Central American Seismic Center (CASC)

Transcription

1 Central American Seismic Center (CASC) E. Alvarenga Centro de Investigaciones Geot~cnicas, San Salvador, El Salvador R. earquero, I. aoschini Instituto Costarricenses de Electricidad, San Jose, Costa Rica J. Escobar Departamento de F sica, Universidad Nacional Autenoma de Honduras, Tegucigalpa, Honduras J. Fernandez Escuela Centroamericana de Geologia, Universidad de Costa Rica, San Jos~, Costa Rica P. Jayol Instituto Nacional de Sismologfa, Vulcanologia, Meteorologia e Hidrologfa, Guatemala J. Havskov Institute of Solid Earth Physics, University of Bergen, Norway N. G~lvez Centro de Investigaciones Geotecnicas, San Salvador, El Salvador Z. Hern~ndez Instituto Nacional de Estudios Territoriales, Managua, Nicaragua L. Ottem~Jller Institute of Solid Earth Physics, University of Bergen, Norway J. Pacheco Instituto de Geofisica, Universidad Autenoma de M~xico, Mexico C. Redondo, W. Rojas Escuela Centroamericana de Geologfa, Universidad de Costa Rica, San Jose, Costa Rica F. Vega Observatorio Vulcanolegico y Sismolegico, Universidad Nacional, Heredia, Costa Rica E. Talavera Instituto Nacional de Estudios Territoriales, Managua, Nicaragua W. Taylor Instituto Costarricenses de Electricidad, San Jos~, Costa Rica k. Tapia Instituto de Geociencias, Universidad de Panama, Panam~i C. Tenorio Departamento de Fisica, Universidad Nacional Autenoma de Honduras, Tegucigalpa, Honduras J. Toral Instituto de Geociencias, Universidad de Panama, Panam~ INTRODUCTION On May 1, 1998 the Central American Seismic Center (CASC) was opened in Costa Rica. The center has the purpose of permanently storing the most important data from seismic stations in Central America, as well as collecting data from Central American stations in semirealtime in order to issue daily bulletins. All data in the database will be available for access through the Internet. BACKGROUND In 1988 a cooperative project in disaster prevention between all Central American countries was started and organized through CEPREDENAC (Centro de Prevenci6n de Desastres Naturales en America Central). One of the most important components in the program was seismology, and for all six countries reinstallation and/or improvement of existing analog field stations and central recording equipment was initiated. Data acquisition systems were installed in all the Central American country networks, and by the end of digital stations were in operation. Today the number is 138. At the same time a joint Central American effort was made to collect, process, and distribute the seismic data for Central America. The initial idea was that the data center would move between the countries, which then would be responsible for the processing. This was done for 1992 and 1993, but it soon became clear that a permanent data center was needed. It is estimated that a substantial part of the data recorded since 1992 has been lost (see below). In 1996 the installation of broadband stations was started in the countries, and it was decided by CEPREDENAC that a permanent seismic center should be created in Costa Rica. This paper gives an overview of the current situation of instrumentation in Central America, status of the database, and the daily operation of the center. 394 Seismological Research Letters Volume69, Number5 September/October1998

2 90~ 84~ 78~ 0~0~'~: ~ ~ I~',~ ~ " iiiii ;: CARIBBEAN SEA ~=~ 18~ ELS, 12~ PAIIIFIC OCI=IkN,- 6ON 0 Broadband Sta ton 6~ 9 Short-period St ation 90ow 84ow 78ow 9 Figure 1. Seismic stations digitally recording in Central America, May 1, Triangles represent short-period stations and circles broadband stations. A total of 138 stations is shown. Many of the networks are so small in extension that the individual stations cannot be distinguished9 SEISMIC NETWORKS IN CENTRAL AMERICA Seismic networks in Central American countries consist mostly of analog transmitting stations varying between three in the case of Honduras and 67 in the case of Costa Rica (Figure 1). In addition there are eight permanent broadband or extended short-period stations. The data are sent from the field stations by radio using conventional FM modulation, or in a few cases (Panamfi and Costa Rica) by telephone. At the local recording centers, the signals are passed on to the SEISLOG data acquisition systems (Utheim and Havskov, 1997). Some of the channels are at the same time recorded on paper. The processing is done with the SEISAN (Havskov, 1997) on Sun workstations. At seven of the central recording sites, a three-component accelerometer, Kinemetrics FBA23, is connected directly to the SEISLOG system and recorded at two gain levels, thus covering a dynamic range of 105 db. Each country has at least one permanent broadband or near-broadband station. Panamfi and Costa Rica have IRIS stations, while the remaining countries have SEISLOG sys- tems with 24-bit digitizers and various types of sensors ranging from LP sensors to Guralp 40T. Currently 230 channels of digital data are recorded in Central America in a uniform format (SEISAN) from 138 seismic stations (Figure 1). DATA CENTER OPERATION The data center has two functions: (1) automatically collect data for large events (M > 4) from all regional stations with modem or Internet connection and determine a preliminary location and magnitude in near realtime, and (2) be a permanent archive for seismic data in Central America. The near-reahime system is based on the network datacollection system SEISNET (Ottem~511er, 1998). This system connects to the IRIS or SEISLOG systems in the region using Internet or modem. At regular time intervals, detection parameters are downloaded to the central computer and events are declared based on the number of networks having detections in a given time window. The SEISLOG systems will give P- and S-trigger times for each channel as well as duration of the event. Based on this information, a prelimi- Seismological Research Letters Volume69, Number5 September/October

3 nary location and magnitude is calculated. After the event has been declared, waveform data for the event are automatically downloaded for selected stations and networks. The intention is that the event detection process shall run continuously so that automatic preliminary epicenters and magnitudes will be available within minutes. The limitation in location speed is the Internet capacity, which also limits the amount of waveform data transferred. At the moment continuous operation cannot be done because of the costs involved in using the modem stations. It is planned that at least one center in each country will be Internet-connected within one to two years. The operator at the center will, on a daily basis, check all network detections and manually make preliminary regional locations for events larger than magnitude 4.0. It is estimated that about two to three events will be declared daily, and these events will be used to make a regional catalog of earthquakes. Since the center cannot permanently collect and store waveform data for all events in the region due to storage limitations, only those center-declared events will be reprocessed and waveform data collected. The preliminary locations and magnitudes for these events will be sent daily to all interested parties. When individual centers have finished processing their data, they will send the parameter data to the regional center together with waveform data for the regional events. The data center will store all readings and magnitudes made available by the individual networks in separate network databases without doing any reprocessing. The regional database will be made using all phase data from the region, which will be merged and checked with the help of the waveform files. A monthly bulletin for the complete regional catalog will be issued. Thus, the aim is that the center will have nearly all data available from the whole region for larger events and most of the phase data for the local networks. Additionally, Nicaragua is currently installing several tide gauges which transmit signals to Managua in realtime, and has a program to issue tsunami warnings. The intention is that the near-realtime locations from CASC will be sent directly to Nicaragua in order to help in determining the tsunami potential of an event. DATABASE STRUCTURE AND FORMATS The data are stored in the SEISAN database system. This is a simple file-based hierarchical system where the parameters are stored in ASCII and waveforms are stored in binary formats. Each database for local networks and the regional catalog consists of yearly and monthly directories in which there is one ASCII file with parameter data (hypocenters, magnitudes, phase readings, etc.) for each event. The file also gives the names of the binary waveform files available either in the regional center or at the local centers. In addition, the parameters are also stored in monthly files for easy access to larger data sets. It is thus very easy to extract one event, one month of events, or a whole database of events. The wave- form files are also stored in a hierarchical system divided into years and months. Outside users can get access to the data using anonymous FTP to the address in Costa Rica. When logging in, the user will get information about the file structure. If logging in using an Internet browser, the ASCII files can be displayed and the user can find which waveform files are associated with a given event. Response information and station coordinates are also available. For more details on the database structure, see the SEISAN manual. The SEI- SAN software and manual are available on ftp.ifjf.uib.n0 or on ISC CD-ROM, Volume 5. The near-realtime locations and recent waveform data are also available on the above FTP address in Costa Rica. STATUS OF DATABASE The authors of this communication held a workshop at the data center from April 29 to May 8, Participants brought available data, and a skeleton of the database, covering mainly the years 1992 to 1997, was set up (Table 1). However, the database has data back to 1505 (Rojas et al., 1993). Although much data were missing, the database now contains parameter data for more than 120,000 events and about 10,000 waveform files, of which only 125 are from broadband stations. Data for the regional catalog were preliminarily processed in order to update the regional catalog for In this first update, event selection was based on events with magnitude larger than 4.5 reported by PDE. Figure 2 shows epicenters of events in the regional database of coda-wave or body-wave magnitude larger than or equal to 4.5. The aim is to complete the database for all parameter and waveform data back to However, it seems that a substantial amount of data, which were stored on tape, has been lost. It is also clear from Figure 2 that the location accuracy is poor, so a complete revision has to be done. Currently, the events in the regional database were located using on average two networks. This number will probably increase to three networks once all currently available data have been included. In general, about 15% of all events in the region are recorded by two or more networks. Thus, most events are recorded by only one network, which is also the reason to keep most of the parameter data in separate network databases. It turned out during the workshop that many networks had no parameter data for large regional events (M > 5). It is currently not known whether the data have been considered too distant for some networks to process, the local systems have not triggered, or the data are lost. However, it is hoped that the declaration of regional events by the data center will prevent this from happening in the future. One of the main purposes of sending the daily mail of declared events is to ensure that data for important events are collected. Most countries have several days of data in the ring buffers on the data-acquisition systems, so it should be possible to retrieve most of the data. 396 Seismological Research Letters Volume69, Number5 September/October1998

4 TABLE 1 Content and statistics of CASC database as of May 1, In addition to the earthquake databases shown below, there is also information about station coordinates and calibration curves, as well as data from recent locations. Database code and operator Time period # events # waveforms Network databases GUA: Guatemala National Seismic Institute SAL: Geotechnical Research Center, El Salvador HON: National University of Honduras NIC: Nicaraguan Institute of Geophysics RSN: University of Costa Rica UNA: National University of Costa Rica ICE: National electric company of Costa Rica UPA: University of Panamd Other databases CAM: Regional database made by CASC HIS: Historical seismicity FOC: Fault plane solutions ACC: Accelerometer data Number of events in the network databases for the years 1992 to Data is available for all agencies in the time period, so in most cases, zero events means that the data has not arrived yet CAM GUA SAL HON NIC RSN UNA ICE UPA CASC will contain all data submitted by the participating countries. However, only data in the area ~ N and ~ W will be reprocessed for the regional database. CURRENT PROCESSING PARAMETERS The regional crustal model used for location is: P velocity (km/sec) Depth to interface (km) The model is partly based on experience from the individual countries and partly on testing using the complete data set. Generally, coda-wave magnitudes are used, based on a preliminary Central American coda-wave magnitude scale developed by calibrating the coda lengths with PDE-ma magnitudes (Marroquin and Ariola, 1992; Vega, 1993): M c = log (coda) * dist where coda is the coda length in seconds and dist is the hypocentral distance in km. Comparing the coda magnitudes to the M c magnitudes for the years for the regional database (364 events with m b and Me), the average rn b and M~ are 4.75 and 4.95, respectively. Thus, the M c scale gives reasonably accurate preliminary estimates of magnitude, although slightly overestimating the magnitudes. Seismological Research Letters Volume69, Number5 September/October

5 90ow 84~ 78~ 18~ i~i:i ~,, ~ 18~ ~-~.~,~-~~i!~iiiii~iiii! ~iiiiiiif!~ " - CARIBBEAN SEA ~i~'~ ~ :~ :~::::~::~i~: ~i~ ~: ~ 9 ~ :i :~::~-~.-!~i!iiiiiiiii~ :!:: ~::::~:~ ~:~. i :::,:~:~: :~ i.~ i ~iiii::i!::-: "~i::!:::.i:~i::~::... 9,.-:":~:~i~::: iii~iliii~! ~, i~ii# ~:~ ~ ',ii::i ~ ( ~ ~.~! 9 :~/:.~iii!iiiii~!iiii~ ~. # ' 12oN r EL SALVADO:Ie % ~.,.~:~, ~: i~i~ i: r ~....~i::i::i~/':~: 90oW 84~ 78~ 9 Figure 2. Regional earthquakes in the CASC database with coda or m b magnitude > 4.5 in the time period Richter local magnitudes M E are also calculated using the modified Richter attenuation relation given by Hutton and Boore (1987). The scale seems to give reasonable values (Vega, 1993) using the regional calibration information (Escobar et al., 1993). However, the practice of using only coda magnitude is dominant, since only 404 single network locations out of 44,000 in the period reported M E. The regional center will calculate coda magnitude, ML, and moment magnitudes based on spectral analysis for the regional events. Clearly, current regional magnitudes scales and crustal structure are preliminary. However, it is reasonable to expect that improvements will be made once the large regional database is available. DISCUSSION Station quality and spacing are very uneven (Figure 1). This means that in some cases relatively important events are only recorded in one country. In terms of rapid and accurate location, the local networks will often provide the fastest and most accurate solution, particularly if the event is within the local networks. However, the many subduction zone events, as well as events in the Caribbean, will often be badly located by local networks since the events are too far outside the networks or the networks are too small. A regional solution could be better, and if available within a short time could also be important in case of large earthquakes. However, it is also clear that the current regional database must be completely revised, since there are too many bad locations. The broadband data have been little used in the region, and only few data are available at CASC. There is obviously room for large improvement both in terms of data collection and processing of broadband data. In Central America and the Caribbean area there is currently a network of broadband stations cooperating in the MIDAS (Middle America Seismographic Network) project in which the Panamanian and Costa Rican IRIS stations participate. MIDAS is planning to set up a data center in Puerto Rico at the end of MIDAS and CASC cover overlapping areas, and there are therefore intentions to cooperate closely in data collection and processing (D. Novello, personal communication). Despite the unevenness of the Central American Network and the incompleteness of the CASC database, the creation of the Central American Seismic Center represents a 398 Seismological Research Letters Volume 69, Number5 September/October1998

6 very significant improvement in practical seismology for Central America, particularly by collecting the data in one place in a standard format. It is obvious that CASC has a good potential for uniformly defining the regional seismicity in Central America and for providing a good basis for future research in seismology for the region. El ACKNOWLEDGEMENTS CEPREDENAC is supported by the Swedish and Norwegian governments through development grants given by the Swedish Development Agency (SIDA) and the Norwegian Agency for Development (NORAD). We also thank governments in all Central American countries for supporting cooperation through CEPREDENAC. CASC's address: att. Carlos Redondo Central American Seismic Center School of Geology University of Costa Rica Costa Rica Phone Fax casc@carn.egeot.ucr.ac.cr credondo@cariari.ucr.ac.cr REFERENCES Escobar, J., B.M. Storheim, C. Aranda, and J. Havskov (1993). Calibration of Central American seismic stations. Report #11 under the project Reduction of Natural Disasters in Central America, Institute of Solid Earth Physics, University of Bergen, Norway, 52 pp. Havskov, J. (1997). The SEISAN earthquake analysis software for the IBM PC and Sun Version 6.0 manual, Institute of Solid Earth Physics, University of Bergen, 236 pp. Hutton, L.K. and D.M. Boore (1987). The M L scale in Southern California, Bull Seism. Soc. Am. 77, Marroquin, G. and L.A. Arriola (1992). Estudio de magnitud coda para la regi6n centromearicana. Report under the project Reduction of Natural Disasters in Central America, Institute of Solid Earth Physics, University of Bergen, Norway, 14 pp. Ottem611er, L (1998). SEISNET user manual, version 1.1, Institute of Solid Earth Physics, University of Bergen, 30 pp. Rojas, W., H. Bungum, and C. Lindholm (1993). A catalog of historical and recent earthquakes in Central America. Report under the project Reduction of Natural Disasters in Central America, NOR- SAR, Norway, 76 pp. Utheim, T. and J. Havskov (1997). The SEISLOG data acquisition system version 7.0 user manual, Institute of Solid Earth Physics, University of Bergen, 101 p. Vega, E (1993). Large earthquakes (mb > 5.0) recorded by the Central American Seismic Network in the period January 1992 to April Report #9 under the project Reduction of Natural Disasters in CentralAmerica, Institute of Solid Earth Physics, University of Bergen, Norway, 192 pp. Institute of Solid Earth Physics University of Bergen Bergen, Norway (].H.) Seismological Research Letters Volume69, Number5 September/October