Different seismic source models are used to estimate regional seismic hazard. Commonly used point, line, and area seismic sources are considered in addition to a new method which is obtained by modifying the line source model to take into account the uncertainty associated with the exact location of the line (i.e., fault). The results are presented in terms of cumulative functions of peak ground acceleration for major sites in the Azerbaijan Province of northwest Iran. Iso-acceleration maps for two different return periods are also developed for each seismic source model and a comparison is made among the results of the models.
The point source model is shown to be unrealistic when used to model large shocks (Ms > 6.5), which correspond to long ruptures. The model cannot incorporate the fault length, thus ignoring possible spatial migration of seismicity along the fault. In addition, the actual attenuation of ground motion departs considerably from that associated with point source assumption.
The conventional line source model, while providing a good representation of vertical strike-slip faults, cannot accurately model the seismicity in other cases, such as reverse faults in general, and thrust (low angle reverse) faults in particular. Epicenters for these latter cases do not lie along a line, as they do in case of vertical strike-slip faults.
The area source model is used for those cases where the distribution of earthquake epicenters in a region does not follow any identifiable geological fault pattern. The spatial migration of seismicity along an active fault during a given exposure time is of vital importance in seismic hazard analysis. An analysis based on an area source model corresponds to assuming this migration will be equal in all directions. The theory of plate tectonics, however, suggests an elongated narrow zone corresponding to each fault.
A fault line model is developed which exhibits less sensitivity of near-field ground motion to precise fault location than the line source model. This model is referred to as the strip source model. According to this model, the seismicity on a fault is spatially distributed in a long and narrow zone along the margins of the corresponding plates or microplates, and decreases with distance from the fault on either side. It is believed that this kind of modeling closely represents the seismicity corresponding to interplate earthquakes, especially when the type of faulting is thrust. Uncertainties due to the location and orientation of faults will be considerable, particularly for the buried faults, and these uncertainties can be incorporated in the strip source model.