��

GNGTS 2017 SeSSione 1.1

towards a fullY automatIC and modular ProCedure for generatIng a hIgh-relolutIon earthQuaKe Catalog of the 2016-17 Central ItalY seIsmIC seQuenCeM. Cattaneo2, D. Spallarossa1, D. Scafidi1, S. Marzorati2, P. De Gori2, C. Chiarabba2, L. Chiaraluce2,G. Ferretti1, M. Segou3, B. Baptie3, D. Hawthorne3, V. Lane4, M. Moretti2

1 DISTAV – Dipartimento di Scienze della Terra, dell’Ambiente e della Vita, Università di Genova, Italy2 Istituto Nazionale di Geofisica e Vulcanologia, Roma/Ancona, Italy3 British Geological Survey, Edinburgh, England4 SEIS-UK, University of Leicester, England

Data acquisition. The 2016-17 central Italy sequence activated an about 60-km-long normal-fault system composed by a set of SW-dipping normal fault segments (Chiaraluce et al., 2017). Following the first mainshock of August 24 (MW 6.0), a very intense seismic activity developed in space and time, and is still active at the time of preparation of this note (September 2017). The hit area was regularly monitored before the sequence onset by the Italian National Seismic Network (RSNC, Amato et al., 2006) and by additional stations of the RESIICO regional network (Marzorati et al., 2016). Starting from the day of the first mainshock, the SISMIKO emergency team of INGV began to install a dense array of temporary seismic stations composed by 22 stations deployed complementarily to the permanent ones (Moretti et al., 2016). After few days, at the beginning of September, colleagues of the British Geological Survey (BGS) and of the School of Geosciences at the University of Edinburgh offered a scientific collaboration consisting of the deployment of additional 24 BB stations. The 10th of September, the seismic network counts 60 station, with a mean inter-distance of 6-8 km, comparable to the earthquakes distribution at depth.

This network configuration was kept working until the end of August 2017, producing a final dataset of continuous waveforms, recorded at 155 stations located within 50 km of distance from the epicentral area, of more than 2.5 TB of data.

This dataset of continuous recording represents the base for generating a high resolution catalog of earthquakes by means of a fully automatic procedure including modules for event detection, P- and S-waves arrival times, location and magnitude computation.

Automatic detection and picking. The whole dataset of continuous recordings was submitted at first to an automatic detection procedure based on a STA/LTA analysis (empirically calibrated for each station as a function of site’s ambient noise, sensor type, etc.) and on a coincident system (empirically calibrated for the network) defining the number of data channels which must be triggering coincidentally within a coincident window in order to declare the start of a potential event. The event detection phase performances has been optimized considering separately sub-networks. The detected events are analysed through an automatic picking procedure. For automatically wave arrival time detecting and locating earthquake an evolution of the RSNI-Picker (Spallarossa et al., 2014; Scafidi et al., 2016) has been used. The RSNI-

GNGTS 2017 SeSSione 1.1

��

Picker is based on an iterative procedure for the automatic identification of phase arrival times by calculating AIC functions (Akaike Information Criterion; Akaike, 1974). Iterations consist of different steps, separately performed for P- and S-phases, where pick identification is checked and refined based on computed locations. NonLinLoc software (Lomax et al., 2000) has been used for locating events, using a 1D model calibrated for the area (De Luca et al., 2009).

The picking analysis is still in progress; two computers are working in parallel, one analysing the sequence from the beginning and the other one starting on 1 January 2017. Fig.1 compares the number of rather-good locations obtained for each day by the automatic picker as compared with the data reported by the ISIDe database (ISIDe working group, 2016). With rather-good quality we intend locations obtained with at least 8 phase readings and computed horizontal location error less then 5 km. As a mean term, the number of automatically located events is, for each day, from 4 to 5 times that of the ISIDe database; overall, taking into account just the overlapping time windows, ISIDe reports 47815 events, while the automatic system obtained 229674 locations (corresponding to a ratio of 4.8).

Fig. 1 - Daily distribution of the number of events of the ISIDe catalogue (black line) and of the new automatic picking catalogue (blue line), when available.

Preliminary quality analysis. A first comparison of the automatic locations has been performed with respect to the locations of the ISIDe database (Fig. 2a). As regards the epicentral estimate, the two datasets show a very good coherence, with about 93% of epicentral differences lying within 2 km. On the contrary, the depth differences are larger, and show a significant offset (the automatic locations are shallower): this could be also due to the different propagation model used in the two procedures.

This comparison is not complete, because it is relevant just to the events reported by the national monitoring of ISIDe, thus limited to higher magnitude events. In order to check the accuracy of location of low-magnitude events, we hand-picked as set events, detected by the automatic procedures within a few hours long time window with a mean-term rate of seismicity and a mean-term percentage of working stations. We selected the first hours of 9 January 2017. The comparison of locations, limited to the best solutions, is reported in Fig. 2b: nearly all the events are located within 2 km of epicentral differences, while about 92% of the events show a difference below 1 km. As regards the depth difference, it seems that for some events

100

GNGTS 2017 SeSSione 1.1

the distribution is quite narrow and centered around 0, while for other events there is a trend indicating automatic location deeper that those derived from manual analysis. Of note, in this test, the propagation model is the same, and such differences must be related to the picking only.

Preliminary results. In Fig. 3 a map showing the best-quality automatic preliminary location is presented, together with the stations location (black triangles). In this stage, it is just worth to note that the lowed magnitude seismicity detected by the automatic procedure, appears to completely fill the gaps of events highlighted in the central part of the activated fault system, as shown by Chiaraluce et al. (2017).

Fig. 2 - a) Comparison of the automatic picking locations and of the ISIDe catalogue locations. b) Comparison of the automatic picking locations and of the manual picking locations, for the first hours of 9/1/2017.

GNGTS 2017 SeSSione 1.1

101

ReferencesAkaike H. (1974). A new look at the statistical model identification. IEEE Transactions on Automatic Control, 19(6),

716:723,doi: 10.1109/TAC.1974.1100705Amato A., L. Badiali, M. Cattaneo, A. Delladio, F. Doumaz, F. Mele (2006). The real-time earthquake monitoring

system in Italy. Geosciences-Revue du BRGM.Chiaraluce L., Di Stefano R., Tinti E., Scognamiglio L., Michele M., Casarotti E., Cattaneo M., De Gori P.,

Chiarabba C., Monachesi G., Lombardi A., Valoroso L., Latorre D., Marzorati S. (2017). The 2016 Central Italy Seismic Sequence: A First Look at the Mainshocks, Aftershocks, and Source Models. Seism. Res. Lett., 88/3, doi:10.1785/0220160221

De Luca G., M. Cattaneo, G. Monachesi and A, Amato (2009). Seismicity in the Umbria-Marche region from the integration of national and regional seismic networks. Tectonophysics, 476(1), 219-231. doi: 10.1016/j.tecto.2008.11.032

ISIDe working group (2016). version 1.0, doi: 10.13127/ISIDeLomax, A., J. Virieux, P. Volant, and C. Berge-Thierry (2000). Probabilistic earthquake location in 3D and layered

models: introduction of a Metropolis–Gibbs method and comparison with linear locations. In: Advances in seismic event location, ed. C. H. Thurber and N. Rabinowitz, 101–134. Dordrecht and Boston: Kluwer Academic Publishers.

Marzorati M., Cattaneo M., Frapiccini M., Monachesi G., Ladina C. (2016). Recent seismicity before the 24 August 2016 Mw 6.0 central Italy earthquake as recorded by the ReSIICO seismic network. Ann. Geophys., 59/5, doi: 10.4401/ag-7191.

Moretti M., Pondrelli S. et al. (2016). SISMIKO: emergency network deployment and data sharing for the 2016 central Italy seismic sequence. Ann. Geophys. 59/5, doi:10.4401/ag-7212

Scafidi D., Spallarossa D., Turino C., Ferretti G., Viganò A. (2016). Automatic P and S-Wave Local Erathquake Tomography: Testing Performance of the Automatic Phase Picker Engine “RSNI-Picker”. Bull.Seism.Soc.Am., doi:10.1785/0120150084

Spallarossa D., Ferretti G., Scafidi D., Turino C., Pasta M. (2014). Performance of the RSNI-Picker. Seism. Res. Lett., 85/6, doi:10.1785/0220130136

Fig. 3 - Distribution of the locations of the preliminary automatic catalogue, and geometry of the analyzed seismic network (black triangles).