Insights into biology’s role in future ocean carbon storage from CMIP6 models

Biotic processes in the ocean play a crucial role in driving and mediating natural long-term ocean carbon storage. IPCC assessment exercises find high uncertainty, and therefore low confidence, around the magnitude and sign of change in future ocean carbon storage. This uncertainty is due to a lack of mechanistic understanding of relevant biological processes and/ or a paucity of observational data which limits robust parameterisations in global ocean biogeochemical models. Our study aims to identify and prioritise the processes that have a strong impact on future ocean carbon storage, with tractability from both a modelling and observational perspective. These processes could be the focus of future studies that aim to improve parameterisations in global biogeochemical models used in Earth System Models. We undertook a gap analysis to identify key processes and highlight future research priorities around three areas: net primary production (NPP), interior remineralisation and alkalinity. Here we evaluate CMIP6 model projections to 2100 under the high emissions SSP5-8.5 scenario to determine both the spread and uncertainty in NPP, particulate organic carbon transfer efficiency through the ocean interior and surface salinity-normalised alkalinity. We undertook a model interrogation of which processes are represented, their level of parameterised complexity and the variability in the parameterisation approach. Our analysis shows that CMIP6 models generally agree on the sign of change for transfer efficiency, but display a wide spread for NPP and salinity-normalised alkalinity by the end of the 21st century. Combining our analysis of CMIP6 models and gaps in knowledge allows the potential key processes and uncertainties driving future changes in key biological components of the ocean carbon cycle to be identified. By highlighting the potential gaps that require attention, the representation of biological processes in global ocean biogeochemical models can be improved in future modelling efforts.