Total meltwater volume since the Last Glacial Maximum and viscosity structure of Earth's mantle inferred from relative sea level changes at Barbados and Bonaparte Gulf and GIA-inducedJ2

Inference of globally averaged eustatic sea level (ESL) rise since the Last Glacial Maximum (LGM) highly depends on the interpretation of relative sea level (RSL) observations at Barbados and Bonaparte Gulf, Australia, which are sensitive to the viscosity structure of Earth's mantle. Here we examine the RSL changes at the LGM for Barbados and Bonaparte Gulf (RSL Bar L and RSL Bon L ), differential RSL for both sites ( RSL Bar,Bon L ) and rate of change of degree-two harmonics of Earth's geopotential due to glacial isostatic adjustment (GIA) process (GIA-induced J 2 ) to infer the ESL component and viscosity structure of Earth's mantle. Differential RSL, RSL Bar,Bon L and GIA-induced J 2 are dominantly sensitive to the lower-mantle viscosity, and nearly insensitive to the upper-mantle rheological structure and GIA ice models with an ESL component of about (120-130) m. The comparison between the predicted and observationally derived RSL Bar,Bon L indicates the lower-mantle viscosity higher than 2 10 22 Pa s, and the observationally derived GIA-induced J 2 of -(6.0-6.5) 10 -11 yr -1 indicates two permissible solutions for the lower mantle, 10 22 and (5-10) 10 22 Pa s. That is, the effective lower-mantle viscosity inferred from these two observational constraints is (5-10) 10 22 Pa s. The LGM RSL changes at both sites, RSL Bar L and RSL Bon L , are also sensitive to the ESL component and upper-mantle viscosity as well as the lower-mantle viscosity. The permissible upper-mantle viscosity increases with decreasing ESL component due to the sensitivity of the LGM sea level at Bonaparte Gulf (RSL Bon L ) to the upper-mantle viscosity, and inferred upper-mantle viscosity for adopted lithospheric thicknesses of 65 and 100 km is (1-3) 10 20 Pa s for ESL130 m and (4-10) 10 20 Pa s for ESL125 m. The former solution of (1-3) 10 20 Pa s is consistent with the inferences from the postglacial differential RSL changes in the Australian region and also inversion study of far-field sea-level data. The inference of the viscosity structure based on these four observational constraints is, however, relatively insensitive to the viscosity structure of D layer.