Anisotropic behavior of GRIP ices and flow in Central Greenland

https://doi.org/10.1016/S0012-821X(97)00193-3Get rights and content

Abstract

Mechanical tests have been performed on strongly textured ice samples coming from a wide range of depths (from 1328 down to 2868 m) of the GReenland Ice core Project (GRIP) for different sample orientations with respect to the prescribed stress. In this way, two directional viscosities, corresponding to the “easy glide” and to the “hard glide” orientations, were determined along the core. The viscoplastic anisotropy gradually increases down to a depth of ∼2600 m and slightly decreases below, revealing a clear correlation between rheology and texture. The experimental mechanical response compares well with that predicted by the ViscoPlastic Self-Consistent (VPSC) model. The VPSC model is also applied to ice samples that exhibit an axisymmetric texture to show in more detail the sensitivity of the rheology to specific texture parameters. This leads to a number of recommendations for future mechanical tests on anisotropic samples. A large-scale ice flow model is finally used to estimate the influence of ice anisotropy on the flow along the GRIP–GISP2 flow line. The particular mechanical behavior of deep GRIP ices in the stress regime corresponding to an ice divide leads to deformation rates that are highly sensitive to bedrock topography and texture pattern. This feature is likely to favour the formation of stratigraphic disturbances in deep ice layers, as observed in the last 300 m of both GRIP and GISP2 cores.

Introduction

This paper deals with the influence of ice rheology on the flow of ice sheets. In particular, the effects of the anisotropic behavior of polar ices will be investigated by: (1) analysing results of several mechanical tests performed on deep ices of the GReenland Ice core Project (GRIP); (2) modelling the mechanical behavior of polar ice with a micro–macro approach; and (3) modelling the large-scale flow of ice in the vicinity of the GRIP drill site (present summit of the Greenland ice cap).

Polar ice, as found in large ice masses like Greenland and Antarctica, is a polycrystalline material, meaning that it consists of an arrangement of crystals (grains). The macroscopic (bulk) mechanical behavior of an ice sample is directly linked to the microscopic (local) behavior of individual grains. Thus, it directly depends on the orientation of the crystallographic axes of grains with respect to the prescribed macroscopic deformation. Ice crystals exhibit an extreme viscoplastic anisotropy (they deform essentially by shear in basal planes). This means that the magnitude and direction of the deformation rate is highly dependent on the orientation of the prescribed stress with respect to the crystallographic axes. This microscopic anisotropy induces a strong macroscopic anisotropy when the polycrystalline sample exhibits a preferred orientation of the c axes (called “texture” in materials science). This is for example the case for Vostok (Antarctica), Dye 3 and Law Dome (Greenland) deep ices 1, 2, 3 and also for GRIP deep ices as will be shown here. Textures develop by the rotation of crystallographic planes during the plastic deformation that the ice is subjected to, and also when polygonization and migration recrystallization mechanisms are active 4, 5. When polar ice does not exhibit a particular orientation of the crystallographic axes (randomly oriented texture), as is the case for example for ice from the surface of ice sheets, the sample behaves identically in all directions and is said to be macroscopically isotropic.

The present paper raises the question of how the observed anisotropy may affect the flow of ice in ice sheets. In particular, we will show that it can make the flow near an ice divide extremely sensitive to bedrock topography and texture pattern. This feature could explain (at least partially) why the stratigraphy of the last 300 m of both GRIP and GISP2 ice cores is disturbed [6], a question of importance when attempting to reconstruct past climate from the study of deep ice cores.

We will present here results of several mechanical tests performed on GRIP ice samples of different depths, therefore presenting different textures. These tests were carried out in two different laboratories (LGGE-Grenoble, France, and Kitami Institute of Technology, Japan) for different orientations of the samples, so that different steady-state (secondary creep) directional viscosities could be determined at each depth (tertiary creep behavior of GRIP ices was presented by Dahl-Jensen et al. [7]). The comparison of all tests will show the good correlation between texture pattern and mechanical response. To understand in a more general way how textures influence the behavior of samples, we apply the ViscoPlastic Self-Consistent (VPSC) model of Lebensohn and Tomé [8] to GRIP ices. This model, which has already been used by Castelnau et al. 9, 10, 11, calculates an approximate solution of the stress and strain rate fields within the whole polycrystal, and thus estimates the mechanical response of a polycrystal for any prescribed stress when only the texture of the material is known. Using the texture of our GRIP samples as input to the model, we will compare the modelled rheology of GRIP ices with that determined experimentally. This is to our knowledge the first time that a rheological model for anisotropic ices is tested quantitatively on mechanical tests performed in the laboratory. The VPSC model enables us also to investigate how the sample behavior is sensitive to specific texture parameters and in particular to texture symmetries, an important question as shown by Azuma and Goto-Azuma [12]. In the last part of this paper, we study how ice anisotropy may affect the large-scale flow of ice in Central Greenland. For this, we apply the large-scale ice flow model developed by Mangeney et al. 13, 14 to the GRIP–GISP2 flow line. This model takes into account an anisotropic constitutive relation which gives the rheology of the ice for a given texture [15]. At present, the model is restricted to textures that exhibit an axisymmetry around the in situ vertical axis (i.e. the vertical axis is a revolution axis for the texture). This is due to the fact that all analytical anisotropic constitutive relations for polar ice have so far been developed for ices presenting orthotropic transversally isotropic behavior. In fact, such a texture symmetry is found only approximately in natural samples [16]. We will show that this approximation should lead to significant inaccuracy in flow prediction. Consequently, the flow of ice near an ice divide cannot be realistically modelled unless a model is developed that precisely describes the development of texture with deformation and recrystallization.

The present paper is structured as follows. In Section 2, we present the results of mechanical tests. In Section 3, we compare the results of the VPSC model with those obtained experimentally, and we show in more detail how texture symmetries affect polycrystal behavior. In Section 4, we apply the large-scale ice flow model to Central Greenland to show how anisotropy influences the in situ flow.

Section snippets

Description

Several mechanical tests have been performed on GRIP ice samples taken from depths ranging from 1328 to 2868 m. The texture of GRIP ices, described in detail by Thorsteinsson et al. [17], changes along the core from a random orientation at the ice sheet surface to a strongly preferred orientation in the deeper parts, with c axes aligned approximately parallel to the in situ vertical direction. This vertical single maximum texture becomes pronounced at 2200 m and persists down to a depth of 2800 m.

Modelling mechanical behavior

Mechanical tests have been used to determine the response of GRIP ice samples for two different orientations of the stress state with respect to texture orientation. To understand in a more general way the relation between texture and rheology, we use a micro–macro model, i.e. a model that relates the behavior of the polycrystal to that of individual grains. We shall now concentrate on the comparison between the experimental rheology of our GRIP samples and that predicted by the model.

Influence on in situ flow

We will now use the large-scale ice flow model of Mangeney et al. 13, 14 to calculate the flow of ice in the vicinity of the GRIP drill site. This 2-dimensional model does not use the shallow ice approximation generally adopted in glaciology, which is based on the existence of a small aspect ratio in ice sheet geometry. The model presents the particularity of completely solving the set of stress equilibrium and incompressibility equations, which renders it applicable to any ice sheet and

Conclusions

We have presented results of several experimental tests performed in two different laboratories (LGGE-Grenoble and Kitami Institute of Technology) on GRIP ice samples. These tests make it possible to estimate the directional viscosities corresponding to a nearly vertical in situ compression and to an in situ horizontal shear combined with a horizontal uniaxial tension. We observe a gradually increasing anisotropy from 1000 down to 2600 m depth, and a slightly decreasing anisotropy below. A

Acknowledgements

This work is a contribution to the GReenland Ice-core Project, carried out at Summit, central Greenland. Financial support was provided by the national funding bodies of Belgium, Denmark, France, Germany, Iceland, Italy, Switzerland, and the United Kingdom, under the auspices of the European Science Foundation. This work was also supported by the PNEDC (Programme National d'Étude de la Dynamique du Climat) CNRS, and by the CEC (Commission of European Communities) Environment Program. We are

References (34)

  • R.B. Alley et al.

    Comparison of deep ice cores

    Nature (London)

    (1995)
  • D. Dahl-Jensen, Th. Thorsteinsson, R. Alley, H. Shoji, Flow properties of the ice from the GRIP ice core — the reason...
  • O. Castelnau et al.

    Viscoplastic modelling of texture development in polycrystalline ice with a self-consistent approach; comparison with bound estimates

    J. Geophys. Res.

    (1996)
  • O. Castelnau et al.

    Modelling texture development along the GRIP ice core (central Greenland)

    Ann. Glaciol.

    (1996)
  • O. Castelnau et al.

    Modelling viscoplastic behavior of anisotropic polycrystalline ice with a self-consistent approach

    Acta Mater.

    (1997)
  • N. Azuma et al.

    An anisotropic flow law for ice-sheet ice and its implications

    Ann. Glaciol.

    (1996)
  • A. Mangeney et al.

    Isothermal flow of an anisotropic ice sheet in the vicinity of an ice divide

    J. Geophys. Res.

    (1996)
  • Cited by (48)

    • Deformation of polycrystalline D<inf>2</inf>O ice: Its sensitivity to temperature and strain-rate as an analogue for terrestrial ice

      2020, Earth and Planetary Science Letters
      Citation Excerpt :

      Analysing deformation in naturally deformed ice masses, by contrast, is an inverse problem: glaciologists use the available microstructural evidence to reconstruct the deformation histories and infer the paleo-conditions. Experimental deformations also typically involve high-temperatures and fast strain-rates compared to natural deformations, and thus major extrapolations are required to make inferences (Castelnau et al., 1998; Duval et al., 2010). Overall, the link between experimental deformations, where the physical conditions are well known; and deformation in nature, remains poorly established.

    • Mechanics and microstructure of deformed natural anisotropic ice

      2018, Journal of Structural Geology
      Citation Excerpt :

      Dahl-Jensen et al. (1997) performed compression tests on ice from the Greenland Ice core Project (GRIP) with varying strengths of CPO, deforming samples for up to three years at vertical stresses of 0.64 MPa, and concluded that once the tertiary strain stage had been reached and a new CPO developed according to stress and temperature conditions, the original flow conditions of the ice had no remaining mechanical effects on behaviour. Similar tests on ice from the same cores by Castelnau et al. (1998) found that ice with a stronger CPO (taken from a greater depth) displayed rheological behaviours with stronger anisotropy. While all of those studies focussed on ice with a compressional CPO (c-axis cluster or cone) which is close to optimally oriented for the applied stresses, Dierckx et al. (2014) is an exception, applying unconfined uniaxial compression to marine folded ice from the Nansen Ice Shelf, Antarctica, cut at different angles to a fold axis.

    • The down-stress transition from cluster to cone fabrics in experimentally deformed ice

      2017, Earth and Planetary Science Letters
      Citation Excerpt :

      Glaciers and ice sheets are major rock bodies on the Earth's surface that develop very strong CPOs as they deform (Alley, 1992; Budd and Jacka, 1989; Hudleston, 2015; Wilson et al., 2014). These CPOs may impart significant anisotropy to ice flow (Azuma, 1995; Castelnau et al., 1998; Duval et al., 2010) which may have an important influence on the development of large-scale ice structures (Bons et al., 2016; Hulbe and Whillans, 1997; Pettit et al., 2007). Such effects thus need to be understood mechanistically, quantified and incorporated into robust constitutive models of ice flow (flow laws) that can be employed in models of glaciers and ice sheets (Gagliardini et al., 2013).

    • Evidence of unfrozen liquids and seismic anisotropy at the base of the polar ice sheets

      2015, Polar Science
      Citation Excerpt :

      The anisotropy assumed in the lower layer corresponds to the maximum anisotropy that can be considered, i. e. when all the c-axes are aligned in the same direction. This anisotropy is higher than the one that can be deduced from the diagrams of c-axis orientations observed in the ice cores (Thorsteinsson et al., 1997; Castelnau et al., 1998). The values of the anisotropy parameters that we introduced in the model in order to match the computed with the observed amplitudes can be physically explained by strong preferred orientation of the c-axes along the vertical in the deeper part of the ice sheet.

    • Multiscale modeling of ice deformation behavior

      2014, Journal of Structural Geology
      Citation Excerpt :

      The fabric obtained after 50% shortening strain compares qualitatively well with the one measured along the GRIP ice core at a depth where the strain is similar (1293 m) (Thorsteinsson et al., 1997). The VPSC model in its “tangent” version was applied to simulate the fabric development along ice cores (Castelnau et al., 1996b,a; 1998). In (Castelnau et al., 1996a), a comparison was made with bound estimates (Reuss and Voigt).

    View all citing articles on Scopus
    1

    At LGGE–CNRS as this work was done.

    2

    At LGGE–CNRS as this work was done.

    View full text