Application of compositional data analysis in geochemical exploration for concealed deposits: A case study of Ashele copper-zinc deposit, Xinjiang, China

One of the top priorities of ore body exploration geochemistry is to effectively identifying “anomalies” and separating them from “background” in geochemical data analyses, during which an accurate understanding of geochemical distribution patterns of elements and anomaly separation has been the focus of ore body hunting. However, geochemical data are compositional data with closure effect and require necessary logratio transformation to open closed data for the following statistical analysis. The invested dataset of 13 elements derived from 2264 surface rock samples from the northwestern Ashele mining area was processed and respectively analyzed as the raw data, logarithmically-transformed data, and isometric logratio (ilr) transformed data with exploratory data analysis (EDA) technology. Of which, the ilr transformed data yields an ideal data structure and scale distribution. The data then was analyzed with principal component analysis (PCA) and robust principal component analysis (RPCA), along with the compositional biplots and the first principal component score maps to identify the spatial distribution of element assemblages. After that, a comparative study of the geochemical distribution patterns based on the raw data of mineralized elements and the principal component scores based on ilr transformation data is conducted to explore the connection with mineralization. As a result, four prospecting prediction targets of two types are delineated in the study area