Fig. 1
(a) Model domain and annual mean surface circulation of the NEMO-Nordic OGCM. The red box indicates the region of temperature local anomaly perturbation in experiment (Section 4.1). The red cross indicates the station where time series have been sampled (0.5°W; 57.9°N). (b) Model domain of the regional atmospheric model RCA4. Bathymetry has been taken from the NEMO-Nordic ocean model. Contour lines are −50, −30, 200, 500 and 1000 m.
Fig. 2
Experimental set-up and exchange variables between the ocean and atmosphere model components. ICA=interactively coupled atmosphere, ICO=interactively coupled ocean, PCA=passively coupled atmosphere, PCO=passively coupled ocean.
Table 1
Model simulations
ICOOceanNEMO1961–2009 ICAAtmosphereRCA41961–2009Passively coupled PCOOceanNEMO1961–2009 PCAAtmosphereRCA41961–2009Perturbation experimentsPeriod Local anomaly12°C<50 m1.1–31.12 1990a Basinwide anomaly+5°K<50 m1.1–31.12 1990a Lateral anomaly+5°K1990–1994a
Fig. 3
Time series from the local anomaly experiment: (a) sea surface temperature, (b) temperature at 72.5 m, (c) difference in downward heatflux (interactive minus passively coupled run). Positive values indicate higher heat loss/lower heat uptake in the uncoupled run. (d) Potential energy anomaly. Red lines indicate interactive coupling. Black lines indicate passive coupling. Dotted lines indicate the control runs without pertubation.
Fig. 4
Heat content anomaly (difference of perturbation experiment minus unperturbed control run) calculated for the North Sea and Baltic Sea for: (a) the basinwide anomaly experiment and (b) the lateral boundary anomaly experiment. Red lines indicate interactive coupling. Black lines indicate passive coupling. Green lines in (a) indicate the difference between the red line and the black line.
Fig. 5
(a) Difference in effective wind stress between the basinwide anomaly experiment and the unperturbed control run. Left: interactively coupled model. Right: passively coupled model. (b) Difference in effective wind stress between lateral boundary anomaly experiment and the unperturbed control integration. Top: Average over the last year of model integration (1994). Bottom: Average over June, July and August of the last year of model integration.
Fig. 6
Comparison of simulated SST in the Baltic Sea with interactive coupling (left), passive coupling (middle) and the ERA40 SST (right) used as lower boundary condition for the atmosphere only simulation. Displayed is the difference to the observed climatology of the BSH (model minus observation).
Fig. 7
Difference between the ICO minus PCO simulation in winter mixed layer thickness (a) and potential energy anomaly (b; Simpson and Bowers, 1981). Mixed layer thickness was calculated according to the 0.03 kg/m3 criterion (de Boyer Montegut et al., 2004).
Fig. 8
Comparison of simulated SST in the North Sea with interactive coupling (left), passive coupling and the ERA40 SST used as lower boundary condition for the atmosphere only simulation. Displayed is the difference to the observed climatology of the BSH (model minus observation).
Fig. 9
Wind lead correlation analysis for a station in the Baltic Sea (17.9°W; 19.9°N) and a station in the stratified northern North Sea (3.2°W; 58.5°N). The lead correlation has been carried out on 6-hourly 10 m wind and SST data during June and July 1990. Red line indicates the interactively coupled run IC. Black line indicates the passively coupled run PC.
Fig. 10
(a) Correlation between times series of monthly mean January heat contents with time series of monthly mean heat contents of other months. Red: interactively coupled model. Black: passively coupled mode. Dashed lines indicate the heat content of the well-mixed southern North Sea only (i.e. south of 54°N). (b) Standard deviation of daily mean SST time series (1990–1999). Note the seasonal cycle has been subtracted before analysis.
Fig. 11
(a) Annual mean SST averaged over the North Sea. Red line=interactive air–sea coupling (IC). Black line=passive air–sea coupling (PC). (b) Difference of annual mean SST between the PC minus the IC simulation. (c) Difference in volume integrated heat content between the PC minus the IC simulation.
Fig. 12
Difference between interactively coupled simulation and passively coupled simulation for winter (DJF) and summer (JJA): (a) 2 m air temperature and (b) precipitation. The right panel shows only differences that are significant at the 95% confidence level (tested with a two-sided t-test. The white line indicates the zero contour line).
Fig. 13
(a) Difference between interactively coupled simulation and passively coupled simulation for strong precipitation. Only precipitation above the 90% percentile was considered. (b) Difference above the 95% confidence level. (c) Relative differences expressed as percentage. (d) Absolute values of precipitation for the interactively coupled simulation and (e) distribution throughout the year of precipitation model outputs above the 90% percentile for the interactively coupled simulation.
Fig. 14
(a) Difference between interactively coupled simulation and passively coupled simulation of 10 m wind velocity for winter (DJF) and summer (JJA). The white line indicates the zero contour line. (b) Same as (a) but considering only strong winds (above the 90% percentile). Also shown: distribution of strong winds throughout the year.
Fig. 15
Comparison of passively coupled (PC) modelled, interactively coupled (IC) modelled and ERA40 re-analysis geopotential height. Displayed is the standard deviation of analysed half-daily fields between 1990 and 1999.