Composite source model parameters for large earthquakes (M greater than 5.0) in the Mexican subduction zone

We propose that synthetic seismograms calculated using synthetic Green's functions and a composite source model may be accurate enough to use in seismic resistant design calculations in the future for subduction zone regions.;In this case study we computed synthetic seismograms for 16 earthquakes (M 5.0-8.1) to compare to data collected from the Guerrero Accelerograph Array located over the Mexican subduction zone. Comparing the synthetic to the observed seismograms tested the accuracy of the synthetic seismograms.;Source parameters that were varied to find the best synthetic seismograms are subevent stress drop, maximum subevent radius and the perturbation to kappa. The subevent stress drop decreased with increasing magnitude. The maximum subevent radius increases approximately exponentially up to about 25 km (limited by the width of the fault) with increasing magnitude. The log of the ratio of observation to prediction of peak acceleration is defined as the residual for that component. The standard deviation of the residuals for the horizontal components are comparable to the standard deviation in a recent regression analysis. This supports that these predictions may be useful for future seismic hazard calculations. Subevent stress drops for the Guerrero earthquakes were larger than the subevent stress drops for the Imperial Valley, San Fernando, Whittier Narrows and Northridge, southern California earthquakes. Thus when this model is applied to new tectonic regions, it is necessary to calibrate the parameters before the prediction can be used.