Search

Snow-slurry lahars, containing a mixture of water, snow, ice and debris are highly mobile for their typical volumes and occur at a rate much higher than the geologic record indicates. This hazard may cause significant proximal impacts at snow-clad volcanoes even at small volumes. However, the occurrence, footprint and intensity of snow slurry lahar hazards are generally ill-quantified due to the lack of a reliable geologic record, as most evidence disappears after melting. A potential avenue to hazard quantification is through numerical modelling. Here, we quantify snow slurry lahar hazard at Mt. Ruapehu, New Zealand through simulations to obtain lahar intensities and footprints.The most recently active vent at Ruapehu is located underneath the crater lake, which has resulted in a predisposition towards generating snow slurry lahars (e.g. in 2007, 1995, 1975, 1969). The volcano is located within a popular national park and has several ski fields on its slopes, with many assets and visitors potentially impacted by lahars. We applied a multi-phase numerical model to simulate the snow slurry lahars with a high-resolution DEM to estimate the impact of lahars on Ruapehu. Calibration and benchmarking of the model and parameters against 2007 and 1995 snow slurry lahars was used to measure simulation performance, as well as providing insights into lahar initiation and dynamics. Simulations of lahar scenarios using the calibrated parameters were then used to analyse potential impacts at Ruapehu though aggregated lahar pathways, impact footprints and measures of lahar pressure and velocity., The 28th IUGG General Assembly (IUGG2023) (Berlin 2023)

The complexity and level of detail in numerical models for geosciences have increased as computational power has become more available and parallelism and new computational techniques have reduced the computational burden of numerical modelling. However simulations still pose a large computational challenge where a large number of simulations are required to adequately sample the input space. One approach to alleviating these limitations is through construction of a surrogate model. A surrogate (or metamodel) is a model of the simulation results within a defined parameter space. Surrogate models typically learn from simulation results to create a fast approximation to the numerical model. We have been using Generative Adversarial Networks (GAN’s) as a surrogate model to depth-averaged simulations of debris avalanches at Ruapehu volcano, New Zealand. This GAN appears highly successful, creating inundation outlines of volcanic debris avalanches (VDA) with high precision (>90% similarity to simulations in some cases) and more than 130,000 times faster than VDA simulations. Our trials of GANs so far highlight a few key points: (1) well-trained GAN surrogates have high accuracy in reproducing model footprints, (2) the surrogate GANs show better accuracy when constrained to groups with similar simulation inputs (i.e. it is not generalizable), and (3) different loss functions (mean-square error vs. mean absolute error) affect trainability and accuracy. These results, and our attempts to push the limits of GAN’s to problems that consider time-varying inputs, such as bushfire spread, will be highlighted., The 28th IUGG General Assembly (IUGG2023) (Berlin 2023)

Lahars are a class of gravity driven mass flows containing a mixture of volcanic debris and water which can cause severe damage to exposed populations in their path. They are generated through a range of initiation mechanisms during both eruptive and quiescent phases of volcanic activity. Lahars also vary in composition from hyper-concentrated streamflowsto large, water saturated debris avalanches. Flow behaviour, volume and erosional processes differ as a consequence of this variability, affecting robust estimates of lahar hazard and destructive effects. Qualitative and semi-quantitative hazard estimates often rely on geologic and historic records to prescribe lahar volume, type and frequency. This is an important limitation on hazard estimates, as historic data is often incomplete and can be irrelevant under different environmental conditions or when eruptions alter topography and hydrology of volcanic slopes. Lahar risk assessment, which relies on a quantification of lahar hazard and damage, is therefore difficult to estimate. In this thesis, new approaches to quantify the hazard and vulnerability components of lahar risk are developed. For lahar hazard assessment, change points in volcanic eruption record completeness are systematically calculated, along with estimations of uncertainty in the date. The frequency and composition of rain-triggered lahars estimated through a lahar susceptibility model that accounts for the mechanics of lahar initiation. This susceptibility is combined with lahar flow models in a new methodology to determine lahar hazard. The primary cause of building losses is identified using detailed numerical modelling in urban areas. This quantification of vulnerability identified that exposure may have a larger effect on lahar risk than vulnerability. The approaches presented here, while limited at times by scarce input data, can be applied to better understand and quantify lahar risk.