Seismologists' access to waveform data from many high-quality stations all over the world has become easy and fast over the last couple of years. Using relatively friendly Internet access techniques, we can acquire a few waveforms from a recent event via FTP or have huge amounts of data from years of continuous recording mailed to us on tape. We can send an e-mail message to have data e-mailed back to us, or use the Web to preview data segments or fill in forms to have custom or preselected data sets generated and sent to us. Major data collection centers are spending lots of effort to make our access to the data easy and efficient. But what must go on behind the scenes to make this work so well?
This month the Electonic Seismologist hosts a guest column describing the inner workings of a major data-collection center that uses an innovative approach to collecting seismic waveform data in near real-time. The IRIS/IDA center at U.C. San Diego operates over thirty stations of the Global Seismograph Network, many of which are in very remote locations. Data from these stations are available within seconds to minutes of real-time via their "Near Real-Time System" (NRTS). Rather than build a specialized and very expensive telemetry system for this network they have made extensive use of the Internet to access their stations. The Electronic Seismologist is particularly impressed with the use of standard TCP/IP protocols for both Internet and dial-up access and the UNIX operating system for the station processors. Both of these nonproprietary components are very familiar to many seismologists. The authors of this column have been involved with the develpment of this system from its design and coding to station installation and along with the rest of the IDA team operate a global near-real-time seismic network at reasonable cost.
THE IDA NEAR REAL TIME SYSTEM
Since 1992, IRIS/IDA stations of the Global Seismographic Network have been configured as nodes on the Internet for the purpose of near-real-time data access. The IDA Station Processor (ISP), a UNIX workstation connected to the IDA Data Acquisition System (IDAS), provides rapid data access to local and remote users. Additionally, it is used to monitor station state of health, and to unpack and copy data tapes for the station operators. The ISP may be installed at the same location as the IDAS or, should conditions require, at a different location connected via a dedicated telecommunication circuit. The software for real-time data collection and transfer, called the IDA Near Real Time System (NRTS), is designed to provide both continuous and occasional data telemetry from IRIS/IDA Stations to remote (from the station) users with the following goals in mind:
As presently implemented, the ISP and NRTS have these features:
The IDAS produces 1,024 byte fixed-length packets, containing from 4.8 seconds to 40 minutes of data, depending on the sample rate of the particular data stream. These packets are immediately written to disk upon receipt from the IDAS, and once on disk they are automatically available for local analysis or remote transfer. In practice, this results in the ability for remote systems anywhere in the Internet to receive and process data seconds to minutes after ground motion at any station.
The NRTS is designed to be replicated on a remote machine and input data from one or more stations. In this manner, it possible to have immediate central access to all data from multiple stations.
THE IDA WIDE-AREA NETWORK CONCEPT
The TCP/IP protocol suite provides reliable data connections between computers. By developing the data acquisition and transmission around these protocols, the task of data collection from remote locations reduces to one of bringing Internet to the station, a problem for which a multitude of off-the-shelf solutions exist. The application software on both ends of the circuit can be designed without the need for knowledge of the details of the communication links. Figure 1 illustrates the IDA wide area network (WAN), with the ISP's appearing as nodes on the Internet using a variety of interconnects.
The connection of the ISP to the Internet is either through a local Internet provider, if such exists, or more typically through a WAN connection established as part of the IDA station installation. The nature of this interconnection is quite station-dependent and determined by economic and technical factors. Table 1 and Figure 2 provide some details of the WAN connectivity at operational IRIS/IDA Stations.
The TCP/IP protocol suite that constitutes the Internet connectivity utilizes the PPP protocol for communications over serial lines. With data compression, almost any analog telecommunications circuit has sufficient bandwidth to make continuous data feeds available to remote users. However, for those sites which employ a dial-up connection, the line charges make it too expensive to maintain a full-time connection. In those cases, the stations are routinely contacted only once per day to bring over state-of-health information, and the data are kept on the station ISP. Specific waveform segments around events of interest are then retrieved upon demand, using one of the exchange protocols described below.
The IDA Data Acquisition System--ISP Interconnect
The IDA Data Acquisition System (IDAS), also designated the IRIS-3 data acquisition system (Given et al., 1992), uses hardware similar to the IRIS PASSCAL system. It digitizes, time-stamps, and records data locally on magnetic tape or disk. It also provides a serial port through which all data may be accessed in near real-time.
When the ISP is located at the same site as the IDAS, the interconnect is simply a serial line. In addition to the data line, the console port of the IDAS is connected to one of the external serial ports of the ISP. Both the local operator and the IDA network operations staff have access to the IDAS console by logging in to the ISP. This remote console access has proven to be very helpful for station maintenance and configuration.
If conditions at the site do not permit the ISP to be located near the IDAS (say due to environmental or power limitations) then the connection between the data port and the ISP can be over a dedicated telemetry link (normally via low-power spread spectrum radio modems).
At unattended stations, the ISP is either a low-end sparc workstation (commonly a Sun Microsystem IPC) with an ASCII terminal for the console, or an inexpensive Intel-based personal computer running the Solaris operating system. At those stations where there is local interest in the waveforms, the ISP is enhanced to include a bit-mapped display and the Datascope package for analysis of the online data, plus hardware and software for copying and unpacking station tapes.
NRTS Software Overview
The IDA Near Real-time Data Collection System (NRTS) collects, reformats, and transfers data from an IDAS to one or more remote recording sites. The ISP is preconfigured to hold a fixed amount of data (typically one week) in a disk loop, and as new data are acquired the oldest data are overwritten. The ISP reads the data from the IDAS serial port and demultiplexes the data into the disk buffer. These data are written in such a manner that they are directly accessible in read-only form from disk via the Datascope utilities. Local users who desire private copies of selected data segments for offline analysis, as well as all remote users, can connect to the ISP using one of several protocols described below to extract the portions of interest. Alternatively, with sufficient bandwidth, it is possible to execute a copy of the NRTS on a remote machine which inputs data from one or more stations.
A set of cron jobs maintains the various log files and sends email to the IDA Data Center if anything unusual occurs. In the absence of problems such as power outages or operator error, the ISP and NRTS can run unattended indefinitely. For example, the Icelandic station BORG has run without interruption and with zero loss of data since July 1995.
The NRTS Data Exchange Protocols
Each NRTS listens on a standard port for connections from clients talking in the NRTS Data Exchange Protocol. This protocol includes negotiation for connection time-outs, TCP/IP buffer sizes, and data formats plus station/channel and start/stop time specifications. Requests can be for either a specific time window or for a continuous data feed. This protocol is used to communicate data between NRTS stations and hubs, internally by the email-based AutoDRM, and by organizations interested in monitoring selected IRIS/IDA Stations in near real-time. At present, there are two supported data-return formats. Clients can ask to receive either the raw IDAS packets or a generic format that masks the details of the underlying data-acquisition system. The former is used exclusively by NRTS hubs, which acquire data from multiple stations, while the latter is used by various application programs that process the data in near real-time and by users outside of IDA. In addition, there are hooks in the protocol so that in the future data can be returned in other formats, such as mini-SEED or as RefTek PASSCAL records.
Currently there are three organizations other than IDA which use this protocol to maintain a continuous near-real-time feed from one or more IRIS/IDA stations. They are the National Earthquake Information Center (Golden, CO), the Alaska Tsunami Warning Center (Palmer, AK), and the Russian Geophysical Seismic Survey (Obninsk, Russia). The Pacific Tsunami Warning Center (Honolulu, HI) is completing developmentof a client tap program which is expected to be in production in the near future.
Other organizations interested in establishing connections to IRIS/IDA stations are encouraged to do so, and may obtain software upon which to build their custom applications from ftp://idahub.ucsd.edu:/pub/pickup/idatap.tar.Z. This toolkit contains library routines that should run on any Unix system that supports sockets, and will permit the user to establish a connection with the server NRTS and to request configuration information and waveforms. This software also handles the details of data compression/decompression and possible byte reordering in a manner transparent to the application. A simple example program is included that illustrates the programming interface. Before using this service, you will need to send email explaining your needs to firstname.lastname@example.org in order to get your host on the list of allowed clients.
In addition to the above protocol, all NRTS systems also include software which allow them to serve as GSETT-3 (Group of Scientific Experts Technical Test) Alpha- or Beta-level stations. The U.S. National Data Center (NDC) in Florida and the Center for Monitoring Research (CMR) in Virginia both maintain full-time data feeds from selected IRIS/IDA stations using the Alpha Protocol. These protocols, with minor changes, are planned to be the data exchange protocols for the CTBT International Monitoring System.
All IRIS/IDA stations with continuous links to the Internet also implement the GSETT AutoDRM email data request system (see Kradolfer, 1996 for an overview of AutoDRM). Those stations with dial-up links have AutoDRM support via idahub.ucsd.edu. To get the current User's Guide, send email with the word "help" in the body of the message to email@example.com, e.g.:
In addition to AutoDRM, the NRTS supports a proprietary e-mail-based data request service (IDADRM) that was developed before the AutoDRM was standardized, and which offers a greater choice in data formats and delivery mechanisms. You may get the current IDADRM User's Guide by sending a "help" request as shown above to firstname.lastname@example.org. As the AutoDRM standard evolves, it is likely that many of the features currently available only through IDADRM will be implemented in some form and IDADRM will be retired.
A World Wide Web interface to the IDA NRTS system can be found at http://www-ida.ucsd.edu/public/nrts.html.
Kradolfer, U. (1996). AutoDRM: The first five years, "Electronic Seismologist", Seism. Res. Lett. 67(4), 30-33.
Malone, S. (1996). SPYDERS on the Internet, "Electronic Seismologist", Seism. Res. Lett. 67(4) 58-63.
Given, H.K., J. Berger, J.-F. Fels, D. Horwitt, and C. Winther. The IRIS-3 High Resolution Data Acquisition System, Scripps Institution of Oceanography Reference Series, 92-23, November, 1992.
SRL encourages guest columnists to contribute to the "Electronic Seismologist." Please contact Steve Malone with your ideas. His e-mail address is email@example.com.
Posted: 22 January 1999