Sensitivity analysis of a one-dimensional model of a volcanic plume with particle fallout and collapse behavior

• We tested the sensitivity of eruption plume height and atmospheric mass loading to source conditions and parameters.
• Vent radius and initial speed have a more profound effect on model outputs than does grain size distribution.
• Rise height and mass loading are sensitive to entrainment parameters, but also wind speed.
• Efforts to better characterize entrainment parameters are important.
• Accurate, local meteorological data should be collected.

We run a volcanic plume model with uncertain boundary conditions and entrainment related model parameters. Output variables tested for their sensitivity to the inputs are total rise height, and mass flux of particles into the umbrella cloud or downwind plume. Boundary or source conditions are vent radius, initial velocity, grain size mean and grain size standard deviation. Model parameters are entrainment rate, α, wind entrainment rate, β, and wind speed.Five sensitivity metrics were considered. Three of these are calculated for each given point in the input parameter space, by perturbing the input variable around fixed points. Two global sensitivity measures quantify the impact on the output of the input over its entire uncertain domain.We find that vent radius and initial speed have a much more profound effect on both outputs than does total grain size distribution. Plume rise height and particle mass flux are sensitive to the entrainment parameters, α and β, but these parameters are not of greater importance than the wind speed. This suggests that while efforts to better characterize entrainment parameters through laboratory experiments is important, similar efforts should be made to collect appropriate meteorological data for the region near the site of the eruption.