Teleseismic event detection using the NORESS array, with special reference to low-yield semipalatinsk explosions

Abstract The small-aperture NORESS array in Norway has been designed for improving signal-to-noise ratios at high frequencies (1 to 15 Hz). While the main motivation for this has been to enhance the capabilities for detecting and characterizing weak seismic events at regional distances, it has been found that the array is also very effective in the teleseismic distance range, in particular for Eurasia. This results from a combination of two factors: (1) due to high Q paths, Eurasian earthquake and explosion recordings generally show high dominant frequencies, typically in the range of 1.5 to 4.0 Hz, where the noise level at the site is low; and (2) the array provides a particularly high signal-to-noise ratio gain of 12 to 14 dB (0.6 to 0.7 m b units) at these frequencies. Analysis of NORESS on-line detection performance shows that there are large regional variations, even within small epicentral areas. This is interpreted as resulting mainly from signal focusing effects underneath the source areas, in analogy to the receiver focusing effects earlier observed across large arrays. A detailed study of NORESS recordings of underground nuclear explosions at the Semipalatinsk Test Site has shown that relative to worldwide m b , NORESS has a significant positive m b bias. It is highest (1.0 m b units) for the eastern part (Shagan River) and somewhat lower (0.4 m b units) for the western parts (Degelen/Konystan), although there is a fair amount of scatter within each area. Analyzing observed signal-to-noise ratios at the NORESS site for a set of low-yield nuclear explosions at Semipalatinsk, with published yields available in the Soviet literature, it is estimated that fully coupled explosions of yields as low as 0.1 kt would be detectable by NORESS under normal noise conditions. However, to give a precise threshold is difficult because of the low number of reference events available and the significant amount of extrapolation involved. It is emphasized that this detection level will not be achieved in cases when the noise level is abnormally high (e.g., in the coda of a large earthquake) or if coupling conditions are not optimal (e.g., in the case of full or partial cavity decoupling). It must also be noted that the event identification threshold is necessarily higher than the signal detection threshold.