Search for: All records

Abstract Pyroclastic density currents (PDCs) are the most lethal volcanic process on Earth. Forecasting their inundation area is essential to mitigate their risk, but existing models are limited by our poor understanding of their dynamics. Here, we explore the role of evolving grain-size distribution in controlling the runout of the most common PDCs, known as block-and-ash flows (BAFs). Through a combination of theory, analysis of deposits and experiments of natural mixtures, we show that rapid changes of the grain-size distribution transported in BAFs result in the reduction of pore volume (compaction) within the first kilometres of their runout. We then use a multiphase flow model to show how the compressibility of granular mixtures leads to fragmentation-induced fluidisation (FIF) and excess pore-fluid pressure in BAFs. This process dominates the first ~2 km of their runout, where the effective friction coefficient is progressively reduced. Beyond that distance, transport is modulated by diffusion of the excess pore pressure. Fragmentation-induced fluidisation provides a physical basis to explain the decades-long use of low effective friction coefficients used in depth-averaged simulations required to match observed flow inundation. 

Abstract The April 2021 La Soufrière of St Vincent eruption generated several pyroclastic density currents (PDCs) during the 2 weeks of the crisis, from 9 to 22 April. To support the hazard assessment team during this eruption, numerical simulations were performed in real time and generated rapid scenario-based PDC invasion maps with the two-phase version of the code VolcFlow, which was able to simulate both the concentrated and dilute regime of PDCs. To generate the maps, only the source properties (shape and location) and the initial volume used to generate the PDCs were varied, all other input parameters were kept constant and estimated from previous simulations. New simulations were then performed based on the field-based deposit map to assess the code's ability to simulate such PDCs. Results show that the syn-crisis invasion maps satisfactorily mimic the observed valley-confined PDCs, while unconfined dilute PDCs were overestimated. The results also highlight that simulation results are greatly improved with additional field-based data, which help constrain the PDC sequence. Numerous lessons were learned, including (1) how to choose the most critical input parameters, (2) the importance of syn-eruptive radar imagery and (3) the potential of this two-phase model for rapid hazard assessment purposes.