A modeling framework to understand transient ocean climate change in large coupled ensembles

. The ocean responds to climate change through modifications of heat, freshwater and momentum fluxes at its boundaries. The role of these contributors in changing the thermohaline structure of the ocean and its circulation has been partly addressed by modeling studies using idealized CO 2 forcings. The question of timescales for these individual contributions during transient climate change is however lacking. Here, we propose a novel modeling framework to isolate these contributions during the entire historical period and projections of coupled climate models. We present the framework in the context 5 of the IPSL-CM6A-LR model and its ocean component NEMO3.6. We start by reproducing a coupled pre-industrial control simulation with an ocean-only configuration, forced by fixed fluxes at its interface diagnosed from the coupled model. We then add a perturbation to each flux component, extracted from the historical+ssp ensemble of simulations of IPSL-CM6A-LR. With this configuration, we successfully replicate the ocean’s response to transient climate change in the coupled model during 1850-2100. This full response is then decomposed in sensitivity experiments in which the perturbations are applied individu- 10 ally to the heat, freshwater and momentum fluxes. Passive tracers of temperature and salinity are implemented to discriminate the addition of heat and freshwater flux anomalies in the ocean from the redistribution of pre-industrial heat and salt content in response to ocean circulation changes. This framework brings new opportunities to precisely explore the mechanisms driving transient within is without a doubt better at reproducing the temporal evolution of the piControl, which ensures minimal drift from our reference simulation. Furthermore, one of the goal of these experiments is to investigate the evolution of the added and redistributed heat components by implementing a passive anomaly tracer forced with identical surface flux perturbations as the prognostic temperature (see section 5). Modifying the prognostic temperature internally means creating discrepancies in the relationship between temperature change and the passive tracer. We thus chose to run all the sensitivity experiments without 625 any treatment of the temperature below freezing other than in the equation of state and Brunt-Vaisala frequency. The surface Arctic ocean is thus not a region of choice to analyse these experiments and should be considered with care. These technical choices are made in response to our scientific constrains; however, other scientific interests might have led to different choices. Acknowledgements. Y.S. wishes to thank Saenko and for their help with the experimental design and for 635 answering questions about the fixed-flux forcing of the ocean model. We also thank de Lavergne and Jan Zika for helpful scientific ideas and discussions. This work was granted access to the HPC resources of IDRIS under the allocation 2020-A009017403 and 2020-A0080107451 made by GENCI. This work also benefited from the ESPRI computing and data centre (https://mesocentre.ipsl.fr) which is supported by CNRS, Sorbonne University, Ecole Polytechnique and CNES and through national and international grants. We acknowledge funding from the ARCHANGE project of the “Make our planet great again” program (ANR-18-MPGA-0001, France) as well as from the 640 European Union’s Horizon 2020 research and innovation program under grant agreement N°821001.