Visible stratigraphy

At sites QSD, QS and KT, where summer melting regularly takes place and forms distinctive melt layers, visible stratigraphy of the pits and the shallow core allowed easy identification of the end of the previous summer (summer 2000) melt season. Hereafter, we use the term ‘summer surface’ for the end of the summer melt season. Visible stratigraphy at QSD, QS and KT is presented in Figure 2.

Fig. 2.

Visible stratigraphy at QSD, QS and KT.

At QSD, the following sequence was observed: (1) compacted snow, with 10 mm thick new snow on top (0– 1.91 m); (2) a mixture of loose hoar layers and faceted-crystal layers containing thin (1 mm) ice layers (1.91– 3.27 m); and (3) thick iced firn down to the bottom of the pit (3.27–3.60 m). Layers (1), (2) and (3) correspond to winter/spring 2000/01, late summer/fall 2000 and midsummer 2001, respectively. Similarly, visible stratigraphy at QS exhibited the following sequence: (1) winter/spring layers, consisting of new snow, lightly compacted snow and compacted snow (0–1.67 m); (2) fall layers consisting of faceted crystals with no signs of melt (1.67–2.79 m), where the boundary between fall and winter layers was not clearly located; (3) late-summer layers, consisting of loose and coarse faceted crystals and hoar, with thin (1–5 mm) ice layers (2.79–3.17 m); and (4) the mid-summer layer of 2000, consisting of iced firn, ice lenses and ice pipes (3.17m to bottom of the pit). At 85 mm depth, there was a visible dirt layer, which was 15 mm thick.

KT had the following seasonal sequence: (1) the uppermost 0.48 m, consisting of faceted-crystal snow, compacted snow, thin ice layers (1–5 mm) and a 60mm thick visible dirt layer at 0.21 m depth (Fig. 3a); (2) compacted snow with no signs of melt (0.48–1.11 m); (3) faceted-crystal snow with no signs of melt (1.11–2.51 m); (4) iced firn containing 2–15mm ice layers (2.51–2.91 m); (5) faceted-crystal snow containing 10–50mm ice layers and ice pipes (2.91–3.51 m); (6) fine-grained firn (3.51–5.39 m); and (7) iced firn including a 10mm ice layer (5.39 m to bottom of the core). Layers in (1) are those affected by the surface melting of late spring or early summer 2001. We located the seasons as follows: ( 1 , 2) winter/spring 2000/01; (3) fall 2000, where the boundary between fall 2000 and winter 2000/01 was not clearly defined; (4) summer 2000; (5) spring 2000, where snow in this layer had probably been affected by the meltwater drained from the summer layer of 2000; (6) winter/spring 1999/2000 and fall 1999; and (7) summer 1999.

Fig. 3.

(σ) The Asian dust layer in the pit at KT. (b) An aerial photograph taken above Quintino Sella Glacier on 27 June 2001, illustrating the patchy snow surface formed by the heterogeneous melt rate due to the dirt cover. Though the scale of the photograph is not known, the scale of the patchiness is on the order of 2–3 m.

At sites KTU and KC, intermittent summer melting forms thin melt layers, which usually allow for identification of the previous summer layer (Reference Goto-AzumaGoto-Azuma and others, 2003). At KTU we located the 2000 summer surface just above a 5 mm ice layer at 4.91 m depth. On the other hand, very little or no summer melting in 2000 at both KC (where limited melting occurs in some summers) and PRC (in the dry snow zone where only very slight melting occurs in some summers due to shortwave radiation) meant that there were no visible signs of the 2000 summer surface at either site.

When members of the drill team reached the Mount Logan area on 8 May 2001, they found widespread reddish-brown dirt on or just below the snow surface (Reference ZdanowiczZdanowicz and others, 2006). This turned out to be visible evidence of a major fallout of Asian dust that occurred in April 2001 (Reference Jaffe, Snow and CooperJaffe and others, 2003; Reference UnoUno and others, 2004; Reference ZdanowiczZdanowicz and others, 2006). At the lower sites (<3000 m) QSD and QS, the dust was dispersed on the snow surface as a thin (0.01–0.03 m) and patchy reddish-brown layer (Reference ZdanowiczZdanowicz and others, 2006). Later in the season when pit sampling was done at QS, the dust was found buried under the snow. On the other hand, at QSD, when the sampling was done in late May, the dust was partially buried under the snow but still visible on the surface, due to heterogeneous wind scour. Higher up at King Trench (KT and KTU) and KC, the dust was found as a thicker layer (0.05–0.13 m) below the snow surface. Figure 3a displays the dust layer observed in the pit at KT.

At sites QSD, QS, KT, KTU and KC, the dust was heterogeneous, being visible in some places and not visible only a few meters away. The dust was visible in the pits at QS, KTand KTU, whereas it was not visible in the pits at QSD, KC and PRC. But even where the dust was not visible, the dust layer was always recognizable because it was denser and more compact than snow immediately above or below it (Reference ZdanowiczZdanowicz and others, 2006) due to somewhat warmer temperatures and/or moist conditions during dust deposition, a post-deposition albedo (shortwave) effect or both. This distinctive dust layer served as an excellent time marker (Reference Goto-AzumaGoto-Azuma and others, 2003). During the melt season, below about 3000 m, melting exposed the layer at the surface and lowered the albedo, thereby increasing the melt rate to produce a patchy surface in terms of dirt cover (Fig. 3b).

Chemical stratigraphy

Figure 4 compares profiles of δ¹⁸O, Ca²⁺ and NO₃ ⁻ from the six sites on Mount Logan. Note that since sites QSD and QS are located in the wet-snow zone, δ¹⁸O and ion concentrations below the previous summer surface would have been disturbed with respect to those at the time of snow deposition.

Fig. 4.

Profiles of δ¹⁸O, Ca²⁺ and NO₃ ⁻. Corresponding peaks at different sites are connected with thin dashed lines. The Asian dust layer is shown by either a thick solid line (where it was visible) or a thick dashed line (where it was not visible). The thickness of each thick line corresponds to that of the dust layer. The location of the end of the 2000 summer melt season at each site is denoted by a dotted line.

The Asian dust layer was associated with high ion concentrations. Though it was not visible in the pits at QSD, KC and PRC, it could be recognized from its ion signature. With the exception of NO₃ ⁻ and SO₄ ^2– at KC, concentrations of Ca²⁺, NO₃ ⁻, SO₄ ^2–, NH₄ ⁺, K⁺ and Mg²⁺ (K⁺ and Mg²⁺ not presented here) were all high in the Asian dust layer, particularly where it was visible. Concentrations of Na⁺ and Cl⁻ were also high in the Asian dust layer at most of the sites.

What we consider corresponding peaks are marked with thin dashed lines in Figure 4. We estimated the depth of the summer 2000 surface at both KC and PRC, where lack of melting did not allow the development of a distinct summer layer, by comparison among the profiles of δ¹⁸O and various ion concentration profiles. At KC, the 2000 summer surface was located at 5.185 m. This is consistent with a first-order estimate of snow deposition between summer 2000 and spring 2001, deduced from the fact that the automatic weather station, which had been installed at KC in late July 2000, was found buried in snow when KC was revisited in May 2001. At PRC, the 2000 summer surface was located at 0.75 m, yielding 0.37 m w.e. snow deposition between summer 2000 and spring 2001, which seems to be reasonable because this is about 60% of the annual mean accumulation rate at the site (Reference FisherFisher and others, 2004). By normalizing the depth between the Asian dust layer (spring 2001) and the summer 2000 surface at each site, we calculated correlation coefficients of δ¹⁸O and ion profiles between sites QSD, QS, KT, KTU and KC. PRC was excluded from the correlation analysis because the site had significantly fewer samples than other sites.

Based on the δ¹⁸O correlation analysis (Table 2), we can divide the lower five sites into two groups: QSD and QS at the lower altitudes, and KT, KTU and KC at the mid-altitudes. Though we did not include PRC in the correlation analysis, PRC is likely on its own at the highest altitude, taking into account the much lower accumulation rate here than at the lower five sites, which is discussed in a later section. Correlation coefficients for ions show a spatial tendency similar to those for δ¹⁸O, while they were generally lower than those for δ¹⁸O. Division into three groups is likely due to the difference in moisture source and its transport pathway: sites in the planetary boundary layer (<3000m a.s.l.) being affected mainly by local sources; sites in the upper geostrophic flow region (>5000 ma.s.l.) being affected by distant sources; and sites in the mixed layer (3000–5000m a.s.l.) being affected by both sources (Reference Holdsworth, Fogarasi and KrouseHoldsworth and others, 1991; Reference HoldsworthHoldsworth, 2001; Reference Holdsworth and KrouseHoldsworth and Krouse, 2002). The occasional difference in the timing of precipitation at different altitudes would also contribute to the differences between the higher and lower sites. In fact, weather reports from QS, KC and PRC were frequently different during the 2001 and 2002 field seasons, when the three stations were manned most of the time. However, it should be noted that though a series of depth profiles of δ¹⁸O and ions shows spatial variations, characteristic peaks can usually be traced across all the profiles. This fact suggests that there is a commonality between snowfall events and deposition of aerosols at the lower and higher sites; they do not take place in a completely independent manner. Frontal systems and, in particular, occluded fronts of stalled cyclones, which are common in this region, bring moisture from both local and distant sources to Mount Logan during the same storm events, where predominance of local vs distant sources depends on the altitude (Reference Holdsworth, Fogarasi and KrouseHoldsworth and others, 1991). The variation in local snowfall at different altitudes would produce noise superimposed on the dominant signal from the larger storm events which affect all altitudes.

Table 2.

Correlation coefficients for δ¹⁸O between the lower five sites

Seasonal variations of δ¹⁸O and ions

We now discuss the variations with depth of δ¹⁸O, Ca²⁺, Na⁺, Cl⁻, NO₃ ⁻, SO₄ ^2–, MSA and NH₄ ⁺ to see if there are any variations associated with seasonality (Fig. 5). At QSD, QS, KT and KTU, where visible stratigraphy allowed easy identification of the 2000 summer surface, the δ¹⁸O maximum was seen right above it, indicating that δ¹⁸O peaks in late summer/fall. On the other hand, the δ¹⁸O minimum was observed in winter layers. At KC, where we could not identify the summer 2000 layer from visible stratigraphy, high correlation of δ¹⁸O between KC and KTU suggests that δ¹⁸O at KC shows seasonality similar to that at lower sites. Additionally, a pit dug at KC in 2002, where the previous summer (the summer 2001) layer was easily identified by visible stratigraphy in this particular year (Reference Goto-AzumaGoto-Azuma and others, 2003), exhibited the same seasonal variation of δ¹⁸O (Goto-Azuma and others, unpublished information). This fact supports the same seasonality of δ¹⁸O at KC. It would be reasonable to assume the same seasonality also at PRC, if we consider the resemblance of δ¹⁸O profiles at KC and PRC, with the assumed depth of the 2000 summer surface. The seasonality of δ¹⁸O observed in this study (late-summer/fall maxima and winter minima) is similar to that reported for sites on the Mount Logan plateau (where the elevation is everywhere >5000 m) (Reference Holdsworth, Fogarasi and KrouseHoldsworth and others, 1991). However, Reference Wake, Yalcin and GundestrupWake and others (2002) and Reference Yalcin, Wake and GermaniYalcin and others (2003, Reference Yalcin, Wake, Kang, Kreutz and Whitlow2006b) reported summer maxima of δ¹⁸O at nearby Eclipse Icefield (altitude 3017 m). The reason for this difference in seasonality between Mount Logan sites and Eclipse is as yet unknown.

Fig. 5.

Profiles of δ¹⁸O, Ca²⁺, Na⁺, Cl⁻, NO₃ ⁻, SO₄ ^2–, MSA and NH₄ ⁺. Location of the Asian dust layer is shown by either a thick solid line (where it was visible) or a thick dashed line (where it was not visible) in each δ¹⁸O diagram. The thickness of each thick line corresponds to that of the dust layer. The location of the end of the summer 2000 melt season is denoted by a dotted line in each δ¹⁸O diagram.

The ion records also display seasonal variations, with more noise than δ¹⁸O. At most of the sites, MSA peaks slightly before δ¹⁸O does, i.e. in late summer/early fall. Other ions show higher concentrations in the Asian dust layer at most of the sites. The dust layer was formed by an episodic event and it somehow overwhelms the seasonal cycle of the various chemical species we are studying. However, most of the ions show peaks in the late winter/ spring broader than the thickness of the dust layer, with the exception of PRC. As Asian dust events, though they are non-periodic, are primarily springtime phenomena (e.g. Reference Merrill, Uematsu and BleckMerrill and others, 1989; Reference JaffeJaffe and others, 1999), winter/ spring peaks of dust-derived ions, such as Ca²⁺, might then be considered as a potential annual marker in Mount Logan snow. Furthermore, sea-salt ions Na⁺ and Cl⁻, along with Ca²⁺ (Fig. 5), at all the sites above 3000 m (i.e. KT, KTU, KC and PRC), show peaks only in winter/spring. At the sites below 3000m (i.e. QSD and QS), seasonal input timing of these ions is similar to that at the higher sites, with additional minor peaks. The seasonal variations of Ca²⁺, Na⁺ and Cl⁻ observed in this study are similar to those at Eclipse Icefield, reported by Reference Yalcin and WakeYalcin and Wake (2001) and Reference Yalcin, Wake and GermaniYalcin and others (2003, Reference Yalcin, Wake, Kang, Kreutz and Whitlow2006b).

Ammonium generally shows a spring maximum, with a few spikes added to this overall trend. Episodic events such as forest fires (Yalcin and others, 2006σ) could produce such spikes and occasionally disturb the general seasonality. Reference Yalcin, Wake, Kang, Kreutz and WhitlowYalcin and others (2006b) also reported the similar seasonality of NH₄ ⁺ at Eclipse Icefield. Concentrations of NO₃ ⁻ and SO₄ ^2–, peaking in winter/spring together with other ions, also show peaks in the late summer/fall. Contradictory to our finding, Reference Yalcin, Wake, Kang, Kreutz and WhitlowYalcin and others (2006b) reported late-spring/summer peaks only. Though our study covers only a single year, seasonality found here might have important implications, since anthropogenic inputs of NO₃ ⁻ and SO₄ ^2– have added a late-winter/early-spring peak to the natural summer peak in the North Atlantic sector of the Arctic (e.g. Reference Finkel, Langway and ClausenFinkel and others, 1986; Reference Goto-Azuma, Koerner, Nakawo and KudoGoto-Azuma and others, 1997). Further investigation is needed to draw a firm conclusion.

Altitudinal variations of δ¹⁸O and ions

We calculated the snow and ion fluxes, and averages of δ¹⁸O and ion concentrations for the period between the end of the summer 2000 melt season and the April 2001 Asian dust event, where there are good time markers. Snow and ions deposited in the Asian dust layer have not been included in the calculation. The results are plotted vs altitude in Figures 6 and 7. Methanesulfonate (MSA), Mg²⁺ and K⁺ data at QS and KC have not been included due to the very low concentrations which could not be measured with sufficient accuracy in some depth intervals of the pits.

Fig. 6.

Deposition fluxes of snow and ions between end of the summer 2000 melt season and April 2001, vs altitude.

Fig. 7.

Mean δ¹⁸O and ion concentrations between end of the summer 2000 melt season and April 2001, vs altitude.

Snow deposition is clearly not a simple function of altitude (Fig. 6). Between the end of summer 2000 and April 2001, snow deposition was highest at KC and lowest at PRC. King Col and King Trench usually receive higher snow accumulation, possibly because these sites are more sheltered and less affected by scouring, but also because local orographic effects tend to enhance snowfall there (Reference KeelerKeeler, 1969), in contrast to the plateau, which receives much less precipitation (Reference KeelerKeeler, 1969; Reference Holdsworth, Pourchet, Prantl and MeyerhofHoldsworth and others, 1984, Reference Holdsworth, Krouse and Peake1988; Reference Holdsworth and PeakeHoldsworth and Peake, 1985). KC is also subject to the local effects of larger storms because it is situated in the lee of the summit of Schoening Ridge, which would be another factor contributing to a high accumulation rate.

As was reported in earlier papers (Reference Holdsworth, Fogarasi and KrouseHoldsworth and others, 1991; Reference HoldsworthHoldsworth, 2001; Reference Holdsworth and KrouseHoldsworth and Krouse, 2002), altitudinal variation of the mean δ¹⁸O between summer 2000 and April 2001 (Fig. 7) follows a characteristic staircase structure with a significant step between 3200 and 4000–5000m. There is no clear and consistent relationship between altitude and most of the ions in terms of either concentration or flux. In general terms, the highest site (PRC) lies separate from all the lower sites. This is discussed by Reference FisherFisher and others (2004) who conclude that the Logan plateau is in the path of air-mass trajectories of much greater length than the areas below it. Below the plateau, fluxes of some of the ions, such as Ca²⁺, NO₃ ⁻, SO₄ ^2– and NH₄ ⁺, show weak positive relationships with elevation (Fig. 6). In contrast, fluxes and average concentrations of Na⁺ and Cl⁻ (Figs 6 and 7) all decrease with altitude, though high snow deposition at KTU and KC due to the local orographic effects somewhat obscures the general decreasing trend in fluxes of Na⁺ and Cl⁻ (Fig. 6). As the Cl⁻/Na⁺ ratio in most of the pit samples is close to that of sea water (1.8), suggesting that sea salt is the major source for Cl⁻ and Na⁺, we would expect the atmospheric concentrations of these ions and hence the fluxes and concentrations to decrease with altitude, i.e. distance from the source effect.

Fluxes of Ca²⁺, NO₃ ⁻, SO₄ ^2– and NH₄ ⁺ increase with snow deposition, whereas those of Na⁺ and Cl⁻ have no relationship with snow deposition (Fig. 8). This suggests that the main scavenging process for Ca²⁺, NO₃ ⁻, SO₄ ^2– and NH₄ ⁺ in this area is ‘rain-out’, i.e. wet, rather than dry, deposition. A similar result has been reported by Reference Yalcin, Wake, Kang, Kreutz and WhitlowYalcin and others (2006b). Although we did not include the Asian dust layer in the flux calculation, wet deposition would be an important scavenging process also for the Asian dust at the sites with high snow accumulation rate, because we noticed a thicker Asian dust layer at sites with higher snow deposition.

Fig. 8.

Deposition fluxes of ions between end of the summer 2000 melt season and April 2001, vs snow deposition.

The distinctly Asian dust layer found in Mount Logan snow indicates that Asian dust is an important source of the dust deposited on Mount Logan (Reference ZdanowiczZdanowicz and others, 2006). Since Ca²⁺ is mainly of continental dust origin, Asian dust may also be an important source of Ca²⁺ in Mount Logan snow.

Major sources for NO₃ ⁻ and SO₄ ^2– may vary with altitude. At a lower-altitude site, Eclipse Icefield, anthropogenic pollutants are reported to be major sources (Reference Yalcin and WakeYalcin and Wake, 2001). On the contrary, no anthropogenic impacts have been observed at a higher site (5340 ma.s.l.) on North West Col (Reference Holdsworth and PeakeHoldsworth and Peake, 1985), where snow contains aerosols scavenged from the middle and/or upper troposphere (Reference Monaghan and HoldsworthMonaghan and Holdsworth, 1990). Since the sampling sites in this study cover both altitudes, effects of anthropogenic sources are expected to vary with altitude. Forest fires would be important sources for NO₃ ⁻ and NH₄ ⁺ (Reference Holdsworth, Krouse and PeakeHoldsworth and others, 1988, Reference Holdsworth1996; Reference Yalcin, Wake, Kreutz and WhitlowYalcin and others, 2006a).

Seasonal variations of heavy metals in Alpine ice have been reported to be largely linked with the changing vertical structure of the regional troposphere, which prevents the transfer of pollutants to higher altitudes in winter but not in summer (Reference Van de VeldeVan de Velde and others, 1998). If this is the case for ions in Mount Logan snow, though little is known, seasonal variations of ions could differ with altitude.