Skip to main content Skip to footer content

How good are shear wave velocity models obtained from inversion of ambient vibrations in the Lower Rhine Embayment (N.W. Germany)?

M. Ohrnberger, F. Scherbaum, F. Krüger, R. Pelzing and S.-K. Reamer

Abstract: 

We present the results of independent tests on the reliability of shear-wave velocity models derived from ambient vibration measurements in the Lower Rhine Embayment (LRE) in north-western Germany. For one site, we compare the shear-wave velocity model obtained from active downhole measurements directly to our result of the inversion of ambient vibration data. We find a remarkably good agreement between the independently derived shear-wave velocity profiles for this site. At another location, we tested the proposed shear-wave velocity model derived from ambient vibration measurements by using local earthquake data recorded at a 350 m-deep borehole station and at a strong motion station at the surface. We tested the accuracy of the shear-wave velocity profile by modifying the shallow part of a minimum 1 D reference model in the LRE, to theoretical model seismograms for a local earthquake (ML = 4.1) and compared those to the earthquake records observed. We further computed the top/bottom spectral ratios for 11 local earthquakes and compared them to the SH transfer function predicted from the ambient vibration model. These tests show some deviation of the proposed shear-wave velocity model and the evidence obtained from the local earthquake records. The mismatch in fundamental peak frequency from 0.62 Hz predicted for the ambient vibration model to approximately 0.50-0.55 Hz obtained from the spectral ratios can be explained by an underestimation of the sediment thickness. Two strong spectral peaks were observed at 3.8 Hz and 10.2 Hz, which the model fails to predict. The limitations of the model derived from ambient vibrations can be attributed to the simplified model parametrization (smooth power-law function) used in the inversion procedure. A more flexible parametrization of the model in the inversion procedure may allow for a better resolution of the very shallow and deeper parts of the velocity profiles.