Ground-motion model is not given here since it requires evaluation of a complex equation that cannot be
summarised. Derive model using an adaptive network-based fuzzy inference system and the random-effects
algorithm to split the variability into inter- and intra-event components. Provide details but these are not
given here.
Characterise sites by Vs,30: values 110–1540m∕s.
Use data from the NGA-West2 database. Exclude earthquakes with focal depth > 25km and aftershocks.
Exclude non-free-field stations. Also exclude sites without measured Vs,30. Most remaining data from
Japan, Taiwan and California.
Provide cumulative distribution functions of dataset. 80% of data from 4.0 ≤ Mw≤ 7.0, 10 ≤ rjb≤ 200km
and 210 ≤ Vs,30≤ 650m∕s.
Data from 580 different sites.
Examine residuals on normal probability plots and find that they conform to the lognormal distribution.
Test robustness of result by re-deriving model using only half of records. Find predictions are almost
unchanged when using only half the records.
Find limited evidence for nonlinear site amplification.