k-Means Clustering to Enhance the Petrified Wood Composition Data Analyses and Its Interpretation

Abstract

Geologically, the fossilization of wood materials into fossils requires appropriate conditions, some of which have been preserved for millions of years. In nature, the organic mass of wood must be quickly replaced by inorganic elements before it decomposes under harsh geological conditions. Anorganic oxides such as silica-oxide, are known to be the main components of most wood specimens (up to 80%). The presence of alkaline oxides such as sodium and potassium oxide seems to play a major role in the presence of dissolved silica during petrification. However, their significance in the petrification phenomenon that occurs in fossilized plant wood is not yet known. Therefore, in this study, cluster analysis was conducted to determine the relationship between the presence of silica and alkaline compounds in petrified wood fossils. The approach used was -means clustering supported by the Elbow Method, which aims to review and order a complex set of data into subsets, thus allowing interpretation. The results showed that the clustering of the fossil wood composition data was optimal at = 3. There is a fair correlation between the presence of silica and alkali oxide compounds (-0.504 to -0.387), as well as with another inorganic compounds (+0.957). The presence of sodium and potassium is strongly correlated during silicification (+0.905). Additionally, the results of data clustering made the wood fossilization process susceptible to describe, especially through data regression. The data visualization provides more facts and proper explanations of the role of alkaline oxides in wood silicification. This study furthers our understanding of wood fossilization, especially the diagenesis of wood chemical composition in geological history.