APPLYING THE RELATIVE HYPOCENTER LOCATION APPROACH - WHERE WAS THE 1980 NOVEMBER 23 IRPINIA EARTHQUAKE

The purpose of this work is to evaluate under what conditions it is fe asible and with what accuracy it is possible to locate the nucleation point of a large earthquake, given the availability of aftershocks loc ated with high precision by the deployment of a local network. We expe riment with several approaches and apply them to the location of the e picentre of the 1980 November 23 Irpinia earthquake (M(W) = 6.9). Firs t we use local P-g phases selected to optimize the azimuthal coverage, obtaining a well-constrained location with a small statistical error, which typically underestimates the true hypocentre uncertainty. We th en exploit the relative location technique, obtaining stable, almost c oincident solutions under three conditions: (1) using multiple indepen dent master events to derive an average epicentre; (2) fitting simulta neously the larger data set for all available master events, using a f orward approach; (3) conducting an a priori evaluation of the statisti cs of station and master events to separate model uncertainties and im prove the statistical accuracy of the relative locations. Moreover, on ly by introducing station statistics can we achieve the desired accura cy of congruent to 1 km in constraining the rupture nucleation point o f this large earthquake, and we show that the application of the relat ive location technique to uncleaned, unweighted data for a single mast er event provides only a crude epicentre with a confidence ellipse dec eivingly smaller than the true hypocentre uncertainty. The revised epi centre for the 1980 November 23 Irpinia earthquake (48.803 degrees N-1 5.302 degrees E) validates the class of multidisciplinary reconstructi ons of the source process such as the model of Valensise et al. (1989) , based on the hypocentre of Westaway & Jackson (1987), and is shifted by almost 13 km to the NW of the epicentre recently proposed by Westa way (1992).