Ground ice as environmental archive

Ground ice, defined as all types of ice contained in frozen or freezing ground [van Everdingen, 1998], is either directly fed by meteoric water sources or by recycled water (e.g. surface water, ground water) that has been subject to post-depositional transformations. Similar to glacier ice, ground ice is a natural environmental archive as it captures short-lived meteorological conditions as

well as long-term climatic trends due to its preservation potential of atmospheric precipitation, air and hydrological conditions at its source, transport pathway and place of deposition. It can be described by numerous physical, chemical and biological parameters (i.e. environmental proxies) that are related to climate and environmental conditions on different scales and allow the reconstruction of valuable information on past climate and environmental changes [NCDC, 2011].

Therefore, ground ice can be studied as a paleoenvironmental archive [Mackay, 1983; Vaikmäe, 1989, 1991; Vasil'chuk, 1991] using analytical methods similar to those applied to ice cores from glaciers and ice caps. This is especially meaningful in polar regions without current glaciation, such as the western Canadian Arctic.

One of the most promising archives for paleoclimate reconstructions are ice wedges that arise from the episodically repeated filling of thermal contraction cracks, mainly fed by snow meltwater, which percolates into the frost fissure, refreezes immediately [Lachenbruch, 1962] and therefore retains its original environmental information [cf. Michel, 1982]. Tabular massive ground ice bodies (excluding ice wedges) are defined as laterally and vertically extensive subsurface ice masses [Mackay, 1972b; 1989] with an ice content exceeding 250 % (i.e. on an ice-to-dry-soil weight basis [van Everdingen, 1998]) and are among the most striking features of permafrost areas.

The occurrence of massive ground-ice bodies has often been related to the former presence of Pleistocene ice caps, since many massive ice exposures have been found within the limits of Quaternary glaciations. Since the end of the 19^th century, early explorers speculated that these ice bodies consisted of relict glacier ice [von Toll, 1897; Lorrain and Demeur, 1985; Kaplyanskaya and Tarnogradsky, 1986; Astakhov and Isayeva, 1988; Ingólfsson et al., 2003; Murton et al., 2005;

Fritz et al., 2011], or whether the ice has a segregation origin [Mackay, 1971, 1973; Rampton, 1991, Mackay and Dallimore, 1992], with glacial meltwater delivering the huge amount of water required for their formation [Rampton, 1988; French and Harry, 1990].

A variety of permafrost landscape features have their origin in the aggradation or degradation of ground ice (e.g. polygonal nets, pingos, thermokarst lakes, thermoerosional valleys and retrogressive thaw slumps), which is a major component of permafrost dynamics. Ground ice locally makes up to 50 % of the volume of near-surface permafrost in the western Canadian Arctic [Mackay, 1971]. Thermokarst phenomena that represent major threats for arctic infrastructure are often associated with the melting of massive ice [e.g. Murton, 2001; Burgess and Smith, 2003;

Lantuit and Pollard, 2008]. Besides massive ice types there exists a variety of non-massive ice types of various origins depending on the origin of water prior to freezing and the principle process of water movement towards the freezing plane [cf. Mackay, 1972b]. For simplification this chapter uses the term “non-massive intrasedimental ice” (NMI) for all types of pore ice or segregated ice [cf. Murton and French, 1994] within surrounding permafrost-affected sediments. NMI might also be used for paleoclimatic studies [Burn et al., 1986; Vaikmäe, 1989; Schwamborn et al., 2006].

Whereas ice wedges are mainly fed by winter precipitation [Vaikmäe, 1989; Vasil'chuk, 1991], NMI often consists of refrozen water, which is a mixture of waters of various origins (i.e. summer and winter precipitation, surface water, and last season's ground water) [Schwamborn et al., 2006].

Even though preservation of soil moisture in NMI occurs in a complex way, i.e. through repeated seasonal freeze and thaw that adds numerous cycles of phase change and therefore promotes isotopic fractionation, it can still reflect environmental and climatic changes [Schwamborn et al., 2006]. Murton and French [1994], Vardy et al. [1997], Kotler and Burn [2000], and Schwamborn et al. [2006] have shown that major changes in paleotemperature and hydrology can be resolved by interpreting the NMI record.

Pioneering work in the field of paleoclimate studies based on ground ice has primarily involved ice wedges and focused on oxygen isotope (δ¹⁸O) variations as an indicator for winter temperature changes [Michel, 1982; Mackay, 1983; Vaikmäe, 1989; Vasil'chuk, 1991]. This was later amended by mutual considerations of δ¹⁸O, δD and deuterium excess (d-excess), which provided additional information for paleotemperature reconstruction, for the identification of the precipitation source, and for unravelling post-depositional fractionation processes [Dansgaard, 1964; Merlivat and Jouzel, 1979; Souchez, 2000; Meyer, 2002a, b; Lacelle et al., 2004; Lacelle, 2011; Opel et al., 2011].

We adapted the approach from Bradley [1999] that is based on ice cores towards ground ice in order to obtain paleoenvironmental information. This involves the analysis (1) of the ice's physical characteristics, (2) of stable water isotopes, (3) of dissolved and particulate matter and (4) of entrapped gas bubbles.

(1) Physical characteristics of the ice such as ice content, sediment inclusions and cryostratigraphic relationships to the surrounding deposits constrain a distinct ice origin and enable the identification of thaw unconformities [Mackay, 1971, 1989; Burn et al., 1986, 1997; Murton and French, 1994; Murton et al., 2004, 2005].

(2) The stable water isotope composition of precipitation is strongly temperature-dependent [e.g.

Dansgaard, 1964; Rozanski et al., 1993] and can therefore be used as a proxy for local to regional temperature regimes [e.g. Mackay, 1983; Kotler and Burn, 2000; Meyer et al., 2002a, b, 2010]. Additional information on moisture origin, water source and freezing conditions can be obtained that way [e.g. Michel, 1986; Lacelle et al., 2004, 2007, 2009b; Fritz et al., 2011; Lacelle, 2011].

(3) Analyses of the dissolved ion content and radiocarbon dating of particulate organic matter can be used for the distinction of water sources [e.g. Mackay and Dallimore, 1992; Fritz et al., 2011] and for providing a chronological context of environmental change deduced form ground ice [Vasil'chuk and Vasil'chuk, 1997; Vasil'chuk et al., 2000, 2001; Meyer et al., 2010; Opel et al., 2011].

(4) Air bubbles in ground ice may represent samples of a former atmospheric composition if they derive from buried glacier ice that consists of firn and allows the direct analyses of the paleo-atmosphere during ice formation. In this context, though not being a topic of this study, Cardyn et al. [2007] have shown that analyzing molar gas ratios of air entrapped in ground ice (O2/Ar and N₂/Ar) may provide a powerful tool for clearly distinguishing between atmospheric gas in glacial ice and gases from intrasedimental ground ice to determine the origin of relict massive ground-ice bodies.