A Rare Window Into Magmatic Conduit Processes: Time Series Observations From Active Lava Lakes

Abstract:

Time-lapse thermal images of the lake surface are used to investigate the circulation and cooling patterns of three lava lakes: Kilauea’s Halema’uma’u crater, Mount Erebus, and Nyiragongo. We report results for the time-dependent, two-dimensional velocity and temperature fields of the lake surface. These data sets constrain the locations of flow divergence (upwelling) and convergence (downwelling), the distribution of distinct “plates” and “rifts”, the dominant time scales for changes in flow pattern at each lake, and the physical properties of the magma.

Upwelling and downwelling locations are strikingly different between the three lakes. Upwelling at Nyiragongo and Erebus occurs dominantly in the interior of the lake, where it is occasionally interrupted by catastrophic downwellings. At Halema’uma’u upwelling and downwelling occur consistently along the perimeter. It remains to be seen whether these differences are dictated merely by the system’s geometry or are indicative of intrinsic factors such as melt viscosity, temperature and volatile and crystal content, or of conduit processes such as gas pistoning or slug flow.

The availability of high resolution data at Halema’uma’u allows as us to document the evolution of crustal plates and rifts and to investigate the physical properties of the lava and the crust. The physical properties of the lake’s surface control lake cooling rates, and thus need to be included in lake circulation and thermal evolution models. We produce time-temperature cooling curves from surface temperature profiles normal to surface rifts and by tracking the cooling of intra-plate bubble bursts. By comparing observations to analytical cooling models, we estimate a porosity of > 80% during the high stand of the lake, slightly higher than estimates of 70% for the upper 120 meters based on gravity data, and close to the porosity of clasts ejected from the lake during recent minor explosions. Furthermore,we find that the number of surface plates is correlated with the lake surface velocity, and that velocity increases preceding the break-up of plates, favoring a gas accumulation and collapse model over a slug model.