DURATION AND AMPLITUDE OF SW JAPAN TREMOR AND THEIR RELATION TO SLOW SLIP EVENT CHARACTERISTICS

Despite the observed correlation between slow slip events (SSEs) and seismic tremor along the Cascadia and SW Japan subduction zones, it remains unclear what is responsible for this association. If, as accumulating evidence suggests, tremor is produced by shear failure on the thrust interface driven by aseismic slip, we might expect to see a relationship between the size and/or frequency of tremor activity and accumulated slip. Reduced displacement calculations for tremor in Cascadia indicate that average values for tremor reduced displacement are not correlated to the amount of slip observed during ETS events (McCausland et al., 2005). However, recent results from Cascadia hint at a relationship between the duration of tremor and the inferred amount of aseismic slip in ETS events (Aguiar et al., 2006). We present a detailed investigation of subduction tremor from SW Japan in order to assess whether similar relationships between slip magnitude and tremor amplitude and duration exist in this region.

We calculate reduced displacement for bursts of tremor during SSEs in western Shikoku, and the Kii and Tokai peninsulas. Average values for reduced displacement range between 0.23 - 0.58 cm2, much smaller than most observations of volcanic tremor and similar to values calculated for subduction tremor in Cascadia (McCausland et al., 2005). Average amplitudes for subduction tremor in Japan do not vary significantly between episodes of peak activity in a given region and do not appear to scale with magnitude of slow slip. There is a slight variation in average amplitude between the three regions studied, which may reflect regional differences in frictional properties.

To investigate how the productivity of tremor relates to slip during episodes, we take two approaches. In the first we treat each burst of tremor as a discrete event and compare the number of events with SSE parameters. In the second approach we consider total durations of tremor, under the assumption that each burst is actually the superposition of many small events. In addition, we investigate magnitude-frequency (and duration-amplitude) relations for SW Japan tremor to assess whether they are better fit by exponential or power law distributions, representing scale-bound and scale-invariant source processes respectively.