Nested modelling of the Greenland ice sheet in support of the dating and the interpretation of the NEEM ice core record (NEEM-B)

Project Details


1. Background and importance of the project Starting with the initial projects in the 1960s (Camp Century and Byrd), deep ice cores have become a crucial pillar of knowledge about Late Quaternary palaeoclimatic changes. The current state of the art is represented in Greenland by the detailed records drilled at Summit (GRIP - Greenland Ice core Project - and GISP2 - Greenland Ice Sheet Project 2) and at NGRIP (NorthGRIP), cf. Dansgaard et al. (1993), NGRIP project members (2004). The most compelling message from the Greenland cores has been that of the very abrupt, millennial-scale, climatic flips of the last glacial period, known as Dansgaard-Oeschger events. Understanding the cause of these events, and their implications for future change, has become one of the hottest topics in climate studies, with significant policy implications. Despite their great significance, the existing Greenland cores are deficient in one very important respect. The last interglacial between 115 to 130 kyears before present (also known as the Eemian) has proved to be a tantalising target: Eemian ice is present but highly garbled in the Summit cores, incomplete due to basal melting in the NGRIP core, and too compressed to use in the Camp Century core. The last interglacial period is however critical for understanding climate change, because it offers an analogue to the current period of warmth in which there was unequivocally only natural forcing. Climate records in marine and terrestrial sediments suggest that the Eemian in Greenland may have been up to 5°C warmer than today, and therefore this period allows us to investigate what might occur in a climate like the one we are approaching. In particular, model studies suggest that the Greenland ice sheet will waste away under such warmer conditions at a rate depending on the amplitude and duration of the warming (Huybrechts et al., 1991; Huybrechts and de Wolde, 1999; Cuffey and Marshall, 2000; Greve, 2000; Huybrechts, 2002; Gregory et al., 2004; Otto-Bliesner et al., 2006). The new North Greenland Eemian Ice Drilling (NEEM) site (77.5°N, 50.9°W, 2650 m a.s.l) has been selected through analysis of available surface elevation data, ice thickness and ice radar data as the most promising site on the Greenland ice sheet for obtaining an undisturbed ice core record covering the Eemian period as well as the previous glacial. The NEEM site is located about 300 km north of NGRIP along the ice divide in an area with rather low accumulation rates above a relatively flat bedrock surface. The ice thickness has been established by radar at 2542 m. Preliminary inverse modeling indicates that the Eemian period will be found at a depth between 2265 and 2345 m at an average annual layer thickness of 7 mm, enough for meaningful climate studies (Buchardt and Dahl-Jensen, 2008). The NEEM record is therefore expected to chart the full course of an interglacial from termination to inception at very high resolution in numerous parameters. Additionally, it will confirm whether Greenland was indeed significantly warmer than at present and whether any rapid climate changes occurred in such a warmer climate. This will allow for an improved assessment of the state of the Greenland ice sheet under such a warmer climate. It is also expected to bring to light whether Dansgaard-Oeschger events occurred in the previous glacial period. From the NEEM findings, it will be possible to relate climate variations from the current and the last interglacial period to predicted global warming scenarios. A correct interpretation of the ice core record requires that two basic problems are adequately solved. First of all, the key problem of any climate reconstruction is accurate dating. The second problem which needs to be addressed results from the dynamics of the Greenland ice sheet itself during the entire time period covered by the ice core. Most importantly, the surface elevation at the time of deposition of the ice has varied under the influence of changes in accumulation rate, ice temperature, ice sheet extent, horizontal ice flow, and possibly other factors (Huybrechts, 2002). Such elevation changes cause non-climatic biases in the temperature records retrieved from the ice core which need to be properly assessed. Moreover, the NEEM site is situated in a flank position along the axis of a gently sloping ridge. As a result, the ice in the core was not deposited locally but came from an upstream location. Deeper ice was deposited further inland at a progressively higher, and therefore colder, location. In addition, during the warm phase of the Eemian the ice sheet geometry around NEEM may have been very different to today. On the basis of currently available model simulations of the Greenland ice sheet (e.g. Cuffey and Marshall, 2000) it is very unlikely that the current geometry and elevation of summits and ice divides was preserved during the Eemian. These constraints are likely to complicate the interpretation of the ice core record and need to be thoroughly assessed. This calls for ice-dynamic modeling supported by geophysical observations from both the vicinity of the ice cores as from the ice core itself. 2. Design and methodology Our intended approach to obtaining the chronology and non-climatic biases is to accurately model the ice sheet history and its flow dynamics for at least the full time span covered by the ice core. This will be done over the entire region where ice particles ending up at NEEM are believed to have originated. We will nest a high-resolution higher-order ice dynamic flow model for the NEEM area into a comprehensive 3D thermomechanical model of the whole Greenland ice sheet. The reconstructed high-resolution 3-D velocity field from a forward experiment with the nested model will subsequently be used in a Lagrangian backtracing algorithm to establish the trajectories of ice particles back to their respective time and places of deposition. The latter information can be directly linked to a wealth of spatio-temporal parameters required for a correct interpretation of the ice core. The procedure directly yields the depth-age distribution, surface conditions at particle origin, and a suite of relevant parameters such as initial annual layer thickness. The procedure is able to fully account for time-dependent changes in such crucial parameters as ice thickness, flow direction, flow velocity, accumulation rate, and basal melting rates. The large-scale component of the nested model will be an existing fully-fledged Greenland ice sheet model (Huybrechts and de Wolde, 1999; Huybrechts, 2002). This model will be run on a 20 km (possibly 10 km) horizontal resolution with 30 layers in the vertical and another 9 layers in the bedrock for the calculation of the heat conduction in the crust. The model has components describing grounded ice flow and basal sliding according to the usual simplifications made within the shallow-ice approximation, the solid Earth response, and the mass balance at the ice-atmosphere interface. The main inputs to the model are the surface climate (mass-balance and surface temperature), the geothermal heat flux, and the eustatic sea-level stand which controls the coastline and the water depth around the continent. The melt- and runoff model is based on the positive degree-day method, and takes into account the process of meltwater retention by refreezing and capillary forces in the snowpack (Janssens and Huybrechts, 2000; Hanna et al., 2005). Within this project, the large-scale model will be further upgraded by incorporating the latest compilations for surface elevation, ice thickness, bedrock topography, and accumulation/ precipitation rate. As the basal melting rate is a major obstacle to retrieving old ice and largely depends on the geothermal heat flux, additional efforts will be done to incorporate various spatial distributions of the geothermal heat flux rather than using a constant value (Huybrechts, 1996; Shapiro and Ritzwoller, 2004; Greve, 2005). The current version of the large-scale Greenland ice sheet model uses a Eulerian approach to obtain first-order estimates of ice age and isotopic content (Huybrechts, 1994). Purely Lagrangian methods have been developed and tested in earlier work (Rybak and Huybrechts, 2003) but have known limitations due to the distribution of tracers, especially close to the bedrock. Within this project we will further improve the large-scale tracer module of the model by implementing a semi-Lagrangian approach (Clarke and Marshall, 2002; Clarke et al. 2005). The fine-scale model will be based on the higher-order model code of Pattyn (2003). This model includes both longitudinal and transverse gradients in the force balance equations. These additional terms improve the velocity solution at ice divides, near to the ice-sheet margin, and in areas with pronounced relief or high velocity gradients. They are also required for a more realistic solution on numerical grids where the horizontal resolution is of the order of the ice thickness. In this project we will use a further development of this ice flow code employing a numerical scheme based on a staggered grid to relieve known problems with convergence under certain circumstances (Bert De Smedt and Oleg Rybak, personal communication). The local flow model will be implemented at a horizontal resolution of 2 km or less with 100 layers in the vertical. At first experiments will be performed with a straightforward one-way 'downward' coupling scheme between the large-scale and fine-scale models. Exchange of information between the two models is expected to take place in anomaly mode (Huybrechts et al., 2007). However further developments are planned with twoway coupling schemes in which the high-resolution flow solution is able to feed back onto the large-scale flow. Apart from the chronology and the non-climatic biases for the NEEM ice core, the nested model will also provide appropriate variables to characterise the strain regime at different depths and places in the Greenland ice sheet. As was done previously in other studies (Marshall and Cuffey, 2000) this will enable to assess the risks of flow disturbances and ice core disruptions, especially in the deeper layers, that could possibly result from divide mobility and flow regime shifts. The time-dependent modeling will be continuously refined as more measurements become available from the drilling itself and from the area around the borehole. Fine-grid airborne and ground-based radar sounding data as well as shallow ice cores are already available for the NEEM site and will be further processed to refine reconstructions of the bed topography and the internal layering both around the borehole and along a flowline upstream from NEEM towards NGRIP. The semipermanent drilling camp was set up at the NEEM location during the summer of 2007. In addition the Alfred-Wegener-Institut für Polar- und Meersforschung (AWI) made a surface traverse from NGRIP to NEEM along the ice divide to perform Ground Penetrating Radar (GPR) mapping of internal reflectors down to 40 m depth to evaluate the accumulation rate. The AWI team also set up GPS fixed measurement stakes for elevation and ice surface velocity measurements, as well as several snow pit and shallow ice cores to approx. 70 m depth en route. Finally we will rerun the 3-D thermomechanical model of the entire Greenland ice sheet to investigate its evolution during the last two climatic cycles and to re-evaluate the implications for global sea level change. This modeling will incorporate refined forcings from the NEEM ice core itself, and also use better constraints on the geothermal heat flux and the relationship between accumulation and temperature change. It is expected that these model runs will shed more light on the fate and stability of the Greenland ice sheet during the Eemian period and the fast climatic changes during the glacial periods in particular.
Effective start/end date1/01/0931/12/12


  • Geographics

Flemish discipline codes

  • Mathematical sciences
  • Civil and building engineering
  • Pedagogical and educational sciences
  • Environmental sciences
  • Earth sciences
  • Social and economic geography