Improved handling of uncertainties in fault zone hazard estimation

Technical Abstract

Uncertainties in data and parameter values have often been ignored in hazard estimates based on recurrence-time modelling of fault zone rupture. Here, a mixture of distributions approach is used to handle uncertainties in parameters estimated from the geological and historical earthquake record of a fault zone, while a mixture of hazards approach is used for parameters estimated from a set of similar faults and for data uncertainties. The former approach admits updating of the distributions for uncertainty as time passes, whereas the latter approach does not.

The proposed methods are described in detail for the exponential and lognormal models. A formula for the expected hazard, when the time of the most recent event is uncertain, is derived for the lognormal model. The methods are applied, by way of illustration, to selected faults, namely the Mojave segment of the San Andreas fault, California and the Wellington-Hutt Valley segment of the Wellington fault, New Zealand. The resulting hazard is presented as a single value which takes account of both data and parameter uncertainties, not a range of values.