels of hydrothermal rising and resulted in abundant fractures and fissures in the area,which could have provided favorable spaces for uranium concentration.The larger uranium ore bodies cut through the early Yanshanian rock mass,Indosinian rock mass and Cambrian rock mass,with the characteristics of "three floors".On SKUA-GOCAD,a 3-D geological software platform,a discrete smooth interpolation DSI method is applied to construct 3-D models of strata,faults,rock mass and ore bodies.The 3-D visualization of uranium metallogenic environment and ore bodies has been completed,which plays an important role in expressing the interspersed relationship of geological bodies in underground space.Based on the studies on the uranium contents of the Cambrian strata and rock mass of each stage,it is determined that the main host units of the deposit are the Cambrian strata,Indosinian rock mass and the third stage of the early Yanshanian rock mass.Through the analysis of 3-D distance field space,the effecting ranges of faults and lithological contacts have been determined: the best buffer distance of faults is 60 m,and the best buffer distance of contact zones is 60 m,too.Through a 3-D morphological field space analysis method,the abnormal azimuth interval of fault plane was determined as60� ? 80� and 200� ? 260�.And the geophysical and geochemical anomaly information was extracted.The extracted quantitative anomalous information was assigned to each gridding prediction unit to establish the prediction model of the deposit.The Pearson correlation coefficient was calculated between variables.The maximum correlation coefficient is 0.36,which implies that there is no correlation between the predictive variables.Using 3-D weights of evidence and 3-D information quantity,and supporting four machine learning methods,i.e.,vector machine,random forest,logistic regression,and artificial neural network,3-D quantitative prediction of uranium resources in deep and periphery of the Lujing deposit has been carried out,and five favorable uranium targets have been delineated,which will provide the basis for the expansion of uranium resources in deep and periphery of the study area.Uranium metallogenic modelling,3-D geological modeling,multi-dimensional information correlation analysis method,and joint delineation of target area by multiple prediction methods are proved to be a key to quantitative prediction of uranium resources and new breakthroughs in uranium prospecting.