Application of Statistical Analysis in Discrimination and Interpretation of Geochemical Data: A Case Study of Labunwa Area, Southwestern Nigeria

Abstract

Soil geochemical data from Labunwa area, bounded by latitude 7 1' 00'' N to 7 5' 00''N and longitude 5 15' 00''E and o 5 18' 00''E, were used to interpret the mineralisation potential of the area. The area is underlain by two rock types: granite gneiss and pegmatite. Twenty-one soil samples taken from the B-horizon were sun-dried, disaggregated, pulverised and sieved with a 63 micron sieve. The sieved samples were partially - digested using modified aqua-regia digestion and then analysed for twenty-two elements using inductively coupled plasma mass spectrometer (ICP-MS). The resulting raw geochemical data were first subjected to exploratory data analysisusing histogram with normal curve, box plot and scatter plot. Due to the compositional nature of geochemical data, the data were clr-transformed to deal with the closure effects and eliminates spurious relationships between compositions. The clr-transformed data were then subjected to multivariate statistical analyses such as, R-mode principal component analysis, compositional biplot and hierarchical cluster analysis. The mean concentrations of the elements in the soil are: Cu (39.98 ppm), Pb (24.69 ppm), Zn (56.33 ppm), Ag (0.03 ppm), Co (21.83 ppm), Mn (1099 ppm), As (0.93 ppm), U (2.66 ppm), Au (0.0004 ppm), Th (18.95 ppm), Sc (8.93 ppm), Cs (1.41 ppm), Nb (0.85 ppm), Rb (69.06 ppm), Rb (69.06 ppm), Sn (2.03 ppm) and Li (3.85 ppm). The results of the box plots, scatter plots and histograms of the untransformed raw data revealed that five of the sixteen analysed elements viz: Zn, Ag, As, Au and Sc, showed positive skewness. The histograms of the elements also showed breaks in distribution revealing multiple populations. Application of R-mode factor analysis produced a five-factor model which accounted for 83.57% of the variability of the data. These include Sc-Cs-Rb-Sn-Li, reflecting the presence of pegmatite bearing these trace elements and Zn-Au reflecting probable presence of gold mineralisation in the area. Like the PCs, the compositionalbiplot and dendogram produced five associations reflecting environmental, lithological controls and probable presence of Au mineralisation within the underlying gneiss in the study area

Publication Info:

Author: Adisa, A.L.* and Asowata, I.T.

Volume: 60

Issue: September

Published By: Journal of Mining and Geology Vol. 60(2) 2024. pp. 171 - 183 © Nigerian Mining and Geosciences Society (NMGS), 2024-09-01