Heaton [1991] reviewed near-source models of earthquake rupture and pointed out that, in contrast to the prevailing view, the duration of slip at a point on the fault (the rise time) is short in comparison to the overall duration of faulting. Heaton argued that slip on the fault was ``self-healing'' leading to phenomena such as partial stress drop, negative stress drop, and very high velocity pulses in the near field.
This observation and interpretation has lead to interesting studies on the effect of velocity-dependent friction on dynamic rupture [ Cochard and Madariaga, 1993; Madariaga and Cochard, 1993], which can self-heal and propagate as a pulse. Brune et al. [1993] have argued that interface separation observed in foam rubber sliding experiments may occur during earthquakes and could provide an explanation for both short rise times and the stress-heat flow paradox. In their model, little frictional heat would be generated because the fault interface would separate during slip. Castro et al. [1991] cite observations of high spectral amplitudes for P waves relative to S waves as evidence of a component of tensile failure during earthquake faulting.
Heaton [1991] has made a convincing case for short rise times and it is clear that the rise time can not be controlled by the overall width of the fault. What has not yet been demonstrated is whether the rise time is short compared to the scale-length of slip heterogeneity during rupture. This possible explanation for observations of short rise times has been suggested by several authors [ Bouchon, 1978; Beroza and Spudich, 1988]. Progress in this area is most likely to come from what might be called quasi-dynamic models of earthquake rupture.