In a fault rupture model used in seismic hazard mappng, changes in assumptions regarding fault and rupture lengths, the Gutenberg-Richter b value, minimum and maximum magnitudes, and acceleration variability may significantly affect accelerations estimated at a site, but the effects vary greatly depending on the location of the site relative to the fault and on the acceleration level. For a given attenuation function, modeling earthquakes along a linear fault as finite length ruptures, rather than as point sources, may cause the acceleration with a specified return period to increase by 25 to 50 per cent at a site near the center of the fault, but only negligibly at a site near the end of the fault. However, if the fault is fairly short, the treatment of rupture lengths may also affect accelerations determined at sites near the end of the fault. Increasing the minimum magnitude m0 by one unit from m0 = 4.0 to m0 = 5.0 may decrease the accelerations for short return periods at a site near the fault by a factor of 2 to 10; the same change in minimum magnitude will have a smaller effect on the return period of a fixed acceleration for a site further from the fault. Including acceleration variability may increase the return period of lower accelerations at a site, but decrease the return period of higher accelerations by a factor 1.4 to 2.0, and by considerably more at the highest accelerations. If the total number of earthquakes is held constant and the Gutenberg-Richter b value is varied, an increase of 10 per cent in the absolute value of b may cause a 20 to 30 per cent decrease in the acceleration calculated for a given return period. Changing the assumed maximum magnitude may significantly affect return periods of higher accelerations at a site; the size of the effect depends upon the attenuation function used.
In the fault rupture model being analyzed, ground motion at a site is a function of earthquake magnitude and closest distance from the site to a rupture along a linear fault segment; ruptures must be contained wholly within the fault, and occur with equal probability at all allowable rupture locations. The model is compared with a rupture model in which the rupture may extend beyond the end of the fault by one-half rupture length, and also with a point source model in which earthquakes occur as points along a linear fault.