A seismic alert management system for the Campania region (southern Italy): Development and experimentation results

Since few decades, the experimentation of seismic early-warning systems is ongoing in several active seismic area of the world. Successful prototype systems have been developed and implemented in Taiwan, Japan, USA and Mexico where alert signals from a dense seismograph network located in the earthquake source area are sent to nearby urban settlements. With about 6 million of inhabitants, and a large number of industrial plants, the Campanian region, is highly exposed to the seismic risk, related to a moderate to large magnitude seismicity originated by active fault systems in the Apenninic belt. The 1980, M=6.9 Irpinia earthquake was the most recent destructive earthquake occurred in the region causing more than 3000 causalties, huge and widespread damages to buildings and infrastructure on the whole regional territory. In the frame of an ongoing project financed by the Regional Department of Civil Protection, a prototype system for seismic early and post-event warning is being developed and tested, based on a dense, wide dynamic seismic network under installation in the Apenninic belt region (Irpinia Seismic Network). Considering a seismic warning window ranging from tens of second before to hundred of seconds after an earthquake, several public infrastructures and buildings of strategic relevance (hospitals, gas pipelines, railways, railroads, ...) of the Regione Campania can be considered as potential test-sites for experimenting innovative technologies for data acquisition, processing and transmission. A potential application of an early warning system in the Campania region based on the Irpinia network, should consider an expected time delay to the first energetic S wave train varying between 14-20 sec at 40-60 km distance to 26-30 sec at about 80-100 km, from a crustal earthquake occurring in the source region. The latter is the typical time window available for mitigating earthquake effects through early warning in the city of Naples (about 2 million of inhabitants including suburbs). Trying to minimize the time to be spent for event detection, location and magnitude estimate is a true technological challenge mostly depending on the storage and real-time computational capacities of data-loggers, on the efficiency and robustness of automatic algorithms and on the connectivity architecture of the seismic network. The prototype Seismic Alert Management System (SAMS) for the Campania Region is conceived as a system where the levels of analysis and decision are distributed over the seismic network nodes. This is realised through the implementation of virtual sub-nets managed by data concentrators (Local Control Centers). Each node of the network has to be able to process and analyse in real-time the first-P wave signal, provide the measured quantities (arrival time, frequency, amplitude?) to its closest LCC. As more stations will record the seismic signal, the new measurements are sent to and processed by the LCC, which crosschecks the information coming from different stations, and outputs a progressively refined estimate of earthquake location and magnitude, along with their uncertainty. We developed a method for real time location based on the station triggering order. We pre-calculate the arrival time from each point of the location grid to each station. Then we look at all the points of the grid which are closer in time to each given station: these points define an area called Voronoi cell. When an earthquake occurs, we can use the Voronoi cell associated to the station that triggers first as the possible hypocentral area. This method can be extended looking not only at the first station, but also at the triggering order, thus defining more strictly the hypocentral region. The real time magnitude estimation is performed by taking advantage from the high spatial density of the network in the source region and the wide dynamic range. Moment estimates will be primarily obtained both by P-wave peak ground motion measurement and attenuation laws at stations close to the source, and by spectral amplitude measurements at given frequencies on narrow band, band-pass filtered P-wave signals. The latter method tries to estimate the low frequency spectral level through a time-domain band pass filtering around a frequency that becomes smaller and smaller as the recorded time window gets longer. In this presentation the concepts and the general architecture of SAMS will be illustrated, by focussing on the required technological innovation of different components of the seismic network (data logger, sensors, data transmission). Concerning the structural engineering applications, as the safety shut-down or off-lining of high-risk industrial plants or infrastructures, it will be discussed the problems of interfacing SAMS with a dedicated structural control system, aimed at protecting the target structure from strong ground motion shaking. Figure 1. Expected detection capability of the Irpinia Seismic Network given the occurrence of an earthquake at 8 km depth underneath the network. The figures represent the number of stations, which would record the first P-arrival at increasing times after the event occurrence. After 3.5s since the origin time of the earthquake, more than two first-P readings should be available for earthquake location, while one second more is needed to obtain first moment-magnitude estimates.