Development of a suite of stochastic ground-motion models for the United Kingdom

Since 1995, various estimates of stochastic ground-motion parameters have been computed for UK earthquakes and a few UK stochastic models proposed. These models have been developed by inverting the available weak-motion data to estimate ranges for the key parameters and using expert judgement and evidence from other regions when data are insufficient. The resulting ground-motion models have been used within site-specific seismic hazard assessments for critical infrastructure and for the 2020 UK National Seismic Hazard Model developed by the British Geological Survey. Often stochastic models have been given a lower weight within these assessments than empirical models from other regions, particularly due to doubts over how the stochastic models scale to larger magnitudes. As part of a broader project to develop a backbone ground-motion model using a hybrid stochastic-empirical method, here we present a summary of analysis conducted using an expanded ground-motion database from the UK and surrounding region to determine stochastic parameters. The ground-motion data have been adjusted to a single rock condition using an approximate technique. We used an approach to determine the stochastic models that is appropriate for their final use, namely within a scaled backbone approach that provides a suite of consistent models with appropriate weights. Due to the trade-off amongst the key parameters (e.g., stress (drop) parameter, geometrical spreading and site attenuation), constraints from the literature and expert judgement are applied. The resulting suite of models captures the uncertainties inherent in the inversion owing to the limited magnitude, distance and structural period range of the ground-motion data. These models will be the basis of a UK ground-motion model due for completion in 2023.