Viscosity structure of Earth’s mantle inferred from rotational variations due to GIA process and recent melting events

We examine the geodetically derived rotational variations for the rate of change of degree-two harmonics of Earth’s geopotential, J₂, and true polar wander, combining a recent melting model of glaciers and the Greenland and Antarctic ice sheets taken from the IPCC 2013 Report (AR5) with two representative GIA ice models describing the last deglaciation, ICE5G and the ANU model developed at the Australian National University. Geodetically derived observations of J₂ are characterized by temporal changes of −(3.7 ± 0.1) × 10⁻¹¹ yr⁻¹ for the period 1976–1990 and −(0.3 ± 0.1) × 10⁻¹¹ yr⁻¹ after ∼2000. The AR5 results make it possible to evaluate the recent melting of the major ice sheets and glaciers for three periods, 1900–1990, 1991–2001 and after 2002. The observed J₂ and the component of J₂ due to recent melting for different periods indicate a long-term change in J₂—attributed to the Earth’s response to the last glacial cycle—of −(6.0–6.5) × 10⁻¹¹ yr⁻¹, significantly different from the values adopted to infer the viscosity structure of the mantle in most previous studies. This is a main conclusion of this study. We next compare this estimate with the values of J₂ predicted by GIA ice models to infer the viscosity structure of the mantle, and consequently obtain two permissible solutions for the lower mantle viscosity (η_lm), ∼10²² and (5–10) × 10²² Pa s, for both adopted ice models. These two solutions are largely insensitive to the lithospheric thickness and upper mantle viscosity as indicated by previous studies and relatively insensitive to the viscosity structure of the D'' layer. The ESL contributions from the Antarctic ice sheet since the last glacial maximum (LGM) for ICE5G and ANU are about 20 and 30 m, respectively, but glaciological reconstructions of the Antarctic LGM ice sheet have suggested that its ESL contribution may have been less than ∼10 m. The GIA-induced J₂ for GIA ice models with an Antarctic ESL component of ∼10 m suggests two permissible lower mantle viscosity solutions of ηlm ∼ 2 × 10²² and ∼5 × 10²² Pa s or one solution with (2–5) × 10²² Pa s. These results suggest that the effective lower mantle viscosity is larger than ∼10²² Pa s regardless of the uncertainties for an Antarctic ESL component. We also examine the polar wander due to recent melting and GIA processes, suggesting that the observed polar wander may be significantly attributed to convection motions in the mantle and/or another cause, particularly for permissible lower mantle viscosity solution of (5–10) × 10²² Pa s.