Keith F. Priestley and Howard J. Patton
Calibration of m _b (P _n ), m _b (L _g ) scales and transportability of the M ₀ :m _b discriminant to new tectonic regions
Bulletin of the Seismological Society of America (October 1997), 87(5):1083-1099

Abstract:
We have developed M ₀ :m _b relationships for earthquakes and explosions in central Asia using regional waveform data from the Chinese broadband, digital station, WMQ. Regional magnitude scales, m _b (P _n ) and m _b (L _g ), were established by (1) determining the attenuation rates of P _n and L _g waves, (2) calibrating the m _b (P _n ) against the teleseismic m _b , and (3) applying Nuttli's method to determine m _b from the L _g waves. Seismic moments for earthquakes and explosions were obtained by applying the Bolt and Herraiz (1983) method calibrated using CMT moments for earthquakes and regional moment estimates of the Soviet Joint Verification Experiment (JVE) explosion. The results show that M ₀ :m _b (P _n ) relationships for central Asian earthquakes and explosions are in excellent agreement with the relationships for the western United States. This demonstrates that the M ₀ :m _b discriminant is transportable to other regions and supports theoretical claims (Patton and Walter, 1993) that M ₀ :m _b are directly comparable for explosions detonated in different emplacement media. Successful application depends upon correcting the m _b (P _n ) for the magnitude bias known to exist between the Shagan River test site and the Nevada Test Site (NTS). In general, region-specific m _b bias must be accounted for whenever regional magnitude scales are calibrated against teleseismic m _b , as pointed out by Douglas and Marshall (1996). Transporting M ₀ :m _b (L _g ) can be problematic because attenuation rates based on coda waves or L _g waves from earthquakes may not necessarily apply to L _g waves from explosions, and vice versa. This is demonstrated for the western United States where the coda Q measured from explosion data is much lower than the coda Q from earthquake data. We show that m _b (L _g ) for earthquakes located close to NTS is greatly overestimated using the Q values determined from the coda of NTS explosions. This is apparently not a problem for L _g calibration in central Asia, perhaps because of differences in velocity and/or Q structures for the two regions. Other important considerations are the character of L _g propagation on short paths (<400 km) for NTS explosions versus long paths (1000 km) for East Kazakh explosions and the properties of coda-wave spreading as a function of distance and velocity structures for different tectonic regions.