This article presents the results of the seismological evaluation of the first 18 months (January 1995 to June 1996) of the Group of Scientific Experts third Technical Test (GSETT-3). Performance measures such as completeness, estimates of detection thresholds, and equivalent circle areas as an estimate of location accuracies are obtained by comparing the REB (Reviewed Event Bulletin) of GSETT-3 to global and regional reference bulletins, and also by intrinsic methods.
Incremental detection thresholds are stated on the PDE (Preliminary Determination of Epicenter) mb scale: magnitude conversion formulae are derived to convert REB mb and other reference bulletin magnitudes to this scale. At the 90% confidence level, an incremental detection threshold of about mb (PDE) 3.5 or better is attained in northwest Europe and parts of North America, Central Asia, and Australia. Apart from these areas, the threshold is generally around 4.0 in the Northern Hemisphere. In the Southern Hemisphere it is 4.2 or better in most of South America and northern and equatorial Africa, and between 4.2 and 4.6 in other regions with the highest values being attained in some remote parts of the southern oceans. Still at the 90% confidence level, equivalent circle areas for the mb (PDE) 4–4.5 magnitude range come close to 1,000 km2 in northern Europe. The global average is about 8,000 km2. At the 50% level (the median of all detected events), most northern hemisphere regions have equivalent circle areas at or below 1,000 km2, and even in the southern hemisphere 5,000 km2 are not exceeded.
Observed performance is used to calibrate network simulation codes, which allow to interpolate system capabilities over aseismic regions and to extrapolate for the capabilities of the future International Monitoring System (IMS) seismic network. It is concluded that at the 90% confidence level, except in remote ocean areas, a global incremental detection threshold of mb (PDE) 4 and a location accuracy of the order of 1,000 km2 are attainable with the IMS seismic network. However this will require a full calibration of the network and an optimized usage of the auxiliary station network. New computational methods as well as calibration will help to improve the depth determination.