The Bulletin of December 1999 events was posted to the ISC web site in late October and the CD for all of 1999 was distributed a few weeks later. Analysis of 1999 was finished ahead of schedule thanks to improved automatic processing and a scheme for concentrating manual analysis on events that require attention.

For the first time, this year's Bulletin includes data for contributed events that do not warrant manual analysis at the ISC. Results from automatic processing were reviewed for 52,000 events, but the total number of events for which the Bulletin includes data jumped from 66,000 in 1998 to 155,000 in 1999. The events integrated into the Bulletin but not reviewed by ISC seismologists were generally too small to be detected by more than a single network.

For events in common with other global summaries, the ISC Bulletin often includes stations sufficient for a more accurate or more reliable solution.

While manual review at the ISC focuses on events with a reported magnitude of 3 or more, contributors are encouraged to send data at least for all M>2 events. Even with this threshold, further improvement in collection could result in a several-fold increase data volume.

The ISC computed mb for 13,500 events of 1999. The worldwide completeness threshold indicated by roll-off of the magnitude-frequency relation was near mb 4.3, as it has been since 1995. Also typical for the late 1990's, about half of the amplitudes used to compute ISC mb are from the prototype International Data Centre (pIDC), mainly because many other agencies do not measure teleseismic amplitudes. Very few events for which the ISC did not compute mb were detected at all teleseismically, but completeness to mb<4 in well-monitored regions suggests that improvement is possible elsewhere.

Processing of readings from disparate contributors reveals the existence of fewer previously unrecognised mid-ocean events than in years past. Nevertheless, the ISC continues to discover dozens of new events each year between seismological networks, even near international borders within Europe.

The reviewed events in the Bulletin have an average of 21 initial phases each. This is close to the average for the last 10 years, reflecting a balance between the growing number of small events that are included and collection of more phase data. The number of later phases equalled about 51% of the number of initial phases, which is within the range of 43% to 53% that holds for almost every year of the Bulletin since 1964.

Preliminary Bulletin Posted

The ISC web site and e-mail servers now offer a preliminary Bulletin that includes even the most recently contributed data. Periodic re-processing excludes most duplicate readings and hypocentres, groups independently computed hypocentres for the same events, and associates phase readings with events.

Many smaller events are found only in regional catalogues that are not available in final form until more than a year later, but most large earthquakes (M>5) are reported soon after they occur. ISC's automatic grouping and association are subject to review by ISC seismologists, but only a few per cent of processing results are changed in analysis so the preliminary Bulletin is generally reliable.

The on-line Bulletin includes links to further data for most of the larger events, including details regarding calculation of moment tensors and source time functions, as well as sets of Spyder® and FARM waveform segments from the IRIS DMC and the Orfeus Data Centre.

Further development in the near future is planned to include maps and links to other on-line databases, especially when criteria used in selecting events from the ISC Bulletin suggest interest in further data that the ISC knows of elsewhere.

Station Coverage Maps Posted

Readings from more than 2500 seismic stations helped to ensure reliability of the Bulletin of 1999 events. But the density of stations varies widely around the world, partly because the ISC aims to be comprehensive rather than uniform. The effect of station density is important but difficult to quantify. It can be hard to distinguish between true changes in seismicity and artefacts from changes among reporting stations, for example, especially if magnitude is not computed for some events.

A general view of detection capability can be gleaned from station distribution statistics such as azimuthal gap, i.e. the largest angle between two adjacent stations as viewed from an event. If the stations that detect an earthquake leave an azimuthal gap more than 180° then it may not be possible to compute an epicentre and the event may fail to be included in the Bulletin.

Within a sufficiently small distance from any point a seismic network includes no more than two stations, which unavoidably have an azimuthal gap of at least 180°. Conversely, when stations at sufficiently large distances are included then azimuthal gap is less than 180° around any point within a network. The coverage of a network at a particular location can be characterised by the distance beyond which one must go to find stations with an azimuthal gap less than 180°.

If arrival times at two or more stations in any semicircle around an event are used to compute a location then no single arrival time controls a trade-off between epicentre and origin time. Thus the distance at which the second azimuthal gap (i.e., azimuthal gap ignoring the most isolated station) falls below 180° is sometimes regarded as a better measure of network coverage. A global map of this distance among stations that contributed to the 1999 ISC Bulletin is show below, and regional maps of station coverage are available at http://www.gxjydl.com/doc/misc/gapmap/index.html

It is unsurprising that detection at teleseismic distances is required to reliably locate events in the oceans, but the maps also show that island stations can be critical to effectively illuminate our view of seismicity off shore from even extensive continental networks. Coverage is understandably sparse in parts of some continents where large earthquakes occur only infrequently, such as eastern South America and Saharan Africa. In addition, however, the maps show that reporting stations were insufficient for good coverage during 1999 in parts of western South America, and northernmost Africa, and much of central Asia.

Members of the ISC staff are working to improve data exchange with earthquake monitoring agencies around the world, especially where coverage is sparse. New catalogues have been contributed from Vietnam and Iran, and additional bulletins are hoped for from Peru, Algeria, Kyrgyzstan, Korea and Iceland. Nevertheless, Bulletin users are encouraged to suggest other agencies that might contribute a catalogue or bulletin to the ISC.

ISC Paper in BSSA

"Global test of seismic event locations using three dimensional earth models" was published in the December 2001 Bulletin of the Seismological Society of America. The paper summarises the work of Chen Qi-fu of the China Seismological Bureau during his Royal Society fellowship at the ISC. Qi-fu showed that each of several 3D models improves agreement between computed hypocentres and well-known locations of explosions and earthquakes. Reprints can be requested from Qi-fu (chenqf@seis.ac.cn) or the ISC.

ISC Posters at AGU

At the AGU Annual Fall Meeting ISC seismologists Dmitry Storchak, Mamy Andrianirina and Esmeralda Banganan authored a poster summarising the 1999 Bulletin. Among many features of the Bulletin, they showed how the ISC's comprehensive data collection results in hypocentres that are generally more reliable than those in other global catalogues for comparable events.

Ray Willemann presented an analysis of residuals from Chen Qi-fu's relocation of 3800 ISC events around the world using a tomographic model, and from a further relocation using the one-dimensional ak135 model. He showed that ak135 improves residuals compared with those from the Jeffreys-Bullen tables that are used in ISC operations, and that further improvement from the tomographic model is also statistically significant. But Ray concluded that each increment is marginal in any particular region, and only the combination of a both a better 1D base model and 3D perturbations provides important improvement in many areas.

Robin Adams in Zürich

Robin Adams was invited as a Resource Expert to the first workshop of the PEGASOS project, a Probabalistic Seismic Hazard Study for nuclear power plants in Switzerland, carried out by the Swiss National Cooperative for the Disposal of Radioactive Waste for the Swiss Federal Nuclear Safety Inspectorate. Robin attended the meeting near Zürich in October and gave a presentation "Experience of the ISC Related to Location and Magnitude Errors and Uncertainties in European Earthquake Catalogues"