Modelling of Tsunamis Generated by Pyroclastic Flows (Ignimbrites)

Pyroclastic flows entering the sea played a major role in generating the largest tsunamiwaves, arising from the 1883 eruption of Krakatau, Indonesia, which caused a considerabledeath toll, most deaths resulting from the tsunamis. The potential exists for similar eventsto occur in Indonesia and New Zealand. Processes leading to tsunami generation by pyroclastic flows, especially those associatedwith Krakatau-type eruptions, are reviewed. The major processes include: 1. Deposition at the shoreline causing a lateral displacement as the zone of depositionmoves offshore. 2. Upward and lateral displacement of water caused by the propagation of a watersupported mass-flow. 3. Downward and lateral displacement of water caused by the sinking of debris from a segregated flow travelling over the water surface. 4. Upward displacement of a large volume of water due to the deposition of acaldera-infill ignimbrite or pyroclastic flow deposit. The pyroclastic flow is modelled as a horizontal piston forcingwater displacement. The flow behaves as a wedge of material displacingseawater horizontally and vertically as it moves outwards from the source.Individual pyroclastic flows are treated as linear features that travel alonga specific direction from the volcano, exhibiting limited lateral spreading.The event duration for the formation of a large pyroclastic flow and thedeposition of the ignimbrite is taken as 200–400 s, with flow velocitiesdependent on the volume of material erupted. For simulations it is assumed that the ignimbrite deposit is elliptical with relativelyuniform thickness and the principal axis orientated along the flow direction. Therefore the tsunami is generated by defining an elliptical source region and defining an effective displacement behaviour at each node within that region. The effective displacement is defined by a start time, a finish time and a vertical velocity. These three parameters determine when the seafloor starts to rise and how far it travels during a model time step. The result is a seafloor disturbance that propagates away from the source. The major difficulty with this approach is determination of the appropriate verticalvelocity. With a real pyroclastic flow the effective vertical velocity at any point isvery high. However the model needs to average the displacement spatially andtemporally. Accordingly we apply the model to pyroclastic flows from Mayor Island, New Zealand to examine the influence of model parameters. To further calibrate the numerical model this study is being undertaken in conjunction with physical modelling of the Krakatau 1883 eruption at the Indonesian Tsunami Research Center, BPPT, Jakarta. Historical data will also be used to refine and calibrate the pyroclastic flow model. Copyright Kluwer Academic Publishers 2001