A Critical Examination of Near-Field Accelerograms from the Sea of Marmara Region Earthquakes

by Sinan Akkar and Polat Gülkan

Abstract

In 1999, Turkey was struck by two major earthquakes that occurred 86 days apart on the North Anatolian Fault system. Both earthquakes had right-lateral strike-slip mechanisms with moment magnitudes greater than 7. The number of strong-motion records obtained from the Kocaeli earthquake (17 August 1999, M_w 7.4) was 34. The second event, designated as the Bolu–Düzce earthquake (12 November 1999, M_w 7.2), triggered 20 instruments. Among the records that we have from these earthquakes, seven are from near-source ground-motion data. These records were obtained from the cities of Gebze (GBZ), Yarimca (YPT), İzmit (IZT) (capital city of the province of Kocaeli), Adapazari (SKR) (capital of the province of Sakarya), Düzce (DZC) (shaken strongly in both the events), and Bolu (BOL). In many of these urban centers, extensive structural damage was observed. Although these near-field data have greatly expanded the strike-slip near-source ground-motion database worldwide for M_w >7 earthquakes, they represent a blurred image of the actual severity of the ground motions in the epicentral area because of the sparseness of the national strong-motion network and the unrepresentative geologic conditions at the recording sites. We examine the records to determine whether they provide clues about the extensive damage on the housing stock in the epicentral region. The goal is tackled with earthquake structural engineering criteria in mind, using the drift spectrum as the primary yardstick. There appears to be conflicting evidence that the fault-normal (FN) direction should represent a greater damage-causing potential when this potential is based on ground-story drift spectra. The component with larger ground velocity does correlate better with the component with larger drift demand, but this does not always coincide with the FN direction. The period of the peak velocity pulse matches the structural period where the drift demand is the largest. Further refinements of code requirements that consider this effect are in order.

Return to Table of Contents