Comparing Loss Estimation with Observed Damage: A Study of the 1999 Kocaeli Earthquake in Turkey

Loss estimation from future earthquakes is of growing importance in planning earthquake protection strategies in high-risk areas. Loss models based on the spectral displacement approach are now widely used because of generally acknowledged deficiencies in earlier approaches using macroseismic intensity or peak ground-motion parameters. However, there has been to date rather little earthquake damage data by which the new generation of models can be assessed and which can be used to calibrate the parameters involved. The availability of several detailed damage surveys carried out following the 1999 Kocaeli earthquake in Turkey, provides a rare opportunity for such an assessment. In this paper the losses which would be predicted from two different approaches to loss assessment – one using predicted macroseismic intensity, the other using the spectral displacement method – are compared with actual observed losses in the Kocaeli event at two different locations where surveys were carried out. One of these sites was very close to the surface fault rupture (< 3 km distance), the other at a distance of about 4.5 km. It is shown that the predictive methods available generally overestimated the losses at these distances, and a number of possible reasons for these discrepancies are considered. The sensitivity of loss estimates to variations in the key parameters governing the estimation in each case are explored, in particular with respect to modifications in the parameters of the attenuation relationships and the vulnerability parameters. The implications of these results for estimating future losses are discussed. This revised version was published online in July 2006 with corrections to the Cover Date.