Regional Coda Magnitudes in Central Asia
LOS ALAMOS NATIONAL LAB ALBUQUERQUE NM
Pagination or Media Count:
Coda waves are known to be influenced less by path and radiation effects than are direct waves, especially for local earthquakes. This is attributed to the averaging effects of the scattered waves that make up the coda. Furthermore, the duration of coda allows multiple independent measurements to be taken. Thus, coda potentially provide high precision estimates of source size such as magnitude and moment, with less path bias than direct waves. For purposes of nuclear explosion monitoring, we investigated the use of Lg coda to estimate magnitude at regional distances in central Asia. To do this, we processed coda data for over 3000 recordings made at stations MAKZ, WMQ and LZH. The study area is geologically heterogeneous, and includes Tibet, the Tarim basin, the Tian Shan and platform regions to the north. In tests using 1-Hz data, the scatter between coda amplitude levels obtained from stations WMQ and MAK measured 0.12 standard deviation in the log10 domain. This is over a factor of 2 lower than the scatter observed for direct Lg. However, in this region, bias enters the coda magnitudes through the effects of lapse time and path. The lapse time bias enters through the inability of standard curves to fit coda decay over long time intervals. We reduce this effect by limiting coda lengths to between 30 and 150 s following the Lg arrival. Coda also showed path dependence similar to direct Lg. We corrected for this using an interpolation kriging method. Corrections for lapse time and path had little effect on inter-station scatter because these effects were correlated between stations. We conclude that the early coda yields a more precise measure of event size than does direct Lg, but coda results must still be corrected for path effects in this heterogeneous region.