Site-dependent spectra from the 1999 Turkey earthquakes considering different sets of strong-motion data
Abstract
In the last decade extensive strong-motion measurements were carried out by the reconnaissance team of the German TaskForce for Earthquakes. Aftershocks were recorded by a mesh of temporarily installed strong-motion recorders during the field surveys to Turkey in 1998 (Adana/Ceyhan), 1999 (Izmit/Kocaeli and Duezce/Bolu) and 2002 (Sultandagi/Afyon). Particularly the two missions in 1999 provided a unique and comprehensive database of recorded aftershocks at sites where building damage occurred. In addition, post-earthquake investigations were initiated around the provinces Adana and Kocaeli in October 2000 to gain more insight into the effect of local site conditions. Therefore, microtremors were recorded at sites of the previously installed strong-motion accelerographs and locations of evident concentration or remarkable scatter of building damage. By applying H/V-spectral ratio method on microtremor data, a classification of the recording sites (into soft soil, stiff soil, rock-type conditions) was performed. For the derivation of attenuation laws different databases were applied. In addition to the aftershock records of the strong-motion stations of German TaskForce for Earthquakes (Schwarz et al., 2002), a small dataset of main- and aftershocks from the 1999 Turkey earthquakes provided by the Kandilli Observatory (KOERI, 2002) and the General Directorate of Disaster Affairs (AFET) could be implied. The magnitude-distance composition of both datasets are quite different in terms of the covered magnitude and distance range. While the dataset of the German TaskForce mainly consists of small magnitude aftershocks (at the present state of elaboration: Ml <= 4.9) being recorded in epicentral distances Re < 70 km, the KOERI dataset is featured by larger magnitudes (Ml = 4.8-7.2) and by a distance range of about 10--250 km. The aftershock database of German TaskForce consists of 538 triaxial acceleration records (rock 53, stiff 52, soft 433), while that of KOERI comprises 145 triaxial acceleration records (rock 6, stiff 36, soft 103). On this basis, three sets of strong motion records were investigated by an one-step as well as a two-step regression analysis (similar to the approach by Ambraseys et al., 1996). Furthermore, parametric studies with respect to the lower bounds of magnitude and source depth were performed. Results indicate that the composition of the dataset and in particular the decision on the lower bound magnitude significantly determine the qualitative spectrum shape. In comparison to the spectra determined by Ambraseys et al. (1996) attenuation functions, the recorded data lead to significant lower accelerations for the borizontal as well as for the vertical components. Furthermore, no significant differences between data from soft and stiff soil recording sites can be observed, supporting the authors opinion that the severity of ground motion during earthquakes is less responsible for the high extent of building damage than the vulnerability of the building stock (which seem to be quite different for one particular building type in dependence on the date of construction). Therefore more attention should be attributed to the investigation of the building inventory.
- Publication:
-
EGS - AGU - EUG Joint Assembly
- Pub Date:
- April 2003
- Bibcode:
- 2003EAEJA.....6839S