Centre Universitaire d'Informatique

Mme Kae Tsunematsu soutiendra, en vue de l'obtention du grade de docteur ès sciences, mention sciences de la terre, sa thèse intitulée

New numerical solutions for the description of volcanic particle dispersal

Abstract :

When volcanoes erupt explosively they inject large amounts of particles of various size (microns to meters) and shape into the atmosphere that eventually deposit back to the ground (i.e., tephra). Numerical models for the transport and dispersal of volcanic particles are important for both the assessment of associated hazards and the understanding of eruptive dynamics. The largest particles (blocks and bombs) typically decoupled from the gas phase at an early stage and follow independent parabolic trajectories. Lapilli and ash (less than 64mm) are typically transported within a vertical buoyant plume that eventually spreads horizontally (i.e. umbrella cloud) at the level of neutral buoyancy. The coarsest lapilli and ash particles sediment back to the ground within hours, whereas the _nest particles (_ne ash) can be suspended in the atmosphere for days or weeks.

Studies of total grainsize distributions of explosive eruptions provide important insights into eruption dynamics and fragmentation mechanisms and are necessary to numerical simulations of particle transport and dispersal. Nonetheless, total grainsize distributions are difficult to constrain mainly due to poor preservation of tephra deposits. Based on quantitative comparisons of field data and numerical simulations we have shown how the tephra deposits of large explosive eruptions should be sampled at least up to 100-300 km from the vent in order to derive complete grainsize distributions that are not depleted in fines.

Many tephra dispersal models already exist in the volcanic literature, but important aspects of particle transport and sedimentation still require better parameterizations. Cellular Automata and Lattice Boltzmann models provide simple numerical solutions to very complicated physical systems and can also be easily implemented by adding new physical processes. In particular, Cellular Automata (CA) has been shown to be more computationally efficient than Lattice Boltzmann (LB) for the description of particle transport and dispersal.

We have firstly developed a two-dimensional (2D) CA model with particle release from the plume corner, which has shown good agreement with field observations for large explosive eruptions. In contrast, the description of dispersal from bent-over plumes and of aggregation-driven sedimentation needs a better parameterization. The 2D CA model was then expanded to three dimensions (3D).

In our 3D CA model, particles are released from the vent, are transported within the volcanic plume and are dispersed through the atmosphere. Particle diffusion is described based on a random velocity which corresponds to the velocity fluctuations of turbulence. The implementation of plume dynamics significantly improves the description of particle in proximal area (less than 50km from vent). CA strategies combine advantages of both Lagrangian and Eulerian methods as they can both track particle trajectories (as Lagrangian models) and be easy to parallelize (as Eulerian models).

Ballistic trajectories are modeled in three dimensions based on a Discrete Event Simulation (DES) method. Multiparticle simulations have been implemented that include particle-particle collisions. We have shown how collision between particles can either increase or decrease travel distance. Ballistics represents a significant volcanic hazard due to their high velocities and temperatures, which can damage vegetation and infrastructures. Vulcano Island (Italy) based on probability maps and energy distributions. Various hazardous thresholds were defined for both roof penetration and roof collapse and indications of the most dangerous areas on the island have been given based on the implementation of model results within the Geographical Information System platform.

Date: Vendredi 27 janvier 2012 à 14h00

Lieu: Bâtiment des Maraîchers - Salle 001

À la Une