Sensitivity of the glacial marine biological pump to particle sinking and dust deposition in MPI-ESM

The marine biological carbon pump substantially contributes to the glacial-interglacial CO2 change. Compared to the late Holocene, proxy data for the Last Glacial Maximum (LGM) generally agree on an increased export production, associated with an enhanced marine biological carbon pump, in the subantarctic region of the Southern Ocean (SO). By contrast, global export production during the LGM is poorly constrained due to the sparseness and uncertainty of proxy data. The efficiency of the biological pump is mainly controlled by phytoplankton growth, ocean circulation and the sinking and remineralisation of organic matter. Previous modelling studies primarily focused on the sensitivity regarding the former two factors. By far, few studies have discussed the impact of marine particle sinking on glacial ocean biogeochemistry.

In this study, we examine the impact of two different sinking schemes for biogenic particles on the LGM ocean biogeochemistry in the Max Planck Institute Earth System Model (MPI-ESM). In the default sinking scheme, sinking velocities of particulate organic matter (POM), biogenic minerals (CaCO3 and opal) and dust are prescribed and kept the same between LGM and pre-industrial (PI) state. Such a scheme is also widely applied in other ocean biogeochemical models. In a new Microstructure, Multiscale, Mechanistic, Marine Aggregates in the Global Ocean (M4AGO) sinking scheme, the size, microstructure, heterogeneous composition, density and porosity of marine aggregates, consisting of POM, CaCO3, opal and dust, are explicitly represented, and the sinking speed is prognostically computed. We discuss the effect of the two particle sinking schemes under two LGM circulation states: “deep LGM AMOC” with a similar NADW/AABW boundary compared to PI, which is produced in many existing models, and “shallow LGM AMOC” with a shallower NADW/AABW boundary, which agrees better with proxy data. Furthermore, we conducted sensitivity studies regarding LGM dust deposition as the latter is subject to considerable uncertainties.

We find that for the deep LGM AMOC, the difference between the impact of the two particle sinking schemes on the ocean biogeochemical tracers is small. On the contrary, for shallow LGM AMOC, the M4AGO scheme yields more remerineralised carbon in the deep ocean and, therefore, better agreement with δ¹³C data, suggesting the quantitative impact of particle sinking schemes strongly depends on the background LGM circulation state. For the default sinking scheme, increased glacial dust deposition increases iron fertilisation and thus leads to a rise in both primary production and export production. For the M4AGO scheme, however, the iron fertilisation effect is surpassed by the ballasting effect that reduces the surface nutrient concentration, and LGM primary production decreases with dust deposition. This preliminary result shows that the new marine aggregate sinking scheme adds further complexities to the marine biological carbon pump response to the climate states. Our further analysis will encompass the other nutrients and dissolved oxygen, as well as the comparison to corresponding proxy data.