Abstract
This paper describes a comprehensive study of the attenuation of Modified Mercalli intensity in New Zealand earthquakes, which has resulted in significantly different and new findings when compared with those of earlier work. The current study used recently revised magnitudes for the 30 events in the carefully selected data set. Magnitudes ranged from M~L~ = 5.0 to M~s~ = 7.8. Special effort was also put into establishing the depths of the events, which ranged from very shallow to 65 km. An expression of a form also used in peak ground acceleration attenuation studies was adopted, I = a + bM + cr + __d__log~10~r with r being taken as the mean distance from the centre of the fault rupture surface to each isoseismal. A twoβstep stratified regression analysis was used because it results in more realistic estimates of standard errors, although its mean curves were virtually the same as those derived from the standard oneβstep method. It was found possible to model accurately the attenuation of the deepest events from regressions of the data of the shallowest events, but the inverse was not true. In a study of the influence of source mechanisms it was found that the attenuation was the same for events with normal and strikeβslip faulting. In contrast, events with reverse fault mechanisms were found to have higher intensities (for the same M and r) than normal and strikeβslip events, by a factor corresponding to that found by Campbell in a study of peak ground accelerations, although the statistical test was narrowly short of the conventional significance level.