Detecting and mapping compositional global outliers to identify mineral exploration targets, Case study: Khusf district, East of Iran
- Autores: Majid Keykha Hosseinpoor, Hamid Moini, Farhad Mohammad Torab
- Localización: Proceedings of the 6th International Workshop on Compositional Data Analysis: Girona, 1-7 de juny de 2015 / coord. por Santiago Thió Fernández de Henestrosa, Josep Antoni Martín Fernández, 2015, ISBN 978-84-8458-451-3
- Idioma: inglés
- Enlaces
-
Resumen
To identify mineral promising targets in khusf area located in the east of Iran, a geochemical stream sediment study was done and 652 samples were analyzed for 23 elements. Although there are some different known classical convenient methods to delineate the univariate anomalies, the most reliable one is to deal with the geochemical anomalies through compositional global multivariate outliers that on this case study has been addressed. After preparation of the raw data (missing and censored values imputation on compositional data using zCompositions package in R), the global outliers of the data were detected and marked in the coordinated map of the samples using mvoutlier package in R. Robust principal components of the compositional data were biplotted as well. The biplots showed the elements affecting the global outliers. Compositional univariate outlier plot is another useful tool in the package that was used to visualize each outlying observation position element-wise. Comparing the resulted global outliers with the officially reported mineralization indexes and field observations confirmed that mvoutlier package is the fastest and most reliable tool amongst the other convenient ways of locating geochemical anomalies provided that the sampling and data preparation stages are done properly
