Modelling the Roermond earthquake of 1992 April 13 by stochastic simulation of its high‐frequency strong ground motion

Abstract

The Roermond earthquake of 1992 April 13 is one of the most important seismic events for assessing potentially destructive strong ground‐motion parameters in central Europe. Owing to the lack of unclipped near‐source records, however, the source depth as well as the high‐frequency signal contents are only known to within large uncertainties. Using a stochastic simulation technique, site‐dependent synthetic acceleration seismograms and simulated Wood‐Anderson records for the Roermond Earthquake were calculated for the epicentral region and a site in 60 km distance. For the distant site, three different damping models were modelled while for the epicentral region, a tertiary and quaternary coverage of 1500 m was assumed. Both constant‐ and frequency‐dependent Q were considered. the resulting simulations were tested for their consistency with (a) epicentral acceleration between 1.0 and 2.0m sec−2 as indicated by the low observed epicentral intensities of slightly above VII, (b) the observed peak ground accelerations at a distance of 50‐60 km which are reported to be 0.3 ± 0.1 m sec−2, and (c) the observed Wood‐Anderson magnitude of MWA≈ 5.9±0.2. While the stochastic simulation does not result in a single unique model, it allows the comparison of certain models (including their uncertainties) in terms of their consistency with the observations. the most pronounced result shows that of the three different source depths (13, 21 and 25 km) which were investigated, the shallow source is by a wide margin the least consistent with the observations, independent of the remaining model assumptions. On the other hand, peak accelerations of 0.3 ± 0.1 m sec−2 in 50‐60km distance can be explained by reasonable combinations of seismic moments and corner frequencies for all the site‐damping models. While under the constant‐Q assumption, the most consistent M0/fc combinations show a clear trade‐off between fc and M0 for M0≈ (5‐10) × 1016 Nm, this trade‐off is strongly reduced if Q is assumed frequency dependent as Q=Q0×fα. For an average seismic moment of M0= 7.5 × 1016 Nm, a corner frequency of 1.0–1.25 Hz provides the maximum joint consistency scores independent of the other model assumptions. Copyright © 1994, Wiley Blackwell. All rights reserved