From Spectra to Geography: Intelligent Mapping of RRUFF Mineral Data

TL;DR

This study introduces a novel machine learning framework aimed at geolocating mineral specimens at the country level using a one-dimensional ConvNeXt1D neural network architecture to classify mineral spectra based solely on their spectral signatures.

Abstract

Accurately determining the geographic origin of mineral samples is pivotal for applications in geology, mineralogy, and material science. Leveraging the comprehensive Raman spectral data from the RRUFF database, this study introduces a novel machine learning framework aimed at geolocating mineral specimens at the country level. We employ a one-dimensional ConvNeXt1D neural network architecture to classify mineral spectra based solely on their spectral signatures. The processed dataset comprises over 32,900 mineral samples, predominantly natural, spanning 101 countries. Through five-fold cross-validation, the ConvNeXt1D model achieved an impressive average classification accuracy of 93%, demonstrating its efficacy in capturing geospatial patterns inherent in Raman spectra.

Figures (9)

• Figure 1: Sample Density by Country
• Figure 2: Top 20 Countries by Sample Count
• Figure 3: Top 20 Mineral Species in the Dataset
• Figure 4: A pair of example spectra from the RRUFF dataset
• Figure 5: An example of a spectrum from Mexico being processed. On the left, the spectrum is raw, while on the right we can see the reinterpolated version over a much larger window, which is appropriately selected to cover the entirety of the dataset and padded to represent the spectrum in the correct window.