基于GWR的丁家山铅锌矿控矿因素影响研究  

Research on the ore-controlling factors in the Dingjiashan Pb-Zn deposit based on GWR

作　　者：黄继先[1,2] 毛先成[1,2] 陈进[1,2] 邓浩[1,2] HUANG Jixian MAO Xiancheng CHEN Jin DENG Hao(Key Laboratory of Metallogenic Prediction of Nonferrous Metals and Geological Environment Monitoring （ Central South University）, Ministry of Education, Changsha 410083, Hunan, China School of Geosciences and Info-physics, Central South University, Changsha 410083, Hunan, China)

机构地区：[1]有色金属成矿预测与地质环境监测教育部重点实验室(中南大学),湖南长沙410083 [2]中南大学地球科学与信息物理学院,湖南长沙410083

出　　处：《地质学刊》2017年第3期401-408,共8页Journal of Geology

基　　金：国家重点研发计划项目(2017YFC0601503);国家自然科学基金项目(41472301);湖南省自然科学基金项目(10JJ6060)

摘　　要：在成矿预测领域,探索各控矿因素对成矿的影响规律,有助于提高成矿预测模型的精准度。隐伏矿体矿产分布具有真三维特性,利用经典统计模型和计量方法进行成矿预测存在可靠性低的问题。三维地理加权回归将真三维空间的位置信息加入普通线性回归模型,更有利于探究矿产分布及其控矿因素之间的内在联系。以福建丁家山铅锌矿Pb品位及其控矿因素数据为例,在排除各控矿因素变量之间存在的多重共线性问题以及证实福建丁家山铅锌矿Pb品位数据具有显著的空间自相关的基础上,对该组数据分别进行了普通线性回归分析和三维地理加权回归分析。结果表明,采用地理加权回归模型比采用普通线性回归模型具有更高的显著性,验证了各控矿因素对成矿的影响具有空间非平稳性。同时,通过采用自适应带宽对地理加权回归方法不同权函数运算结果的对比,证明Tricube权函数更能体现各控矿因素对成矿的影响随距离衰减的规律。从不同核函数的地理加权回归模型在三维空间的作用效应来看,各控矿因素的三维空间非平稳性显著,表明三维空间的地理加权回归模型在地下矿体建模与预测等方面还有很大的发展空间。In metallogenic prognosis,it is helpful in improving the precision of metallogenic prognosis model by studying the ore-controlling factors＇ influence on mineralization. The hidden mineral distribution is of real three-dimensional feature,so there is the problem of low reliability of the classic statistical model and measuring method in metallogenic prediction. It is better to explore the internal relation between mineral distribution and the ore-controlling factors using the 3D geographically weighted regression（ GWR） model which can import the spatial position information into generalized linear regression model. This paper takes an example of the Dingjiashan zinc ore grade and its controlling factors data. Firstly,the multi-collinearity of all controlling factors is resolved and spatial autocorrelation of zinc ore grade is confirmed,and then the generalized linear modeling（ GLM） and 3D GWRs are applied. The results show that GWRs have a higher fitting precision than GLM and thus the spatial non-stationarity of the controlling factors is verified. The results comparison of different weight functions of GWR by adaptive bandwidth suggests that Tricube weight function can explain the attenuation law of the controlling factors with increasing distance. Finally,the effect of GWR of different kernel functions in 3D space shows a significant non-stationarity of all ore-controlling factors in 3D space,which indicates GWR in 3D space has great potentials in modelling and prediction of underground ore bodies.

关 键 词：地理加权回归 空间非平稳性 三维地质空间 普通线性回归 空间自相关 铅锌矿床 福建丁家山 

分 类 号：P628.1[天文地球—地质矿产勘探] P618.4[天文地球—地质学]