Virtual Seismic Networks

Frank Vernon, University of California, San Diego and
Terry Wallace, University of Arizona

Figure 1. A feasibility test to collect data from seven different regional and global seismic networks and integrate them into a "Virtual Seismic Network". Maps on the left show station locations with the name of each network. Data from each network are processed to produce epicentral locations which are shown on the maps on the right. Collated seismograms from an earthquake near Fiji, recorded on the VSN, are shown on the two waveform plots in the lower right.

The GSN and PASSCAL datasets currently provide most of the broadband data used in seismological research, although this situation will be evolving in the future as data from permanent regional and national broadband stations become more widely available. The 1999 deep Mw = 7.1 Russia-China border event, which did not have a co-located PASSCAL experiment, was well recorded by the Kyrgyz Broadband Network (KNET), the Austrian National Seismic Network, and many other regional and national networks. Many of the data for these events were available almost immediately through the Internet.

GSN stations routinely send data back to their respective data collection centers in San Diego and Albuquerque using the NRTS and LISS real-time data delivery systems. The GSN currently includes facilities in more than 80 countries. On a national scale, twenty-three IRIS member universities, along with the USGS, operate real-time regional seismic networks in every seismically active region in the United States (http://www.cnss.org). These networks range from the large aperture University of Alaska network (with nearly 300 stations covering all of Alaska including the Aleutian Islands) to the smaller local networks, such as the Montana Regional Seismic Network.

The most striking characteristic of the global distribution of seismographic stations is it's incredibly heterogeneous nature. Many agencies from many countries support the complex infrastructure. These different agencies have missions that range from seismic monitoring to detailed hazard assessment at local levels. They are all united, however, by the data they collect.

The transmission of real-time data from seismic field stations to a primary data collection center is being done in a variety of ways according to each network's specific mission and based on unique hardware, communication systems, number of stations, and areal coverage requirements. While each individual network may have different missions and use different types of equipment, all network data centers have one thing in common: access to the Internet. With this existing infrastructure, we can now consider the integration of all accessible real-time data.

Figure 2. The seismogram in the upper left is a helicorder style plot showing a full day's record at Charters Towers in Australia, showing the motions caused by a significant earthquake in Indonesia. The record section in the lower left is from the same earthquake recorded at many stations around the world. The cross-section of the earth gives a representative sample of ray paths. The map in the upper right show the location of the event and samples of the first part of the seismogram at selected stations.

Starting at the end of 1998, we conducted a feasibility test for real-time data integration from multiple disparate seismic networks to create a "Virtual Seismic Network" (VSN). In the test, data from the IRIS GSN network, the PASSCAL Broadband Array, four US regional networks (University of Alaska, UC Berkeley, UC San Diego, and University of Nevada, Reno), and the Kyrgyzstan National Broadband Network were integrated into one common data processing system (Figure 1). This test successfully demonstrated that over 150 seismic stations from seven different primary data collection centers could be accessed through the Internet and processed in real-time. The level of processing accomplished during the test included: data assembly, automated phase picking, event location, and display of event location and magnitude information. In a parallel development, the USGS at Golden has developed their "Virtual Data Logger" as a mechanism to use real-time data from many network data sources to build their global catalog.

The VSN is ushering in a new era in seismology where global data are available to any researcher in real-time at nominal cost. But, how will the "real-time" era change seismology? The most immediate impact is that individual researchers will be able to design experiments which will be able to maximize existing resources. For example, if a scientist is interested in operating a portable experiment in Chile, they can construct their own virtual network from the global inventory of stations that maximizes the monitoring capability for that region of interest. Individual researchers can also receive continuous data streams (not event segmented), which means that new discoveries will be made which in the past would have been most likely unexamined (if you don't see it, how can you order it from the DMS?).

The following are two examples that show the utility of customized "VSNs":

Network 1: Study of the physics of deep focus earthquakes.

An experiment could use several broadband regional networks as seismic arrays to form continuous beams on regions were deep earthquakes occur. The network would have the advantage of observing smaller events than the current global catalogs record, provide much finer relative locations between events, and yield significant information about the rupture properties of the larger events.

Network 2: Designing a real-time seismicity module for an Earth science course.

Students could be allowed to design their own seismic networks using Internet-available stations. The virtual networks could be used, for example, to study a currently active aftershock region. Figure 2 shows a project which presents the record section views of seismograms, showing the source and site locations with a cross-section of the Earth for large teleseisms.

This is unquestionably only the beginning. As more scientists obtain access to real-time data streams it is to be expected that entirely new functions will be devised to extract more information from these data in real-time as they design their own Virtual Seismic Networks.

Acknowledgments
Bob Woodward provided Figure 2. The seismic networks operated by the University of Alaska, University of California at Berkeley, University of Nevada, Reno, the IDA project at UCSD and USGS Albuquerque Seismic Lab, kindly provided access to their real-time waveform data.