October - December 1999

BP-Amoco Royal Society Fellow Joins ISC

Dr. Chen Qi-fu of the China Seismological Bureau arrived at the ISC in December to begin an investigation of earthquake locations using 3-dimensional earth models. Qi-fu's one-year visit is funded by a BP-Amoco Royal Society Fellowship. Various researchers have investigated this topic before, but almost all routinely published earthquake locations are based on laterally invariant travel time models.

Harvard University software that computes travel times in a long-wavelength model "SP12" will be used with the ISC's existing location algorithm to compute some of the new locations. Locations using other global models, such as "BSE" from Utrecht University may be computed as well. An important element of Qi-fu's work will be comparing differences between biases of the various models, and the effect of different crustal corrections.

The ISC's traditional locations, based on primary arrival times and the Jeffreys-Bullen travel time tables, will continue to be published in the Bulletin. Based partly on Qi-fu's work, however, the hypocentres available on the ISC CD's and web site may eventually be supplemented with the 3D-based locations.

Anthony Hughes
Receives O.B.E.

Congratulations are due to former ISC director Anthony Hughes, who was appointed an Officer of the Order of the British Empire in the New Year honours list for "services to global seismology". Anthony will be presented with the award by Her Majesty the Queen at Buckingham Palace.

Alison Bird Returns to Canada

After completing a 2-year appointment, Alison Bird has returned home and accepted a position with the Canadian Geological Survey at the Pacific Geoscience Center. Alison was an important part of our efforts to accelerate Bulletin publication over the last year, and will be missed by us all.

New Bulletin Interface
on ISC Web Site

ISC's database manager James Harris has created an important new capability for ISC's web site. The new page at www.gxjydl.com/search/bulletin/ offers selection of events and readings from long time periods. Previously, it had been possible to select events only from one calendar month at a time.

Some universities have kept ISC data on-line in order to select data from long time spans. But a linear search of files for more thirty years may take tens of minutes, exceeding time-out periods for Internet connections. Taking advantage of indexing available in our relational database, we are able to provide a much quicker selection.

Nevertheless, there is still a risk that queries may overwhelm the ISC's resources. In order to minimise the chance of problems we first estimate the number of events based on the geographic, depth, magnitude and date criteria of a request. We attempt to fulfil all requests except those likely to return hypocentres for more than 5000 events or readings for more than 500 events. Users requiring larger data sets are suggested to purchase the CDs.

NSF Awards
New Grant

The National Science Foundation has awarded the ISC a 3-year grant to expand the types of data available from the ISC on its CDs and web site. The two most important new data are the "EHB" hypocentre catalogue and the readings previously reported to the ISC but unassociated with earthquakes in the Bulletin.

Bob Engdahl and his colleagues computed the EHB hypocentres from readings in the Bulletin. The distinct elements of their location procedures are a different travel-time model, ak135, and a statistically-based phase identification designed, in part, to remove bias in identifying depth phases. The new hypocentres have more tightly clustered depths, and they may have advantages over hypocentres in the Bulletin. When these new locations become available from the ISC, some researchers will be able to dispense with their own re-computation of hypocentres before beginning analysis of seismicity or details of regional earth structure.

The readings reported to the ISC but unassociated with published locations have not previously been easily accessible. They are stored in binary-format files originally written on IBM, DEC and Sun computers over several decades. The raw files have been copied to modern tape cartridges, however, so the data are certainly recoverable.

Replacement
CD Prepared

In late 1999 a review of ISC data collection programs discovered that amplitudes, periods and slownesses reported in GSE2.0 format files were being truncated. ISC's GSE2.0 parser has been in use since January 1997, and affects data for earthquakes since January 1996.

The truncated slownesses have had no effect on hypocentres, since they are not used by the ISC. Amplitudes of teleseismic body waves are generally small enough to avoid truncation, but many regional phases and surface wave amplitudes were effected. Truncated regional phase amplitudes have had no further effect in the Bulletin because ISC does not compute its own ML. The ISC does compute its own MS, but only from LR readings with periods between 10s and 60s. Since the effected LR readings also had their periods truncated, to periods less than 10s, they were not used by the ISC to compute MS.

Despite the limited impact on earthquake parameters, we are making every effort to correct the mis-stated data that have been published. The only CD with effected data is volume 7, and we plan to distribute replacements for every copy purchased so far. In addition, we plan to distribute copies of the replacement CD to every subscriber of Bulletin volume 33, which reports 1996 earthquakes. Further details have been posted to http://www.gxjydl.com/truncations.

ISC Presentations at
AGU Fall Meeting

The ISC presented three posters at the AGU meeting in San Francisco last December on differences between independently picked arrival times, the future of seismic bulletins, and accessing data from the ISC.

The poster on differences between arrival time picks was "one scary poster", according to Paul Richards of the Lamont-Doherty Earth Observatory. In this poster Ray Willemann compared onset times at the same station but picked independently by station operators and by analysts at the prototype IDC (agency code EIDC in the Bulletin). Though based on waveform data from the same or co-located sensors, EIDC readings are earlier on average by several tenths of seconds for each station and phase reviewed. The average differences are small compared with the scatter, which is typically about 1 second for primary arrivals and larger for secondary arrivals. Nevertheless, the systematic differences have high statistical significances owing to the large number of contributing data. This concerned Professor Richards because efforts to improve earthquake locations using elaborate station correction functions might come to nought if station timing biases depend on record readers more than earth structure.

The poster on seismic bulletins asked what priorities ought to be set for the coming era of ubiquitous broadband digital data. The important data types to include might depend on whether the Bulletin is intended to serve primarily study of earth structure, earthquake physics or seismic hazard. The most controversial part of the poster may have been the suggestion that ISC might consider computing higher-order moment tensors in the future. Higher-order moment tensors are a representation of distributed sources that avoids favouring any particular kinematic model. While some seismologists are concerned that such computations are inherently under-determined, Jeff McGuire and Tom Jordan of MIT presented a paper at the meeting showing that positivity constraints that only require the source to be physically meaningful often stabilise the inversion.

The most pleasing feature of the ISC presentations, however, may be that the poster on data access has already become out of date! Among the "future plans" described in the poster, "historical" hypocentres from 1904-1963 were posted to the ISC web site in late 1999 and the new page for queries spanning multiple calendar months is announced in this newsletter. Soon from the ISC, promised the poster, are selections from the Bulletin by e-mail in response to AutoDRM messages and recently received readings, i.e., those being collected for future issues of the Bulletin.

Scary Differences: The distribution of 937 differences between Pn times reported for the Noress Array by the Norsar and by the prototype IDC, and associated by the ISC with the same earthquake. Excluding outliers > +/-2s, the standard deviation is 0.64s, so the uncertainty of the mean is 0.02s. Thus, even though the mean of 0.11s for this asymmetrical distribution is smaller than the mode, it significantly non-zero.