It's time to revisit one of the more successful and popular seismic data access systems available on the Internet. The Automatic Data Request Manager (AutoDRM) is approaching ten years old but is going strong, providing access to ever-increasing amounts of seismic data for more and more users. A fairly complete description of the development of the AutoDRM was one of the first guest columns of the Electronic Seismologist back in the summer of 1996. The system has undergone relatively few changes and enhancements since then, attesting to its original good design and thoughtful execution. If anything, its success is also one of its current problems; there are so many data available via the AutoDRM mechanism that it is hard to keep up with what is available when and from where. To help with this problem the father of AutoDRM is back at it, having now produced a system for summarizing data availability from the myriad of AutoDRM's and providing this summary via a simple set of Web pages and, yes, via an AutoDRM e-mail request method as well. Urs Kradolfer is again our guest columnist to explain this new service.
Waves4U: Waveform Availability through AutoDRM's
Seismograms are the key data for many seismological researchers, and since seismic waveforms are now widely recorded and archived in digital form, global communications facilities, such as the Internet, potentially allow the seismological community to access lots of data. An important step toward making seismic signals available for research was the foundation of international data centers such as those at IRIS and ORFEUS. These centers collect and provide waveforms for significant earthquakes originating from many stations worldwide.
Mainly in order to allow access to near-real-time waveform data, but also to facilitate international data exchange, many observatories have implemented an AutoDRM, which is an e-mail-based data exchange system (Kradolfer 1993, 1996). Today this tool is widely used, especially for retrieving waveforms from individual stations or observatories, and from regional or international data centers. However, for the single user it has not been easy to get an overview on which stations and channels were provided by the numerous AutoDRM's. Therefore, already in its early stages, lists of AutoDRM's and their functionalities were regularly updated during the meetings of the GSE (Group of Scientific Experts) at the Conference on Disarmament in Geneva. Other lists were maintained at Web pages (e.g., http://seismo.ethz.ch/autodrm/autodrm_list.html), and in 1998 and 1999 the European-Mediterranean Seismological Center published detailed lists on the availability of seismic waveforms in its EMSC Newsletter. The obvious drawback of any such list is that it is soon outdated as new stations are built or old ones are closed. In addition, some observatories reported on questionnaires their planned rather than their actual data availability. In May 1999, a colleague entered my office and told me that he had just sent a waveform request to an AutoDRM address which was published one month earlier and that he got an error message, because the e-mail address did not exist. This was a frustrating moment and was the ignition point to develop what I had thought about for two years but never had found the time to do (this may sound familiar to many readers ...). A tool was needed that quickly gives accurate and updated information on which stations and channels are provided through what AutoDRM's. With Waves4U we now have such a tool. With only the input of a list of AutoDRM's, including their e-mail addresses, the software automatically and regularly sends requests for two seconds of waveforms to all listed AutoDRM's. Both the station and the channel list are set to the wildcard "*", thus requesting data from all available channels of the AutoDRM's. All returned e-mails are then processed and analyzed, and for each AutoDRM a list of stations and channels, where waveform data were actually received, is automatically updated and its results are made available on the Web at URL http://www.seismo.ethz.ch/Waves4U. This allows anyone to get an overview of what data are available worldwide in near real-time.
For the researcher who would like to study specific earthquakes, it is usually less important to have immediate access to waveform data, while scientists responsible for rapid earthquake determination and fast alert systems are interested in sources where they can retrieve waveform data in near real-time. To assist the different needs of users, it is therefore desirable that Waves4U also indicates what waveform archiving system is behind a specific AutoDRM. There are many different data acquisition systems used these days in seismology (see, e.g., Malone, 1999), and there are also different archiving strategies at the various observatories: Some store only segmented waveforms (event files), and some store continuous waveforms. Of the latter, some provide waveform data in near real-time, while others collect the data with some delay from their own stations. To make things further more complex for the user, some observatories maintain a limited waveform buffer, going back seven or fourteen days in time for example, and some keep continuous data permanently. On top of all these different environments, we have to take into account that both data availability for different stations and availability of the computers, where the data are stored and the AutoDRM software is running, can vary substantially from one site to the other.
To somehow bring some clarity into this tricky situation, Waves4U requests both recent data (two hours back in time) and older data (24 hours and 48 hours back in time) and displays for each channel how old the newest waveform data were that were returned for each day. The history currently covers seven days (the current day and six days back in time). In Figure 1 an example of three channels is displayed; for ease of explanation, the channel is assumed to be BHZ in all three cases.
In the first line we immediately see that station CONT1 has provided recent waveforms on each of the seven days, indicated by a green ball. We may therefore assume that these data are available in near real-time and from a continuous data buffer. In the second line the result for station CONT2 suggests that data for this channel are available from a continuous data archive but not in near real-time, because the yellow balls indicate that on each day data were received that were more than twelve hours old. Both stations seem to be available from a reliable system, since there were responses on each day. The third line represents the results for station SEGM: No waveform data were received for the first three days of the week shown (gray balls). Three days back from the current day, recent waveforms were received (green ball); a day later only data older than twelve hours but newer than 48 hours were received (red ball). On the current day, no data were received (because Waves4U requests data only back to 48 hours; gray ball). This suggests that data from station SEGM are available from a segmented archive where data for earthquakes are available, but no continuous data are stored.
These three different types of results show that it is often easy to get a visual impression relatively quickly on the data availability for individual channels. Assigning a number to each color (green = 0, yellow = 1, red = 2, gray = 3) allows one to compute an average availability value (av), ranging from 0.0 (green on all days) to 3.0 (gray on all days), which is useful when the data availability for a specific channel must be determined automatically by software. This can be done easily by sending a request with the key word "Waves4U" to our AutoDRM in Zurich (email@example.com) and giving the station and channels for whichthe result should be reported in the usual station and channel list environment. Such a request mail could appear as follows:
and may return a response as follows:
The recipient of this mail will detect that waveform data for the stations CONT1 and CONT2 are available from one AutoDRM (autodrm@xyz) only, while data for the station SEGM can be retrieved from two AutoDRM's and specifically that the second AutoDRM (autodrm@abc) obviously provides continuous waveforms for station SEGM, while the autodrm@xyz provides only event data for this station.
The Waves4U main Web page features a list of all AutoDRM's actually serving waveforms, and for each AutoDRM the results are displayed for all stations and channels. In addition, a user may enter a station code and will then get a list of all channels available for this station, sorted according to the AutoDRM's providing it. Furthermore, response/calibration information is given, if available. It is also possible to search a geographic area for stations providing waveform data through one or several AutoDRM's. Upon all status changes new GMT maps (Wessel and Smith, 1995) are produced, showing the distribution of stations providing data (Figure 2). At the time when this article was written (late October 1999), 1,016 stations were providing a total of 2,715 channels worldwide.
The critical user may wonder whether the periodic requests transmitted to all known AutoDRM's will cause a burden for the systems running an AutoDRM. Compared with the ten thousands of requests sent each month to AutoDRM's, the few hundred additional ones by Waves4U are a small percentage. In addition, if scientists consult Waves4U before requesting data, it is likely that fewer requests will be made by individual users in order to find out for themselves what data are available at which site. Any additional traffic and CPU usage caused by Waves4U should be minimal. However, the author is flexible and any complaints (sent to firstname.lastname@example.org) will be taken seriously. Even more welcome are announcements of new AutoDRM installations. Once their e-mail addresses are added to the Waves4U list, these sites will automatically be polled and the results will immediately appear on the Waves4U Web pages.
Jochen Braunmiller and Stephan Roethlisberger provided valuable suggestions for the development of Waves4U, and Caroline Kradolfer helped prepare this article. Their support is much appreciated.
Kradolfer, U. (1993). Automating the exchange of earthquake information, Eos, Trans. Am. Geophys. U. 74, 442.
Kradolfer, U. (1996). AutoDRM: The first five years, Seism. Res. Lett. 67, 30.
Malone, S. (1999). Seismic network recording and processing systems I, Seism. Res. Lett. 70, 175.
Wessel, P. and W. H. F. Smith (1995). New version of the Generic Mapping Tools released, Eos, Trans. Am. Geophys. U. 76, 329.
http://www.iris.washington.edu (Web page of IRIS)
http://orfeus.knmi.nl (Web page of ORFEUS)
http://www.emsc-csem.org (Web page of EMSC)
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: 24 February 2000