ISSN  2096-3955

CN  10-1502/P

Reproducing past subduction and mantle flow using high-resolution global convection models
JiaShun Hu, LiJun Liu, Quan Zhou
doi: 10.26464/epp2018019
Plate subduction drives both the internal convection and the surface geology of the solid Earth. Despite the rapid increase of computational power, it remains challenging for geodynamic models to reproduce the history of Earth-like subduction and associated mantle flow. Here, based on an adaptive approach of sequential data assimilation, we present a high-resolution global model since the mid-Mesozoic. This model incorporates the thermal structure and surface kinematics of tectonic plates based on a recent plate reconstruction to reproduce the observed subduction configuration and Earth-like convection. Introduction of temperature- and composition-dependent rheology allows for incorporation of many natural complexities, such as initiation of subduction zones, reversal of subduction polarity, and detailed plate-boundary dynamics. The resultant present-day slab geometry well matches Benioff zones and seismic tomography at depths < 1500 km, making it possible to hindcast past subduction dynamics and mantle flow. For example, the model produces a flat Farallon slab beneath North America during the Late Cretaceous to Early Cenozoic, a feature that has been geodynamically challenging to reproduce. This high-resolution model can also capture details of the 4-D evolution of slabs and the ambient mantle, such as temporally and spatially varying mantle flow associated with evolving slab geometry and buoyancy flux, as well as the formation of shallow slab tears due to subduction of young seafloors and the resulting complex mantle deformation. Such a geodynamic framework serves to further constrain uncertain plate reconstruction in the geological past, and to better understand the origin of enigmatic mantle seismic features.
Keywords: global subduction model, data assimilation, Earth-like subduction, Farallon flat slab, mantle flow

Supported by Beijing Renhe Information Technology Co. LtdE-mail: info@rhhz.net