From needles to plates
: extreme volcanic ash shapes and implications for dispersion modelling

  • Jennifer Saxby

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)


Volcanic ash is hazardous to aircraft and can remain in the atmosphere for days or longer after an eruption. The rate of removal depends on meteorology and particle terminal fall velocity, which is sensitive to particle shape. However, most volcanic ash forecasts assume spherical particles; shape is difficult to quantify, and while the velocity of spheres can be determined analytically, modelling non-spheres relies on empirical drag laws. I assess the accuracy of drag laws for non-spheres for the range of flow regimes anticipated for volcanic ash falling in air, determine ash particle shape ranges, and assess the sensitivity of ash forecasts to shape.

Measurements of ash from Icelandic eruptions show that shape descriptors based on surface area are highly sensitive to imaging resolution and the particle size fraction used. I compare calculated terminal velocities to those measured in a settling column for volcanic ash and analogues; shape dependent drag laws produce more accurate results than a spherical approximation. Particle shape also impacts the method of size measurement, as size for non-spherical particles can be measured in several ways.

Using the NAME model, I determine the sensitivity of ash forecasts to particle shape. Model particle trajectories are sensitive to shape where sedimentation velocities exceed atmospheric vertical velocities; a non-spherical 100 μm particle can travel 44% further than an equivalent volume sphere. Therefore, the sensitivity of ash concentration forecasts to particle shape is dependent on the input particle size distribution.

I use these insights on measuring and modelling particle shape to discuss how best to parameterise shape in operational dispersion modelling systems; I identify drag laws which are accurate for the whole range of flow regimes expected for volcanic ash falling in air and compile particle shape data in order to recommend default values for use with these laws.
Date of Award28 Nov 2019
Original languageEnglish
Awarding Institution
  • University of Bristol
SupervisorAlison C Rust (Supervisor) & Katharine Cashman (Supervisor)

Cite this