Progress in Deriving Upper-Mantle Structure beneath Western U.S.

Recent upper-mantle triplication data recorded by USArray display sharp variations in both travel time and differential times between branches similar to that reported by Song and Helmberger (2006). Jumps between branches by up to 8 second are common in S-data. Generally, the recent tomography images of P- travel time data by the EarthScope community, Burdick et al (2008), predicts the horizontal geometry quite well, but does not predict the waveform triplications because of the choice of reference model. The present western U.S. reference triplication P-model, GCA, has the 410 discontinuity at 395 km, which is incompatible with recent S-models. Hence, we propose a modified P-model that can be used with S to construct a travel-time-delay map of paths sampling above the 410 and in the transition zone similar to that derived by Chu and Zhu (2008) for Tibet. Such data can then be added to conventional datasets using a hybrid tomographic method correcting for realistic path corrections in the upper mantle. Models with low- velocity zones and high-velocity zones should greatly help in producing accurate synthetic modeling. Deep events provide the best sources, such as the November 26, 2007 Guerrero, Mexico earthquake. We first inverted the focal mechanism and earthquake depth from teleseismic seismograms using a grid search algorithm. The best-fit solution shows a normal faulting at the depth of 52 km. This event provided a complete map of triplications from 10 to 30 degree. The AB and CD branches of P waves cross at the distance of 17 degree, which is 0.5 degree larger than that predicted by the GCA velocity model. The separation between CD and EF branches is about 0.5 second smaller at 20 degree compared with the GCA prediction, while the separation agrees with the GCA prediction at larger distances. Based on this observation, a 1D P-wave velocity model has been constructed for western U.S.. In this model, the 410 velocity discontinuity is at 420 km, which is 25 km deeper than the GCA. The 660 velocity discontinuity is at 648 km, 12 km shallower than the GCA model. Velocity jump at the 660 discontinuity is 4.20%, compared to 5.78% for the GCA model.