The new Swedish Seismic Network
Reynir Böðvarsson| Department of Earth Sciences, Uppsala University, S-752 36 Uppsala, Sweden |
Introduction -
Operation scheme -
Site processing
Central processing -
References
Introduction
The history of instrumental seismology in Sweden began 1904 when the 1000 kg Wiechert horizontal pendulum was installed in Uppsala.
Fig. 1. Station locations in the new Swedish network.
The new Swedish National Seismic Network (SNSN), which is a digital broadband seismic network, is now under construction. The first part of the network was put into operation in 1998. This part of the network consists of six station at approximately the same locations as the stations in the old analog network constructed by Marcus Båth in the 1960s and are shown red squares in figure 1.
Additionally twelve stations are under construction along the coast of the Gulf of Bothnia. This is a separate project financed by the Swedish Natural Science Research Council (NFR), the Knut and Alice Wallenberg Foundation and the Swedish Nuclear Fuel and Waste Management Co (SKB). The main purpose of this network is to study microearthquake activity along the Coast of the Gulf of Bothnia to gain better understanding of the ongoing deformation processes in that area. These stations are shown as blue triangles in figure 1.
All stations are equipped with Güralp CMG-3ESPD broadband seismometers with digital output. These seismometers are flat to velocity in the period range from 0.02 to 30 seconds. The digital data is time stamped within the sensor using the GPS satellite system. The sampling frequency will be 100 sps at all stations.
The data acquisition system used is the so called SIL system which was developed within the SIL project, a joint Nordic project for earthquake prediction research in Iceland, 1988 through 1992 (Stefánsson et al 1993, Böðvarsson et al 1996, 1999). The main achievement of the SIL-project was to establish an automatic earthquake data acquisition and evaluation system, the SIL-system. As detailed plans were made for the SIL project, the importance of microearthquakes for understanding the ongoing deformation processes within the crust were recognized. It was recognized that the recording of earthquakes down to magnitude and retrieval of source information from these events would require a new seismic network design (Stefánsson et al 1986, Böðvarsson 1987). The rapid evolution in computer and communication technology and the introduction of inexpensive but powerful personal computers allowed for such a design of the SIL network (Böðvarsson et al 1996, 1999).
The network operation scheme
Teleseismic events will primary be recorded at the old 6 station locations. During some periods, teleseismic events will probably also be recorded at the remaining stations when research funds are available. Teleseismic events will primary be recorded using information available on Internet. The so called ``E'' type messages from USGS and NEIC, containing a single line of hypocenter and magnitude information on recent earthquakes, are received via electronic mail. Event information from different European networks will also be used for earthquakes occurring at closer distances. A selection programme reads the messages and selects events that fulfill certain criteria of magnitude and epicentral distance. The programme will use the iasp91 traveltime tables (Kennett and Engdahl, 1991) to compute the first arrival time at each station. The teleseismic body wave data are fetched with a sampling rate of 20 samples per seconds and the surface wave data with a sampling rate of 4 samples per second.Regarding local and regional earthquakes, all 18 stations will be operated as a single seismological network providing automatic location and fault plane solution of all located earthquakes. As in the SIL system in Iceland the automatic analysis performed by the system will be divided into four categories: single- and multi-station analysis, multi-event analysis and the alert analysis. Single-station analysis is performed at each site on data recorded by that station. Multi-station and multi-event analysis is done at the center where data from more than one station are available. The alert monitoring is also done at the center, using parameters derived from the single- and multi-station analysis. A schematic description of the data flow in the system is given in figure 2.
Fig. 2, Processes and data flow in the SIL data acquisition system. From Böðvarsson et al (1999).
Processing at the site stations
The software at the site can be divided into two categories: utility processes and application processes. The utility processes are general data management processes, designed for flexibility and valid for any type of data acquisition. The application processes read a channel of data stream as if it were an endless file. Channels are opened as regular files would be, by a call to the specific function in the utility library. The most recent part of the data are kept in shared memory for fastest possible access.The communication between the center and the stations is designed to be independent of the physical way it is realized. Unix utilities are used throughout, providing the best possible portability of the software. To minimize data transmission costs the SIL system uses single-station phase detections and multi-station event selection. The basis of this concept is to treat all transients detected at the stations as if they were phases associated with real earthquakes. The detector uses a simple comparison of power in two adjacent windows in six frequency bands of the seismic trace. This is similar to the STA/LTA approach but in our case the time-windows used are short and both of the same length. Selected windows around the detected transients are processed in a manner one would process a true seismic phase and the results stored in a compact structure, called a phase log. Each phase log entry is only 128 bytes long and is therefore inexpensive to transmit to the center. The detection thresholds can therefore be set very low, allowing smaller earthquakes to be detected. The phase logs will be transmitted to the center once every hour. Each phase log includes onset time, duration, reference to previous and following phases, type of phase (P or S), signal and noise averages, maximum amplitude, azimuth and coherency (Roberts et al 1989) and spectral parameters including DC-level and corner frequency.
Processing at the center
Phase association and event location.
Selection of waveform data to be transferred from the stations is carried out automatically by the selector software at the center. At the center, the phase logs from different stations are merged into a single time-ordered list. The first step of the selection process is to search for time intervals which contain two or more phase detections that may originate from the same seismic source. The phase detections in this time interval are then submitted to the iterative location, phase association and phase truncation procedure as explained below. The principles for the step from a list of phase detections to a list of earthquakes or seismic events are described by (Slunga 1980, Böðvarsson et al 1999). In short each combination of three observations (onset times of P or S phases and azimuths of P phases) is taken as defining the initial location of an earthquake and is then followed by iterative location and phase association and truncation. This procedure may lead to a ''kinematic event`` (no dynamic constrains) defined by three or more observations. The list of kinematic events contains a large proportion of false events due to random coincidences of observations. Therefore each event is assigned a quality measure. Ideally the quality of an event should measure the probability that the event is a true seismic event. The computation of quality is based on both kinematic considerations and analysis of the amplitudes of the detected phases (dynamic information).Fault plane solutions.
Apart from locating the earthquake, the routine analysis performed on every recorded event will include estimation of the fault plane solutions for the earthquake. The estimation of focal mechanism and source parameters are based on results of the spectral analysis of short data segments containing the direct P and S wave arrivals. The spectral estimation is done at the site stations, using windows around the automatic time picks, and repeated at the center after manual refinement of arrival time readings. The low frequency amplitude of each phase is determined by fitting a three parameter model to the observed spectra (Boatwright 1978). To estimate the fault plane solution for the earthquake a systematic search over strike, dip and rake is performed. For each combination of the three source angles, the misfit between observed and predicted spectral amplitudes is calculated. In addition to the single best fitting solution, all solutions that fit the observed polarities and have amplitude misfit less than a predefined threshold value are taken as acceptable (Slunga 1981).The alert system.
The alert system is a collection of routines for monitoring extracted parameters in selected regions and sites. For this purpose, Sweden will be divided into a number of regions and different alert thresholds assigned to each region. The parameters are extracted from the results of the analysis described above and from dedicated alert detectors at the sites. The alert system will be started at regular intervals and for each event defined by the multi-station analysis. Five parameters will be monitored for each region. These are M , the local magnitude of individual earthquakes, N , the number of earthquakes in a time interval, S, a dimensionless measure of moment release during the same time interval and time-weighted measures of the number of events and accumulated moment release (Böðvarsson et al 1999). The purpose of the SNSN alert system is to provide information about the seismic activity in different regions for increased attention of the network operators.Multi-event absolute and relative locations.
At the SNSN center the algorithm described by Slunga et al (1995) will be used to simultaneously determine absolute and accurate relative locations of clusters of similar earthquakes. An example of the application of the relative location algorithm to a group of earthquakes in the Tjörnes fracture zone in Iceland is shown in Figure 3. After relocation the epicenters of the 18 successfully located events lie on an approximately 1 km long line segment (Figure 3.a). Assuming that all the earthquakes occurred on the same fault, the attitude of the fault can be estimated by fitting a plane through the accurately determined hypocenters. The strike of the best fitting plane through the group is , similar to the strike of the main transform faults of the TFZ.
Fig. 3. The relative location of a group of 18 earthquakes in the Tjörnes fracture zone. (a) shows a mapview of the epicenters after relocation, X is east, Y is north. In (b) the hypocenters are viewed along the strike of the best fitting plane through the group. Z is depth and X' is horizontal and orthogonal to the strike. (c) shows the poles to all planes through the hypocenter group, such that the mean distance of the 18 earthquakes from the plane is less than 50~m, plotted on an equal area projection of the lower hemisphere. From Böðvarsson et al (1999).
Acknowledgement
I would like to thank my colleagues Ragnar Slunga, Björn Lund, Conny Holmqvist, Sverker Olsson, Hans Palm and Stefán Böðvarsson who participate actively in the network construction process. Special thanks to the land-owners and others that are housing our stations.References
Boatwright, J., 1978. Detailed spectral analysis of two small New York State earthquakes. Bull. Seism. Soc. Am., 68, 1117-1131.Båth, M., 1974. Spectral analysis in Geophysics. Elsevier, 563 pp.
Böðvarsson, R., 1987. Design of the data acquisition system for the South Icelandic Lowland (SIL) project. Icelandic Meteorological Office, report.
Böðvarsson, R., Rögnvaldsson, S.Th., Jakobsdottir, S.S., Slunga, R., Stefánsson, R., 1996. The SIL data acquisition and monitoring system. Seismol. Res. Lett., 67, 35-46.
Böðvarsson, R., Rögnvaldsson, S.Th., Slunga, R. Kjartansson, E., 1999. The SIL data acquisition system - at present and beyond year 2000. Phys. Earth Planet. Inter., 113, 89-101.
Kennett, B. L. N. and Engdahl, E. R., 1991. Traveltimes for global earth quake location and phase identification. Geophys. J. Int.i, 105, 429-465.
Kulhanek, O..1990. Anatomy of Seismograms. Elsevier, 178 pp.
Kulhanek, O. and R. Wahlström, 1996. History of Instrumental seismological observations in Sweden. in: R. Wahlstr&oiuml;m (editor). Seismolograph recording in Sweden, Norway-with arctic regions, Denmark-with Greenland and Finland. Proc. The Uppsala Wiechert Jubilee Seminar, Uppsala.
Roberts, R. G., Christoffersson, A. and Cassidy, F., 1989. Real-time event detection, phase identification and source location estimation using single station three-component seismic data. Geophys. J., 97, 471-480.
Slunga, R., 1980. International Seismological Datacenter. An algorithm for associating reported arrivals to a global network into groups defining seismic events. Tech. Rep. C 20386-T1, Swedish National Defence Res. Est., Stockholm.
Slunga, R., 1981. Earthquake source mechanism determination by use of body-wave amplitudes - an application to Swedish earthquakes. Bull. Seism. Soc. Am., 71, 25-35.
Stefánsson, R., Bungum, H., Böðvarsson, R., Hjelme, J., Husebye, E., Johansen, H., Korhonen, H. and Slunga, R., 1986. Seismiskt datasamlingssystem för södra Islands lågland. Icelandic Meteorological Office, report. In English with Icelandic and Swedish summaries.
Stefánsson, R., Böðvarsson, R., Slunga, R., Einarsson, P., Jakobsdóttir, S., Bungum, H., Gregersen, S., Havskov, J., Hjelme, J. and Korhonen, H., 1993. Earthquake Prediction Research in the South Iceland Seismic Zone and the SIL Project. Bull. Seism. Soc. Am., 83, 696-716.