Mechanisms of ocean heat uptake, transport and storage in ACCESS-OM2

The global ocean plays a dominant role in the net increase of the Earth's energy inventory observed since the 1970s, storing more than 90% of the additional anthropogenic heat. While subsurface ocean heat gain slows down the warming of the surface atmosphere, it also causes global mean sea level rise through thermal expansion of the ocean's volume. About one third to half of the global mean sea level rise projected by climate model simulations towards the end of this century is due to thermal expansion. However, dynamic sea-level projections from climate models have a large spread in global mean amplitude and regional distribution. The ocean's response ‚ÄövÑvÆ that is how efficiently the ocean uptakes the excess heat from the atmosphere and transports it to deep layers ‚ÄövÑvÆ is responsible for roughly half of the uncertainties in dynamic sea-level projections. A quantitative understanding of the key physical processes involved in ocean heat uptake and redistribution in models used to project dynamic sea level is now feasible, timely, and needed to reduce the currently large range of uncertainties. In this thesis, key mechanisms involved in heat uptake, transport and storage were examined through the use of tracer budgets applied to climatological and idealised climate change simulations from the Australian Community Climate and Earth System Simulator Ocean Model (ACCESS-OM2). The tracer budgets (for both heat and salt) were estimated using model diagnostics that were directly resolved by the 1! ACCESS-OM2 horizontal resolution and indirectly represented via parameterisations. A quasi-steady state was achieved by forcing ACCESS-OM2 with the JRA55-do surface flux repeat-year forcing product for 1000 years. The climate change experiments were then run for the successive 80-year period, forced by doubled atmospheric CO2 scenario perturbations, from the Flux-Anomaly-Forced Model Intercomparison Project (FAFMIP) protocol, based on the ensemble mean of 13 Coupled Model Intercomparison Project Phase 5 (CMIP5) simulations. In addition, the FAFMIP protocol has a large set of experiments, with single and combined forcing perturbations (momentum, freshwater, heat, and all forcings) and a design that allows the partition of ocean heat content changes into added heat‚ÄövÑvp(to reveal where uptake of heat at the surface is stored in the ocean due to passive component only) and redistributed heat‚ÄövÑvp (associated with changes in ocean circulation, diffusion, eddies, mixing, etc.) For the quasi-steady experiment, the last 20 years of the 1000-year run were averaged and analysed to investigate the net effect of key physical processes in the global vertical heat and salt transport balance, under the perspective of the recently proposed super-residual transport (SRT) framework. In this framework, the ACCESS-OM2 results show that vertical tracer transport is dominated by downward fluxes associated with the large-scale ocean circulation and upward fluxes induced by mesoscale eddies, with two distinct physical regimes. In the upper-ocean, where high-latitude watermasses are formed by mixed-layer processes, through cooling or salinification, the SRT balances those processes by transporting heat and salt downward. In contrast, in the ocean interior, the SRT opposes dianeutral diffusion via upward fluxes of heat and salt, with about 60% of the vertical heat transport occurring in the Southern Ocean. Overall, the SRT is largely responsible for removing newly-formed watermasses from the mixed-layer into the ocean interior, where they are eroded by dianeutral diffusion. Unlike the classical advective-diffusive balance, dianeutral diffusion is bottom-intensified above rough bottom topography, allowing an overturning cell to develop in alignment with recent theories and observations. Thus, this SRT framework can be considered a contemporary interpretation of the classical advective-diffusive balance in the ocean interior. The SRT also proved to be an essential concept to progress intercomparisons with the aim to understand and reduce uncertainties in climate and sea-level projections, as it allows model comparisons independent of grid resolution (coarse, eddy-permitting or resolving), and to further theoretical development of simple climate models from which simulations are used for policy and decision-making. For the climate-change experiments, the ACCESS-OM2 simulations in response to the doubled CO2 scenario perturbations were de-drifted by a parallel control run and averaged over years 61-80 (centred at year 70). The key mechanisms driving ocean changes were analysed for the subpolar Southern Ocean (sSO, south of 60!S) and for large-scale basin changes. Focusing on these two regions offers a comprehensive view of the global heat uptake; the response in the sSO corresponds to global changes below 2000 m, while changes in other regions largely affect the upper-2000 m. Details are presented below. The sSO around Antarctica is the origin of the densest water mass of the global ocean, Antarctic BottomWater (AABW). In the CMIP5 coupled climate model simulations this region shows a relatively high spread in dynamic sea level projections. AABW is linked with the lower cell of the Meridional Overturning Circulation and occupies about 60% of the ocean's volume. Despite its relevance, climate models are well-known to have difficulties in representing the shelf mixing processes that form AABW in the sSO. Similar to other coarse resolution models, AABW originates at both shelf and open ocean areas in ACCESS-OM2. How the mechanisms of AABW formation ‚ÄövÑvÆ particularly in the Ross and Weddell Gyres ‚ÄövÑvÆ respond to changes in surface forcing was investigated by comparing FAFMIP's single (momentum, freshwater, and heat) and all forcing perturbations. Results show that changes in wind stress (i.e. poleward shift and intensification of the Southern OceanWesterlies) dominate the response of AABW formation, relative to the surface heat and freshwater forced changes. Changes in the wind stress forcing enhance bottom water formation and accelerate the lower branch of the meridional overturning circulation (MOC). This response occurs due to an increase of ocean deep convection as open-ocean polynyas become more widespread. The effect of surface warming and freshening only partially compensates the changes due to wind stress, reducing AABW formation. The overall message is, in agreement with previous studies, that processes that enhance convection (i.e. winds) will cool the deep ocean, while processes that stratify the ocean (i.e. input of freshwater or positive surface heat anomalies) inhibit convection and lead to a build-up of heat below the surface. The dominance of wind changes is a meaningful result as perturbations are derived from CMIP5 models. Importantly, the response to surface perturbations is time-dependent, where the non-linear evolution of the polynyas has a dominant role. These results help to understand the sSO response in models to climate change scenarios, where somewhat unrealistic open-ocean polynyas control ocean heat uptake. The sSO response in ACCESS-OM2 and in a past study based on an ensemble of coupled CMIP models for individual forcings are similar, suggesting that future intercomparisons, including high-resolution models which better represent shelf scale processes, can reveal differences on how AABW formation and the lower cell of the Meridional Overturning Circulation respond to surface perturbations under the FAFMIP protocol and SRT framework. For the large-scale ocean basin changes, ocean heat storage due to local addition of heat (added‚ÄövÑvp) and due to changes in heat transport (redistributed‚ÄövÑvp) were quantified in ocean-only 2xCO2 simulations. While added heat storage dominates globally, redistribution makes important regional contributions, especially in the tropics. Heat redistribution is dominated by circulation changes, summarised by the superresidual transport, with only minor effects from changes in vertical mixing. While previous studies emphasised the contribution of redistribution feedback at highlatitudes, this study shows that redistribution of heat also accounts for 65% of heat storage at low latitudes and 25% in the mid-latitude (35‚ÄövÑv¨50!S) Southern Ocean. Tropical warming results from the interplay between increased stratification and equatorward heat transport by the subtropical gyres, which redistributes heat from the subtropics to lower latitudes. The Atlantic pattern is remarkably distinct from other basins, resulting in larger basin-average heat storage. Added heat storage is evenly distributed throughout the mid-latitude Southern Ocean and dominates the total storage. However, redistribution stores heat north of the Antarctic Circumpolar Current in the Atlantic and Indian sectors, having an important contribution to the peak of heat storage at 45!S. Southern Ocean redistribution results from intensified heat convergence in the subtropical front and reduced stratification in response to surface heat, freshwater, and momentum flux perturbations. These results highlight that the distribution of ocean heat storage reflects both passive uptake of heat and active redistribution of heat by changes in ocean circulation processes. The redistributed heat transport must therefore be better understood for accurate projection of changes in ocean heat uptake efficiency, ocean heat storage and thermosteric sea level. The main findings contained in this thesis are important contributions for the understanding of ocean heat uptake, transport and storage in general circulation models. The super-residual framework has been presented in Chapter 3 [Dias et al., 2020] and further applied to investigate the ocean response to idealised climate change scenarios in Chapters 4 and 5, where it is found to be an important tool for elucidating the role of key physical mechanisms. Under the climate change scenari...