Improved Focal Depth Determination for Use in Ctbt Monitoring

Seismic event location remains as one of the most important discriminants for separating natural tectonic and explosive events. However, in order to be useful for discrimination purposes, the uncertainties associated with seismic locations must be well defined and reliable, and this has proven to be difficult to accomplish to the required degree of accuracy. In particular, high-confidence estimation of focal depths remains as an outstanding monitoring problem. During the past year, we have continued to pursue a research program which is directed toward the development of improved detection and identification procedures for the depth phases pP and sP, as well as with formulation of a new algorithm for computing more reliable confidence intervals on focal depth estimates determined from P-wave first arrival times. With regard to depth phase identification , we have continued to investigate the utility of the fully automatic network stacking algorithm which maps the IDC post-P detection times at a station into candidate depth phases using the pP P and sP – P delay times predicted by the IASPEI travel-time tables for that epicentral distance and then combines the individual station results as a function of candidate source depth. This automatic algorithm has now been applied to data from about 150 REB events in the Hindu Kush and in the Hokkaido and central Honshu regions of Japan. Prominent candidate pP and sP peaks have been identified in the resulting network detections stacks for a majority of these events, including some with mb values as low as 3.7 and depths as shallow as 50km. Current effort on this phase of the project centers on the incorporation of the Pearce algorithm (Pearce, 1977; 1980) into the depth phase identification procedure. In this approach, the relative amplitudes of P and any candidate pP and sP phases are determined for the various observing stations and processed by the Pearce algorithm to define the range of earthquake focal mechanisms which is consistent with these relative amplitude observations. This permits the relative amplitude characteristics of candidate depth phases to be assessed for consistency with the predictions for characteristic earthquake focal mechanisms in that source region, thereby greatly increasing the confidence in the depth phase identification. The investigation of improved confidence intervals on focal depth estimation determined from P-wave first arrival times has also continued, with Monte Carlo simulations being used to define precise confidence regions corresponding to generalized model assumptions. This model is currently being carefully evaluated using data from earthquakes for which the focal depths are well constrained by verified depth phase observations and is being extended to include the formulation of an hypothesis test which will be suitable for event screening purposes.