Observations from regressions on ground motion data from Britain and north-western Europe : ID 1741 [abstract]

Regressions were carried out on two peak ground acceleration datasets. The first dataset contains observations from UK earthquakes ranging from 0.7 to 5.1 ML recorded over distances of 1-441 km. The second contains data from earthquakes of similar magnitude (1.9-5.9 ML) compiled by Free et al. (1998) for north-western Europe. Individual regressions on these datasets suggest that attenuation in these two regions is similar. In light of this, the datasets were combined and further regressions carried out. We compared these results with the relations of Ambraseys et al. (1996), which are based on data from larger earthquakes and which have been applied in UK hazard assessments. At magnitudes less than around 4.9 ML, our relation estimates realistic ground motions whereas those calculated using Ambraseys et al. (1996) are signicantly higher. At 5.5 ML and above, our relation predicts unrealistically high accelerations (up to three times higher than those estimated using Ambraseys et al., 1996). However, between 4.9-5.0 ML, the two relations coincide for distances of 50 km or less. This result implies that there is some effect of magnitude that neither relation accounts for. For the realistic prediction of ground motion for the entire range of magnitudes, a hybrid approach may be necessary