5:15 PM - 6:30 PM
[SVC30-P07] The effect of dissolved water induced by lithostatic pressure on lava migration
Keywords:Unzen volcano, lava viscosity, dissolved water
The apparent viscosity of Unzen lobe lava had been estimated to be 0.9–4.2×1010 Pa s from the observed lava surface velocity (2~5 m/h) in 1991, using Jeffreys model with lava density 2300 kg/m3, lava thickness 40~70 m and slope angle 25~30 °, based on the assumption that the lava was a Newtonian or Bingham fluid (Fukui et al., 1991; Suto et al., 1993). In contrast, the viscosity of synthetic groundmass glass derived from Unzen lava (~79 wt.% SiO2) had been measured in the laboratory at expected lava temperatures (780°C–880°C: Nakada and Motomura, 1999) to be 1011–1013 Pa s (Goto, 1999). On the basis of Einstein–Roscoe equation, Goto (1999) concluded that ~50 vol.% crystallized Unzen lava lobes probably did have a few orders of magnitude higher viscosity than was observed, and he speculated that the lava-migration mechanism was different from that of Newtonian flow. On the other hand, Cordonnier et al. (2009), who obtained higher viscosity values from laboratory experiments than the field-based values for Unzen natural-lava samples, pointed out that, based on Hess and Dingwell's (1996) hydrous-melt viscosity model, 0.2 wt.% water in the groundmass glass (Nakada and Motomura, 1999) compensates for the gap between observed and experimental viscosity. In the present study the effect of dissolved water on lava migration was numerically investigated, based on synthetic dry Unzen groundmass glass viscosity data by Goto (1999), with the aids of water solubility model by Liu et al. (2005) and the degree of viscosity decrease by dissolved water derived from hydrous rhyolitic melt viscosity by Goto et al. (2005). In the following we fix the lava temperature to be 825°C.
Liu et al. (2005) provided empirical water solubility model for rhyolitic melt as functions of pressure and temperature. Using this model water solubility in rhyolitic groundmass glass was calculated for the interior of 70 m-thick Unzen dacite lava at each depth: e.g., 0.44 wt.% at the base of lava (1.71 MPa), 0.29 wt.% at 30 m form the surface (0.79 MPa) and 0.10 wt.% at the surface under atmospheric pressure (0.1 MPa). Empirical viscosity from hydrous rhyolitic melt (~77 wt.% SiO2) by Goto et al. (2005) was fitted well as a function of water content at fixed temperature, expressing exponential decrease in logarithmic viscosity with increasing water content: -1.89, -1.62 and -0.83 by 0.44, 0.29 and 0.10 wt.% dissolved water, respectively. Supposing that the effect of dissolved water on the degree of rhyolitic melt viscosity decrease is identical, hydrous Unzen groundmass glass viscosity is estimated from dry melt viscosity by Goto (1999) to be 1.4×1010 Pa s at the base of lava, 2.7×1010 Pa s at 30 m from the surface and 1.7×1011 Pa s at the surface. These values are similar to or higher than the field-based viscosity, but integrated shear velocity at each depth gives surface velocity ~5 m/h, similar to the highest end of the observed velocity, mainly from the contribution of low viscosity zone on the bottom. These are the cases if the lava is crystal-free, but Unzen lava contains ~50 vol.% crystals. It is difficult to evaluate the effect of crystals on Unzen lava viscosity correctly because there is no universal viscosity model applicable to differing suspensions of crystals of various size and shape distributions, but simple Einstein–Roscoe equation expects a few order increase in viscosity by suspended crystals. The highest end of estimated effusion temperature (880°C) gives one order lower groundmass glass viscosity than the above evaluations, but insufficient to compensate for the viscosity increase by suspended crystals.
The present study indicates the importance of dissolved water induced by lava lithostatic pressure and resultant viscosity decrease, but insufficient to compensates for the gap between observed and experimentally-predicted viscosity of Unzen lava.
Liu et al. (2005) provided empirical water solubility model for rhyolitic melt as functions of pressure and temperature. Using this model water solubility in rhyolitic groundmass glass was calculated for the interior of 70 m-thick Unzen dacite lava at each depth: e.g., 0.44 wt.% at the base of lava (1.71 MPa), 0.29 wt.% at 30 m form the surface (0.79 MPa) and 0.10 wt.% at the surface under atmospheric pressure (0.1 MPa). Empirical viscosity from hydrous rhyolitic melt (~77 wt.% SiO2) by Goto et al. (2005) was fitted well as a function of water content at fixed temperature, expressing exponential decrease in logarithmic viscosity with increasing water content: -1.89, -1.62 and -0.83 by 0.44, 0.29 and 0.10 wt.% dissolved water, respectively. Supposing that the effect of dissolved water on the degree of rhyolitic melt viscosity decrease is identical, hydrous Unzen groundmass glass viscosity is estimated from dry melt viscosity by Goto (1999) to be 1.4×1010 Pa s at the base of lava, 2.7×1010 Pa s at 30 m from the surface and 1.7×1011 Pa s at the surface. These values are similar to or higher than the field-based viscosity, but integrated shear velocity at each depth gives surface velocity ~5 m/h, similar to the highest end of the observed velocity, mainly from the contribution of low viscosity zone on the bottom. These are the cases if the lava is crystal-free, but Unzen lava contains ~50 vol.% crystals. It is difficult to evaluate the effect of crystals on Unzen lava viscosity correctly because there is no universal viscosity model applicable to differing suspensions of crystals of various size and shape distributions, but simple Einstein–Roscoe equation expects a few order increase in viscosity by suspended crystals. The highest end of estimated effusion temperature (880°C) gives one order lower groundmass glass viscosity than the above evaluations, but insufficient to compensate for the viscosity increase by suspended crystals.
The present study indicates the importance of dissolved water induced by lava lithostatic pressure and resultant viscosity decrease, but insufficient to compensates for the gap between observed and experimentally-predicted viscosity of Unzen lava.