Module B - Summary of the first phase of MiKlip

The aim of Module B is to enhance the understanding of decadal variability, to improve existing model components and to incorporate additional climate subsystems that are relevant for decadal climate predictions. This is done by concentrating on different processes like soil-atmosphere interactions, Arctic processes, atmosphere-ocean coupling, impact of natural external forcing, chemical processes and stratospheric processes.

Improvement of existing model components and coupling of additional subsystems

A new parameterisation was developed in project SPARCS accounting for surface fluxes over inhomogeneous sea ice (Lüpkes et al. 2013, Lüpkes and Gryanik 2015). Two different regimes were considered: the marginal sea ice zone with drifting floes and the inner Arctic regions during summer with melt ponds and leads. SPARCS showed that the differences in the morphology of these regimes require in general different parameterisations of the transfer coefficients for the turbulent transport of heat and momentum in the atmospheric boundary layer over sea ice (Fig. 1). As the most detailed parameterisations can only be applied by atmospheric climate models when they are coupled with sophisticated sea ice models, SPARCS developed also a simplified parameterization, which describes the transfer coefficients as a nonlinear function only of sea ice and melt pond concentrations. The new parameterisation, which involves also a new stability correction, has been tested in cooperation with project TORUS in the atmosphere-ocean-sea ice model ECHAM6-FESOM. It was shown that the new parameterisation has a large impact, especially on the momentum fluxes. But also other atmospheric variables (2m air temperature, surface pressure) as well as the sea ice concentration are influenced (Fig. 2). The impact on the ensemble average of the sea ice concentration was relatively small, however, in some regions statistically significant effects were found and there was a large variability between ensemble members.

Module B Summary - Figure 1
Figure 1 (SPARCS): Schematic illustrating the derivation of the parametrisation of the neutral form drag coefficient Cdf for the ice regime (top) with drifting floes and (bottom) for the inner summer Arctic with melt ponds and leads. Arrows indicate dynamic pressure (wind) on edges. hf and hp are ice freeboard, Di and D’w are edge lengths, Sc and S’c are sheltering functions, Ce and C’e are geometry dependent coefficients, z0,w is the roughness length over open water, and A is the sea ice concentration.
Module B Summary - Figure 2
Figure 2 (SPARCS/TORUS): Effect of the new parameterisation of transfer coefficients over sea ice in the Arctic. Differences in (left) sea level pressure (SLP, unit hPa) and (middle) 2 m air temperature (TEMP2, unit K) based on an ensemble of 10 20-year-long ECHAM6-FESOM runs with the new and with the default parameterisation. (right) The same for sea ice concentration in spring, but results of one ensemble member are shown.

The dependence of decadal predictions on the employed ocean component is tested and analysed within the project TORUS by the application of an ocean model alternative to the MPI-OM. Their ocean-sea ice model FESOM has an unstructured grid that allows for increased horizontal resolution in key regions like the Arctic and the tropics (Sidorenko et al. 2015). The long control run under present-day forcing showed that ECHAM6-FESOM performs as well as other climate models and has similar shortcomings in the North Atlantic circulation, leading to a too weak deep convection in the Labrador Sea, with phases of stronger convection related to ice-free conditions and weaker convection related to ice-covered conditions (Fig. 3, blue curve, R1). The decadal climate variability of global mean temperature, atmospheric teleconnection patterns, large-scale oceanic variability patterns and El Niño Southern Oscillation (ENSO) is well reproduced. An increased horizontal resolution in the tropics leads to an even more realistic representation of ENSO. First results from a long simulation under present-day forcing over 250 years with a new version of FESOM with better refinement in the northern North-Atlantic, the Canadian Arctic archipelago, in the Labrador Sea, and over the whole Arctic in general showed an improvement (strengthening) of the deep convection in the Labrador Sea (Fig. 3, red curve, R2).

Module B Summary - Figure 3
Figure 3 (TORUS): (left) Annual maximum of the mixed layer depth [m] averaged over 300 years of the present-day control simulation with reference grid R1, and (right) time series of the annual Labrador Sea mixed layer depth maximum with reference grid (R1, blue curve) and refined grid (R2, red curve).

The memory effect of soil moisture has a strong impact on the hydrological and energy cycles at the land surface on decadal time scales. To account for these soil-atmosphere interactions a realistic representation of subsurface hydrodynamics is needed but it is computationally expensive. Therefore, project MCRA developed a model complexity reduction approach to correct the simplified climate model with the information from a full-physics model that was applied over specified catchments (Shrestha et al. 2014). They have run their simplified regional climate model for the European CORDEX domain at a high resolution of 0.11° (Fig. 4) for short time periods, e.g. including the European heat wave in August 2003, and analysed the influence of the lower energy boundary conditions on the water and energy cycle at the land surface. They also developed similarity indices that are needed for the hydrologic parameterisations in the complexity reduction approach.

Module B Summary - Figure 4
Figure 4 (MCRA): Increment of total water storage [mm] over the European CORDEX domain during the European heatwave in August 2003. The increment is calculated as the difference of total water storage from August 2003 to July 2003.

To consider chemical processes in the atmosphere in detail, chemistry climate models need to be applied. In Module B MAECHAM5/HAM and EMAC were used for different problems. They are based on the atmosphere model ECHAM that is part of the MiKlip prediction system. As the inclusion of interactions between chemistry and climate is computationally expensive, the project LiCoS optimised a Rosenbrock integrator for chemical reactions achieving a speed-up of 2 to 3 times in EMAC compared to the standard integrator.

Another possibility to drastically reduce computation time when accounting for stratospheric chemistry is realized in the project FAST-O3 by developing the fast stratospheric chemistry scheme SWIFT to simulate important interactions between climate and ozone (Rex et al. 2014). They implemented SWIFT as an alternative chemistry module into a chemistry and transport model and compared it to the full chemistry version and to satellite observations. The comparison indicated that the computationally much faster parameterisation yields similarly good results as the full chemistry version (Fig. 5). In the last year of the project SWIFT was also implemented in EMAC.

Module B Summary - Figure 5
Figure 5 (FAST-O3): Performance of the polar chemistry module. North polar vortex averages of (top) ozone concentrations and (bottom) HCl mixing ratios as a function of time in 2004/2005 and altitude. From left to right: ATLAS Chemistry and Transport Model with SWIFT as chemistry module, ATLAS CTM with full chemistry, MLS satellite observations.

EMAC was also applied in the project STRATO to analyse the response of the coupled stratosphere-troposphere-ocean system to decadal variability of the stratosphere including the natural external solar forcing. The comparison between EMAC and Baseline1-LR showed a good agreement in structure and magnitude of the decadal solar signal. However, the ozone and solar temperature signals are overestimated in both Baseline1-LR and MR compared to EMAC and to observations (Fig. 6). In the lower stratosphere the ENSO temperature signal is reproduced in both Baseline1 versions but weaker than in EMAC and in observations.

Module B Summary - Figure 6
Figure 6 (STRATO): Annual mean tropical (25°S-25°N) solar regression coefficients for (top) temperature in Kelvin per 100 units of the 10.7 cm solar flux in (left) the Baseline1-LR, Baseline1-MR and EMAC models (1960-2007), and (right) in CCMVal-2 simulations and observations (black) (1960-2004); (bottom) same for ozone in % per 100 units of the 10.7 cm solar flux. Note different axes.

As a further natural external forcing the impact of large volcanic eruptions is analysed in the project ALARM. They provided a volcano module with a two-step approach. First, the volcanic radiative forcing is calculated with MAECHAM5/HAM. Second, this forcing is used in the MiKlip prediction system. With this setup they performed simulations with a Pinatubo-like eruption in 2013 and showed that the impact is similar to the reconstructed patterns after large historic eruptions. To assess the impact of volcanic eruptions on the predictive skill, they analysed Baseline0-LR decadal hindcasts with and without major volcanic eruptions and showed that especially in the first year the prediction skill over Eurasia is significantly improved if the eruptions are considered (Fig. 7).

Module B Summary - Figure 7
Figure 7 (ALARM): Difference in the ensemble mean hindcast skill (MSSS) of near surface air temperature between volcanic and non-volcanic Baseline0-LR hindcast simulations over the first prediction year. Squares indicate anomalies which are significant at the 95% confidence level. The plot is produced with the MiKlip Evaluation System (MurCSS tool).

Understanding of decadal variability


The impact of clouds on decadal climate variability was analysed by the project LiCoS. They found that key players in the interannual to decadal variability of large-scale climate patterns are atmospheric processes that are caused by cloud feedbacks (Bellomo et al. 2014, 2015). These feedbacks are likely to be responsible for a big part of the ENSO variability by changing global circulation. They also showed that a positive feedback among sea surface temperature (SST), cloud cover, and large-scale atmospheric circulation can explain decadal climate variability in the Pacific Ocean. In addition, LiCoS quantified the ozone pollution prediction skills of the MiKlip prediction system by identifying the atmospheric stagnation index as a good indicator for the occurrence of high ozone days in industrialized regions (Fig. 8). As all necessary parameters are part of the standard output of the MiKlip prediction system, the atmospheric stagnation index can be used as a proxy for an ozone pollution index.

Module B Summary - Figure 8
Figure 8 (LiCoS): (top) Ratio of days with a high ozone concentration (>50 ppbv) and an Air Stagnation Index (ASI) equals to 1 and all high ozone days (corresponding to a hit-rate for the ASI) for (left) Europe and (right) the Eastern USA. (bottom) As above but difference between hit-rate and false-alarm rate. In all cases an additional condition for the surface temperature >290K has been applied.
Module B Summary - Figure 9
Figure 9 (ATMOS): Explained variance (computed using the ANOVA technique) of convective precipitation due to the SST variability in a 5 member ensemble using HadISST forcing from 1870-2007 (a) for summer (JJA) and (b) for winter (DJF). The contours show the climatological SST.

For a better understanding of atmosphere-ocean processes the project ATMOS concentrated on the North Atlantic and found in atmosphere-only simulations that a significant fraction of the convective precipitation over and south of the Gulf Stream can be explained by the variability of the underlying SST, especially in summer (Fig. 9, Hand et al. 2014). In winter nearly all of the anomalous precipitation is connected to passing atmospheric fronts. They also found a resolution dependence of the intensity, tilt and north-eastward extension of the North Atlantic storm tracks and of precipitation along the path of the Gulf Stream. This demonstrates the importance of reducing the misplacement of the Gulf Stream and the North Atlantic Current and therefore the cold bias in coupled climate models like the MiKlip prediction system. The need to reduce the coupled model bias was also shown in the project MultiCliP by the importance of the subpolar North Atlantic and its interaction between gyre and overturning circulation for the northward oceanic heat transport (Müller et al. 2015). Beside the North Atlantic they also focused on the Pacific and found a sensitivity of North Atlantic and tropical SSTs to external natural forcing like volcanic eruptions, which is absent for the extratropical North Pacific SSTs. The understanding of the circumstances and mechanisms favouring specific phasing between Atlantic and Pacific SST modes could lead to a reduction of uncertainty in decadal climate predictions.

Module B Summary - Figure 10
Figure 10 (MultiCliP): Annual Northern Hemisphere–average surface air temperature evolutions around the 1815 Tambora eruption from reconstructions (red line: ensemble mean, red shading: 5th–95th percentile range of the 521-member raw calibration ensemble by Frank et al., (2010)) and from an ensemble of full-forcing climate simulations performed with ECHAM5/MPIOM (black dashed line: unsmoothed ensemble mean, black solid line: smoothed (11-year moving average) ensemble mean, lines from green to blue: individual simulations). Anomalies are with respect to the 1799–1808 period. Vertical dashed lines indicate the occurrence of the 1809 eruption of unknown location, and of the Tambora and Cosiguina eruptions.

The analysis of the impact of volcanic eruptions by MultiCliP revealed that Northern Hemisphere temperature anomalies around eruptions depend on the initial state (Fig. 10, Zanchettin et al. 2013). Together with the project ALARM they expect an improvement of decadal predictions including volcanic forcing when sea ice is initialised. Both projects also showed that the expansion of Arctic sea ice extent after a volcanic eruption is robustly simulated. According to ALARM quite accurate aerosol forcing fields would be necessary to improve predictions of the dynamical response to stratospheric sulphate aerosol loading for a Pinatubo-like eruption (Fig. 11, Toohey et al. 2014). They also showed that the variability of the temperature anomalies in mid- and late winter in the northern polar stratosphere is significantly reduced under the strong Tabora-like eruption and not significantly changed under the weaker Pinatubo-like eruption. However, weaker eruptions do produce significant global cooling.

Several projects in Module B investigated the effect of the different resolutions in Baseline1-LR and MR. In general, the stratosphere should be included, as MultiCliP found that this improves the simulation of extra-tropical tropospheric variability. This is the case for both versions, LR and MR. However, the higher resolution in MR is necessary if an improved representation of specific processes like the quasi-biennial oscillation is required. Concerning stratospheric variability STRATO found a gradual improvement from LR to MR, but both versions overestimate variability during winter in southern hemisphere polar regions.

Modulre B Summary - Figure 11
Figure 11 (ALARM): First post-eruption NH winter ensemble mean anomalies of (top) temperature and (bottom) zonal wind for four Pinatubo forcing data sets in historical-LR simulations. Two forcing sets are based on satellite observations of the Pinatubo aerosol (S98: Stenchikov et al., 1998 and CCMI: SAGE_4λ, Arfeuille et al., 2013), while two others are constructed from independent simulations with MAECHAM5/HAM (SVC: strong vortex composite, WVC: weak vortex composite). Anomalies which are significant at the 95% confidence level (as determined by comparison with the variability of a control run) are hatched.


Bellomo, K., A. Clement, T. Mauritsen, G. Rädel, and B. Stevens (2014): Simulating the role of subtropical stratocumulus clouds in driving Pacific climate variability., J. Clim., 27(13), 5119-5131, doi:10.1175/JCLI-D-13-00548.1

Bellomo, K., A. Clement, T. Mauritsen, G. Rädel, and B. Stevens (2015): The influence of Cloud Feedbacks on Tropical Atlantic Variability. J. Clim., 28, 2725-2744. doi:10.1175/JCLI-D-14-00495.1

Hand, R., N.S. Keenlyside, N.-E. Omrani, and M. Latif (2014): Simulated response to inter-annual SST variations in the Gulf Stream region., Clim. Dyn., 42, 715–731, doi: 10.1007/s00382-013-1715-y

Lüpkes, C., V. M. Gryanik, A. Rösel, G. Birnbaum, and L. Kaleschke (2013): Effect of sea ice morphology during Arctic summer on atmospheric drag coefficients used in climate models., Geophys. Res. Lett., 40, 446–451, doi:10.1002/grl.50081

Lüpkes, C., and V. M. Gryanik (2015): A stability-dependent parametrization of transfer coefficients for momentum and heat over polar sea ice to be used in climate models. J. Geophys. Res. Atmos., 120, doi:10.1002/2014JD022418

Müller, W.A., D. Matei, M. Bersch, J.H. Jungclaus, H. Haak, K. Lohmann, G.P. Compo, P.D. Sardeshmukh, and J. Marotzke (2015): A twentieth-century reanalysis forced ocean model to reconstruct the North Atlantic climate variation during the 1920s, Clim. Dyn., 44, 1935-1955, doi:10.1007/s00382-014-2267-5

Rex, M., S. Kremser, P. Huck, G. Bodeker, I. Wohltmann, M. L. Santee, and P. Bernath (2014): Technical Note: SWIFT – a fast semi-empirical model for polar stratospheric ozone loss. Atmos. Chem. Phys., 14, 6545–6555, doi:10.5194/acp-14-6545-2014

Shrestha, P., M. Sulis, M. Masbou, S Kollet, and C. Simmer (2014): A Scale-Consistent Terrestrial Systems Modeling Platform Based on COSMO, CLM, and ParFlow., Monthly Weather Review, 142 (9), 3466 - 3483, doi:10.1175/MWR-D-14-00029.1

Sidorenko, D., T. Rackow, T. Jung, T. Semmler, D. Barbi, S. Danilov, K. Dethloff, W. Dorn, K. Fieg, H.F. Gößling, D. Handorf, S. Harig, W. Hiller, S. Juricke, M. Losch, J. Schröter, D. Sein, and Q. Wang (2015): Towards multi-resolution global climate modeling with ECHAM6–FESOM. Part I: model formulation and mean climate., Clim. Dyn., 44, 757-780, doi:10.1007/s00382-014-2290-6

Toohey, M., K. Krüger, M. Bittner, C. Timmreck, and H. Schmidt (2014): The impact of volcanic aerosol on the Northern Hemisphere stratospheric polar vortex: mechanisms and sensitivity to forcing structure., Atmos. Chem. Phys. Discuss., 14, 16777-16819, doi:10.5194/acpd-14-16777-2014

Zanchettin D., O. Bothe, H.-F. Graf, S.J. Lorenz, J. Luterbacher, C. Timmreck, and J.H. Jungclaus (2013): Background conditions influence the decadal climate response to strong volcanic eruptions., Journ. of Geoph. Res.: Atm., 118, 4090–4106, doi:10.1002/jgrd.50229

More on Module BNews-Icon

This description summaries the achievements of Module B during the first phase. Module B continues in MiKlip II, read more here.