Some first order observations on magma transfer from mantle wedge to upper crust at volcanic arcs

The viscosity of lavas erupted at volcanic arcs varies over orders of magnitude. A comparison of the relative abundance of viscous lava dome eruptions indicates that the average viscosity of arc lavas also varies considerably between arcs. It is shown that for continental or transitional arcs with little within-arc crustal deformation and without underlying slab windows or tears, average lava viscosity is anti-correlated with average surface heat flux. The latter may be influenced by crustal thickness and crustal magma throughput. To constrain the relative contributions of these parameters, variations of average lava viscosity with average crustal thickness and plate convergence rate are assessed. While crustal thickness appears to have little effect on average lava viscosity, a good anticorrelation exists between average lava viscosity and plate convergence rate, with the exception of two arcs that show significant intra-arc crustal deformation. If plate convergence rate is a good proxy of the rate of melt generation within the mantle wedge [cf. Cagnioncle et al. 2007], these first order observations indicate that where the rate of mantle melting is high, crustal magma throughput is rapid and efficient, resulting in low viscosity melts migrating through a hot overriding crust; in contrast, where the rate of mantle melting is low, crustal magma transfer is slow and inefficient, resulting in high viscosity melts that may frequently stall within a cool overriding crust prior to eruption. Uranium series geochemical evidence from dome lavas is presented and lends support to this interpretation. Finally, some explanations are offered for the observed average viscosity variations of arcs with underlying slab windows or tears and/or significant intra-arc crustal deformation. ■