To a first approximation, much of what happens in the Earth system depends on the dynamics and interactions of the two key geophysical fluids that make it up: the atmosphere and the ocean. As anyone familiar with the UK’s weather will attest, the atmosphere is flighty and dynamic, with perturbations in temperature, humidity or chemical composition quickly created and dissipated on timescales of days to weeks or months. The ocean, by contrast, is a much more sedate fluid. While its surface readily exchanges heat and freshwater with the atmosphere, its interior has a vast thermohaline circulation that sluggishly transports water and chemical constituents into the deep and around its basins. Depending upon where it does so, water leaving contact with the atmosphere can take decades to centuries or even millennia before it again comes into contact with the atmosphere.
Consequently, with an average ventilation timescale of more than 2000 years, the ocean component of the Earth system has a long memory – one that can “remember” environmental perturbations far longer than other components such as the atmosphere and land surface. In addition, the ocean serves as the largest reservoir of carbon in the Earth system. Relative to the atmosphere – where the concentration of carbon dioxide (CO2) has been of long-standing interest – the ocean stores more than 50 times as much. A consequence of this, both for Earth system models and the real Earth system, is that even small imbalances in the air-sea exchange of CO2 can lead to large changes in atmospheric CO2. This is a particular concern for fully coupled ESMs where both the climate and CO2 are free to evolve. Since both our understanding of marine biogeochemistry is far from complete and observations of carbon in the ocean are relatively under-sampled, there is a risk of substantial – and unwelcome – drift in ESM-simulated atmospheric CO2 towards unrealistic values.
As part of the UK’s preparations for CMIP6 – and to specifically address the ocean’s carbon cycle – spin-up simulations of the low resolution version of UKESM1 (UKESM-lr) have been planned and now begun. To this end, the marine carbon cycle of UKESM1-lr has been initialised from estimated pre-industrial conditions and has been spinning up in ocean-only mode using atmospheric forcing (including winds, radiation, heat and freshwater fluxes) derived from an atmosphere-land simulation that employed sea surface temperatures (SST), sea-ice (SIC) and atmospheric gases, including CO2, representative of pre-industrial conditions. To date, almost 700 years have been completed, comprised of more than 20 cycles of the 30-year atmospheric dataset.
The accompanying graphs show representative time-series of ocean circulation, biological activity in the ocean, and – most pertinently – the exchange of CO2 between the ocean and the atmosphere. While the ocean’s circulation is broadly stable, some deficiencies have been detected that will hopefully be remedied as the spin-up progresses. Similarly, ocean productivity is found to euilibriate within the spin-up duration so far, though somewhat at the lower end of that observed. Mean while, the air-sea CO2 flux is behaving as we expect: a long-term transient steadily approaching, but not reaching, zero. However, the aim of spin-up is to decrease drift, rather than eliminating it completely.
The next stage of UKESM1 spin-up will involve the transition of the ocean’s simulated state – physical and biogeochemical – to a fully coupled simulation. Because of the much-increased cost of the coupled simulation, after a multi-decadal period of spin-up in this mode, the ocean’s end state and the atmosphere’s output will be used in a further ocean-only phase. Depending on progress, it is planned that iteration between ocean-only and coupled phases will continue, before a final – and extended – coupled phase will be used to provide an initial condition for CMIP6. It is expected that this will be produced by late-2016 or early-2017.
Figures here are provided by the Biogeochemistry Evaluation Suite, developed by Lee de Mora (Plymouth Marine Laboratory). This is a Python-based tool operating on the JASMIN facility and producing web-based output accessible to the UKESM1 community.