Nonparametric Approach to Analysis of Space-Time Data on Earthquake Occurrences

As an alternative to traditional, parametric approaches, we suggest nonparametric methods for analyzing spatial and temporal data on earthquake occurrences. Nonparametric techniques are particularly adaptive to anomalous behavior in the data and provide a new way of accessing a variety of different types of information about the way in which both intensity and magnitude of events evolve in time. They can be employed to estimate the spatial trajectory of event clusters as a function of time, and to define quiescent and active periods. The latter application suggests new approaches to forecasting high magnitude events. Our methods are founded on multivariate techniques for curve and surface estimation, particularly in contexts where curves or surfaces are unbounded at points or along lines.