The Earth's climate is changing. This is a conclusion of the Third Assessment Report published by the International Panel on Climate Change. The increasing concentrations of greenhouse gases and sulphate aerosols alter the energy path through the atmosphere. In the future, a likely consequence of the induced global warming is an increased melting of the Greenland ice sheet. This could lead to abrupt climatic modifications associated with the collapse of the thermohaline circulation in the North Atlantic. To investigate this issue, a coupled model of the climate system has been developed. Most components of this system are taken into account by this model. LMD5, an atmospheric general circulation model originating from the Laboratoire de Météorologie Dynamique in Paris, simulates the atmosphere. CLIO2, an oceanic general circulation model set up at the Institut d'Astronomie et de Géophysique G. Lemaître at the UCL, accounts for the ocean and the sea ice. Finally, GISM is a Greenland ice-sheet model developed at the Vrije Universiteit Brussel (VUB). The atmospheric and oceanic components had already been coupled. A new version has been elaborated during this thesis to enable long-term realistic simulations. To restrain the initial drift, adjustements have been made in atmospheric, oceanic and sea-ice parameters in collaboration with the research teams that set up the models. Other modifications have been performed in the framework of climate change experiments to separately handle greenhouse gases and sulphate aerosols. Finally, LMD5-CLIO2 has been coupled to GISM in collaboration with VUB researchers. Validation of LMD5-CLIO2 implies a 150-year long control simulation under constant 1970 forcings. The validation is twofold: on the one hand, the LMD5-CLIO2 results have been contrasted with observational estimates and on the other hand, these have been compared to other coupled models results. This leads to the conclusions that LMD5-CLIO2 simulates relatively well the present- day climate and that it performs as well as many other coupled models. Two climate change experiments have been carried out using the IPCC SRES B2 scenario. The first one deals with the impact of greenhouse gases and sulphate aerosols on the 21st century climate. The globally averaged warming reaches 2.4°C at the end of the 21st century compared to 1970. The hydrological cycle is amplified: precipitation increases by 3.3 % over the same period. This corroborates the projections of other coupled models. Consequently, it improves our confidence in the LMD5-CLIO2 simulations. The second climate change experiment performed with LMD5-CLIO2-GISM investigates the impact of the Greenland ice-sheet freshwater flux on the North Atlantic thermohaline circulation. To our knowledge, it is the first time that such a complex coupled model is used to analyse this issue. In this simulation, deep convection in the North Atlantic ocean shuts down at the end of the 21st century, which is an unusual result. The strong and abrupt reduction of the intensity of the thermohaline circulation implies significant mainly local modifications of sea ice and of the atmosphere. Cautions should be exercised to this projection. Firstly, LMD5-CLIO2 has deficiencies which probably influence the results. Secondly, modifications of the initial conditions could lead to another projection: such spectacular event not occurring or differences as to when it happens. Consequently, other ensemble simulations should been performed. Thirdly, the current simulation is too short to know if the reduction is temporary or not. Longer experiments are therefore required.

A new Earth system model of intermediate complexity, LOVECLIM, has been developed in order to study long-term future climate changes. In particular, LOVECLIM includes an interactive Greenland and Antarctic ice sheet model (AGISM) as well as an oceanic carbon cycle model (LOCH). Those climatic components can have a great impact on future climate. However, most studies investigating future climate changes do not take them into account. The few studies in recent literature assessing the impact of polar ice sheets on future climate draw very different conclusions, which shows the need for developing such a model. The aim of this study is to analyse the possible perturbations of climate induced by human activities over the next millennia. A particular attention is given to the evolution of the oceanic thermohaline circulation. A series of numerical simulations have been performed with LOVECLIM over the next millennia using various forcing scenarios. The global equilibrium warming computed by the model ranges from 0.55°C to 3.75°C with respect to preindustrial times. The model does not simulate a complete shut down of the oceanic thermohaline circulation but a transient weakening followed by a quasi-recovering at equilibrium. In most of the projections, the Greenland ice sheet undergoes a continuous reduction in volume, leading to an almost total disappearance in the most pessimistic scenarios. The impact of the Greenland deglaciation on climate has been assessed through sensitivity experiments. The removal of the Greenland ice sheet is responsible for a regional amplification of the global warming inducing a total melt of Arctic sea ice in summer. The freshwater flux from Greenland generates large salinity anomalies in the North Atlantic Ocean that reduce the rate of North Atlantic Deep Water formation, slowing down the oceanic thermohaline circulation.

A new Earth system model of intermediate complexity, LOVECLIM, has been developed in order to study long-term future climate changes. In particular, LOVECLIM includes an interactive Greenland and Antarctic ice sheet model (AGISM) as well as an oceanic carbon cycle model (LOCH). Those climatic components can have a great impact on future climate. However, most studies investigating future climate changes do not take them into account. The few studies in recent literature assessing the impact of polar ice sheets on future climate draw very different conclusions, which shows the need for developing such a model. The aim of this study is to analyse the possible perturbations of climate induced by human activities over the next millennia. A particular attention is given to the evolution of the oceanic thermohaline circulation. A series of numerical simulations have been performed with LOVECLIM over the next millennia using various forcing scenarios. The global equilibrium warming computed by the model ranges from 0.55°C to 3.75°C with respect to preindustrial times. The model does not simulate a complete shut down of the oceanic thermohaline circulation but a transient weakening followed by a quasi-recovering at equilibrium. In most of the projections, the Greenland ice sheet undergoes a continuous reduction in volume, leading to an almost total disappearance in the most pessimistic scenarios. The impact of the Greenland deglaciation on climate has been assessed through sensitivity experiments. The removal of the Greenland ice sheet is responsible for a regional amplification of the global warming inducing a total melt of Arctic sea ice in summer. The freshwater flux from Greenland generates large salinity anomalies in the North Atlantic Ocean that reduce the rate of North Atlantic Deep Water formation, slowing down the oceanic thermohaline circulation.