Global simulations of temporal gravity due to mantle flow and their sensitivity to the mantle rheology

With the plans of the MAGIC/NGGM mission approved, there will be several decades of satellite gravity  data available. Both periodic and secular mass changes can be studied with this data, mostly surface mass changes like hydrology, ice melt, glacial isostatic adjustment, and large earthquakes. With the increasing time period of the gravity data set, smaller processes in the signal can be detected. Therefore, we conduct sensitivity analysis on small temporal gravity signals which can be related to mass change due to mantle convection.

We perform various sensitivity analysis studies to understand the added benefit of detecting mantle flow with satellite gravity change observations. A fast stoke solver (FLAPS) is developed that is based on an axisymmetric half annulus geometry. The model evolves over 50 years after which the difference between the initial and final state to compute the rate of change. Realistic Earth models (PREM) as well as synthetic models are tested to better understand the sensitivity of the gravity change data. To understand 3D variations in structure and viscosity, we use the open-source mantle flow software ASPECT and incorporate interior models related to ESA's 4D Dynamic Earth project. For the upper mantle the WINTERC-G model incorporates multi data types information in a joint inversion. New analysis show data sensitivity down to the transition zone. For the lower mantle, we use available global tomography models.

The gravity change observations are sensitive to the absolute viscosity state of the mantle. This is contrary to dynamic topography and geoid data, which do not have this sensitivity and studies using these data always have an ambiguity wrt. viscosity state. Moreover, it seems that the gravity change data is more sensitive to the lower mantle of the Earth. 3D calculations need HPC resources and we show that the mesh resolution needs high computational demands to consistently account for the temporal gravity due to mantle flow. Nevertheless, the modelled magnitude of the gravity change linked to global mantle convection seems to be larger than the formal error estimates of the GRACE and GRACE-FO instrumentation. A longer acquisition period will reduce the secular errors in the ocean, atmosphere and tidal correction models, such that eventually mantle convection can be studied directly by satellite gravimetry.