Array Analysis of Regional-Distance P-Coda in South Asia

Wepresent a comprehensive studyof regional distanceP-coda recordedby amedium-aperture array station located in ChiangMai, Thailand (CMAR). Using array processing techniques, we analyzed data from 955 events that occurred at distances of 13°–30° with primarily continental paths. Nearly all significant, coherent arrivals in the P-coda traveled along the great circle path with ray parameters within 2–3 sec=deg of the direct arrival. Most of this energy was created near the source, probably via Lg; Rg → P scattering. This was especially true at the larger distances of 28°–30° in which coda decay rates decreased and large-amplitude significant arrivals occurred throughout the first 30 sec of the coda. The codas from approximately 40 intermediatedepth events in our data set were generally less energetic than shallow-event coda at comparable distances. In some cases, the coherent later arrivals we observed were related to uppermantle triplications. At distances of 14°–17°we observed clear secondary arrivals related to the 410 km discontinuity. They appeared on average 2 sec later than expected relative to the first arrivals, which may be partially explained by a depressed 410 beneath Burma. We isolated the diffuse component of the P-coda by subtracting array beams from individual elements to create residual seismograms. The diffuse, residual energy was highly variable from trace to trace and became equal in amplitude to the coherent component after only 20–30 sec of lapse time. We used a high-resolution maximum-likelihood technique to compute frequency–wavenumber spectra of the residual seismograms but did not conclusively observe any deterministic P → Rg scatterer locations. The majority of the diffuse energy is created quite close to CMAR and does not travel as a plane wave across the array.