have worked out a new attenuation relationship applicable to northwestern Turkey based on recent ground motion data recorded during the 1999 Turkey earthquakes and a few of their following events. Because of our own work along similar lines [2,3] we believe the evolution of ground motion predictions for the same or similar regions is of interest even though our own models are intended to be applicable to the whole of Turkey.
Ground motion predictions based on records made in Turkey are affected by two influences. One is the relative uncertainty that shrouds the local site character. A good many sites do not have borehole or other data to establish their local site descriptor. The other is the interference from the buildings where the sensor may have been placed. In spite of these shortcomings, we have utilized available data from 47 stations that recorded 93 horizontal components during 19 distinct events to report prediction expressions in [2]. We have now updated this study by considering 57 earthquakes during 1976-2003 containing 112 records with 223 horizontal acceleration time series, to be reported in [3]. In 1999, Kocaeli earthquake data comprises 46 and 35% of our database assembled in [2,3], respectively. Ozbey et al. is based on