Time-varying changes and uncertainties in the CMIP6 ocean carbon sink fromglobal to local scale

As a major sink for anthropogenic carbon, the oceans slow the increase in carbon dioxide in the atmosphere and regulate climate change. Future changes in the ocean carbon sink, and its uncertainty at a global and regional scale, are key to understanding the future evolution of the climate. Here we report on the changes and uncertainties in the historical and future ocean carbon sink using output from the Coupled Model Intercomparison Project Phase 6 (CMIP6) multi-model ensemble and compare to an observation-based product. We show that future changes in the ocean carbon sink are concentrated in highly active regions - 70 % of the total sink occurs in less than 40 % of the global ocean. High pattern correlations between the historical uptake and projected future changes in the carbon sink indicate that future uptake will largely continue to occur in historically important regions. We conduct a detailed breakdown of the sources of uncertainty in the future carbon sink by region. Consistent with CMIP5 models, scenario uncertainty dominates at the global scale, followed by model uncertainty and then internal variability. We demonstrate how the importance of internal variability increases moving to smaller spatial scales and go on to show how the breakdown between scenario, model, and internal variability changes between different ocean regions, governed by different processes. Using the CanESM5 large ensemble we show that internal variability changes with time based on the scenario, breaking the widely employed assumption of stationarity. As with the mean sink, we show that uncertainty in the future ocean carbon sink is also concentrated in the known regions of historical uptake. Patterns in the signal-to-noise ratio have implications for observational detectability and time of emergence, which we show to vary both in space and with scenario. We show that the largest variations in emergence time across scenarios occur in regions where the ocean sink is less sensitive to forcing - outside of the highly active regions. In agreement with CMIP5 studies, our results suggest that for a better chance of early detection of changes in the ocean carbon sink and to efficiently reduce uncertainty in future carbon uptake, highly active regions, including the northwestern Atlantic and the Southern Ocean, should receive additional focus for modeling and observational efforts.