GIS支持下的线性回归证据权法扩展及成矿预测

GIS-based extended weights of evidence method based on linear regression in metallogenic prognosis

本文提出一种利用线性回归分析来指导成矿有利证据层筛选及最佳临界值获取的方法,并以桂西-滇东南地区锰矿为例,在多元地学空间数据库的基础上进行了基于线性回归的证据权法的矿产资源预测研究。实例表明,线性回归分析能利用成矿信息估计矿化分布的总体趋势,其回归分析结果能反映成矿信息与矿化分布的正负相关性,结合统计分析方法指导获取的最佳临界值,不仅具有地质意义,而且具有统计意义,能在证据图层二值化时最大限度减少信息丢失。

Focusing on the issue of the loss of metallogenic information when converting the evidence layers into binary layers, a method of using the linear regression analysis, i.e. the weights of evidence method based on linear regression, was proposed to guide the selection of evidence layers which are advantageous to ore-formation and obtaining the best critical value, and conducted an instance study of manganese resources prediction with this method based on the multivariate geo-spatial database by taking the western Guangxi and southeastern Yunnan as an example. The instance study indicated that the linear regression analysis could use the metallogenic in- formation to estimate the whole trend of mineralization distribution, and the result of regression analysis could reflect the positive and negative correlation between the metallogenic information and mineralization distribution, and the best critical value obtained by combi- ning the statistical analysis method has statistical significance as well as accurate geological significance, which would reduce the loss of metallogenic information at the greatest degree when converting the evidence layers into binary layers.