BRIDGING THE GAP BETWEEN NATIONAL AND LOCAL NETWORKS: A DISCUSSION

Salvatore Barba and Franco Mele

ISTITUTO NAZIONALE DI GEOFISICA -Via di Vigna Murata, 605 - 00143 ROMA - ITALY

E-Mail: barba@nettuno.ingrm.it - mele@marte.ingrm.it

ABSTRACT

The Italian Seismic Network managed by the Istituto Nazionale di Geofisica grew rapidly in the last years: the telemetered seismic stations exceeded the number of 80; in the meantime some local networks were installed and others were planned. The total amount of data flow from those networks and the cost of transmission on leased telephone lines increased proportionally, as expected. Moreover, the main bottleneck of the analogue telemetry, i.e. the very low dynamic range, has limited the application of recorded data. We are facing the following problems that demand a solution:

1. we need to broaden the dynamic range in the transmission system;
2. the total cost of leased lines has to be kept limited;
3. real time data analysis should be distributed and the number of false detections should be reduced, in order to minimize the workload of both automatic data processing and off line analysis and revision;
4. a larger number of seismic stations should be managed without an equivalent increase of the number of leased telephone lines or computing facilities.

In this work we discuss possible solutions to these major problems, planning a digital network characterized by remote detection of seismic signals and transmission of triggered data on demand. These features, joined with algorithms based on a combination of both single station and network triggers, can reduce the amount of data transmitted to the acquisition center, saving both transmission cost and workload, without compromising the time response of the network for civil defense purposes. Such a system can override the separation, well evident in Italy, between the National Network and the Local ones if all of them share the same algorithms.

PROBLEMS

The dynamic range of the short period Italian Seismic Network is nowadays limited to 13 bits by our analogue transmission system. We need to provide researchers with more reliable data; this aim can be achieved by improving both the dynamic range and the signal to noise ratio.

Because of the continuous transmission, increasing the number of stations would result in a proportional increase in the costs of leased transmission lines. We need to improve the instrumental coverage in areas of seismological interest while keeping costs limited.

Advanced numerical filtering for STA/LTA computation and further increase of the number of stations would result in an unbearable CPU workload for our centralized acquisition system. In the meantime, false detections, that is the analysts workload, can be reduced by a centralized triggering procedure.

Collecting in a unique center the records coming from local networks would have two advantages: 1) all the data would be available in quasi real time for civil defense purposes and 2) the use of the same data format would allow an easier and faster data elaboration for research purposes. On the other hand, we emphasize the need for a decentralized management of the local networks.

POSSIBLE SOLUTIONS

As a minimum requirement to improve data quality, each seismic station has to be provided with a 16 or 24 bit A/D converter, GPS timing and stand-alone elaboration facilities.

Detection at the station and transmission of significant signal instead of the continuous one would reduce transmission costs allowing the increase of the number of stations. The use of an elaboration center for each local network would allow a further reduction of transmission costs by exploiting long distance high speed links.

The reduction of the number of false detections can be achieved by using a detection algorithm of mixed type, that merges the single station analysis and a modified version of the network trigger. Data transmission from each station can be required if the trigger is acknowledged by the network, on the basis of spatial and time coincidence.

When a local networks is provided with its own center of data collection and elaboration facilities, its integration into the national network can be achieved by implementing protocols of trigger acknowledgment and data transmission. In this way, stations belonging to a local network can be treated in a transparent manner.

FURTHER REQUIREMENTS FOR CIVIL DEFENSE PURPOSES

As known, the short period Italian National Seismic Network has been developed for both civil defense and research purposes. Any change in the configuration of such a network, originally developed in collaboration with the USGS, must preserve a few characteristics required to monitor the seismic activity in real time.

Seismic signals from several stations spread through the whole Italian country should be continuously transmitted in order to allow a visual check of the seismic activity. Nowadays operators use analogue drums to follow this aim, so that a continuous transmission at low resolution (8 bit and 20 sps) is adequate for this purpose.

The stations that would not implement the continuous transmission should transfer data requested by the acquisition center within 1 minute from the recording.

DESCRIPTION OF THE TRIGGERING PROCEDURE

All the stations belonging to the national and local networks are subdivided into 4 categories hierarchically organized. Each of them is characterized by a different priority in trigger declaration or data storage.

1. Ordinary Station - The stations that currently operate in the Italian Network transmit the signal continuously. Nowadays the acquisition system detects the presence of significant signal and records it without any check on the existence of a coincidence among different stations. As a possible development any ordinary station could behave according to one of the following descriptions.
2. Leading Station - When a station belonging to this group reveals the presence of seismic signal, it transmits this condition to the central acquisition system together with the signal. The acquisition system can also force the station to transmit a segment of buffered signal. This station participates to the event declaration with a weight s_i=1.
3. Cooperative Station - When a significant signal is recognized, the station sends to the acquisition system a trigger status only. The signal is transmitted only if the acquisition system acknowledges the trigger or on specific request. This station participates to the event declaration with an assigned normalized weight s_i that is inversely proportional to the density of stations around it.
4. Passive Station - No elaboration of the signal is made at the station but buffering. The transmission of data occurs only on request of the acquisition system. This station never participates to the event declaration (s_i=0).

EVENT DECLARATION

The acquisition system declares the presence of a probable seismic event in the following cases:

1. When a leading station reveals a trigger.
2. When a set of N cooperative stations declaring a trigger satisfies the following conditions:
   • where k>1, v_min is the minimum apparent p-wave velocity expected in the area, t_i and t_j are the times associated with the triggers declared by the i-th and j-th stations;
   • , where D_i is the distance between the i-th station and a weighted center of the considered set.

COLLECTION OF FURTHER DATA

Once an event has been declared, the acquisition system requires lacking data to all the stations located inside a certain area. The idea is to identify all the stations that could provide useful signals. This may be implemented defining an heuristic function computed at each station j belonging to the network:

where s_i is the weight associated with the i-th station that takes into account the network geometry, D_ij is the distance between the i-th and j-th stations and r_i is a reliability distance of the i-th station that depends on its sensitivity. The acquisition system requires the j-th station to send data if the value F_j exceeds a certain threshold.

CONCLUSIONS

In a country like Italy, with widespread relevant seismicity, instrumental seismology has a key role in the study of seismogenic areas. Local networks made of closely spaced sensors are necessary to find out and keep under control active structures. The decentralized management of local seismic networks is necessary to ensure an efficient maintenance of the equipment and a prompt intervention for troubleshooting. On the other hand, optimum on line data elaboration for civil defense purposes requires all the available information to be managed in a unique procedure. Merging of data from various networks in a unique data-base makes the work of the researcher easier and faster. It is possible to project a system that drives, with a unique procedure, the acquisition of seismic signals belonging to various local networks spread over a regional area.

This aim can be achieved by merging the single station triggering procedure with the network triggering procedure.

ACKNOWLEDGMENTS

Many thanks are due to Dr. Ray Buland who wrote most of the software for the real time data acquisition system running since 1984 at the Istituto Nazionale di Geofisica in Roma.

(Figures are not enclosed here - Captions follow)

Fig. 1 - Italian Seismicity until 1992. This figure, drawn from the ING catalogue, shows seismic events with intensity greater then or equal to VIII MCS. Major seismic events, although spread all over the country, show that active structures should lie along orogenic belts.

Fig. 2 - The Italian National Seismic Network. The Italian Network is constituted by about 80 short period seismometers (Geotech S13, 1 s eigenperiod, 3 components or vertical). Each station transmits on a dedicated analogue link. The A/D conversion and the acquisition are performed at the ING in Rome. An automatic procedure performs in real time a single channel analysis on the continuous stream of data and stores significant portions of the signal. No coincidence is required to record data. When the procedure finds a correspondence among different stations, it tries to locate an earthquake.

Fig. 3 - Italian Seismicity with Md 2.0-2.5 in the year 1993. The distribution of the located events of low magnitude is strongly affected by the station distribution. Only an high density of stations can allow instrumental studies of seismogenic areas.

Fig. 4 - Model of active seismic structure in Southern Italy (after Valensise G., Pantosti D. and Cucci L.; Geometry, segmentation and timing of the central-southern Apennines seismogenic belt, Italy. Terra Cognita, 5:271). (a) Faults active at least in historical times; (b) proposed faults for which the exact location is unknown; (c) faults, or fault segments, with historical records of activity but with unknown location. Patches labeled 'epicentral area' mark instances in which the fault could not be identified; in these cases the active structure responsible for the earthquake should lie within the area indicated.

Back to manuscripts