To establish seismic design criteria for buildings in Thailand, probabilistic hazard analysis requires estimation
of ground motion intensity such as peak ground acceleration (PGA) or spectral acceleration. This estimation
process needs to use an attenuation relationship, which provides PGA estimates as a function of earthquake
magnitude, distance, and other seismic parameters. However, Thailand does not have enough strong motion
records to develop a reliable attenuation model for the region, so existing models from other parts of the world
have to be adopted. Unfortunately, different available attenuation models provide significantly different
estimates at large distance, which is the case for sites in Thailand. Thus, the appropriate model for Thailand
should be determined by comparing PGA estimates by each model to ground motions recorded in Thailand. In
this study, twenty attenuation equations previously developed for shallow crustal earthquakes in both active
tectonic and stable continental regions, and for subduction earthquakes were evaluated using a total of 163
ground motions recorded by Thai Meteorological Department (TMD) from 45 earthquake events between July
2006 and July 2007. PGA estimated by attenuation models for earthquake magnitude ranging from 4 to 7 were
plotted and compared to the field records. The square root of mean of square (RMS) of the differences between
estimated PGA and actual records was also computed for each attenuation model to quantify how well the
model predicts ground motions. It was found that the attenuation models proposed by Idriss (1993), Sadigh et
al. (1997), and Toro (2002) have the lowest RMS for shallow crustal earthquakes. And the model by Crouse
(1991) has the lowest RMS for subduction earthquakes. Therefore, these models are the most suitable
attenuation models for Thailand.