New classification method of volcanic ash samples using statistically determined grain types

TL;DR

This paper introduces a new automated classification method for volcanic ash samples based on statistically determined grain types, improving consistency with eruption styles and applicable to various geological fields.

Contribution

The study presents a novel, automated grain classification technique using cluster analysis on large datasets, eliminating human bias and enhancing eruption style interpretation.

Findings

Using both grain shape and transparency improves classification accuracy.

The method successfully distinguishes eruption types in volcanic samples.

Applicable to stratigraphy, sedimentology, and planetary science.

Abstract

We developed a method to classify volcanic ash samples by introducing statistically determined grain types. Using more than 10,000 numbers of automatically measured grain data (parameters of grain shape and transparency) and the cluster analysis, we made grain types without human eyes. By components of the grain type in each samples, we classified samples from types of basaltic monogenetic volcanoes: 1) Funabara scoria cone, Izu Peninsula, Japan (magmatic eruption origin); 2) Nippana tuff ring, Miyakejima, Japan (phreatomagmatic eruption origin); and 3) rootless cones in Myvatn, Iceland (rootless eruption origin). We tested two cases; using grain shape parameters only, and both of grain shape parameters and transparency values. It is found that the sample classification is more consistent with their eruption style in the case of using both parameters of grain shape and transparency. By sampling several layers of an outcrop, this procedure can be used to interpret changes in eruption/fragmentation style during a volcanic event. Furthermore, this procedure might be applicable to other aims such as sedimentology and planetary science.