July - September 2000

Expanded Internet Data Services

E-mail data retrieval

Data available from the ISC web site can now be obtained by e-mail, using the AutoDRM request protocol. A summary of request directives supported by the ISC can be retrieved from /doc/autodrm/.

Links to event-related products

Optionally, the ISC web Bulletin now includes hypertext links to sets of waveform data and event parameters available from other web sites. Users who check the "include links" box in /search/bulletin will receive a Bulletin containing links to waveforms at IRIS-DMC, ORFEUS and GFZ-Potsdam, and to event parameters at Harvard U., U. of Tokyo, ING-Rome, and U. of Michigan. A description of the links is available from /search/bulletin/descrip.html.

Integrated Bulletin and Bibliography

The Bibliography of Seismology, which is complete through publications in 1995 (/search/bibliography.html) is indexed to notable earthquakes from the ISC Bulletin, as well as offering search by author, title and key words. The web Bulletin now includes citations from the bibliography if any of the notable events are returned.

Phase arrivals by station and date

Users can now retrieve arrivals for specified stations and time intervals, from /search/arrivals/. Station codes are those from the International Registry of Stations, maintained jointly by the ISC and World Data Center for Seismology, Denver. A comprehensive listing of information on these stations and their networks is available from wwwneic.cr.usgs.gov/neis/station_book. Availability of arrivals for frequently reporting stations is available from /search/stations.

Recently contributed data

ISC analyses data two years in arrears, after national and regional agencies have released their final bulletin. Many data are contributed much sooner, however, and as soon as they are contributed arrival data are now included with phase data retrieved by station and date. Users must be prepared to deal with duplicate and/or erroneously parsed data when using arrivals from months that have not yet been analysed at the ISC. Most reports contributed electronically are also available in their original format from /doc/misc/collection.html

ISC Internet data services are freely available to all, thanks to support for the ISC from more than fifty member organisations around the world (/doc/intro/isc_members.html).

Survey of Users

Everyone who uses ISC data is encouraged to complete our survey at /survey.html by the end of this year. The survey is also available by e-mail, or by post or fax. A complimentary copy of the 1997 Catalogue CD (i.e., without phase data) is offered to each person returning a completed survey.

ISC at the AGU Meeting

ISC will have four presentations at the AGU meeting this December. Three of them will update ISC studies of event location, discussed on the back of this newsletter. The fourth evaluates several alternative explanations for year-to-year changes in rates of moderate-size earthquakes (m_b = 4 to 6). The abstracts are at /agu2000.html.

Waveforms at the ISC!

Robin Adams, ISC's senior seismologist until he retired in 1992, described his work as that of a "blind seismologist", because he had to interpret readings without benefit of examining the records himself. The Centre's current senior seismologist Dmitry Storchak carries on this tradition, but its end is nigh.

With a system developed by Matt Evens this summer, ISC seismologists run a program to specify their requirements. The system then composes complete requests using an inventory of available waveforms, which catalogues where waveforms of hundreds of stations are available on-line. A notification that is generated when the requests are fulfilled can be used to alert the person requesting the data, or to start a program to display or process the data.

Waveforms might help to evaluate phase identifications when editing difficult earthquakes. In addition, the ISC is considering whether new measurements there are that should be catalogued along with onset time picks and amplitudes.

Matt Evans, a student in the University of Leeds M.Sc. geophysics programme, created a waveform retrieval system for the ISC.

Preparing to Use New Travel Time Models and Additional Phases

With the advice of the Governing Council, the ISC has been working towards updating its earthquake location procedures to make use of advances over nearly 35 years since the Bulletin was first published. After a review of the ISC's progress to date, the Executive Committee suggested implementation of some improvements for earthquakes from January 2000. To achieve this, the ISC must be confident by late 2001 that it has selected good procedures and that they are working well. ISC seismologists described preparations for this important change at the European Seismological Commission meeting in Lisbon this September.

PKP and S arrival times improve moderate size event locations.

Many types of phases associated with seismic events all over the world are already published in the Bulletin, yet only the P times are used for computing ISC hypocentres. Dmitry Storchak has investigated what size earthquakes would have their locations changed most if the ISC took more advantage of the wide variety of contributed phases.

Moderate sized earthquakes (m_b < 4.5) are shifted more than larger events when additional phases are used, but even some m_b > 5.5 events move more than 30 km.

Using S and PKP arrival times to relocate events of one month of the Bulletin shows that the most benefit comes from the use of S arrivals at close distances and PKP arrivals at distances near the caustic. Relocations using Jeffreys-Bullen travel times for P at distances 0º-105º, S at 0º-8º and PKP at 140º-150º does not result in a systematic shift of hypocentres. Uncertainties in earthquake location generally improve, even though the variance of travel time residuals rises slightly. The number of events with the fixed depth diminishes, since more depths are constrained by PKP times. The greatest location improvements are found for m_b < 4.5 events, since epicentres of larger earthquakes are already well-constrained by a sufficient number of P times.

Regional S arrival times help locate unconstrained events.

Due to poor station distribution the ISC fails to compute a depth or even an epicentre for some earthquakes recorded by a good number of stations. Mamy Andrianirina's study focused on these events, which are troublesome because so many small earthquakes are recorded only at stations in a limited range of azimuths. For an arbitrarily selected month, he found that using secondary phase times, especially regional phases such as Sg and Sn, can cut the number of events without an ISC-computed location by more than half. With large azimuthal gaps, an accurate travel time model remains essential to avoid biased locations. Regardless of the model, however, reliable locations could not be computed for many of these events from P times alone.

Events with no ISC location (pale grey) have large azimuthal gap. With S times, however, ISC locations can be computed in most cases (dark grey); the limit is availability of S times rather than the distribution of stations.

For events where ISC was already able to compute a location from P times, relocating the events using S times usually moved the ISC-computed location closer to that reported by a local agency, indicating better accuracy. In addition, formal uncertainties of ISC locations were reduced with the addition of regional secondary phase times.

Perhaps the most intriguing response to this work is that several agencies have volunteered to begin contributing secondary phase times, now that ISC plans to put them to good use. If so, it may be possible to compute ISC locations for even more of these events.

Travel times from 3-D earth models reduce location bias.

Chen Qi-fu presented results from teleseismic location tests that compared locations using spherically symmetric (JB, PREM, ak135) and laterally heterogeneous models (SP12, KH07), and investigated the effects of elevation and crustal thickness corrections.

Events were relocated with each model using ISC's standard procedures. The algorithm is based on Geiger's method, but weights arrival times iteratively and estimates location uncertainty from results of modern robust regression. The 3-D models give statistically significant improvement in locations over any of the 1-D models for globally distributed ground-truth events, including 26 explosions studied by Smith and Ekström and 82 earthquakes used by Kennett and Engdahl for the IASP91 model. Nevertheless, tests excluding regional and local data showed that with teleseismic data alone 3-D models are insufficient to equal the accuracy achieved in the Bulletin.

Cumulative distributions of explosion epicentre errors for several one- and three-dimensional earth models.

For a one-month set of all earthquakes that could be relocated without manual editing, 3-D model KH07 improved residuals and location uncertainty compared to those in the Bulletin. Further tests using KH07 with selected South American events improved resolution of the subduction zone (especially at h>100 km) compared with the Bulletin, but not as well as Engdahl et al., who used depth phases.