Upper mantle density modelling for large-scale Moho gravity inversion: case study on the Atlantic Ocean

Moho geometry is an important parameter in geodynamic understanding. Seismic methods can provide accurate burial depths, but the coverage is low; the widely used CRUST1.0 model under oceans is greatly restricted by a low coverage of seismic data. Gravity inversion can provide high coverage of Moho undulations, but the accuracy is relatively low. Observed gravity anomalies integrate gravity effects induced by 3-D density perturbations in the Earth. For studies of the geometry of a specific density interface based on gravity, all gravity effects caused by other interfaces should be removed from the total anomaly map based on 3-D density distributions. We use temperature-based and velocity-based methods to model the lithospheric/upper mantle density; two temperature-calculation parameter models (PSM and GDH1), three thermal expansion coefficients (alpha) and three scaling ratio models are tested for each method. PSM and GDH1 result in a similar mantle gravity anomaly (MGA) pattern but different MGA average values when the alpha model is the same. Different alpha and scaling ratio models result in different Moho inversion results, and the optimal result from the temperature-based method has an RMS of 3.2 km with the CRUST1.0 model. A comparison between this inversion result and the CRUST1.0 model shows that the temperature-based method performs well in reproducing Moho geometry in the main part of the Atlantic Ocean, especially in the central basin. The optimal result from the velocity-based method has an RMS of 4.3 km with the CRUST1.0 model. Compared with the temperature-based method, the velocity-based method performs worse in the central basin but better in regions with low-velocity and low-density mantle away from the mid-ocean ridge.