A new strategy for the estimation of plume height from clast dispersal in various atmospheric and eruptive conditions

Plume height is an important parameter routinely used to characterize and classify explosive eruptions. Though the strategies to estimate key eruption source parameters such as erupted volume and mass flow rate have evolved over the past few decades, the determination of plume height of past eruptions is still mostly based on empirical approaches that do not account for the new developments in plume modelling based on the interaction of plume and wind. Here we present a revised strategy for the retrieval of plume height from field data that accounts for key aspects of plume dynamics and particle sedimentation, which include: i) the effect of wind advection on the buoyant plume, ii) a new parameterization of the gravitational spreading of the umbrella cloud for distances smaller than the radius of the plume, iii) the effect of particle shape on particle sedimentation, iv) the effect of different atmospheric profiles in different climate zones, v) three-dimensional wind, temperature and pressure data, and vi) topography. In particular, as wind can affect the dynamics and height of the plume, new computed sedimentation patterns are more complex and result in non-linear relationships between downwind and crosswind deposition. Our method is tested against observations of the 2011 eruption of Shinmoedake (Japan), the 1980 eruption of Mount St Helens (USA), and the 1991 eruption of Pinatubo (Philippines). These are well-constrained examples of small, intermediate, and high intensity eruptions, respectively. Intensity scenarios are introduced to account for the non-unique relation between plume height and particle sedimentation resulting from wind advection of volcanic plumes. We further demonstrate that needle-like and disk-like particle shapes can have downwind distances 36 to 70% larger than the equivalent spheres. In addition, we find that the effect of latitude on the determination of plume height is more significant for low and intermediate intensity scenarios with a discrepancy between 7 and 20%