Long-term monitoring of global mass transport within the Earth system improves our ability to mitigate natural hazards and better understand their relations to climate change. Satellite gravity is widely used to monitor surface mass variations for its unprecedented spatial and temporal coverage. However, the gravity data contain signals from visco-elastic deformation in response to past ice sheet melting, preventing us from extracting signals of present-day surface mass trend (PDMT) directly. Here we present a global inversion scheme that separates PDMT and visco-elastic glacial isostatic adjustment (GIA) signatures by combining satellite gravimetry with satellite altimetry and ground observations. Our inversion provides global dual data coverage that enables a robust separation of PDMT and GIA spherical harmonic coefficients. It has the advantage of providing estimates of Earth's long wavelength deformation signatures and their uncertainties. Our GIA result, along with its uncertainty estimates, can be used in future GRACE processing to better assess the impact of GIA on surface mass change. Our GIA estimates include a rapid GIA uplift in the Southeast Alaska and the Amundsen Sea Embayment, due to the viscoelastic response to recent glacial unloading. We estimate the average surface mass change rate from 2002– 2010 to be −203 ± 3 GT·a−1 in Greenland, −126 ± 18 GT·a−1 in Antarctica and, −62 ± 5 GT·a−1 in Alaska. The GIA low degree spherical harmonic coefficients are sensitive to rheological properties in Earth's deep interior. Our low-degree GIA estimates include geocenter motion and J&2 which provide unique constraints to understand Earth's lower mantle and ice history. Plain Language Summary Surface mass exchange between Earth's “spheres”—atmosphere, hydrosphere, cryosphere, biosphere, and pedosphere— are enormous. Monitoring surface mass change helps to understand climate change and mitigate hazardous effects such as extreme drought or flooding. Measurements of surface mass change are perturbed by subsurface processes, such as mantle flow underneath Earth's crust. Often, a model is used to correct the subsurface signals from observations, and likely to introduce un-modeled process errors into the surface mass estimates. We use a mostly data-driven method to extract present-day surface mass change trends along with their error estimates. The results give an enhanced view of the surface mass processes and can help improving the accuracy of future surface mass estimations. Moreover, we get a more accurate picture of the subsurface processes, that are mainly caused by Earth's viscous response to past ice sheet melting. We find that Earth's crust bounces back more rapidly after glacier melting events than we assumed before, especially in places where the mantle viscosity is low.