| Abstrak/Abstract |
Quaternary arc alkaline magmatism develops in the rear of Java Island, Sunda Arc, at a very different range of slab depth in the central and eastern section (~270-580 km and ~150 km, respectively). We collate published geochemical data of volcanic rocks from four alkaline volcanoes (Muria, Lasem, Bawean, and Ringgit-Beser) and perform statistical analysis to evaluate geochemical characteristics of each suite. The set of major and trace elements of the collated data points is first scaled and transformed using principal component analysis (PCA) for dimensional reduction. This step is then followed by implementing K-Means algorithm to cluster the data points based on Euclidian distances. K-Means clustering of major and trace elements dataset suggest that alkaline volcanics from East Java are geochemically distinct from those of Central Java. Alkalines from Central Java are more elevated in K2O and total alkali, while those from East Java show higher ratio of LILE/HFSE. In terms of major elements, the algorithm further suggests that Muria samples can be clustered into two, owing to the variation in total alkali and K2O content. These two clusters, however, are not well reflected on trace element datasets. Lasem alkalines show distinct cluster as they are highest in Na2O/K2O and SiO2, while Bawean samples are blending in into the two Muria clusters. In terms of trace element ratio, Ringgit-Beser alkalines show two distinct clusters, controlled by two sets of differing components (i.e., La/Nb, Ce/Nb, Zr/Nb versus Ba/Nb and Sr/Y). |
