Constraining crustal structure in sediment dominated regions: An H-k-V stacking method

The extensive coverage afforded by EarthScope USArray in the lower 48 states provides a unique opportunity to explore crustal variation across a range of tectonic environments and geologic ages. The thickness and average composition of the continental crust are related to its tectonic and geologic history. While crustal thickening or thinning indicate the deformation history, changes in average composition carry information on how crust was formed and reworked through different geologic processes. The most common method of constraining the crustal structure is to calculate H-k stacks (Zhu and Kinamori 2000), which use arrival times of P-to-s converted seismic waves across the Moho together with waves reverberating within the crust. However, when H-k stacking is applied to regions covered with a thick sediment layer, reverberations mask the Moho P-to-s conversions making interpretation of the crustal structure difficult, if not impossible. Furthermore, H-k stacks are constructed for an assumed average crustal P or S velocity. Therefore, inferences of crustal composition must rely on the ratio of P to S wave velocity alone (Vp/Vs ratio), and crustal thickness estimates may be biased. For example, in regions where an assumed Vp is higher than the actual Vp, the crustal thickness will be overestimated and the Vp/Vs ratio will be underestimated. We overcome both limitations of standard H-k analysis by using complementary data: S-to-p conversions and SsPmp phases. Because sedimentary S-to-p reverberations do not contaminate the direct Moho-related S-to-p conversion, we incorporate S-to-p data in our H-k stacks. We shows that doing so is useful in identifying crustal thickness in regions where P-to-s H-k stacks are not sufficient. The large amplitude post-critical S-to-p reflection at the Moho (SsPmp phase) has been used to obtain crustal thickness estimates (see Parker E.H. et al. 2013, Yu et al 2013) in regions where an assumed average Vp is known to be inaccurate. This is because the relative timing of the SsPmp phase depends on Vp but not Vs. We show that, with the addition of SsPmp, we can construct 3-parameter H-k-Vp stacks that allow us to pinpoint Vp and average crustal thickness even in areas with thick sedimentary cover.