When network station spacing is large relative to event depths, constraining focal depths using direct arrivals is difficult. To help overcome this problem, we incorporated the SzS, SzP, and PzP Socorro magma body reflections into the earthquake location process. To find a velocity model appropriate for combining direct and reflected arrivals, we jointly inverted for 75 hypocenters and a one-dimensional velocity model. We used over 1000 direct arrivals and over 400 reflected arrivals, solving for P velocity (Vp) and Poisson's ratio (v) above the approximate base of the seismogenic zone, Vp and v below the base of the seismogenic zone, depth to the upper surface of the magma body, and station corrections. High-velocity and low-velocity initial models converged to the same solutions without damping. We found that Vp and v may both decrease slightly below about 10-km depth. Magma body depth is near 18.75 km. A close fit between observed and modeled reflected phase arrival times and an even lateral distribution of positive and negative reflected-phase residuals indicate that the magma body upper surface is nearly flat.
In the Socorro area, combining direct and reflected phases reduces average focal depth error by more than a factor of 3 and average origin time error by a factor of 2 compared with using only direct phases. This is because reflected arrivals are more sensitive to changes in focal depth than direct arrivals, and combining upgoing (direct) and downgoing (reflected) times reduces the trade-off between focal depth and origin time. For this method to be successful, phases must be carefully identified. To help identify secondary phases, we used relative arrival time curves, plotting phase-minus-P times versus S-minus-P times. Picking reflection pairs (SzP and SzS, SzP and PzP, or SzS and PzP identified on a single seismogram) significantly reduces the chances of phase misidentification. The SzP-SzS and SzP-PzP pairs also help prevent trade-offs between reflector depth and focal depths.