Aqueous sulfur speciation possibly linked to sublimnic volcanic gas-water interaction during a quiescent period at Yugama crater lake, Kusatsu–Shirane volcano, Central Japan

After the phreatic eruption in 1982-83, volcanic activities at Kusatsu-Shirane volcano had been decreasing and reached a minimum in 1990, had turned to a temporal rise in activity up to 1994 and then decreased again at least up to 1997. During this low-activity period we observed a relatively short (<1 y) cyclic variation in polythionates (FT) in the Yugama lake water. Spectral power density analysis of the FT time-series by an autoregressive (maximum entropy spectral estimation, MESE) method indicates that the major frequency in the FT variations is 1.0 y -1 and the minor is 2-3 y -1 (1.0 and 0.3-0.5 y in periodicity, respectively). Annual variations in the lake temperature are ruled out for explaining these periodicities. We attribute these cyclic variations to a cyclic magnification-reduction in meteoric-water injection into a hydrothermal regime where volcanic gases from cooling magma bodies at depth and cooler oxidized groundwater come into contact with each other. This interaction may result in a periodical change in the composition and flux of SO 2 and H 2 S gases being discharged into the lake and forming FT. From a phase deviation (2-3 months) in the cycles between the annual precipitation, including snowmelt, and the FT time-series, we estimated the maximal depth of a hydrothermal reservoir beneath the lake assuming a vertical hydraulic conductivity (5 × 10 -3 cm/s) of the volcanic detritus around the summit hydrothermal system. Chloride in the lake water reached a maximum 1.5 years faster than FT. This is most likely due to a gradual elevation of the potentiometric front of a concentrated sublimnic solution in the hydrothermal reservoir. Variations of dissolved SO 2 and H 2 S in the lake water were not consistent with those of the fumarolic gases on the north flank of the volcano. This fact together with additional observations strongly suggests that these fumaroles may have the same origin but are chemically modified by a subsurface aquifer. The FT monitoring at active crater lakes during a quiescent period can provide insight into the annual expansions and reductions of a volcano-hosted hydrothermal reservoir. A simple analytical method in the field for the semi-quantitative determination of dissolved SO 2 and H 2 S in crater-lake waters is given in Appendix 2.