Multivariate analysis of Japanese volcanic rocks: Volatile and major elements.

Abstract

Tokyo152, JapanFour types of multivariate procedures (principal components analysis (PCA), factor analysis (FA), cluster analysis and multiple regression analysis (MR)) were applied to find the relationship between volatile contents and genesis of Japanese volcanic rocks. Twelve major elements and seven volatile minor elements were considered on 44 rock samples obtained from various areas of Japan. The variation in the data was demonstrated by 5 components and 5 common factors in PCA and FA, respectively. The first factor implies similar behavior among the major elements; the second factor represents the contribution of seawater to magma genesis; the third factor corresponds to the contribution of the mantle component; the fourth factor suggests separation of sulfide phase from silicate phase; and the fifth factor shows the contribution of sedimentary materials to magma genesis. INTRODUCTION A knowledge of the volatile behavior in vol canic rocks is of great value for understanding magma genesis in subduction zones. However, little work on volatile contents in volcanic rocks has been done because of the difficulty in the analytical technique and the difficulty in estimat ing the original concentration in magmas. Yoshida and his co-workers gave analytical data on F, Cl, Br, I (Yoshida et al., 1971; Yoshida, unpub lished), S (Arikawa, 1987;Hong, 1988) and B (Isozaki et al., 1973). These works were investi gated on respective element individually or were applied elementary statistics on only two or three elements simultaneously. Further work on the synthetic correlation structure among these ele ments was awaited. In the previous work, we studied the multivariate analysis on these six ele ments to understand their behaviors and found the usefulness of the method (Anazawa and Yoshida, 1994).Multivariate analysis allows us to consider changes in several properties simultaneously and is not affected by the number of objects being studied. This group of statistical methods is very powerful, but only limited examples on geo chemical data have been reported, because of its complicated theoretical strategy and operational methodology.In the previous report, we examined six volatiles and found that these elements form three groups by their correlation: (1) Cl, Br and B; (2) F and S; (3) I (iodine). The obtained factor struc ture, however, contained the possibility of chang ing the characteristics with additional variables for lacking of data such as major element concentra tions. (personal communication, 1991) in our research group determined arsenic for 50 Japanese volcanic rocks. He, however, could not find the correlation between As and S. Instead, he suggested a high correlation between As and B. Because of this considerable interest in As properties, we took As analytical data for the application of statistical analysis.In the present report, we describe new findings from multivariate analysis on 44 samples for 12 major and 7 volatile minor elements. The purpose of the present work is, (1) to find the factor str...

Heavy-Metal Distribution in River Waters and Sediments around a “Firefly Village”, Shikoku, Japan: Application of Multivariate Analysis

“…The details of those statistical methods are found in the literatures. 11,12 Calculation of the conservative mixing…”

Quantification of aeolian dust (Kosa) contribution to the Japan Sea sediments and its variation during the last 200 ky.

“…In order to interpret factor model, Anazawa and Yoshida (1996) proposed to use multiple-regression analysis between geochemical indices and factors. Here, we examine the mineral composition of factors using multiple-regression analysis.…”