Statistical evaluation of compositional changes in volcanic gas chemistry: a case study

The geochemical analysis of fumarolic gases collected at quiescent and active volcanic systems over time is one of the main tools to understand changes in the state of activity for surveillance and risk assessment. The continuous output of chemical species through fumarolic activity, which characterizes the inter-eruptive intervals, has also a major and general influence on the environment. The mobilization of chemical species due to weathering of volcanic rocks, or the input of gaseous components from fumarolic activity, results in some kind of modification of the environment affecting, in particular, water, soils, and the consequent growth of the plants present in these areas. In this paper, an investigation on the chemical composition of fumarolic gases collected at Vulcano island (Sicily, southern Italy) is performed, with the aim to discover how data changes during the monitored period of time, and to design a strategy for the environmental surveillance of volcanic systems taking into account the nature of the analyzed data. In order to summarize the contribution of all the components that can affect the chemical composition of volcanic gases, a multivariate statistical approach appears to be suitable. Since many of those methods assume independent observations, the possible presence of time-dependent structures should be carefully verified. In this framework, given the compositional nature of geochemical data, we have applied recent theoretical and practical developments in the field of compositional data analysis to work in the correct sample space and to isolate groups of parts responsible for significant changes in the gas chemistry. The proposed approach can be generalized to the investigation of complex environmental systems.