Discrimination between Eurasian Earthquakes and Underground Explosions using the mb: M, Method and Short-Period Autoregressive Parameters

The mb: M, criterion has been tested on a data base comprising 61 presumed nuclear explosions and 45 presumed earthquakes from Eurasia recorded at NORSAR between 1971 and 1974. For all events, PDE solutions are available, and mb (PDE) has been compared with M, (NORSAR), which has been obtained for 46 of the explosions and all of the earthquakes. For the remaining explosions Rayleigh waves were not detected, and M, was replaced with a noise measurement representing the upper limit for M,. The detection threshold of Rayleigh waves was estimated at M, = 2-9+0*4, where the large standard deviation results from large seasonal variations of the background noise level. The mb: M, criterion works quite well for this data base, with the exception of a few anomalous events in Western Russia and some of the events for which no Rayleigh waves are detected. Significant improvement in discrimination is obtained when the mb: M, data are combined with short-period auto-regressive parameters used to describe the autocorrelation structure of the signals and the preceding noise, and this specifically applies to those events for which no Rayleigh waves are found. 1