Ocean carbon uptake and storage influenced by wind bias in global climate models
The paper shows that Earth System Model (ESM) simulations with the typical surface wind bias may over-estimate ocean carbon uptake, and thereby under-estimate atmospheric carbon dioxide concentrations in the twenty-first century, relative to unbiased simulations. In short, the ESM simulations being used in support of the IPCC AR5 may not fully reflect the impacts of climate change.
The coupled-climate models used for the IPCC forth assessment report (AR4), did not contain interactive carbon cycles. In these models, the atmospheric CO2
concentration was prescribed. The new generation of models, so called Earth System Models or coupled carbon-climate models, do contain interactive carbon cycles, which allows for carbon to be exchanged between the atmosphere, ocean and land. In this set of models, instead of prescribing an atmospheric CO2
concentration, it is possible to specify only EMISSIONS of CO2
, and to allow the model to figure out where this carbon ends up. We know from observations that only about half of anthropogenic CO2
emmissions remain in the atmosphere, while the remaining 50% of carbon that humans emit is taken up by the land and the ocean. The rate of future uptake of anthropogenic carbon by the ocean and land is a critical question in climate science, because it ultimately determines the increase in atmospheric CO2
concentration and therefore the resulting climate change.
Thus, the new generation of models that allow carbon to flow through the earth system are more realistic (and complicated). Ironically, these better models may lead to greater uncertainties in projections of future climate change. If two Earth System Models differ in their rate of ocean carbon uptake, they will (all other things being equal) end up with different atmospheric CO2
concentrations, and therefore different climates, even though the rate of CO2
emissions was the same for each model. Thus, the rate of ocean carbon uptake can represent an uncertainty in climate change projections. Our paper has identified a process that causes ESMs to differ in their rate of ocean carbon uptake. In essence, errors (biases) in the model simulation of winds over the Southern Ocean has an effect on the ocean circulation and the rate of ocean carbon uptake in the models. Southern Hemisphere wind-stress biases are ubiquitous amongst the AR4 generation of coupled-climate models, and also exist in all the available AR5 models (the new generation). Typically the westerly winds over the Southern Ocean are too strong, and too equatorward in the models, relative to observations. Our work has quantified the uncertainty in the rate of ocean carbon uptake resulting from these wind-stress biases. The wind-biases in the AR4 models result in a 10% uncertainty in the rate of ocean carbon uptake.
In global climate model pre-industrial control simulations the Southern Hemisphere westerly winds show a systematic bias in position and strength relative to estimates of their actual position and strength. These wind-stress biases impact the transport of the Antarctic Circumpolar Current, the nature of Southern Ocean water mass formation, and could affect the rate of meridional overturning. However, it is not known what influence the wind-stress biases have on oceanic carbon uptake and storage. Here we demonstrate using a coupled carbon-climate model that the wind-stress biases reduce equilibrium ocean carbon storage, redistribute carbon within the ocean and increase oceanic carbon uptake during climate change simulations. The wind-stress biases act directly by influencing Ekman pumping dynamics in the Southern Ocean, but also appear to have an indirect effect on the overturning circulation and carbon distribution through the Agulhas leakage and Indo-Atlantic salt flux. These results indicate that carbon-climate model simulations with the typical pre-industrial wind-stress bias will over-estimate ocean carbon sequestration, and thereby under-estimate atmospheric CO2
concentrations in the twenty first century, relative to unbiased simulations. The new generation of coupled carbon-climate models may be subject to these wind-biases, which could alter their carbon-climate response, though it is worth noting that the uncertainty arising from wind-biases that we demonstrate here is one of several uncertainties that affect modelled ocean carbon uptake.
See our paper appearing soon in Nature Climate Change (publications