Climate drift of AMOC, North Atlantic salinity and arctic sea ice in CFSv2 decadal predictions

Abstract

There are potential advantages to extending operational seasonal forecast models to predict decadal variability but major efforts are required to assess the model fidelity for this task. In this study, we examine the North Atlantic climate simulated by the NCEP Climate Forecast System, version 2 (CFSv2), using a set of ensemble decadal hindcasts and several 30-year simulations initialized from realistic ocean–atmosphere states. It is found that a substantial climate drift occurs in the first few years of the CFSv2 hindcasts, which represents a major systematic bias and may seriously affect the model’s fidelity for decadal prediction. In particular, it is noted that a major reduction of the upper ocean salinity in the northern North Atlantic weakens the Atlantic meridional overturning circulation (AMOC) significantly. This freshening is likely caused by the excessive freshwater transport from the Arctic Ocean and weakened subtropical water transport by the North Atlantic Current. A potential source of the excessive freshwater is the quick melting of sea ice, which also causes unrealistically thin ice cover in the Arctic Ocean. Our sensitivity experiments with adjusted sea ice albedo parameters produce a sustainable ice cover with realistic thickness distribution. It also leads to a moderate increase of the AMOC strength. This study suggests that a realistic freshwater balance, including a proper sea ice feedback, is crucial for simulating the North Atlantic climate and its variability.

Appendices

Appendix 1: Oceanic freshwater transport terms

Integrating the freshwater equation Eq. (1) over the water volume bordered by constant latitude circles at the south (φ _s ) and north (φ _n ), longitudes in the west (λ_w) and east (λ_e), as well as from depth h to the sea surface, the freshwater volume transports into the box from the western (U _W ), eastern (U _E ), southern (V _S ) and northern (V _N ) boundaries, as well as through the bottom surface (W _B ) can be derived as the following mathematical expressions:

$$U_{W} = \mathop \smallint \limits_{ - h}^{0} \mathop \smallint \limits_{{\varphi_{s} }}^{{\varphi_{n} }} u\left( {\lambda_{w} ,\varphi ,z,t} \right)\left( {1 - \frac{{S\left( {\lambda_{w} ,\varphi ,z,t} \right)}}{{S_{o} }}} \right)ad\varphi dz$$

(A1)

$$U_{E} = - \mathop \smallint \limits_{ - h}^{0} \mathop \smallint \limits_{{\varphi_{s} }}^{{\varphi_{n} }} u\left( {\lambda_{e} ,\varphi ,z,t} \right)\left( {1 - \frac{{S\left( {\lambda_{e} ,\varphi ,z,t} \right)}}{{S_{o} }}} \right)ad\varphi dz$$

(A2)

$$V_{S} = \mathop \smallint \limits_{ - h}^{0} \mathop \smallint \limits_{{\lambda_{w} }}^{{\lambda_{e} }} v\left( {\lambda ,\varphi_{s} ,z,t} \right)\left( {1 - \frac{{S\left( {\lambda ,\phi_{s} ,z,t} \right)}}{{S_{o} }}} \right)a\cos \varphi d\lambda dz$$

(A3)

$$V_{N} = - \mathop \smallint \limits_{ - h}^{0} \mathop \smallint \limits_{{\lambda_{w} }}^{{\lambda_{e} }} v\left( {\lambda ,\varphi_{n} ,z,t} \right)\left( {1 - \frac{{S\left( {\lambda ,\phi_{n} ,z,t} \right)}}{{S_{o} }}} \right)a\cos \varphi d\lambda dz$$

(A4)

$$W_{B} = {\mathop{{\int\!\!\!\!\!\int}\mkern-21mu \bigcirc} }w\left( {\lambda ,\varphi , - h,t} \right)\left( {1 - \frac{{S\left( {\lambda ,\varphi , - h,t} \right)}}{{S_{o} }}} \right)a^{2} \cos \varphi d\lambda d\varphi$$

(A5)

where u, v and w are the zonal, meridional and vertical components of the ocean current, respectively. φ and λ are latitude and longitude and a is the earth radius. The sign convention for these terms is that volume transport into the box is positive.

Appendix 2: Antarctic sea ice

For completeness, the Antarctic sea ice simulated by the CFSv2 is briefly described. Figure 18 shows the 20-year mean sea ice concentration in the Southern Ocean from May to October for the OI-ICE analysis (Fig. 18a) and the last 20-year means from the two CFSv2 simulations: ORG (Fig. 18b) and CORR (Fig. 18c). In comparison with OI-ICE (Fig. 18a), the extent of ice cover (including open water within the ice pack) seems slightly larger in the model. However, the extent of ice-cover with fraction >0.9 is smaller in both runs (Fig. 18b, c). High ice concentration coverage in the model is mostly in the Weddell and Ross Seas while it surrounds the whole Antarctic in observations. On the other hand, there is a clear improvement from ORG (Fig. 18b) to CORR (Fig. 18c).

Fig. 18

The climatological Antarctic sea ice concentration for May–October from a OI-ICE for 1991–2010 and from the last 20-year simulations of b ORG and c CORR. The contour interval is 0.1. The shading scale is shown below these panels. d shows the time series of total areas of sea ice cover in the Southern Hemisphere for OI-ICE (green), ORG (red) and CORR (black). The unit is 10⁶ km²

ORG (red curve, Fig. 18d) underestimates the total sea ice area (excluding the open water within the ice pack) in both winter and summer seasons, in comparison to OI-ICE (green curve, Fig. 18d). The excessive summer melting in ORG makes the Antarctic Ocean almost ice-free during summer within the first 5 years of simulation (red curve, Fig. 18d). The sea-ice area in CORR (black curve, Fig. 18d) is comparable to the observations in winter but there is little improvement during summer.

The distributions of the mean sea ice thickness also demonstrate large discrepancies between CFSR and the model simulations (Fig. 19). In the former, the thickness of ice covers surrounding the Antarctica is generally 1.6 m and thicker (Fig. 19a). The ORG ice thickness (Fig. 19b) is much thinner throughout the basin, about 0.2–0.4 m around Antarctica except for the Weddell and Ross Seas where it is close to 1 m (Fig. 19b). The CORR thickness (Fig. 19c) improves slightly from that of ORG. However, it is still much lower than that of CFSR. The model runs also seriously underestimate the total ice volumes after the first year of simulation (Fig. 19d). The CFSR ice volume (purple curve, Fig. 2d) is maintained between 20,000 and 30,000 km³ during winters from 1981 to 2008. Within 5 years of the simulations, the ice volume is reduced to around 6,000–9,000 km³ in ORG (red curve, Fig. 19d) and 9,000–12,000 km³ in CORR (black curve, Fig. 19d). During summer, the CFSR volume is about 3,000 km³ while the model values are close to 0 for both runs.

Fig. 19

The climatological Antarctic sea ice thickness for May–October from a CFSR for 1990–2009 and the last 20-year simulations of b ORG and c CORR. Note that the contour intervals are uneven and the unit is m. The shading scale is shown below these panels. d shows the time series of total volume of sea ice in the Southern Hemisphere for CFSR (purple), ORG (red) and CORR (black). The unit is 10³ km³

The ice concentration of the ICE simulation (Fig. 20a) does not show a significant change from that of CORR (Fig. 20c) but the ice thickness is increased, with the ice surrounding Antarctica over 0.4 m thick (Fig. 20b and d). Both the total area and volume are also increased (Fig. 20e, f). Overall, the ice properties in the Southern Ocean are less sensitive to the change in the albedo parameters than in the Arctic Ocean. This is possibly because the latter is more land-locked, so that the oceanic circulation is less effective than the local physics. The larger warm bias in the Southern Ocean may also be a limiting factor. Further improvements of the Antarctic sea ice simulation may rely on a better treatment at the ice-ocean interface.

Fig. 20

The 10-year climatological Antarctic sea ice a concentration and b thickness (unit: m) for May–October from the ICE simulation. The corresponding fields for the same period from CORR are shown in (c) and (d). The contour interval is 0.1 for concentration and is unevenly distributed for thickness. The shading scale for concentration (thickness) is shown on the left (right). The time series of total area (e) and volume (f) of sea ice in the Southern Hemisphere are shown for OI-ICE/CFSR (green), CORR (red), ICE (blue) and sensitivity runs (black, dashed curves). The unit of area is 10⁶ km². The unit of volume is 10³ km³