Ground failure in Adapazari, Turkey during the 1999 Kocaeli earthquake was severe. Hundreds of structures settled, slid, tilted, and collapsed due in part to liquefaction and ground softening. Ground failure was more severe adjacent to and under buildings. The soils that led to severe building damage were generally low plasticity silts. In this paper, the results of a comprehensive investigation of the soils of Adapazari, which included cone penetration test ͑CPT͒ profiles followed by borings with standard penetration tests ͑SPTs͒ and soil index tests, are presented. The effects of subsurface conditions on the occurrence of ground failure and its resulting effect on building performance are explored through representative case histories. CPT-and SPT-based liquefaction triggering procedures adequately identified soils that liquefied if the clay-size criterion of the Chinese criteria was disregarded. The CPT was able to identify thin seams of loose liquefiable silt, and the SPT ͑with retrieved samples͒ allowed for reliable evaluation of the liquefaction susceptibility of fine-grained soils. A well-documented database of in situ and index testing is now available for incorporating in future CPT-and SPT-based liquefaction triggering correlations.

When geotechnical 1D ground response analyses are performed to evaluate site effects in lieu of the use of attenuation relations, it is with the expectation that the attenuation model's standard deviation, and any bias in the median, would be reduced. In this article, we evaluate the degree to which these and other benefits of equivalent-linear, 1D ground response analyses are realized and develop recommendations for implementing the results of such analyses into hazard calculations. This is accomplished by comparing response spectral accelerations (S a ) from recordings to predictions derived using ground response analysis procedures as well as attenuation relationships with and without amplification factors. The results are compiled for 134 motions from 68 sites, and prediction residuals are interpreted to assess the models' relative bias and dispersion. We find that ground response analyses are unbiased for T Յ ϳ1 sec, but underestimate longer period S a in deep basins. For soft soils, ground response analyses reduce dispersion for T Ͻ 1 sec relative to alternative models. This dispersion reduction is not observed for other site categories nor at longer periods. These results suggest that ground response analyses are beneficial for S a predictions at soft soil sites, but generally provide no identifiable benefit for typical stiff soil or rock sites.

No abstract

No abstract

The M w 7.1 Duzce earthquake occurred on 12 November 1999 along the North Anatolian Fault in northwestern Turkey. This paper documents observations from a field reconnaissance team, addressing two principal aspects of this significant earthquake: the recorded ground motions and the distribution and severity of the earthquake effects on the built environment. In general, the recorded ground motions from this earthquake were smaller than predicted by ground motion predictive equations available at the time of the event. One anomalous recording is presented and potential causes for this irregular motion based on observations from field reconnaissance are discussed. The effects of rupture directivity on the near-fault recordings are assessed and the effects of soil conditions on the recorded ground motions are examined. The patterns of building damage based on post-earthquake reconnaissance are presented for the most strongly shaken cities in the near-fault region: Duzce, Kaynasli, and Bolu. Damage in Duzce was concentrated in the southern part of the city, which is underlain by softer sediments. Damage in Bolu was distributed evenly throughout the city; whereas damage was concentrated on more recent alluvial sediments in Kaynasli. No evidence of liquefaction or ground failure was observed in the populated areas surveyed after the earthquake.