Evaluating tomotectonic plate reconstructions using geodynamic models with data assimilation, the case for North America

Reconstructions of Earth's past surface kinematics are traditionally based on a combination of relative plate motions inferred mostly from preserved seafloor information and an assumed absolute reference frame using data from hotspots and/or true polar wander. Recently, plate reconstructions progressively introduced constraints from deep mantle structures like those imaged through seismic tomography. This additional information is utilized through either implicit or explicit fashion, where the lateral location of an imaged mantle slab represents that of its paleo-trench with the age of initial subduction estimated from geological proxies and slab depth. Here we quantitatively evaluated the geodynamic and tectonic implications of three recent global plate reconstructions (Müller et al., 2016, 2019; and Clennett et al., 2020) by focusing on subduction beneath North America. These reconstructions imply different amounts of trench retreat, plate motion and subduction zones, due to their varying dependence on the tomotectonic constraints. We simulated their respective subduction histories since 200 Ma using a sequential data-assimilation methodology. The resulting present-day slab structures based on these reconstructions show clearly diagnostic differences, among which the model based on Müller et al., 2016 best matches seismic tomography and Mesozoic paleotopography constraints within North America, supporting the tomography-implied differential lithosphere motion relative to the mantle. In contrast, modeled results based on the explicit tomotectonic reconstruction of Clennett et al., 2020 match both slab structure/evolution and associated paleotopographic constraints to the least. Consequently, the presented data-assimilation geodynamic modeling exercise, through reproducing the associated subduction history and continental tectonics, could quantify the tectonic implications of different plate reconstructions. We propose that further implementing this exercise through an iterative geodynamic-tomographic-tectonic workflow could serve to improve the tomotectonic reconstruction.