Post-Eocene Subduction Dynamics and Mantle Flow beneath Western U.S.

Abstract:

Both surface geology and mantle seismic images suggest a complex late Cenozoic history of mantle dynamics over western U.S. We try to understand this history by simulating the Farallon subduction since 40 Ma. Forward subduction models assimilating time dependent seafloor ages, plate kinematics and evolving plate boundaries suggest that the present-day 3D distribution of fast seismic anomalies below western U.S. mostly represent late Cenozoic slabs, which experienced multiple phases of segmentation during subduction because of their young age and small mechanical strength (Liu & Stegman, 2011). A major slab segmentation event occurred around mid-Miocene, with the resulting slab tear and induced asthenosphere upwelling correlating with the Steens-Columbia River flood basalts (SCRB) eruption both in space and in time (Liu & Stegman, 2012). This suggests that a mantle plume is not required for the formation of the SCRB. Segmentation of the Farallon slab generates rapid toroidal flows around the newly formed slab edges beneath the Cascadia arc. These mantle flows may affect both the pattern and composition of arc volcanism through transportation of oceanic asthenosphere material into the mantle wedge.

Based on the forward model, we further test the influence of slow seismic anomalies on mantle dynamics. On the one hand, we explicitly input a deep hot anomaly to represent the putative Yellowstone plume. On the other hand, we develop a hybrid scheme that combines the adjoint inverse method with the high-resolution forward simulation approach, so that the present-day mantle seismic structure is entirely consistent with the convection model. Our preliminary results suggest that a hot plume could actively rise up only when it is several hundreds of kilometers away from the slabs, as is the case prior to 20 Ma. Subsequently, the plume is dominated by the surrounding slabs, resulting in an overall downwelling mantle flow. This suggests that a plume might have contributed to the SCRB formation, but has little effects on the subsequent Yellowstone hotspot evolution. The hybrid adjoint models reveal a similarly passive motion of the Yellowstone slow seismic anomaly. We also find that the spatially extensive slow seismic anomalies observed today are mostly originated from the entrained Pacific oceanic asthenosphere.