- Ground-motion model is:
PGA is in cm∕s2, β1 = 0.8807, β2 = 0.7056, β3 = -2.8645, β4 = 0.0923, τ = 0.1977 ± 0.0067
(inter-event), ϕS2S = 0.0915 ± 0.0240 (site-to-site), ϕSS = 0.1164 ± 0.0032 (single-station intra-event) and
ϕR = 0.0577 ± 0.0017 (remaining intra-event). Also report station effects but not reported here due to
- Focal depths, D, between about 0 and 9km.
- Data from 10 different stations of ICEARRAY I, which are all located on a lava layer.
- Remove poor quality records.
- All data from aftershocks of 2008 Ölfus earthquake (Mw6.3).
- Use a Bayesian hierarchical model to determine the various components of ground-motion variability:
event, station, event-station and unexplained effects. Use a Matérn covariance function (with 3 parameters:
decay, smoothness and amplitude parameters) to model a mean-zero Gaussian spatial fields. Use a Markov
chain Monte Carlo to determine the coefficients.
- Verify that the determined coefficients and variabilities are stable.
- Examine inter- and intra-event residuals w.r.t. ML, R, D and back-azimuthal angle and find no trends.
- Examine correlations between station effects. Find effects are highly correlated for close-by stations.