AbstractThe Irpinia Seismic Network (ISNet) is a modern infrastructure at the core of the ongoing Earthquake Early Warning System (EEWS), under development in Southern Italy. The main target of the ISNet is to provide alerts for moderate to large earthquakes (M>4) to selected target sites in Campania Region, and to rapidly estimate ground shaking in the whole region in the immediate post-event. The network covers an area of about 100×70 km2, corresponding to the Apenninic active seismic zone, where several large earthquakes occurred during the last centuries, including the Ms=6.9, 1980 Irpinia earthquake. ISNet comprises 29, 6-components seismic stations equipped with both accelerometers and velocimeters, with real-time telemetry.

The Irpinia Seismic Network (ISNet): a modern facility forearthquake early warning

C. Satriano (1), A. Zollo (2), G. Iannaccone (3), A. Bobbio (3), L. Cantore (2), V. Convertito (3), M. Di Crosta (1), L. Elia (1), G. Festa (2), M. Lancieri (3), C. Martino (1), A. Romeo (1), M. Vassallo (1).

(1) RISSC-Lab, AMRA scarl, Via Diocleziano 328, 80124 Napoli, Italy(2) RISSC-Lab, Università di Napoli Federico II, Via Diocleziano 328, 80124 Napoli, Italy(3) RISSC-Lab, INGV, Osservatorio Vesuviano, Via Diocleziano 328, 80124 Napoli, Italy

Hardware And Data Management SystemTo manage the seismic network we developed a web application, ISNet Manager, and a database (PostgreSQL) that keeps track of the several components that comprise or are produced by the network, such as stations, devices and recorded data. The application is written in JavaServer Pages (JSP,Tomcat) and is accessed through a web browser.

Seismic DataThe interface for viewing events and waveforms. Events can be filtered on time and location, magnitude, and distance to the stations. Waveforms can be filtered on station, component, instrument and quality.

Hardware Monitoring Several Java programs implement an automatic hardware monitoring layer. The devices marked for monitoring are polled at regular intervals. The Java programs perform the device specific query, retrieve the internal variables from the hardware and store them into the database.

Interactive Visualization Of Live Seismic Data

ISNet Monitor is an interactive visualization tool for: real-time seismic data produced by our seismic network (Seedlink streams); near real-time data (PGA, PGV, PGD maps) calculated a few minutes after an interesting event; historical seismic data computed or recorded in the region.The application is written in C++ for its performance, and the rendering makes use of the OpenGL library, the de facto standard for scientific visualization.

References• An Advanced Seismic Network in Southern Apennines (Italy) for Seismicity

Investigations and Experimentation with Earthquake Early Warning. E. Weber, et al. Seismological Research Letters, Vol.78, N.6, December, 2007

• Real-time evolutionary earthquake location for seismic early warning. C. Satriano, A. Lomax, A. Zollo, Bulletin of Seismological Society of America, Vol. 98, No. 3, June 2008, doi: 10.1785/0120060159

• Earthquake magnitude estimation from peak amplitudes of very early seismic signals on strong motion records. A. Zollo, M. Lancieri, and S. Nielsen, Geophysical Research Letters, 33, L23312, 2006

• The cry wolf issue in seismic early warning applications for the campanian region, Iervolino I., Convertito V., Giorgio M., Manfredi G., Zollo A. (2007) in Earthquake Early Warning Systems, P. Gasparini et al. (eds.), Springer-Verlag.

Real Time Earthquake LocationThe evolutionary, real-time location technique is based on an equal differential

time (EDT) formulation and a probabilistic approach for describing the

hypocenter estimation. The algorithm, at each time step, relies on the

information from triggered arrivals and not-yet-triggered stations. With just one

recorded arrival, the hypocentral location is constrained by the Voronoi cell

around the first triggering station constructed using the travel times to the not-

yet-triggered stations. With two or more triggered arrivals, the location is

constrained by the intersection of the volume defined by the Voronoi cells for the

remaining, not-yet-triggered stations and the EDT surfaces between all pairs of

triggered arrivals. As time passes and more triggers become available, the

evolutionary location converges to a standard EDT location.

Following a distributed approach, the network is organized in 6 sub-nets: waveform data is collected and elaborated at local hubs (LCC, Local Control Centers), which, in turn send processed parameters to a Network Control Center (NCC) in Naples, 100 km away from the network center. The network is designed to provide estimates of the location and size of a potential destructive earthquake within a few seconds from the earthquake detection, through a fully probabilistic approach, where the computation results are continuously updated with time. For the real time location we developed an evolutionary, real-time approach, based on the equal differential time formulation (see Real Time Earthquake Location). The size of the earthquake is also evaluated by a real-time, evolutionary algorithm based on a magnitude predictive model and a Bayesian formulation (see Real Time Magnitude Estimation). The warning time for selected sites of the Campania Region is reported in the figure. It takes into account the performance of the seismic network and of the computation time to perform the data analysis and the estimation of the event’s location and magnitude.

Location test for two synthetic events occurring at the center and outside the Irpinia Seismic Network (left and right panel respectively). The three orthogonal views show the marginal values of the relative location probability density function. The true hypocenter is identified by a star. The time from the first trigger is indicated as δt, while Δt is the time from event origin. For each snapshot, stations that have triggered are marked with a circle.

Real Time Earthquake Magnitude

The real time and evolutionary algorithm for magnitude estimation is based on

a magnitude predictive model and a Bayesian formulation. It is aimed at

evaluating the conditional probability density function of magnitude as a function

of ground motion quantities measured on the early part of the acquired signals.

The predictive models are empirical relationships which correlate the final event

magnitude with the logarithm of quantities measured on first 2-4 seconds of

recording. In this application we use the empirical relationship between low-pass

filtered, initial P- and S-peak displacement amplitudes and moment magnitude

(e.g. Zollo et al, 2006).

While the P-wave onset is identified by an automatic picking procedure, the S-

onset can be estimated from an automatic S-picking or from a theoretical

prediction based on the hypocentral distance given by the earthquake location.

At each time step, progressively refined estimates of magnitude are obtained

from P- and S-peak displacement data. Following a Bayesian approach, the

magnitude PDF computed at the previous step is used as a priori information.

Synthetic seismograms for a simulated M 7.0 earthquake. The seismograms are computed using a line source, rupture model (constant rupture velocity) while complete wave field green’s functions in a flat-layered model are computed by using the discrete wave number summation method of Bouchon (1981). Each vertical line indicates the 1-second signal packets examined at each time step. For example, after three seconds from the first P phase arrival, 13 stations are acquiring, and the 2-sec S-phase peak is available at the nearest stations. This observation motivates the use of the S phase information in a real-time procedure: if a dense network is deployed in the epicentral area the nearest stations will record the S-phase before the P phase arrives to the far ones.

Left panel: Bayesian PDFs for magnitude at several time steps from the firstP-phase arrival.

Right top: magnitude estimation with uncertainties as a function of time; the dashed line is the actual magnitude value, the errors represent the 95% of confidence bound evaluated as cumulative PDF integral in the 5-95% range.

Right bottom: probability of exceeding magnitude 6.5 and magnitude 7.5 thresholds as a function of time. The dashed line is the 75% probability level.

EGU2008-A-07558

This work was partially funded by AMRA scarl through the EU-SAFER project FRAMEWORK PROGRAMME CONTRACT N. 036935

www.saferproject.net