Independence, precursors and earthquake hazard

Abstract

Abstract The problem of combining the hazard from precursors of several different types has been found to be tractable only when assumptions of independence between precursors of different types are made. It is shown that the usual conditional independence assumptions are not realistic for continuous-time hazard estimation. An alternative approach is presented in which the main assumptions are that the proportion of earthquakes unrelated to precursors of a particular type is independent of information on precursors of other types, and that false alarms of a particular type occur at a constant rate independently of earthquakes and other alarms. The method applies to hazard estimation within a domain defined by limits (however wide) of location and magnitude and to a general probability distribution for the time of occurrence of an earthquake following a given alarm of a particular type. The problem is simplified when the precursors concerned have widely differing time-scales.