Abstract - Clustering of paleoearthquakes in space and time has for a number of years been discussed in the literature and may be a widely occurring phenomenon (e.g., Wallace 1984, Coppersmith 1989, Marco et al. 1996). In many fault systems however insufficient paleoearthquake data are available to determine whether clustering occurs on most faults and therefore to describe how it varies between faults. In the absence of these data the underlying controls of the clustering process cannot be assessed. The principal aim of this work is to determine what causes clustering of large prehistoric earthquakes over timescales of tens of thousands of years. To address this question we utilise data for active normal faults in the Taupo Rift, New Zealand.
Data on the earthquake history of faulting is provided by displacements measured on up to 11 radiometrically dated horizons (ca. 2-22 kyr in age) in 24 trenches excavated on fault traces and distributed across the rift. The largest faults in the system (displacement rates 0.1-1.6 mm/yr) have been trenched and the strain recorded in the sample population is inferred to record approximately 40% of the total extension across the rift. In each trench older horizons generally display greater displacement, however, the precise form of the relation between displacement and elapsed time varies between faults. These variations are indicated by displacement profiles ranging from step functions, with episodic slip accumulation, to near-linear trends with constant displacement rates. The wide variation in the shape of displacement profiles is inferred to reflect a significant difference in the paleoearthquake history of events on each fault. Those displacement profiles characterised by step-functions, which account for between one third and a half of all faults sampled, are consistent with the notion that earthquakes on these faults experienced some degree of temporal clustering. Time periods of rapid slip accrual typically range from 5-10 kyr and are separated by intervals of similar duration in which little activity occurred. The degree of temporal variability of slip events is matched by spatial fluctuations in fault slip with adjacent faults having both comparable and dissimilar displacement profiles.
In the Taupo Rift the key to understanding how slip accumulates on individual faults lies in the recognition that each is an element of a larger kinematically coherent system. Over timescales of 5-10 kyr spatially local anomalies in displacement are removed when displacements for a given horizon are aggregated across the rift. These data suggest that on timescales of thousands of years all faults in the rift are in communication and support the view that the boundary conditions for the rift are stable. The occurrence of large earthquakes in the fault system are, therefore, highly ordered and appear to be predictable over some temporal and spatial length scales.
Abstract of talk given to:
New Zealand Geological Society, Wellington, June 2004