Linear vs. Non-Linear Earthquake Location and Seismogenic Fault Detection in the Southern Tyrrhenian Sea, Italy

We report a comparison between the performances of linearized and non-linear hypocenter location algorithms working in 3D velocity structures. For this purpose, we used the SIMUL linearized location method by Evans et al. (1994) and the BAYLOC non-linear grid-search probabilistic algorithm by Presti et al. (2004). Comparisons are made using the datasets of P and S readings relative to the two main seismic sequences occurring in the last ten years in the southern Tyrrhenian sea, i.e. the 1998 sequence of maximum magnitude 5.2 near Ustica island and the 2002 sequence of max. magnitude 5.9 offshore Palermo city. We find that in the relatively poor network conditions of both sequences the SIMUL and BAYLOC algorithms produce hypocenter locations of comparable accuracy, while location error estimates from SIMUL are generally less accurate than BAYLOC's. This result is a confirmation in a 3D velocity structure of a finding already reported by previous investigators who compared the performances of linear vs. non-linear location algorithms in 1D structures (Lomax et al., 1998; Lomax et al., 2000; Lippitsch et al., 2005) and further underlines the implications of the linearization process. Also, referring to the problem of detecting seismogenic faults from hypocenter trends delineated in poor network conditions, we introduced a procedure based on BAYLOC's location probability concept with the purpose of establishing when hypocenter trends really mark seismogenic structures and when they simply reflect ill-conditionning of the location process. This procedure (ISO-TEST) showed that while the NE-SW trend of the 2002 sequence can only in minor part be ascribed to ill-conditioning of the location process (what basically means that it effectively marks the orientation of the source), the NW-SE trend of the 1998 sequence is strongly contamined by the location process and source detection is therefore doubtful in this case. Although ISO-TEST is shown to be already capable to bring benefits to seismogenic fault detection in areas where the location problem ill-conditionned, improvements can be expected from wider testing of synthetic earthquake generators and deeper evaluation of misfits between synthetic and real location probability distributions in the space domain. Efforts are currently made in this connection.

References Evans, J. R., Eberhart-Phillips, D. & Thurber, C. H., 1994. User's manual for simulps12 for imaging Vp and Vp/Vs: a derivative of the "Thurber" tomographic inversion simul3 for local earthquakes and explosions, U.S. Geol. Surv. Open-file Rept., 94-431. Lippitsch, R., White, R., & Soosalu, H. 2005. Precise hypocentre relocation of microearthquakes in a high- temperature geothermal field: the Torfajökull central volcano, Iceland, Geophysical Journal International 160 - 371-388. Lomax, A., Cattaneo, M., Bethoux, N., Deschamps, A., Courboulex, F., Deverchère, J., & Virieux, J., 1998. Comparison of linear and non-linear earthquake locations for the 1995 Ventimiglia sequence, Poster presentation at: European Geophysical Society, XXII General Assembly, http://alomax.free.fr/posters/vintimiglia. Lomax A., Virieux, J., Volant, P., & Berge-Thierry C., 2000. Probabilistic earthquake location in 3D and layered model, in Advances in Seismic Event Location, 101-134, Kluwer Academic Publishers, Netherlands. Presti, D., Troise, C. & De Natale G., 2004. Probabilistic location of seismic sequences in heterogeneous media, Bull. Seism. Soc. Am., 94, 6, 2239 2253.