Quantifying spatiotemporal variability in Neogene organic carbon burial: a case for ocean model upsampling

3D biogeochemical ocean models such as cGENIE can explicitly model depth-dependent carbon cycle processes, such as remineralisation of organic carbon. This potential advantage of 3D models (in comparison to box ocean models) can, however, be limited by coarse spatial resolutions. In particular, continental shelves may be underresolved in 3D model bathymetric grids. Such grids are created by downsampling (palaeo)-digital elevation models (DEMs).

We develop an algorithm (here termed ‘DEM-based upsampling’) to project 3D ocean model output onto its associated DEM. This resolves depth-dependent quantities and fluxes at the seafloor at degree-scale and better captures shallow seafloor, including continental shelves. This is critical for modelling organic carbon cycling, as continental shelves receive more than half of the global flux of organic carbon to the seafloor. We validate the DEM-based upsampling algorithm using area-weighted errors between a modern-Earth model run and observational data (World Ocean Atlas 2023). Upsampling yields statistically significant reductions in error in modelled temperature, salinity, oxygen concentration, and phosphate concentration across bootstrap confidence intervals and paired non-parametric tests.

We then derive the first spatially-resolved model record of ocean organic carbon burial from 25 Ma – present using the PhanerO3D framework, driving cGENIE with SCION biogeochemistry and HadCM3L atmospheric physics. We obtain organic carbon burial flux by upsampling cGENIE’s organic carbon export flux and applying a simple burial scheme. We find the global burial rate peaks in the early Miocene, then declines over the remaining Neogene. This trend agrees well with geochemical records until the latest Miocene – Pliocene. We find global variability to be largely driven by regional changes; notably declining North Atlantic margin burial over the Miocene, and rising West Pacific burial in the Pliocene.

These results highlight the advantages of DEM-based upsampling as a tool in palaeoclimate modelling: better constraining depth-dependent ocean processes, facilitating deeper investigation of spatiotemporal patterns, and potentially facilitating more spatially precise proxy-model comparison.