These are examples of predicted PGV for the Landers and Northridge earthquakes, and can be directly compared to the TriNet maps of slide 5.  The contours are in cm/s and the red numbers show the values recorded at local strong motion stations.  The contour maps were constructed from predicted values only at the locations of the actual stations, although it is possible to calculate PGV at points where there are no stations.  The lower panels compare the data (red circles) to predictions (black squares) as a function of distance.  The model predictions are found to describe the overall attenuation quite well however there are deviations due to unaccounted for Earth structure such as Los Angeles Basin in the Landers comparison.
	Generally the regions of strong shaking (PGV greater than 10 cm/s) compare very well between the observed TriNet maps (slide 5) and the predicted maps (above).  The predicted map for Landers tends to over estimate the value at LUC which is in part due to the methodology and in part due to an artifact of our simple single plane model for Landers in which the station is located only 200m from the fault when it should be located approximately 3 km away in reality.  This is a short coming of the method however as we will show later in the generation of combined data/model maps it is a problem that can be circumvented.  The Northridge map (above) is also found to agree well with the TriNet map (slide 5) in terms of the area of the PGV greater than 10 cm/s and the fact that the strongest ground motions are located at the northern edge of the San Fernando Valley and to the north.