Pattern Analysis Of Geochemical Data And Prediction Of Mineral Resources In Covered Area

The prediction of mineral resource in covered area is a difficult and frontier issue. Around this cutting-edge issues, the paper discussed the analysis on the geochemical patterns of covered area and reported some prediction works of mineral resources in Dalaimiao district, Inner Mongolia, where is an area commonly covered by Quaternary sediment.Through computer simulations, principal component analysis and independent component analysis(ICA) were compared in assisting geo-objects identification, the result showed that both PCA and can effectively assist in geological object identification, but both technologies have their advantages under different conditions: when there are significantly differences among rocks in the magnitude of attributes, the PCA is superior to ICA, otherwise ICA is better than PCA. From this conclusion, there is an inference that the PCs of geochemical data cannot work well in indicating mineralization when geochemical landscape consists of rock outcrop(or other geo-objects) associated intensive geochemical signature and mineralization associated weak geochemical anomalies in covered area. Those kinds of patterns of data are common in covered area.In order to find the solution to the problem, the geochemical patterns of Dalaimiao District. By introducing local singularity analysis, KL divergence, ROC curve to the analysis of the geochemical patterns of a geological section and regional exploratory geochemical data. The results show that elements in concealed mineralized rock can penetrate Quaternary sediment layer to the surface in veinlet forms, the singularity index anomaly on the surface can indicate concealed extraordinarily enrichment of elements or minerals. Moreover, different media have relatively similar multivariate statistical model, such as covering layer and underneath rock have rather similar multivariate patterns in singularity. There are similar patterns of singularity index in different depths of covering layer. That is to say singularity obtained from surface is stable and does not vary with the thickness of covering layer. A study of regional geochemical data also showed that covering sediments and conceal geo-objects share similar singularity index patterns. Comparing with the space of concentration, in the space of singularity index, the relationships between geo-objects are clear and well-constructed. Moreover, the study also implied that the first principle component of elements’ singularity index can more effectively indicate mineralization than the first principle component of elements’ concentration.Then, according to the above conclusions, the procedure of exploratory geochemical data process was redesigned and applied. Besides, a novel method named “optimal filtering” was applied on high-precision magnetic data. Moreover, the ICA was applied on remote sensing data to extract the ferric and hydroxyl anomalies. According to the above result, some prospective targets were delineated. the molybdenum mineralization found in a prospective target verified the effectiveness of the prediction works in covering area.