V13E-07:
Anatomy of a strombolian plume: inferences from particle data
Abstract:
We collected thermal video for 31 explosions at Stromboli during 2012 and 2014. We obtain a statistically robust database of size, mass and velocity for 83 000 particles. Statistically, the distribution for all parameters for 2012 is similar to that obtained for 2014. Most particles have sizes of 5 to 15 cm, so that the majority of individual bomb masses are below 0.4 kg. However, 4950 (6 %) of the particles are heavier than 5 kg and represent 59 % of the total mass eruptedThe particle velocity distribution is Gaussian with a mode of 20 to 30 m/s. Explosive activity at Stromboli has been viewed as involving large particles with low velocities (mean = 25 m/s). Recently, various studies have shown that a few lapilli-sized particles can be ejected as part of an initial high velocity gas-jet, which has average velocities of 80 m/s and a maximum of 405 m/s. Our data support this duality. Our spatial resolution allows us to detect particles down to 5 cm. Thus, we assume that all of the bombs are detected and that the bomb velocity distribution has a mode of 20 to 30 m/s. However, we also show that the smallest particles have the highest velocities with a maximum speed of 240 m/s, and a projected maximum of 456 m/s.
Stromboli’s explosions have been classified by visual clustering; depending on whether they are ballistic dominated (Type 1), ash dominated, with (2a) or without (2b) particles, or gas-dominated (Type 0). We used our parameters to cluster the explosions into 3 classes: Type 2a, 2b and Type 0. Correlation with seismic data now help to understand the source mechanism, and support recent models whereby Type 2a and 2b events are a consequence of the presence of a viscous, degassed cap at the head of the magma column, its fragmentation providing 103 to 104 kg of solid material to the plume in each eruption. However, Type 0 are a consequence of slug bursting in a “cleaner” conduit, the cap having been lost by convective overturn.