No abstract

Gridded past millennium climate reconstructions, encompassing the European continent (e.g. Luterbacher et al 2004), based on proxy records, have previously been generated assuming that spatial temperature covariance across the region behaves in the past in the same way as it does in the observational period. This strategy bears the risk of artificially identifying the same patterns of variability as presently observed and overlooking periodically occurring modes of internal climate variability, that are not uniformly spatio-temporally expressed. Here, we construct regional proxy composites for Europe which are not constrained by the modes of variability expressed in the 20th century, and should thus broadly represent coherent regional summer temperatures back through time, independent from present modes of variability. The proxy data set analysed was provided by the efforts of the EU 6th Framework Millennium project. Proxy data are dominated by tree ring width, density and annual height increment. Four composites are used describing summer temperatures in the Arctic, Central, Pyrenean and Alpine zones of Europe. The proxy data sets cover the period AD 1260-1996. We jointly analyse an ensemble of simulations with global climate models participating in the Climate Model Intercomparison Project (CMIP5) included in the 5th Assessment report of the Intergovernmental Panel on Climate Change (IPCC) covering the period AD 850-2005, with the above aims in mind. The climate models were driven by estimations of the main external forcings. The implementations of these forcings may vary among simulations, depending on the different estimates used and on the structure of the models themselves. We analyse three simulations (CCSM4, IPSL, MPI-ESM) that, at the time of writing, provide daily data over the period AD 850-2005 obtained with the very same model version through the whole period. The analysis of both proxy and model data allowed us to more precisely identify the behaviour of the internal climate mode identified.

The long-term variability of an aquaplanet climate is analyzed with a coupled atmosphere–ocean–sea ice general circulation model. The main result of the 20,000 years simulation is a very dominant low-frequency oscillation with a period of approximately 700 years. All compartments of the aquaplanet (atmosphere, ocean, and sea ice) are involved as the climate alternates between warmer and colder states. Comprehensive time series analyses, as well as a comparison between mean states of cold and warm phases, give a detailed picture of the life cycle of the low-frequency oscillation. The warm phases are characterized by ice-free polar waters and a weaker meridional overturning circulation. During cold phases, the poles are completely covered by sea ice (down to 65∘ N/S) and the overturning cells in the ocean are stronger. The climate state changes throughout atmosphere and ocean; however, surface areas in high latitudes are especially affected due to the changing sea ice cover. The meridional energy transport in atmosphere and ocean alternates with the climate regime, since the ocean is more efficient in transporting heat poleward when the poles are ice-free

The dominant mode of inter-annual variability in the tropical Pacific is El Niño-Southern Oscillation (ENSO). ENSO is not symmetric in the sense that El Niño is generally stronger than La Niña. However, many CMIP5 models, including the Max Planck Institute Earth System Model (MPI-ESM), produce an almost symmetric ENSO. This paper shows that, when resolving the intra-daily air-sea interactions by coupling the atmospheric and oceanic model components once per hour, the simulated ENSO asymmetry is improved. The improvement is closely related to the simulated diurnal cycle of sea surface tempreature (SST). In the central tropical Pacific, the simulated diurnal range of SST is about 0.2 °C, up to 10% of SST anomalies of the simulated ENSO events. During El Niño events, the diurnal cycle of SST anomalies enhances the atmospheric moist instability, whereby triggering more convection in the central tropical Pacific. During La Niña and normal years, however, the mean convection is not changed by the included diurnal cycle of SST anomalies. As a result, the anomalies of the trades, which are directly related to the convection, are stronger during El Niño years than that during La Niña years, making El Niño to be stronger than La Niña via Bjerknes feedback. These results obtained with a low resolution MPI-ESM are further confirmed by simulation with the same model at higher spatial resolutions, suggesting that the role of the intradaily air-sea interactions for the ENSO asymmetry is independent of model resolutions.