Limitations of Deterministic Modelling of Slope Stability on Volcanic Edifices

The conditions leading to the 18 May 1980 sector collapse of Mount St Helens have been the subject of a number of detailed investigations. Preservation of the initial failure plane(s) allowed Voight et al. (1983) and Donnadieu et al. (2001) to undertake back analyses and determine a range of possible failure conditions. While the models proposed offer major insights into potential failure mechanisms, we will demonstrate that deterministic analyses are of limited usefulness because many of the model parameters, such as cohesion, internal friction and pore pressure, are very poorly constrained. This creates problems of non-uniqueness in the solution. An alternative approach involves a series of Monte Carlo simulations to identify potential combinations of parameters that will produce the observed failure plane. Initial input ranges are specified for each parameter and the predetermined model is run repeatedly, with the parameter values for each model selected at random from within the input ranges. The interaction between parameters can be examined in detail, providing a better understanding of the potential failure conditions. This approach, which has been tested initially on a theoretical slope with predetermined failure conditions, highlights the fact that it is impossible to generate a unique model that fits the data when the slope has poorly defined strength parameters. This has clear implications for the validity of commonly used deterministic approaches. This probabilistic back analysis approach has been used to reanalyse the conditions that led to the May 18 collapse on Mount St Helens. Donnadieu, F., Merle, O., and Besson, J.C., 2001, Volcanic edifice stability during cryptodome intrusion, Bulletin of Volcanology, vol 63, p61-72. Voight, B., Janda, R.J., Glicken, H., and Douglass, P.M., 1983, Nature and Mechanics of the Mount St-Helens Rockslide-Avalanche of 18 May 1980, Geotechnique, vol 33, p243-273.