Pore fluids and the LGM ocean salinity—Reconsidered

• Deep-sea core pore fluid chlorinity/salinity data present a terminal constraint advection-diffusion problem.
• Inferences can be made using conventional linear control theory methods.
• A long list of assumptions involves the physics, initial and boundary conditions, and the noise level and structure.
• Solutions that inferred a very high LGM abyssal salinity in the Southern Ocean lie within the uncertainty estimates.
• The chlorinity/salinity data alone, however, do not require abyssal salinities above the LGM global maximum.

Pore fluid chlorinity/salinity data from deep-sea cores related to the salinity maximum of the last glacial maximum (LGM) are analyzed using estimation methods deriving from linear control theory. With conventional diffusion coefficient values and no vertical advection, results show a very strong dependence upon initial conditions at −100 ky. Earlier inferences that the abyssal Southern Ocean was strongly salt-stratified in the LGM with a relatively fresh North Atlantic Ocean are found to be consistent within uncertainties of the salinity determination, which remain of order ±1 g/kg. However, an LGM Southern Ocean abyss with an important relative excess of salt is an assumption, one not required by existing core data. None of the present results show statistically significant abyssal salinity values above the global average, and results remain consistent, apart from a general increase owing to diminished sea level, with a more conventional salinity distribution having deep values lower than the global mean. The Southern Ocean core does show a higher salinity than the North Atlantic one on the Bermuda Rise at different water depths. Although much more sophisticated models of the pore-fluid salinity can be used, they will only increase the resulting uncertainties, unless considerably more data can be obtained. Results are consistent with complex regional variations in abyssal salinity during deglaciation, but none are statistically significant.