Abstract
Paleoceanographic proxies indicate that the ocean state during the Last Glacial Maximum (LGM) differed from the modern ocean state. Depth profiles of ocean sediment pore fluid δ18O and [Cl−] have been used to reconstruct the δ18O and salinity at the LGM. Here, it is investigated whether pore fluid profiles can constrain ocean δ18O and salinity at other times and, simultaneously, their ability to constrain the LGM δ18O and salinity. An inverse framework is developed that relies on Bayesian parameter estimation, thus allowing formal separation of prior assumptions from the information in observations. Synthetic problems are used to explore the information about past ocean tracers that can be recovered from pore fluid profiles. It is concluded that prior knowledge of deep ocean mixing time scales is essential to an accurate inverse estimate of LGM ocean salinity and δ18O from modern pore fluid profiles. The most recent 10 000 years of ocean salinity and δ18O and the error in their estimates are better constrained by the pore fluid profiles than are the LGM values. The inverse estimate of salinity and δ18O is strongly correlated with the estimate of diffusivity of oxygen isotopes and [Cl−] in sediment pore fluids. Uncertainty on the diffusivity of oxygen isotopes and chloride in sediments is reduced through inversion of the pore fluid profiles, but simultaneous estimation of δ18O or salinity and diffusivity increases the total uncertainty. The error reported in previous work may underestimate the true uncertainty of LGM deep ocean salinity and δ18O.
Current affiliation: Department of Earth and Planetary Sciences, Harvard University, Cambridge, Massachusetts.
