Kernel estimation methods for seismic hazard area source modeling

Abstract In probabilistic seismic hazard analysis, the representation of seismic sources by area zones is a standard means of data reduction. However, where the association between seismicity and geology is complex, as it is in many tectonic regimes, the construction of zone geometry may become contentiously subjective, and ambiguities may end up being resolved through appeal to the nonscientific rule of conservatism or pragmatism. Although consideration of alternative zonations within a logic-tree framework provides a channel for some of the uncertainty, it does not address the fundamental validity of the zonation procedure itself. In particular, neither the minimal assumption of uniform seismicity within a zone nor the Euclidean geometry of a zone accord with the fractal spatial distribution of seismicity, and the magnitude insensitivity of zonation ignores the spatial extent and correlations of different-sized earthquakes. An alternative procedure for area source modeling avoids Euclidean zones and is based statistically on kernel estimation of the activity rate density inferred from the regional earthquake catalog. The form of kernel is governed by the concepts of fractal geometry and self-organized criticality, with the bandwidth scaling according to magnitude. In contrast with zonal models for intraplate regions, the kernel estimation methodology makes provision for moderate earthquakes to cluster spatially, while larger events may migrate over sizeable distances.