Long-term ice sheet–climate interactions under anthropogenic greenhouse forcing simulated with a complex Earth System Model
- 1.7k Downloads
Several multi-century and multi-millennia simulations have been performed with a complex Earth System Model (ESM) for different anthropogenic climate change scenarios in order to study the long-term evolution of sea level and the impact of ice sheet changes on the climate system. The core of the ESM is a coupled coarse-resolution Atmosphere–Ocean General Circulation Model (AOGCM). Ocean biogeochemistry, land vegetation and ice sheets are included as components of the ESM. The Greenland Ice Sheet (GrIS) decays in all simulations, while the Antarctic ice sheet contributes negatively to sea level rise, due to enhanced storage of water caused by larger snowfall rates. Freshwater flux increases from Greenland are one order of magnitude smaller than total freshwater flux increases into the North Atlantic basin (the sum of the contribution from changes in precipitation, evaporation, run-off and Greenland meltwater) and do not play an important role in changes in the strength of the North Atlantic Meridional Overturning Circulation (NAMOC). The regional climate change associated with weakening/collapse of the NAMOC drastically reduces the decay rate of the GrIS. The dynamical changes due to GrIS topography modification driven by mass balance changes act first as a negative feedback for the decay of the ice sheet, but accelerate the decay at a later stage. The increase of surface temperature due to reduced topographic heights causes a strong acceleration of the decay of the ice sheet in the long term. Other feedbacks between ice sheet and atmosphere are not important for the mass balance of the GrIS until it is reduced to 3/4 of the original size. From then, the reduction in the albedo of Greenland strongly accelerates the decay of the ice sheet.
KeywordsIce sheets Anthropogenic climate change Meridional overturning circulation Earth system modelling Sea level
The melting of glaciers and the thermal expansion of the oceans associated with anthropogenic climate change are expected to produce substantial sea level changes during the next few centuries (Meehl et al. 2007). The continental-size glaciers on Earth, the Greenland and Antarctic ice sheets (GrIS and AIS), store a volume of water equivalent to a sea level rise of 7.2 and 61.1 m, respectively (Church et al. 2001). While half of the current ablation in the GrIS is due to surface melting, almost all ablation happens as calving at the margin of the AIS, due to the very low atmospheric temperatures. Alley et al. (2005) and Gregory et al. (2004) showed that for a uniform warming of more than 3 K, the GrIS would begin to decay, disappearing in between 1000 and several thousand years, depending on the magnitude of the warming.
In addition to producing changes in sea level, future modifications of the ice sheets could produce changes in the climate system via changes in ocean circulation, surface albedo, and/or atmospheric circulation patterns. Such modifications of climate could potentially affect the mass balance of the ice sheets and therefore play an important role in their own evolution. There are a few examples of modelling studies in the literature performed with coupled ice sheet–climate models with the aim of identifying and quantifying these feedbacks between ice sheets and climate for both the cases of past and future climate. For the studies of past climates, Intermediate Complexity Models (EMICs, for an overview see Petoukhov et al. 2005) have been used as tool, due to the long time scales and the limitations imposed by computational resources. For future climate projections, some studies have focused exclusively on the feedbacks between ice sheets and ocean (Huybrechts et al. 2002; Fichefet et al. 2003), while others have investigated the feedbacks between ice sheets and atmosphere as well (Ridley et al. 2005; Driesschaert et al. 2007). Only the studies of Ridley et al. (2005), with HadCM3, and Mikolajewicz et al. (2007b) were performed with General Circulation Models (GCMs) both for the ocean and atmosphere, and with full coupling between the three components ice sheet, ocean, and atmosphere. This type of tool enables the investigation of the feedbacks between ice sheets and both the circulations of atmosphere and ocean. Feedback studies using models of intermediate complexity may allow the estimate of some of the relevant feedbacks but are in general limited by the strong simplifications of atmospheric physics and dynamics and their potential impact on some of the simulated feedbacks.
The albedo-feedback is considered to have played a key role in the growth of ice sheets during past glacial inceptions (Kageyama et al. 2004; Calov et al. 2005). Manabe and Broccoli (1985) found that the albedo and topographic effects of ice sheets alone explain much of the northern hemisphere cooling identified in paleoclimatic records of the last glacial maximum. In Ridley et al. (2005), albedo changes act as a negative feedback for the decay of the GrIS under a constant 4× CO2 scenario. The evolution of convective cells, with rising warm air over the areas becoming ice-free and descending air over the ice sheet, originates the transport of cold air from the interior of the ice sheet into the ablation zones at the margins of the ice sheet, reducing surface melting.
The height-feedback, i.e., the modification of surface temperature due to changes in elevation, can play an important role in the processes of growth and decay of ice sheets, for instance at glacial inception (Gallee et al. 1992; Wang and Mysak 2002), or in the future development of the GrIS. In Huybrechts and de Wolde (1999) a substantial contribution of this feedback to the decay rate of the GrIS is shown. These authors also investigated the effect on the velocity field of changes in topography driven by changes in the surface mass balance. They found a deceleration of the decay rate during a first phase, which turned into acceleration at a later stage.
The freshening of ocean waters at deep convection sites by meltwater from the ice sheets could potentially modify the strength of the NAMOC. Several authors have investigated this issue for past climates and future climate projections. The discharge of meltwater from the northern hemisphere ice sheets has been proposed as the triggering mechanism for abrupt climate change in the past at the time of the Younger Dryas (Broecker et al. 1988; Maier-Reimer and Mikolajewicz 1989). Fichefet et al. (2003) found a substantial weakening of the NAMOC in response to increased meltwater fluxes from the GrIS in a twenty-first century simulation. Driesschaert et al. (2007) investigated the response of the climate system to different greenhouse gas scenarios with a three-dimensional (3D) EMIC including a dynamic ice sheet component. They found a noticeable weakening of the NAMOC due to freshwater fluxes from the GrIS only under the forcing of the most extreme scenario.
A reduction in northward ocean heat transport due to a collapsed or weakened NAMOC potentially modifies the climate of the North Atlantic region (e.g. Stouffer et al. 2006, Fig. 4). This regional climate change could produce a reduction of meltwater fluxes from the ice sheets in the region, acting as a mechanism stabilising the NAMOC. This negative feedback between ice sheet melting and NAMOC changes could modify the time scales of recovery of the NAMOC after strong meltwater fluxes pulses both in past and future climates. This mechanism has not been received much attention in the literature yet. In the greenhouse gas simulation of Swingedouw et al. (2006), with a coupled-ocean atmosphere model and a simple land-ice melting parameterisation, the freshwater forcing from ice sheet melting increases nearly linearly from 0 to 0.2 Sv in 140 years of 1% per year increase of atmospheric CO2. The NAMOC weakens from 10 to 5 Sv, with a substantial contribution of ice sheet melting to this weakening. The negative atmospheric temperature anomaly associated with this decline has its maximum far from Greenland, over the Barents Sea, being its absolute value less than 2 K over Greenland. The mechanism suggested above does not seem relevant in this study, since ice sheet melting does not appear to decrease substantially following the weakening of the NAMOC. Similarly, in the 2× CO2 simulation of Swingedouw et al. (2007), the impact on the Greenland climate of the reduced ocean heat transport accompanying the decline of the NAMOC does not prevent land-ice melting from causing the shutdown of the NAMOC.
Changes in the topography of the ice sheets can modify the general circulation of the atmosphere. Modelling studies of past climates indicate that the path of the jet stream is highly dependent on the topography of the Laurentide ice sheet (Manabe and Broccoli 1985; Cook and Held 1988). The changes in the atmosphere that the absence of the GrIS would cause in the case of its complete deglaciation have been investigated in studies where the ice sheet is absent either from a pre-industrial basic state (e.g., Lunt et al. 2004; Toniazzo et al. 2004; Junge et al. 2005) or from a human-perturbed climate (Ridley et al. 2005). These studies found a regional impact of the absence of the GrIS, with maximal warming in Greenland due to drastic reduction of surface albedo and topographic height, and substantial changes in the path of the storm track and in the mean circulation of the atmosphere in the northern high latitudes.
A key question in the investigation of these feedbacks between ice sheets and climate is whether there exist thresholds for them to begin to operate, as well as the extent to which the interactions of ice sheets and climate behave in a non-linear way. In Ridley et al. (2005) the feedbacks between the GrIS and climate begin to contribute to the mass budget of the ice sheet from the time when the ice sheet has decayed to 2/3 of its original volume. The response of the ocean circulation to changes in freshwater fluxes can be highly non-linear, especially if the model is close to a bifurcation point of the NAMOC. The changes in ocean heat transport resulting from a collapse of the NAMOC are partially compensated by enhanced atmospheric transport of latent heat (and thus moisture) in the North Atlantic drainage basin (e.g. Nakamura et al. 1994; Schiller et al. 1997). Once the model has crossed the bifurcation point of the NAMOC even a reduction of additional freshwater supply does not necessarily lead to a substantial recovery of the NAMOC (e.g. Stommel 1961; Manabe and Stouffer 1988; Mikolajewicz and Maier-Reimer 1994; Rahmstorf and Willebrand 1995). Whether such a bifurcation point of the NAMOC exists under present conditions is subject of debate. Several studies have investigated whether the GrIS would regrow after its complete deglaciation if pre-industrial or present-day levels of atmospheric CO2 would be reestablished (Crowley and Baum 1995; Calov et al. 2005; Toniazzo et al. 2004; Lunt et al. 2004).
In our study, a global ice sheet model has been bi-directionally coupled to a complex climate model with an Atmosphere–Ocean General Circulation Model (AOGCM) as core in order to study the long-term evolution of global ice sheets under anthropogenic forcing and the impact of changes in their mass balance on the climate system. With this approach, changes due to ice dynamics can be taken into account, as well as the feedbacks between ice sheets and climate and other feedbacks between other main components of the climate system (ocean, atmosphere, global vegetation, and carbon cycle). The relatively coarse resolution of the atmospheric component (T21) represents a compromise between the needs of detailed physical representation of processes and the realisation of long simulations, in order to investigate the long-term response of the climate system to anthropogenic greenhouse forcing and to identify the non-linearities and critical thresholds that could arise.
This paper is a follow-up of Mikolajewicz et al. (2007a) (in the following Mik07), where this new Earth System Model (ESM) is introduced with a general description of its components, validation of the model against observations, and results from multi-millennia ensemble simulations describing the long-term response of the Earth System to prescribed carbon emissions in the twenty-first century. These emissions follow three IPCC scenarios of increasing greenhouse forcing: B1, A1B and A2. The paper aims to extend the results from Mik07 by focusing in the role of ice sheets on climate changes. The main goals of this paper are to present a detailed description of the coupling of the ice sheet component to the rest of components in this ESM, and to identify the feedbacks between ice sheets and climate relevant to anthropogenic climate change. In order to accomplish the second goal, two different sets of simulations are analysed: a set of simulations where the atmospheric carbon dioxide concentration is increased by 1% per year until stabilisation at two, three and four times pre-industrial levels, and a second set of longer simulations following the CO2 emission scenario A2 until year 2100 and an exponential decrease of emissions afterwards. The results of the first set of simulations are presented first, in order to explore the existence of thresholds and non-linearities in the climate system by means of the analysis of its response to increasing anthropogenic greenhouse forcing. By the end of a longer simulation with forcing corresponding to the high emission scenario A2, the GrIS almost disappears. This simulation and a simulation with the same setup but not including the feedbacks between ice sheets and climate are compared in order to evaluate the importance of these feedbacks for the mass balance of the GrIS, as well as the impact on the climate system of the disappearance of the GrIS.
The structure of this paper is as follows: first the ESM is described in Sect. 2, focusing on the ice sheet model and its coupling to the atmosphere and ocean components. In Sect. 3, the setup of both stabilisation and A2 simulations is presented and the global changes in the ocean and atmosphere in the stabilisation simulations are described. In Sect. 4, changes in the mass balance of global ice sheets and in sea level in the stabilisation simulations are shown. Section 5 focuses on the feedbacks between ice sheets and the climate system. The results of the A2 simulations are used here in order to complement the findings from the stabilisation simulations. Summary and discussion are given in Sect. 6.
2 The model
2.1 The Earth System Model
The core of the ESM is an Atmosphere–Ocean General Circulation Model. The ESM includes the main physical and biogeochemical components of the Earth System via fully coupled models of ocean biogeochemistry (modelled with HAMOCC, Maier-Reimer 1993), land vegetation (modelled with LPJ, Sitch et al. 2003), and ice sheets (modelled with the 3D model SICOPOLIS, Greve 1995). The ocean biogeochemistry and land vegetation components together with a well-mixed atmospheric box constitute a closed carbon cycle. Therefore, atmospheric carbon dioxide concentrations can be calculated prognostically by the model when carbon emissions are prescribed. The only external forcing used in the simulations presented here with this ESM are carbon emissions or prescribed atmospheric carbon dioxide concentrations.
Long (multi-millennia) simulations can be performed with this model due to its coarse resolution (T21 for the atmosphere model). In order to save further computation time and to facilitate the long-term simulations, a periodically synchronous coupling technique has been applied (Voss and Sausen 1996). The underlying assumption is that the long-term memory resides in ocean and ice sheets and—to a lesser extent—in the land-biosphere, while the atmosphere model consumes by far the most computation time. Assuming that the atmosphere is on time-scales of decades and longer in a statistical equilibrium with the underlying fields of ice sheets, sea surface temperature and sea ice, allows to integrate the coupled model with alternating periods of fully coupled mode and periods where the slower components are driven in stand-alone mode with forcing derived from previous fully coupled periods. For the ocean the heat fluxes are modified using a nonlinear anomaly energy balance model. A detailed description of this technique is given in Mik07. The model with prescribed ice sheets has been applied to several studies of the Eemian and Holocene (Schurgers et al. 2006, 2007; Gröger et al. 2007). The full model with interactive ice sheets has been used for several studies of anthropogenic climate change (Mikolajewicz et al. 2007a; Winguth et al. 2005; Schurgers et al. 2008).
The atmosphere is modelled with ECHAM3 (Roeckner et al. 1992) at a horizontal resolution of T21 (approximately 5.6°) and a vertical resolution of 19 layers. The prognostic variables are vorticity, divergence, temperature, humidity, surface pressure and cloud water. The time step is 40 min.
The ocean model LSG2 is an improved version of LSG by Maier-Reimer et al. (1993), with a horizontal resolution of 5.6° in two overlapping grids (64 × 64 points on an Arakawa E grid). Main modifications from the original LSG are the inclusion of a parameterisation accounting for the sub-grid-scale tracer transport due to eddies (Gent et al. 1995) and the use of a second-order total variation diminishing scheme for tracer advection (Sweby 1984). The vertical resolution is 22 vertical layers, with thickness varying with depth from 50 m in the uppermost layer to almost 800 m at 5,600 m depth. The time step of the model is 5 days, but a time step of 1 day is used for the thermodynamics of the surface layer. A simple dynamic sea ice model is included. The coupling time step between atmosphere and ocean is 1 day.
The model’s simulated climate has a cold bias in comparison to observations in high latitudes and a warm bias in low latitudes. Especially in northern hemisphere winter, the model’s climate is 10 K colder over the Arctic. This bias is related to the underestimation of wintertime clouds over the Arctic. In the southern hemisphere, the temperature errors are smaller and the simulated zonal mean surface air temperatures lie within the range spanned by different climatologies. In summer, the simulated precipitation over the Arctic is about twice as strong as in the observational estimates. Near the South Pole, the model overestimates precipitation as well. More details about the climate of the model are given in Mik07.
In “Appendix” the stability of the model in standard perturbation experiments is discussed and compared to the outcome from a model intercomparison project.
2.2 The ice sheet model
The ISM SICOPOLIS used for this study is a 3D thermomechanical (that is, it includes the dependence of the flow of ice on temperature) model that has been applied to several studies of past, present and future climate (Greve 1997, 2000a, b; Greve et al. 1999; Calov and Marsiat 1998; Savviin et al. 2000).
The ISM integrates the time-dependent equations governing ice sheet extent and thickness, ice velocity, temperature, water content and age for any specific grounded ice as a response to external forcing. This is given by surface temperatures, surface mass balance, sea level and geothermal heat flux.
The model equations are subjected to the Shallow Ice Approximation (SIA), which means that they are scaled with respect to the ratio of typical thickness to typical length, and only first-order terms are kept. The SIA yields hydrostatic pressure conditions and ice flow governed by the gradients of pressure and the shear stresses in horizontal planes. Ice shelves cannot be modelled with this approximation and are not included in this ISM.
Isostatic depression and rebound of the lithosphere due to the ice load are modelled by a local lithosphere relaxing asthenosphere model (Le Meur and Huybrechts 1998), which balances the downward ice weight and the upward buoyancy force exerted by the asthenosphere on the elastic lithosphere. To include the viscosity of the asthenosphere a time lag of τ V = 3,000 years is applied.
For the geothermal heat flux, a global mean value of 55 m Wm−2 (Sclatter et al. 1980) has been taken for the northern hemisphere domain. In the southern hemisphere the same value has been used, except for Antarctica, where a 2-D map of geothermal heat flux has been implemented in order to account for the different age of the bedrock in the East and West sections. The values taken are 70 m Wm−2 for the West AIS and 45 mWm−2 for the East AIS (Sclatter et al. 1980). A transition area between these two regions of width 1,000 km is defined parallel to the Transantarctic Mountains in order to assure a smooth transition in the basal thermal forcing.
The model grid is built on a polar stereographic projection. The model domain consists of two sub-domains covering the entire globe except the tropical areas, from approximately 20°N (S) to the North (South) Pole. The horizontal resolution is 80 km. This is a relative coarse resolution compared to the resolution used in other studies of anthropogenic-forced changes of the mass balance of the GrIS or the AIS performed with coupled models (e.g. Huybrechts et al. 2002; Fichefet et al. 2003; Ridley et al. 2005; Driesschaert et al. 2007). In these studies the model components are simpler and/or the domain of the ice sheet model was constrained to the regions of Greenland and/or Antarctica. In our study, however, the domain is quasi-global (tropical regions are excluded for obvious reasons), the atmospheric and ocean components are GCMs, and multiple simulations have been performed with different scenarios of greenhouse gas forcing. This combined with the fact that the ESM has been designed for the performance of long simulations imposes a limit to the resolution of the ice sheet model due to computational cost. The vertical resolution is 21 levels for the ice column, with increasing resolution with depth, and 11 levels for the bedrock directly underneath.
2.3 Coupling of the ISM to the climate model
The atmospheric forcing of the ISM consists of seasonal 2-m atmospheric temperatures and precipitation rates. Annual mean temperatures from CTRL are too cold in Greenland by approximately 7 K, due to underestimation of cloud cover in the Artic region. In the southeast coast temperatures are too warm. The spatial pattern of temperature agrees well with observations, with colder temperatures in the interior. In Antarctica, the pattern of temperature agrees with observations as well, with the coldest temperatures in the interior of East Antarctica and the highest temperatures in the Antarctic Peninsula. Annual mean temperatures are approximately 4 K lower than observations in the interior of East Antarctica and 3 K lower in West Antarctica. Temperatures are overestimated in the area of the big ice shelves of the Ross and Weddell Seas, because ice shelves are not modelled within the ISM. In Greenland, the pattern of precipitation follows the observed pattern of highest precipitation rates in the southeast, decreasing precipitation rates with height, and a minimum over the northwest. The amount of precipitation is overestimated in the Arctic region, and especially in central Greenland due to the smooth representation of the topography. Total precipitation over the ice sheet amounts to 657 × 1012 kg year−1. The figure given in Church et al. (2001) for total accumulation over the GrIS is 520 × 1012 kg year−1. This overestimation of precipitation over ice sheets is a common problem of coarse resolution AGCMs (Ohmura et al. 1996). The pattern of precipitation over Antarctica follows the observed pattern as well: lowest precipitation rates occur over the interior of the Antarctic Plateau in East Antarctica, and higher rates take place in West Antarctica, with maximal rates in the Antarctic Peninsula. The amount of precipitation is overestimated in the interior of Antarctica due to the insufficient representation of the topographic barrier to the penetration of precipitation, as it happens with Greenland.
Due to the bias in the climate simulated over the ice sheets, a correction has been applied for both the 2-m temperatures and precipitation rates. The use of “corrected” forcing is a commonly used—although far from ideal—approach in many projections of the future mass balance of ice sheets (Huybrechts et al. 2002; Ridley et al. 2005; Driesschaert et al. 2007) and has been for more than a decade the standard approach in coupled AOGCMs. Instead of using the raw atmospheric fields, anomalies from the AGCM are superimposed on the present-day climatology taken from ERA40 (Uppala et al. 2005). Precipitation anomalies have been preferred to corrections based in the ratio between reanalysis and model data in order to conserve water in the ESM. These corrections applied on the atmospheric forcing of the ice sheets will be referred to in the following as “flux corrections”.
Air temperatures are assumed to follow a sinusoidal annual cycle with amplitude equal to the difference between the summer and annual temperatures. Additional variations due to the diurnal cycle and changing weather conditions are treated as normally distributed statistical variations with standard deviation σ stat = 5°C (Greve et al. 1999).
Due to the differences in the size of the atmospheric grid (T21) and the ice sheet grid (80 km), a downscaling technique is needed. The atmospheric data is bi-linearly interpolated onto the ice sheet model grid, and height differences are accounted for using the same algorithm as for the “flux corrections”. The time step for the ISM as well as the coupling time step is 1 year.
The ice sheet model provides topography and albedo changes via changes in the glacier mask to the atmosphere and freshwater fluxes to the ocean. As the AGCM can treat land points only either as glaciers or as ice-free land points, an atmospheric model grid point (resolution T21) is defined as glacier point if at least 50% of its area is glaciated according to the interpolated ice-covered area from the ISM grid. Freshwater fluxes from the ice sheets are treated differently in the case of being supplied to the ocean as ice and in the case of being supplied as liquid water. The first case corresponds to iceberg calving and basal melting due to ocean heat supply, the second case to surface melting and basal melting due to geothermal heat fluxes. The distinction is made in order to account for the heat exchange between calved icebergs and floating ice shelves and the ocean when the former melt.
2.4 Initialization of the model and control ice sheets
For the initialization of the ESM, a 10,000-year spin-up simulation has been performed. The flux corrections for the atmospheric forcing of the ISM were calculated during this simulation.
For the initialization of the ISM, two glacial cycles have been simulated with a simple climatic forcing. A time-dependent temperature anomaly from the central Greenland GRIP ice core (Dansgaard et al. 1993) for the northern hemisphere and from the VOSTOK ice core (Jouzel et al. 1993, 1996) for the southern hemisphere has been superimposed on the present-day climatology from ERA40 (Uppala et al. 2005) for the temperature forcing of the ISM.
A 2,250-year-long control simulation (CTRL) corresponding to pre-industrial climate was performed in order to be used as reference simulation for the anthropogenic climate change simulations. The mean simulated concentration of CO2 is 279.5 ppmv.
The simulated northern hemisphere ice sheets from CTRL have an area of 2.15 ± 0.02 × 106 km2. These ice sheets are located mainly on Greenland. Other glaciated areas are simulated in Svalbard, Iceland, Baffin Island, Ellesmere Island and the Rocky Mountains. These glaciated areas are placed at locations, which correspond to observed glaciers and ice caps.
In the southern hemisphere, all simulated glaciated points are placed on Antarctica. The simulated area of the AIS (12.70 ± 0.02 × 106 km2) exceeds by 3% the measured area (Church et al. 2001). Its volume (66.94 ± 0.19 m SLE) is 9% bigger. All numbers given here correspond only to grounded ice, that is, the area and volume of the ice shelves have not been taken into account. The ice excess is placed in West Antarctica and the area of the Amery Ice Shelf (Fig. 1). The grounding line is correctly located in most of the margins. In the area of the Ronne Ice Shelf it is slightly misplaced towards the interior of the continent. The simulated area of grounded ice in the Antarctic Peninsula is more extended than the observed.
In general, the simulated ice sheets are in good agreement with observations. The described discrepancies are due to the combination of several factors, such as the very crude atmospheric forcing imposed during the spin-up, the coarse resolution of the ISM, and the incomplete representation of the processes taking place at the grounding line.
3 Simulations setup and changes in atmosphere and ocean
3.1 Simulations setup
Two different sets of anthropogenic climate change simulations have been performed with the ESM. The first set consists of stabilisation scenarios where the atmospheric concentration of CO2 is increased by 1% per year, until 2×, 3× and 4× pre-industrial levels are achieved, at years 70, 105 and 140, respectively. Those experiments have a length of 1,000 years. A detailed description of carbon cycle interactions with climate in these stabilisation scenario simulations is given in Winguth et al. (2005). For the second set of experiments, carbon emissions instead of CO2 concentrations have been prescribed, as follows: for the calendar years 1750–2000 historical emissions (Marland et al. 2005; Houghton and Hackler 2002) have been used, followed by several IPCC emission scenarios for 2000–2100 (Nakicenovic et al. 2001), and an exponential decay of emissions (with a time constant of 150 years) for subsequent years. Other greenhouse gases like methane and aerosols have not been taken into account. The outcome of these experiments has been described in Mik07. The present paper will focus on the output of the stabilisation simulations and two multi-millennia simulations for the emission scenario A2.
For each scenario, two simulations have been performed: one where the feedbacks from the ice sheets are passed to the other components (two-way coupling) and another one where those feedbacks are ignored (one-way coupling). The former simulations will be referred to as 2×, 3×, 4× and A2, and the later as 2×_1w, 3×_1w, 4×_1w and A2_1w, with *_1w meaning one-way coupling. The comparison between those two sets allows the identification of climatic changes induced by the ice sheets.
An additional simulation CTRL_NOGr was performed in order to investigate whether the GrIS is bi-stable and the climatic impact of the disappearance of the GrIS in a pre-industrial climate. The setup of this simulation is identical to CTRL, except for the initial GrIS, which is reduced to a small cap in the southern part of Greenland.
List of experiments
Pre-industrial concentration of CO2 (prognostic)
1% increase of CO2 until stabilisation at 2× CO2
1% increase of CO2 until stabilisation at 3× CO2
1% increase of CO2 until stabilisation at 4× CO2
One-way coupling: feedbacks ice sheets-climate not included
As 3×, with fixed topography: the simulation does not allow for the height albedo feedback or changes in the ice flux
As 3×, with fixed advection: velocities are prescribed as in CTRL
Calendar years 1750–9000
Carbon emissions prescribed according to: (1750–2000) historic, (2001–2100) IPCC SRES A2, (2100–9000) exponential decay with a time constant of 150 years
Calendar years 1750–7000 (for 7000–9000 ice sheet model run offline with atmospheric forcing from 6000–7000)
As A2, but changes in freshwater fluxes, glacier mask and topography from ice sheet model not passed to other components
As CTRL, without GrIS as initial condition
3.2 Climate change
The rate of global temperature increase is strongest when the atmospheric CO2 concentration is still rising. After stabilisation, global temperatures continue rising due to the delay associated with the storage of heat by the ocean (Voss and Mikolajewicz, 2001). In the 4× simulation a period of slight global mean temperature decrease occurs between years 200 and 400. This is, as it will be shown in the next paragraph, related to significant changes in global ocean circulation.
Reduced formation of North Atlantic Deep Water (NADW) and a decline in the strength of the meridional overturning circulation take place in all simulations (Fig. 2b). In 2× and 3× this reduction in the strength of the NAMOC is modest and it reaches values only slightly lower than those of the control simulation after a century from the time of maximum reduction at year 150 in 2× and at years 300–400 in 3×. In 4×, on the contrary, the strength of the NAMOC is steadily decreasing with time, until a state without a NADW cell is reached by year 400. It does not recover by the end of the simulation. The cause of this reduction in the NAMOC and the role played in it by the ice sheets is analysed in Sect. 5 of this paper.
The hydrological cycle is enhanced in all greenhouse simulations. In the North Atlantic, smaller anomalies/reduction of precipitation are seen associated with the low warming/net cooling signal in this region. The model shows a mean increase of global mean precipitation of 5.1% for a CO2 doubling (estimated from the ESM), compared to an average of 6.6% from the IPCC 2001 report (Houghton et al. 2001).
4 Evolution of global ice sheets and sea level
4.1 Evolution of the Greenland ice sheet
By the end of the simulations, 4× shows a different 2D pattern of thickness changes than 2× and 3× (Fig. 6), with the middle part of the GrIS showing lower thickness than in CTRL due to reduced accumulation rates (Fig. 7), increased thickness in the southern margins due to reduced melting, and moderate thickness reduction in other margins due to a lesser increase in surface melting than in 3×. The reduction with respect to CTRL in the snowfall rates in the middle part of the ice sheet is linked to the precipitation anomaly associated with the cold anomaly in the North Atlantic. The maximum reduction of snowfall takes place on the east coast. Consistent with the temperature forcing described before, the southern part of the GrIS has lower melting rates than in CTRL. In the rest of the low elevation areas, melting rates are higher than in CTRL but lower than in 3×.
4.1.1 The role of ice sheet dynamics
Dynamical changes can play an important role in the mass balance of ice sheets on different time-scales. Here we will investigate two processes which can potentially contribute significantly to the total mass budget on the time scales which are the subject of this study: the height-feedback and changes in the horizontal transport of ice. The height-feedback is a positive feedback for surface melting: since the height of an area of an ice sheet is lowered due to surface melting, the ice-surface temperatures increase due to the reduced height. The increase in temperatures lead to increased melting rates. The changes in the horizontal transport of ice are caused by changes in topographic gradients, triggered by changes in the surface mass balance, and changes in the thickness of the column of ice.
Modelling studies of changes in the mass balance of ice sheets not including a dynamic ice sheet, as calculations from snowpack models and calculations directly from the output of GCMs (Bugnion and Stone 2002; Wild et al. 2003; Wild and Ohmura 2000; Ohmura et al. 1996), do not include any of the two processes described above. In order to evaluate the importance of these processes to the total mass budget, two off-line simulations 3×_FIXTOP and 3×_FIXADV have been performed. In 3×_FIXTOP, the topography of the ice sheet is kept unchanged, as in the modelling approaches not including a dynamic ice sheet. Effects of the height-feedback and changes in horizontal transport are neglected. This approach could easily be adopted in AOGCMs by calculating the integrated mass balance at each grid point and adding the thickness change to the topography of the control integration. This could then—with some thresholds to estimate the conversion to ice-free points—be used to estimate time dependent changes in surface topography without using a dynamic ice sheet model. In 3×_FIXADV (FIXADV stands for fixed advection), the topography of the ice sheet is allowed to change, but the transport at each grid point is kept constant throughout the simulation. Hence, only the height-feedback is included and not the effect of changes in the ice flux. 3× is taken as reference simulation, since it is the stabilisation scenario simulation where strongest changes occur to the GrIS.
The feedback associated with changes in the horizontal transport changes its sign along the process of decay of the GrIS. This decay is faster in the simulation with fixed fluxes 3×_FIXADV than in 3× until the ice sheet decays by approximately 8% of its original value. The initial lowering of the topography in the low marginal areas of the ice sheet in 3× triggers an increase on the horizontal flux of ice due to increased topographic gradients. This enhanced flux brings extra ice to the low elevation areas, increasing their topographic height. This, via the height-feedback, lowers the melting rates at the margins and acts as a negative feedback for the decay of the GrIS. The high-elevated areas originally in the accumulation area are lowered by this process of increased horizontal ice flux towards the margins of the ice sheet. The lowering of their topography is not sufficient, though, to bring a significant fraction of these regions to the ablation area. Later in the process of ice sheet decay, the reduction of height in these areas is sufficient to cause some summer melting in part of them. This additional surface melting from areas that initially did not experience any summer melting causes an increase in the total surface melting over the ice sheet. This changes the sign of the feedback caused by changes in the horizontal transport of ice to positive. By the end of the simulation, dynamical changes increase the mass loss of the GrIS by approximately 20%.
4.2 Evolution of the Antarctic ice sheet
The volume of the AIS undergoes an increase approximately linear with time in all the greenhouse simulations. By the end of the simulations, the additional volume of water stored in the ice sheet is equivalent to a sea level drop of 50 cm in 2×, 1 m in 3×, and 1.2 m in 4× (Fig. 5b). This shows a larger storage of water in the simulations with larger increases in the concentration of greenhouse gases. The model does not simulate any abrupt change in the evolution of the AIS in any of the simulations.
The surface melting increases only in the Antarctic Peninsula and in some of the low elevated marginal areas of the ice sheet. Changes in the horizontal transport of ice are responsible for minor differences between local surface mass balance and total (vertically integrated) mass balance. No significant changes in the horizontal transport occur for the high elevation areas of the East Antarctic Plateau.
The contribution to the mass balance of changes in the ice dynamics at the margin of the ice sheet cannot be evaluated with our ISM. Ice shelves, ice streams, and outlet glaciers are not included. Therefore the above analysis lacks important components of the mass balance, which could be dominant over the next centuries. More details about the limitations of our analysis are given in Sect. 6.
4.3 Sea level changes
The increased storage of water in the AIS either compensates or exceeds the contribution to eustatic sea level rise from the GrIS. By year 1000, the contribution from the ice sheets to sea level changes is close to zero in the simulation 3×, while it is a net lowering of the sea level in 2× and 4× (Fig. 5a, b). The contribution to sea level by year 1000 from oceanic thermal expansion (Fig. 5c) is 0.6 m (2×), 1.25 m (3×) and 1.5 m (4×). The maximum rate of thermal expansion occurs by year 300 in the case of 4×, and is lower afterwards. The expansion has not reached an equilibrium by the end of the simulations. Maximum net sea level rise (+1.1 m, Fig. 5d) takes place at year 1000 in 3×, and year 600 in 4× (+0.9 m). This is due to the approximately linear increase of AIS volume with time, while the thermal expansion term slowly approaches an equilibrium value, and to the low contribution of GrIS to sea level rise in comparison with 3×.
5 Feedbacks between ice sheets and climate
5.1 Feedbacks between ice sheets and ocean circulation
There are no substantial differences in the strength of the NAMOC between the simulations including the extra meltwater from the ice sheets and the simulations with meltwater fluxes as in CTRL (Fig. 2b). Thus the GrIS does not play a major role in simulated stratification changes. Instead, increased atmospheric moisture transport into the North Atlantic drainage basin as well as ocean surface warming cause the weakening of the NAMOC.
The mass balance of the GrIS in these stabilisation scenario simulations has been shown to be highly sensitive to the climatic effects of changes in the NAMOC, the main effect being a drastic reduction of melting rates when significant weakening of the NAMOC occurs. This effect was shown to be particularly strong in the case of a complete collapse of Atlantic overturning: the contribution to sea level changes from the GrIS in 4×, where the overturning collapses, is much lower than in 3×, where the overturning weakens but recovers afterwards. Thus, in 4×, the regional climate signal associated with a collapse of the NAMOC is dominant over the stronger global warming signal associated with higher CO2 forcing and the collapse of NADW formation leads to reduced melting especially over the southern half of the GrIS.
5.2 Feedbacks between ice sheets and atmosphere
Changes in the area and volume of the GrIS can affect the atmosphere by three main mechanisms: thermodynamic near-surface heating/cooling due to change of topographic height (discussed in Sect. 4.1.1), modification of the general circulation of the atmosphere caused by topographic and thermal contrast change, and changes of surface albedo due to changes in ice sheet area and presence of surface meltwater.
Comparison of 2×, 3× and 4× with the respective simulations *_1w shows no differences in the global mean temperature (Fig. 2a). Thus the ice sheet–atmosphere feedbacks do not have a global impact on the simulated climate.
In A2, a period with strong melting rates occurs between the years 5000 and 5500. During this time, several points of the ice sheet deglaciate on the atmospheric grid. The marginal points of the ice sheet model close to those atmospheric grid points experience increased melting rates due to the warming associated with the changes in surface albedo. By 5500, the area of the ice sheet is reduced to 1/2 of the original area. Most of the northern half of Greenland is ice-free, except the central-east margin, where the bedrock is highest. By 7000, the ice sheet is reduced to 1/5 of its original volume. The small ice sheet is located at the southern tip and on the southeast coast. There the summer temperatures are lower than in the control simulation, due to the regional climate change associated with a collapsed NAMOC. The topography of the ice sheet at the southern tip of Greenland has remained almost unchanged since the beginning of the simulation, with its dome being always higher than 2,500 m. In A2_1w, the area has fallen to only 62% of the initial extent and the volume to 52% by year 7000.
5.2.1 Irreversibiliy of the deglaciation of Greenland
6 Summary and conclusions
This study is one of the first investigating the interactions of ice sheets and climate with a coupled Earth System Model (ESM) consisting of General Circulation Models (GCMs). Our approach permits us to address the long-term effects of freshwater fluxes on ocean circulation and ice sheet topography and albedo changes on the atmosphere. Also, we can evaluate whether these changes in the climate system driven by the ice sheets are important for ice sheet mass balance and therefore whether these feedbacks should be taken into account in studies of ice sheet evolution.
Our results show a relatively small loss of mass from the GrIS compared with other studies. For instance, Alley et al. (2005) obtained sea level rises of 1, 2, and 3 m after 1,000-year simulations with the ice sheet model of Huybrechts and de Wolde (1999) forced with a uniform increase in mean annual temperature obtained from an average of seven IPCC models under scenarios where atmospheric CO2 stabilised at 550, 750 and 1,000 ppmv, respectively. Ridley et al. (2005) obtained an almost complete elimination of the GrIS within 3,000 years under constant 4× CO2 forcing. After the first 1,000 years, the ice sheet decays to 40% of the initial volume. The smaller loss of mass by the GrIS in our study is due to the lower climate sensitivity of this ESM and to the regional effects over the GrIS of a weakened/collapsed NAMOC.
Since our meltwater fluxes from the GrIS are one order of magnitude smaller than atmospheric moisture transport anomalies, they do not play a major role in changes in the NAMOC. Huybrechts et al. (2002) coupled a 3D ice sheet model of the GrIS to the AOGCM LMD5.3-CLIO at a coarse resolution. The changes in the geometry of the GrIS were not passed, however, to the atmospheric component. They applied the SRES B2 scenario to this model and found a GrIS contribution of 4 cm to sea level rise by 2100. About 0.03 Sv of additional freshwater fluxes enters the North Atlantic. They did not find significant changes in the patterns of climate change over the North Atlantic region compared with a climate-change run without Greenland freshwater feedback. Ridley et al. (2005) found a small effect of Greenland melting on the NAMOC. In their model, the freshwater fluxes from Greenland peak at 0.06 Sv, causing an additional 1–2 Sv decrease in the strength of the overturning. By contrast, Fichefet et al. (2003) report a strong influence of the GrIS meltwater discharge on the NAMOC until the year 2100, although the simulated freshwater fluxes (0.015 Sv) are comparable to the results of this study. However, their control simulation showed a considerable drift towards a weaker NAMOC. Therefore their results could be affected by model drift or their model could be close to a bifurcation point of the NAMOC, at which even very small perturbations could have a strong effect. In Driesschaert et al. (2007) the freshwater flow from the melting GrIS remains above 0.1 Sv for 3 centuries for the most extreme forcing scenario selected for their study, but only a slight decline of the NAMOC is simulated. Swingedouw et al. (2006) also report a strong contribution of ice sheet melting to the reduction of the NAMOC in their simulations, where the pCO2 concentration is prescribed to increase by 1% per year. Ice sheet melting increases from 0 to 0.2 Sv from year 0 to year 140. In the 2× CO2 simulation of Swingedouw et al. (2007), the contribution of ice sheet melting brings the NAMOC to a collapse. In the study of Mikolajewicz et al. (2007b), the effect from meltwater fluxes from ice sheets is negligible in the phase of initial weakening of the NAMOC, but turns to be more important during the recovery in subsequent centuries. In their 2× CO2 simulation, ice sheet melting contributes by 0.025–0.03 Sv to total freshwater anomalies, and hinders the recovery of the NAMOC.
Here, changes in the mass balance of the GrIS have been shown to be highly influenced by regional climate change associated with a weakened NAMOC. A substantial reduction of melting rates accompanies the collapse of the NAMOC in a simulation where the greenhouse gas concentrations have been set to four times pre-industrial levels. The contribution of the GrIS to sea level changes is much lower than the contribution of the ice sheet in a simulation with three times pre-industrial levels. The reduction of melting rates over the GrIS as a consequence of regional climate changes caused by a weakening of the NAMOC could act as a stabilising mechanism for the strength of the overturning in the case of a dominant role for freshwater fluxes from the GrIS in weakening the NAMOC, as it reduces the freshwater input into the North Atlantic and this favouring again stronger NADW formation. However, this is not the only mechanism at work and in our model the changes in atmospheric moisture transports dominate over this ice sheet feedback.
In common with Huybrechts and de Wolde (1999) and Huybrechts et al. (2002), changes in ice sheet dynamics are found to act as a mechanism accelerating the decay of the GrIS. Changes in the horizontal transport caused by changes in the topography, however, act to reduce the integrated surface melting in the first stages of decay of the ice sheet, until the increased transport lowers a sufficient area in the interior of the ice sheet to a height with warmer surface temperatures permitting summer surface melting.
According to our results, the positive atmospheric feedbacks associated with the decay of the GrIS are not relevant for the mass balance of the ice sheet until sufficient changes in the topography and area of the ice sheet have taken place. This threshold was found to be at 3/4 of the original volume and area of the ice sheet. Accelerated GrIS decay is due to increased surface melting, and seems to be caused by enhanced warming due to a reduction in summer albedo. However, an additional simulation separating the albedo contribution to the total changes in the climate of Greenland from other effects would be needed to quantify more rigorously the role played by albedo. In Ridley et al. (2005), the threshold was found to be at 2/3 of the original volume. They found that the processes associated with the decay of the ice sheet act as a negative feedback for its decay. This reduction of the decay rate takes place via the development of atmospheric convection due to the strong thermal contrast in their model between the deglaciated margins and the glaciated areas. They suggest that the development of these circulation cells may be a function of GCM resolution. These convective cells do not develop in our ESM, where the resolution is lower and where the thermal contrast between deglaciated and glaciated model points is substantially weaker than in the results of Ridley et al. (2005).
In our model, the GrIS cannot regrow after deglaciation if atmospheric CO2 levels would return to pre-industrial levels. Therefore, the GrIS is, according to our results and in agreement with Crowley and Baum (1995) and Calov et al. (2005), bi-stable.
Regarding the AIS, our results show a net gain of mass of the ice sheet for all the stabilisation scenarios. Church et al. (2001) predict a continuous growth of the AIS, which compensates enhanced melting rates from the GrIS, up to the year 2100. Simulations with the 3D ice sheet model of Huybrechts and de Wolde (1999) passively coupled to a 2D climate model show a moderate and constant lowering of sea level, reaching 30 and 50 cm by the (calendar) year 3000 for doubling and quadrupling of CO2 simulations, respectively. For the 8× CO2 scenario, the contribution of the AIS to sea level rise is positive, due to increased surface melting in the margins and increased basal melting in the ice shelves caused by warmer ocean temperatures. Ice shelves are not properly modelled in the ice sheet model used in our study. Therefore a potential retreat of the grounding line in the West AIS caused by the thinning of ice shelves due to enhanced ocean heat fluxes cannot be simulated. Warner and Budd (1998) and Huybrechts and de Wolde (1999) showed that grounding line retreat along the ice shelves could happen for basal melting rates >5–10 m year−1, destroying the West Antarctic Ice Shelves after a few centuries. Recent measurements show an increase in bottom melting near the AIS grounding lines (Rignot and Jacobs 2002).
The main limitations of this study are associated with the use of the Shallow Ice Approximation (SIA), the coarse resolution of the ice sheet and atmospheric models, and the use of flux corrections at the interface atmosphere-ice sheet.
Most ice sheet models are based on the SIA (Hutter 1983), which is valid for an ice mass with a small aspect ratio (vertical dimension ≪ horizontal dimension). The SIA is not a good approximation at all places in the ice sheet, such as at the ice divide or near the ice margin (Baral et al. 2001). The dynamics of ice streams, outlet glaciers and ice shelves cannot be modelled with our ice sheet model. Recent measurements of acceleration of ice streams and outlet glaciers in Greenland (Rignot and Kanagaratnam 2006; Joughin et al. 2004) and Antarctica (Scambos et al. 2004; Shepherd et al. 2002; Joughin et al. 2003) reveal that rapid dynamic changes can be important, contributing a substantial fraction of the ongoing sea level rise and potentially becoming dominant over ice sheet mass balance changes in the future. Models including the full set of physical processes implicated in the ongoing changes are needed to assess if these ongoing changes represent minor perturbations before stabilisation or a major change affecting sea level substantially (Alley et al. 2005).
The resolution of the atmospheric model is critical for an accurate modelling of precipitation over the ice sheets (Ohmura et al. 1996). The lower the resolution, the higher the overestimation of precipitation due to reduced orographic forcing. Our relatively high growth rates for the AIS are influenced by overestimation of precipitation associated with the coarse resolution and by the use of absolute precipitation anomalies instead of rate anomalies in the flux correction, a choice taken for mass conservation purposes. Therefore our quantitative results for the AIS and its effect on sea level should be interpreted with caution. The effect of precipitation increase on the total mass balance of the GrIS is also probably overestimated due to the same reasons. Besides, higher resolution in the ice sheet model would resolve better the width of the ablation area (Wild et al. 2003), because the margins of ice sheets are very steep.
Another critical issue is certainly the use of anomalies rather than using directly the forcing data from the AGCM, which might distort bifurcation diagrams. First simulations from a new ESM indicate that it seems to be possible to obtain a reasonable ice sheet climate without flux correction if a higher resolution AOGCM with improved physical parameterisations in the atmospheric component is used (Vizcaíno 2006; Mikolajewicz et al. 2007b).
From our results we conclude that ice sheets are an active component of the climate system, with changes in their mass balance being able to modify substantially their own climate and the climate of other regions. Therefore we encourage their inclusion as dynamical components of climate models, in order to obtain more accurate projections of sea level changes and patterns of climate change, and in order to improve our understanding of the feedbacks operating between them and the rest of components of the climate system.
This work was performed within the project CLIMCYC, funded by the DEKLIM program of the German Ministry of Education and Research. Arne Winguth is supported by NASA grant NAG5-11245 and the UW Graduate School Research Funds. The simulations have been performed at the “Deutsches Klimarechenzentrum”. Prof. Ralf Greve is thanked for his support with the ice sheet model and Dr. Hugo Lambert for his advice on the grammar and style of the manuscript. Two anonymous reviewers are thanked for their constructive comments to the manuscript. Dr. Sven Kotlarski, Dr. Helmut Haak and Dr. Felix Landerer contributed with helpful remarks to the first version of the manuscript.
- Braithwaite RJ, Olensen OB (1989) Calculation of glacier ablation from air temperature, West Greenland. In: Oerlemans J (ed) Glacier fluctuations and climatic change. Kluwer, Dordrecht, pp 219–233Google Scholar
- Budd WF, Smith IN (1979) The growth and retreat of ice sheets in response to orbital radiation changes. In: Sea level, ice, and climatic change, Proceedings of the Canberra symposium, IAHS Publ. 131Google Scholar
- Calov R (1994) Das thermomechanische Verhalten des groenlaendischen Eisschildes unter der Wirkung verschiedener Klimaszenarien - Antworten eines theoretisch-numerischen Modells. Ph.D. thesis, stitut fuer Mechanik, Technische Hochschule Darmstadt, GermanyGoogle Scholar
- Calov R, Marsiat I (1998) Simulations of the northern hemisphere through the last glacial–interglacial cycle with a vertically integrated and a three-dimensional thermomechanical ice sheet model coupled to a climate model. Ann Glaciol 27:169–176Google Scholar
- Calov R, Savvin AA, Greve R, Hansen I, Hutter K (1998) Simulation of the Antarctic ice sheet with a three-dimensional polythermal ice-sheet model, in support of the EPICA project. Ann Glaciol 27:201–206Google Scholar
- Church JA et al (2001) Changes in sea level. In: Houghton JT, Ding Y, Griggs DJ, Noguer M, Van Der Linden PJ, Dai X, Maskell K, Johnson CA (eds) Climate change 2001: the scientific basis. Cambridge University Press, Cambridge, pp 640–693Google Scholar
- ETOPO5 (1988) Data announcement 88-mgg-02, digital relief of the surface of the earth. Technical report, NOAA, National Geophysical Data Center, Boulder, Colorado, USAGoogle Scholar
- Driesschaert E, Fichefet T, Goosse H, Huybrechts P, Janssens I, Mouchet A, Munhoven G, Brovkin V, Weber SL (2007) Modeling the influence of Greenland ice sheet melting on the Atlantic meridional overturning circulation during the next millennia. Geophys Res Lett 34(10):L10707. doi: 10.1029/2007GL029516,2007 CrossRefGoogle Scholar
- Gallee H, Ypersele JP, Fichefet T, Marsiat I, Tricot C, Berger A (1992) Simulation of the Last Glacial Cycle by a coupled, sectorially averaged climate-ice sheet model:2. Response to insolation and CO2 variations. J Geophys Res 97(D14):15,713–15,740Google Scholar
- Gent PR, Willebrand J, McDougall T, McWilliams JC (1995) Parameterizing eddy-induced tracer transports in ocean circulation models. J Phys Oceanogr 19:2962–2970Google Scholar
- Gregory JM, Dixon KW, Stouffer RJ, Weaver AJ, Driesschaert E, Eby M, Fichefet T, Hasumi H, Hu A, Jungclaus JH, Kamenkovich IV, Levermann A, Montoya M, Murakami S, Nawrath S, Oka A, Sokolov AP, Thorpe RB (2005) A model intercomparison of changes in the Atlantic thermohaline circulation in response to increasing atmospheric CO2 concentration. Geophys Res Lett L12703 32 (12):L12703. doi: 10.1029/2005GL0232091 CrossRefGoogle Scholar
- Greve R (1995) Thermomechanisches Verhalten polythermer Eisschilde - Theorie, Analytik, Numerik. Ph.D. thesis, Institut fuer Mechanik, Technische Hochschule Darmstadt, GermanyGoogle Scholar
- Greve R (2000b) Paleoclimatic evolution and present conditions of the Greenland ice sheet in the vicinity of Summit: and approach by large-scale modelling. Paleoclimates 2:133–161Google Scholar
- Gröger M, Mikolajewicz U, Maier-Reimer E, Schurgers G, Vizcaíno M, Winguth A (2007) Changes in the hydrological cycle, ocean circulation and carbon/nutrient cycling during the last interglacial. Paleoceanography 22(4):PA4205Google Scholar
- Houghton RA, Hackler JL (2002) Carbon flux to the atmosphere from land-use changes. In: Trends: a compendium of data on global change. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, TN, USAGoogle Scholar
- Houghton J, Ding Y, Griggs D, Noguer M, Van der Linden P, Dai X, Maskell K, Johnson CA (2001) Climate change 2001: the scientific basis. Cambridge University Press, CambridgeGoogle Scholar
- Hutter K (1983) Theoretical glaciology. Kluwer, NorwellGoogle Scholar
- Imbrie J, Martinson J, McIntyre D, Mix A, Morley J, Pisias N, Prell W, Shackleton N (1984) The orbital theory of Pleistocene climate: support from a revised chronology of the marine delta 018 record. In: Berger A (ed) Milankovich and climate, part I, vol 126. D. Reidel Publishing Company, DordrechtGoogle Scholar
- Jouzel J, Barkov NI, Barnola JM, Bender M, Chappelaz J, Genthon C, Kotlyakov VM, Lipenkov V, Lorius C, Petit JR, Raynaud D, Raisbeck G, Ritz C, Sowers T, Stievenard M, Yiou F, Yiou P (1993) Extending the Vostok ice-core record of paleoclimate to the penultimate glacial period. Nature 364:407–412CrossRefGoogle Scholar
- Maier-Reimer E, Mikolajewicz U (1989) Experiments with an OGCM on the cause of the Younger Dryas. In: Oceanography 1988 (pp 87–100). UNAM Press, Mexico, 208 ppGoogle Scholar
- Marland G, Boden T, Andres R (2005) Global, regional, and national fossil fuel CO2 emissions. In: Trends: a compendium of data on global change. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, US Department of Energy, Oak Ridge, TN, USAGoogle Scholar
- Marsiat I (1994) Simulation of the northern hemisphere continental ice sheets over the last glacial–interglacial cycle: experiments with a latitude–longitude vertically integrated ice sheet model coupled to a zonally averaged climate model. Paleoclimates I:59–98Google Scholar
- Meehl GA, Stocker TF, Collins WD, Friedlingstein P, Gaye AT, Gregory JM, Kitoh A, Knutti R, Murphy JM, Noda A, Raper SCB, Watterson IG, Weaver AJ, Zhao Z-C (2007) Global climate projections. In: Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller HL (eds) Climate Change 2007: the physical science basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change.Cambridge University Press, CambridgeGoogle Scholar
- Nakicenovic N, Alcamo J, Davis G, De Vries B, Fenhann J, Gaffin S, Gregory K, Grübler A, Jung TY, Kram T, Lebre La Rovere E, Michaelis L, Mori S, Morita T, Pepper W, Pitcher H, Price L, Riahi K, Roehrl A, Rogner H, Sankovski A, Schlesinger M, Shukla P, Smith S, Swart R, Van Rooijen S, Victor N, Dadi Z (2001) Special report on emissions scenarios. In: A special report of Working Group III of the Intergovernmental Panel on Climate Change. Cambridge University Press, CambridgeGoogle Scholar
- Petoukhov V, Claussen M, Berger A, Crucifix M, Eby M, Eliseev AV, Fichefet T, Ganopolski A, Goosse H, Kamenkovich I, Mokhov II, Montoya M, Mysak LA, Sokolov L, Stone P, Wang Z, Weaver AJ (2005) EMIC Intercomparison Project (EMIP): comparative analysis of EMIC simulations of climate, and of equilibrium and transient responses to atmospheric CO2 doubling. Clim Dyn 25:363–385CrossRefGoogle Scholar
- Reeh N (1991) Parameterization of melt rate and surface temperature on the Greenland ice sheet. Polarforsch 59(3):113–228Google Scholar
- Roeckner E, Arpe K, Bengtsson L, Brinkop S, Duemenil L, Esch M, Kirk E, Lunkheit F, Ponater M, Rockel B, Sausen R, Schlese U, Schubert S, Windelband M (1992) Simulation of the present-day climate with the ECHAM model: impact of the model physics and resolution. Max-Planck-Institut für Meteorologie, Hamburg, Report no. 93Google Scholar
- Schurgers G, Mikolajewicz U, Gröger M, Maier-Reimer E, Vizcaíno M, Winguth A (2008) Long-term effects of biogeophysical and biogeochemical interactions between terrestrial biosphere and climate under anthropogenic climate change. Accepted for publication in a special issue of Global and Planetary ChangeGoogle Scholar
- Sitch S, Smith B, Prentice I, Arneth A, Bondeau A, Cramer W, Kaplan J, Levis S, Lucht W, Sykes M, Thonicke K, Venevsky S (2003) Evaluation of ecosystem dynamics, plant geography and terrestrial carbon cycling in the LPJ dynamic global vegetation model. Global Change Biol 9:161–185CrossRefGoogle Scholar
- Stommel HH (1961) Thermohaline convection with two stable regimes of flow. Tellus 13:224–230Google Scholar
- Stouffer RJ, Yin J, Gregory JM, Dixon KW, Spelman MJ, Hurlin W, Weaver AJ, Eby M, Flato GM, Hasumi H, Hu A, Jungclaus JH, Kamenkovich IV, Levermann A, Montoya M, Murakami S, Nawrath S, Oka A, Peltier WR, Robitaille DY, Sokolov A, Vettoretti G, Weber SL (2006) Investigating the causes of the response of the thermohaline circulation to past and future climate changes. J Clim 19:1365–1387CrossRefGoogle Scholar
- Sweby PK (1984) High-resolution schemes using flux limiters for hyperbolic conservation-laws. SIAM J Num Anal 21:1995–1011Google Scholar
- Uppala SM, Kallberg PW, Simmons AJ, Andrae U, da Costa Bechtold V, Fiorino M, Gibson J, Haseler J, Hernandez A, Kelly G, Li X, Onogi K, Saarinen S, Sokka N, Allan R, Andersson E, Arpe K, Balmaseda M, Beljaars A, van de Berg L, Bidlot J, Bormann N, Caires S, Chevallier F, Dethof A, Dragosavac M, Fisher M, Fuentes M, Hagemann S, Holm E, Hoskins B, Isaksen L, Janssen P, Jenne R, McNally A, Mahfouf JF, Morcrette JJ, Rayner N, Saunders R, Simon P, Sterl A, Trenberth K, Untch A, Vasiljevic D, Viterbo P, Woollen J (2005) The ERA-40 re-analysis. Q J R Meteor Soc 131:2961–3012CrossRefGoogle Scholar
- Vizcaíno M (2006) Long-term interactions between ice sheets and climate under anthropogenic greenhouse forcing. Simulations with two complex earth system models. Berichte zur Erdsystemforschung 30. http://www.mpimet.mpg.de/fileadmin/publikationen/Reports/BzE_30.pdf
- Voss R, Sausen R (1996) Techniques for asynchronous and periodically synchronous coupling of atmosphere and ocean models. Part II: impact of variability. Clim Dyn 12:605–614Google Scholar
- Warner R, Budd W (1998) Modelling the long term response of the Antarctic ice sheet to global warming. Ann Glaciol 27:161Google Scholar