MOVIECLIP is the successor of MultiCliP, which has investigated mechanisms of multi-decadal variability and the relative importance of internal variability vs. the response to external forcing. The focus of MultiCliP has been on the oceanic response that exhibits memory on time scales important for near-term climate predictions. MultiCliP has also investigated oceanic and coupled variability modes and their teleconnections, and contributed to an improved understanding of the stratosphere-troposphere-ocean interaction. MOVIECLIP aims at further enhancement of the physical understanding of decadal climate variability. Therefore, the processes that are relevant on these time scales will be identified and analysed, and their representation in the Earth system model MPI-ESM will be investigated with a regional focus on the North Atlantic and continental European sector. This includes, in particular, the investigation of the role of the ocean heat budget. The project will provide recommendations to the model devellopers, including advising about appropriate horizontal and vertical resolutions to optimise the prediction skill for certain prediction variables and regions, and/or optimising the consumption of computing time resources via a reasonable choice of the model resolution, as well as methods to improve the representation of specific processes (e.g. by the implementation of bias correction techniques). MOVIECLIP will analyse existing climate simulations with different versions of MPI-ESM and its atmospheric component, called ECHAM6, in different resolutions. Additionally, dedicated sensitivity experiments are planned to isolate the impact of specific processes of interest.

The project is structured in three parts:

The first work package will have its focus on the question which mechanisms drive/cause the predictability in the MiKlip prediction system, and explore the impact of ocean variability on large-scale atmospheric circulation modes. Particular modes of decadal ocean variability are analysed with a special focus on the questions, to what extent they can provide predictability on decadal time scales, how the ocean signal is transferred to the atmosphere and what are the teleconnection pathways that transfer the signal from certain regionally pronounced modes through the atmosphere to remote regions.

The second work package will deal with the impact of model biases in the ocean model (e.g. the North Atlantic SST (sea surface temperature) bias) and strategies for online and a posteriori correction techniques. This work package will have a strong link to ATMOS-MODINI. In contrast to the latter, MOVIECLIP will test methods to correct the impact of the SST bias in the atmospheric component of the model (e.g. by correcting the anomalous surface heat fluxes associated with the bias), instead of modifying the ocean circulation itself as to be done in ATMOS-MODINI.

The last work package will investigate the relation of oceanic variability modes with mechanisms of heat storage and release in the North Atlantic, for example through differential variations in the gyre- and overturning heat transports and related meridional heat transport divergence. MOVIECLIP will contribute to coordinated multi-model investigations of a restricted number of mechanisms planned in the framework of the CMIP6 Decadal Climate Prediction Model Intercomparison Project (DCPP) project.
 

Progress so far

Concerning WP 1, we concentrated on understanding the link between the Atlantic Multidecadal Variability (AMV) and the Atlantic Meridional Overturning Circulation (AMOC) in MPI-ESM. We performed a statistical analysis on the MPI Grand Ensemble, consisting of a 2000 year pre-industrial control (pictl) run, a 100-member ensemble of historical simulations and a 68-member ensemble of runs with an incremental CO2 increase by 1%/year.  In the experiments the AMV is lagging the AMOC at 45°N by approx. 4 years and a weakening of the AMOC is following the AMV maximum by approx. 10-15 years. This correlation is robust for all experiments, but the amplitude varies between different periods of the pictl run and within the individual ensemble members of the historical ensemble. Furthermore the amplitude is much weaker towards the end of the 1% experiment than in the other simulations, going along with a much weaker amplitude of the AMOC decadal variability. A positive correlation between upward turbulent heat fluxes and SST was found, indicating that the temperature anomalies are caused by the ocean on the decadal time-scales.

In WP 2 we analysed the atmospheric response to the North Atlantic SST bias simulated in the coupled model system. In a set of SST sensitivity experiments with the atmospheric model component of MPI-ESM the SST anomaly reflecting the SST bias (with respect to observations) was prescribed. The SST-sensitivity experiments show that atmospheric biases in the coupled model compared to observations can be at least partly associated with the simulated SST bias; even though the biases' amplitude is stronger in the coupled system. The geopotential height is reduced (increased) over warm (cold) SST, leading to modified large-scale meridional pressure gradients, resulting in a southward shift of the mean zonal wind system. The according southward shift of the baroclinic regions causes a southerly shift of the North Atlantic storm track. The precipitation response in a first approximation follows the SST pattern (and the according changes in latent heat flux), but is modified by the storminess response in some regions, in particular in the coupled system.