There, the observed chlorophyll concentration, as well as the one simulated in CM5_piCtrl, is lower than 0.05 mg/m3 (e.g. Séférian et al., 2012). As
a result, less heat is trapped in the surface layer in these areas in CM5_piCtrl as compared to CM5_piCtrl_noBio, explaining the cold surface anomalies seen on Fig. 4. Coastal upwellings in equatorial regions, on the other hand, are relatively rich in chlorophyll, and one would expect a net surface warming in CM5_piCtrl as compared to CM5_piCtrl_noBio. This is what is found by Lengaigne et al. (2006) and Patara et al. (2012), two independent studies using similar twin experiments with another coupled climate model and the same oceanic component as ours, namely NEMO. Yet, in our case, the warming effect is very weak or absent (Fig. 4). At mid to high latitudes, previous studies (e.g. Lengaigne et al., 2009 and Manizza, 2005) Sorafenib have suggested that bio-physical feedbacks would result in an intensification of the seasonal cycle: in summer, the presence of phytoplankton MLN0128 increases the surface warming, as more heat is trapped at the ocean surface, while in fall and winter, the deepening of the mixed layer acts to bring
the underlying anomalously cold layers to the surface. This is indeed the case in our simulations for the Southern Ocean and the subpolar North Atlantic and North Pacific that are marked by a warming in local summer in CM5_piCtrl (Fig. 4, right panel), and a moderate to strong cooling in winter (Fig. 4, middle panel). Consistently, the seasonal cycle of SST at mid to high latitudes is slightly enhanced in CM5_piCtrl as compared to CM5_piCtrl_noBio (Fig. 5). Note however that the physical parameterization changes Alanine-glyoxylate transaminase described in Table 1 induce much stronger changes to the seasonal cycle amplitude in CM5_piStart compared to CM5_RETRO than to CM5_piCtrl_noBio (Fig. 5).
Such effect can hardly be seen in forced mode (Fig. 3) and might thus be due to air-sea interactions. In annual mean (Fig. 4, left), ice-free areas at northern high latitudes experience a cooling in CM5_piCtrl as compared to CM5_piCtrl_noBio, which again differs from earlier studies, in particular Lengaigne et al. (2009). These authors have argued that warming associated to phytoplankton blooms occurs concomitantly with the ice retreat along the Arctic coastal shelves in spring and this mechanism is then amplified in summer due to a larger reduction of sea-ice thickness and concentration. In our model, such biologically-induced warming occurs indeed in summer but its global effect is largely counteracted by the winter cooling. Fig. 6 shows the adjustment of the model to the biogeochemical component, helping to understand differences with previous model versions: during the first decade (left panel), the anomalous vertical temperature profile is close to what is expected from the one-dimensional adjustment described above, and broadly agrees with results from Lengaigne et al., 2006 and Lengaigne et al.