Station site effects, uncertainties in seismic source spectrum, and instrument response errors are among the well-known frequency-dependent contaminating factors that limit the reliability of short-period Q measurements of regional phases. For the Pn wave, a regional phase of importance for both magnitude determination and nuclear test ban verification, the problem is made worse by the added uncertainty of its geometric spreading function. For realistic earth models, the Pn geometric spreading function is likely to depart drastically from that expected of canonical head waves. The extent of this departure is sensitively dependent upon the regional crust/mantle structure, making geometric spreading assumption a conspicuous source of disagreement among the published Pn attenuation (QPn) estimates.
We describe a technique, referred to here as the extended reversed two-station method (RTSM), for simultaneous determination of QPn and geometrical spreading function. The formulation, being designed to bring about direct cancellation of the contaminating source, station and instrument effects, is a reliable tool for mapping the Pn propagation characteristics over continental paths, long and short.
The extended RTSM has been tested using Pn spectral amplitude data derived from seismic records of the Eastern Canada Telemetered Network (ECTN). We find the spreading rate coefficient n in the power-law representation of the geometric spreading (d−n, d being epicentral distance) to be frequency dependent, increasing from 1.11 at 1 Hz to 1.77 at 20 Hz. Our QPn model in eastern Canada takes the form of QPn = 189f0.87. The results from eastern Canada suggest that: (a) there exists a significant positive velocity gradient in the uppermost mantle (≧ 0.0037 sec−1); (b) the regionally recorded Pn waves are dominated by the superposition of a series of interfering diving waves bent by the velocity gradient and internally reflected at the underside of the Moho discontinuity; and (c) the very strong frequency-dependence of QPn we found in this study region may not be unique among low-attenuating shield and platform regions.