Project deliverable Open Access
Swingedouw, Didier; Herbaut, Christophe; Garric, Gilles; Devilliers, Marion; Mignot, Juliette; Houssais, Marie-Noelle; Blaizot, Anne-Cecile
The Greenland ice sheet has been melting at an increasing rate for the last few decades. Large rates of melting have also been reported to have occurred in the 1920s, due to a warming of the region during this decade. Recent reconstructions of the ocean circulation in the North Atlantic have suggested that the large-scale overturning circulation might have been reduced over the last century, possibly due to the freshwater released by Greenland ice sheet melting. To evaluate this hypothesis, we have conducted a series of ocean-only and fully-coupled climate simulations where we include or exclude the observed melting of Greenland ice sheet, which is usually neglected in climate simulations of the last century.
Based on a recent estimate of the Greenland ice sheet melting, we have constructed a forcing melting field for the ~100 km resolution IPSL-CM6A-LR climate model (including ocean, atmosphere land and sea ice), and also for a ~2-3 km NEMO regional coupled sea ice-ocean model. The use of two different resolutions for the ocean allows us to evaluate the importance of small-scale oceanic processes for the impact of the Greenland ice sheet melting. We have then integrated these two types of models over two different time periods: 1920-2014 for the climate model, with 10 different members to account for the potential role of intrinsic variability of the climate, and 2004-2017 for the high-resolution ocean-only model, due to its high cost in terms of computing time. We have considered two types of simulations: one where the freshwater release from Greenland ice sheet is computed within the climate model (through a basic freshwater closure) and the other where we apply time-varying observation-based estimates.
The main results highlight several crucial insights. First, the climate model simulations show that the observation-based melting of Greenland ice sheet, even if it has a modest amplitude, does have a significant impact on the state of the North Atlantic, and even induces a slight reduction of the large-scale ocean overturning (by around 0.5 Sv or 2-3%) over the last century. However, such an impact is far lower than the suspected 15% weakening of this circulation by a recent observation-based estimate. It is found in the different members of the ensemble, that intrinsic variability could explain a large part of this observation-based weakening but not the totality. This discrepancy with the observation-based estimate of the weakening may be related to (i) the fact that this estimate of overturning weakening is indirect and therefore subject to a large uncertainty, or, (ii) the fact that the climate models may be missing important processes.
The use of the high-resolution ocean model can help to evaluate the hypothesis (ii). The few years of high-resolution ocean simulation indicate that the spread of fresh water from Greenland towards the center of the Labrador Sea, where oceanic convection occurs, is larger at high than low resolution, due to small-scale processes that increase the lateral exchanges. The larger freshening of the Labrador in the high-resolution model may then limit deep convection, and weaken the oceanic circulation. Thus, the estimate of ~0.5 Sv of weakening in the low-resolution climate model might be a low estimate of the weakening of the Atlantic overturning circulation, which could be larger when the small-scale processes are included. To conclude, based on the climate model simulations analysed here, the observation-based estimate of the weakening of the ocean circulation is more likely due to natural variability than to anthropogenic forcing, but limitations in climate models still preclude high confidence in this result.
Finally, we have estimated the impact that Greenland ice sheet melting may have on the decadal variability of the sea surface temperature in the North Atlantic. Indeed, the melting is also varying in time, which might have an impact on decadal-scale variation of ocean temperature and therefore the associated potential predictability. We have found that the large warming that was observed in the mid-1990s can be explained to a large extent by the Greenland ice sheet melting. This is a very new result, that will deserve further investigations to have a good understanding of the processes at play, which will be the topic of the last year of the research led within the project.