Identifying the Unique Geochemical Fingerprints of Omo Group Beds around Lake Turkana, Kenya by Modified Gaussian Discrimination Analysis

TL;DR

This study applies a modified Gaussian Discriminant Analysis to identify unique geochemical fingerprints of geological beds around Lake Turkana, enabling efficient identification of fossils' origins and demonstrating the method's effectiveness for high-dimensional data classification.

Contribution

The paper introduces a modified GDA method with an optimal search strategy to identify unique geochemical fingerprints in high-dimensional geochemical data.

Findings

Successfully identified unique geochemical fingerprints for five beds

GDA method effectively classifies high-dimensional geochemical data

Potential for rapid fossil provenance identification

Abstract

A modified Gaussian Discriminant Analysis (GDA) is used with an optimal search strategy to identify the unique geochemical fingerprints of six different geological beds in the Lake Turkana area. Three-hundred samples were collected from six different beds in the Omo Group of Lake Turkana, where each sample consisted of the PPMs of 11 different chemical compound. The GDA analysis discovered a unique combination of three tuff compounds that can uniquely identify one of the five beds, where a 6th bed is excluded because only 7 samples were collected from it. These geochemical fingerprints are important because any Turkana hominin fossil in the Omo Group can now be efficiently identified by matching the geochemical fingerprint of a fossil's host rock to the fingerprint of the bed in which it is found. In general, the GDA search strategy can be a powerful tool for efficiently identifying unique class fingerprints of high-dimensional data.