The detection threshold of a network including NORESS-type arrays in the Soviet Union is estimated using the frequency-dependent attenuation of Pn and Lg observed at NORESS to normalize the threshold of individual stations. The normalization begins with a parameterization of the magnitude and range dependence of the NORESS Pn and Lg spectra in terms of seismic moment and apparent attenuation. A relationship between these spectral amplitudes and the time-domain amplitudes used in signal detection at NORESS is then determined. This allows an estimate of NORESS detection capability from the spectral parameterization and a prediction of the Pn detection capability for regions with different wave propagation characteristics. A comparable prediction is not possible for Lg because of dispersion and the non-stationarity of pre-Lg noise. Instead, an empirical relationship for Lg detectability based on observed temporal amplitudes is determined. To validate the results for NORESS, predictions based on our model are compared with observed detection statistics and with results obtained by Ringdal (1986), who compared detections at NORESS to bulletins produced by local seismic networks.

The simulations represent the detection capability of a network of NORESS-type arrays, 20 within and 13 outside the Soviet Union. The 90 % ML threshold for detecting three Pn phases for events in the Soviet Union is estimated to be between 2.4 and 2.7 if the frequency-dependent attenuation and noise throughout the network are the same as those observed at NORESS. This threshold is reduced by 0.1 ML if Pn Q is 50% higher in the Soviet Union than estimated for Scandinavia, and is increased by 0.2--0.3 ML if Pn Q is 50% lower. The ML thresholds are reduced by 0.2--0.3 if the detection criteria include Lg as well as Pn phases. While these simulations are carefully normalized by the actual detection capability of a prototype station for test ban treaty monitoring, observations of attenuation and noise inside the Soviet Union are needed to determine how they actually differ from those at NORESS. Also, it must be noted that the NORESS detection capability estimates (and therefore the simulations) are for average conditions and there are regular variations that must be considered when estimating treaty monitoring capability. ¿ American Geophysical Union 1989