SEB/MB from the AWS driven distributed model 2009-2012

The altitudinal dependence of glacier-wide MB and SEB components for the MB years 2009/10 and 2010/11 as calculated by the MB model is shown in Fig. 2.16. The pattern of annual MB (Fig. 2.16a) is characterized by a distinct steepening of gradients between 5650 and 5800 m a.s.l. that was also found by Mölg et al. (2012) at Zhadang glacier and by Fujita & Ageta (2000) at Xiao Dongkemadi gla-cier in the Tanggula Shan. The altitude range of this steepening differs within these studies as it does for the two balance years considered in Fig. 2.16. In the year 2009/10 the change in slope occurs between 5700 and 5800 m a.s.l., far below the equilibrium line altitude (ELA, 5950 m a.s.l. in 2009/2010) whereas in 2010/11, when ablation on Zhadang glacier was rather weak (Fig. 2.8 and 2.17), the steepening happens slightly below the ELA between 5650 and 5750 m a.s.l. This is in line with the findings of Mölg et al. (2012) and Fujita & Ageta (2000) and was also observed at Morter-atschgletscher by Klok & Oerlemans (2002). A similar pattern can be found in the vertical profiles of SWnet (Fig. 2.16d), Qmelt (Fig. 2.16e) and therefore surface melt (Fig. 2.16b), and surface albedo (α) (Fig. 2.16g). Qmelt has the same shape as surface melt because the two parameters are directly con-nected through LM (see section 2.3.1). As α influences the other variables (see sections 1.3.4.1, 2.3.1) it is the main driving factor for the shape of the vertical MB profile (Braun & Hock 2004). SWin and LWin are the only ground-independent energy sources for the glacier surface (see section 2.3.1). SWin

is nearly constant from around 5600 up to 5800 m a.s.l. due to minimal effects resulting from terrain shading. It shows a first minimum in the uppermost glacier regions mainly because of steep slopes in the eastern and western parts (Braun & Hock 2004) and a second minimum at the glacier tongue, possibly due to the convex shape of the main tongue and the steep slopes west of the second, smaller glacier tongue (Fig. 2.16d). These regions of smallest levels of solar insolation are reflected in the slopes of the vertical profiles of b (Fig. 2.16a) which are slightly steeper in the lowest parts (com-paratively less melt) and less steeper in the uppermost parts (sharp transition to positive values).

Small SWin at the glacier tongue favours a number of feedback processes that are noticable in the lowest glacier areas, e.g. a comparatively higher α (Fig. 2.16g) and thus smaller SWnet (Fig. 2.16d), and minimum Ts (expressed in LWout, Fig. 2.16f) in accordance of findings by Braun & Hock (2004) on King

George Island (Antarctica). These effects are damped but not overcompensated by LWin, Qsens or Qlat

because these values are rather stable or increase only little towards the glacier tongue in the annual mean (Fig. 2.16d,e) (Mölg et al. 2009). Total QG approximates zero or is slightly positive at all eleva-tions (Fig. 2.16e). According to negative Qlat, sublimation is responsible for a small and rather cons-tant glacier wide mass loss that can be larger than annual mean surface melt in the upper parts (Fig.

2.16b, Fig. 2.17b). The total amount of refreezing depends on both surface melt and the depth of the snowpack. Therefore, maximum refreezing happens where the combination of both is optimal, e.g.

where decent amounts of surface melt and a thick snow pack is present. Fig. 2.16c indicates that this is generally the case in the middle part of the glacier. In the MB year 2009/10 the amount of re-freezing increased up to 5750 m a.s.l. due to increasing snow cover and then decreased from 5850 m a.s.l to the top due to decreasing melt water availability. Snowfall does not increase linearly with altitude because in the lower regions Tair exceeds the threshold for solid precipitation more of-ten than in the upper regions (Klok & Oerlemans 2002). In the MB year 2010/11 a snow cover is evi-dent over longer periods even in the lower regions of the glacier tongue (see section 2.5.4). This leads to larger amounts of refreezing in these areas with a maximum in 5650 m a.s.l. (Fig. 2.16c).

Total refreezing in the upper glacier regions in 2010/11 is less than in 2009/10 because of less sur-face melt (Fig. 2.16b). The snow cover pattern and therefore the amount of refreezing are further modified through terrain effects and snow redistribution (Fig. 2.15, see section 2.4.4). They do not necessarily show a dependency on altitude. Within the considered period, the annual mean propor-tion of refreezing reaches 37% in the upper glacier regions and ranges between 2% (2009/10) and 13% (2010/11) in the lowermost altitude band. The less negative annual mean glacier wide b in 2010/11 can be explained by a larger amount of solid precipitation (Fig. 2.16c) which leads to higher net accumulation and additionally to lower surface melt (Fig. 2.16b) through increasing α (Fig. 2.16g) through decreasing SWnet (Fig. 2.16d) and Ts (expressed through LWout, Fig. 2.16f).

The feedback processes stress the importance of the effect of the albedo on glacier SEB and MB. This has been reported in several studies on glaciers in Central Asia (e.g. Aizen & Aizen 1994; Caidong &

Sorteberg 2010; Fujita 2007; Yang et al. 2011) and in other regions (e.g. Giesen et al. 2009; Mölg et al. 2008, 2009; Van den Broeke et al. 2011; Van Pelt et al. 2012).

Fig. 2.17 illustrates annual cycles of glacier-wide mean monthly SEB and MB components as calcu-lated by the MB model. In general, LWin (+249.7 W m^-2) and SWin (+220.0 W m^-2) dominate energy input over the considered period, followed by Qsens (+20.4 W m^-2) and QG (+1.6 W m^-2). Energy sinks at the glacier surface are LWout (-284.8 W m^-2), SWout (-144.1 W m^-2), Qlat (-29.1 W m^-2) and Qmelt

(-22.0 W m^-2), making SWnet (+74.2 W m^-2, see Table 2.4) the most important energy source, highly depending on α. This was also observed by Li et al. (2014b) in the catchment containing Zhadang glacier, by Andreassen et al. (2008) at Storbreen (Norway) and by Mölg & Hardy (2004) at Kiliman-jaro. Some of the lowest values of SWnet are evident in or around September when larger snowfall amounts in context with storm events increase α. In summer 2009, extreme low α causes highest SWnet resulting in large amounts of Qmelt (Fig. 2.17a) and therefore surface melt (Fig. 2.17b, Fig. 2.8a).

LWnet shows a rather regular seasonal cycle playing a smaller role as energy sink in summer than in winter (Fig. 2.17a, see also Li et al. 2014b). This can be explained through the seasonal patterns of LWin, depending on Tair, e and N (see section 2.3.1) (higher in summer), and LWout, depending on Ts

(largest negative values in summer). Averaged over the whole period QC is very small and naturally tends to zero (see also Li et al. 2014b, Zhang et al. 201). In winter, when the surface is colder than the underlying layers, QC becomes positive. Slightly negative QC is evident in spring (Fig. 2.17a) when the surface warms but subsurface layers are still cold from the winter season. QPS is always negative with larger values when α is low.

Fig. 2.16: Modelled vertical profiles of (left) the annual specific mass balance and its components and (right) mean SEB components on Zhadang glacier for the MB years 2009/10 and 2010/11 (1 October – 30 September). Data are averaged for 50-m altitude bands. Note the different x-axis scales of the com-ponents (negative means mass or energy loss at the glacier surface, see section 2.3.1 and Tab. 2.2 for abbreviations).

The generally dry conditions on the TP (Aizen et al. 2002) lead to continuously negative Qlat (Braun &

Hock 2004) on monthly scales with largest values in winter when air humidity is smallest (Fig. 2.2c) and higher wind speeds drive turbulence (Fig. 1.11). In absolute values, Qlat is usually larger than Qsens

and responsible for significant mass loss through sublimation especially in winter and spring prior to the onset of the monsoon (Fig. 2.17b). The glacier-wide MB estimate for the whole period 2009-2012 is -4900 kg m^-2 (Table 2.5). In general, sublimation (-1003 kg m^-2) is the second largest factor of glacier-wide mass loss after surface melt (-6455 kg m^-2) but clearly dominates ablation in winter, when air temperatures are below 0°C and surface melt is nearly absent (Fig. 2.17b). Subsurface melt (-13 kg m^-2) plays only a minor role. Solid precipitation (+1874 kg m^-2) and refreezing (+697 kg m^-2) contribute to mass gain of the glacier. Maximum snowfall amounts occur in summer (Fig. 2.17b, see section 1.4.1). Refreezing happens in spring, when percolating melt water reaches cold winter snow layers, in summer, when a wet snow layer is present and internal water refreezes at night, and in autumn, when capillary water freezes (see section 2.3.1). When solid precipitation is small, the amount of refreezing is even larger than accumulation by snowfall (Fig. 2.17b). On average, 11% of the surface and subsurface melt refreezes, which is in line with Mölg et al. (2012) and less than the amount obtained for Xiao Dongkemadi glacier (20%, Fujita & Ageta 2000, Fujita et al. 2007). A reason

for the latter can be found in the different glacier types according to their climate setting. After Shi &

Liu (2000) Zhadang glacier is a subcontinental (subpolar, polythermal) glacier type whereas Xiao Dongkemadi glacier belongs to the extreme continental (polar, cold) type with lower ice tempera-tures and therefore larger cold content. Nevertheless, the amount of refreezing confirms previous studies that refreezing is a significant process for many Asian glaciers (Ageta & Fujita 1996, Fujita &

Ageta 2000).

Table 2.4: Mean absolute values of energy flux components as modelled for 27 April 2009 – 10 June 2012 and for the two MB years with proportional contribution to total energy flux.

Sum* SWnet LWnet Qsens Qlat QG Qmelt

Table 2.4 lists the absolute and relative contributions of energy flux components to total energy flux.

For the total considered period, SWnet accounted for 44% of the total energy flux, followed by LWnet

(26%). Qlat, Qsens and QG contribute with 17%, 12% and 1%, respectively. In the MB year 2009/10 larger SWnet was responsible for more Qmelt available at the surface, compared to the year 2010/11 (Table 2.4, Fig. 2.17). These results are in agreement with those of Zhang et al. (2013).

Table 2.5: Calculated glacier-wide mass balance components for the total period 27 April 2009 – 10 June 2012 and for the two MB years.

Solid precipitation Surface melt Refrozen water Subsurface melt Sublimation Mass balance*

total [kg m^-2] 1873.5 -6454.7 697.0 -12.5 -1002.9 -4899.5±2040

2009/10 632.3 -2106.5 245.1 -4.5 -333.7 -1567.3±653

2010/11 711.1 -915.7 217.8 -3.5 -278.6 -268.8±653

% of mass loss 88 0.2 12

*Mass balance = solid precipitation - surface melt + Refrozen water - subsurface melt - sublimation

From the AWS forced model period 27 April 2009 – 10 June 2012 two glacier-wide annual MB estimates (1 October to 30 September) for Zhadang glacier can be obtained and compared with previous studies. Due to the gap filling with HAR data (see section 2.2.3) we obtain a continuous time series. Modelled MB is -1567.3±653 kg m^-2 for 2009/10 and -268.8±653 kg m^-2 for 2010/11 (Table 2.5). Given uncertainties are obtained through model calibration (see section 2.4.1). MB values are within the range published by Mölg et al. (2014) with a HAR forced MB model of similar structure as the model applied here (Fig. 2.26). The results from this study are also comparable to Zhang et al.

(2013) who employed an AWS forced MB model based on daily mean input variables (MB 2009/10:

-1969 kg m^-2, MB2010/11: -431 kg m^-2). Table 2.5 shows calculated MB components and their relative contribution to total mass loss for the total period and the two MB years. As it can be derived from Fig. 2.17b, specific MB in 2009/10 was more negative than in 2010/11 due to less snowfall and increased surface melt, subsurface melt and sublimation. In total, effective melt (surface melt – refrozen water) accounts for 88% of total mass loss followed by sublimation (12%). Mass loss through subsurface melt is negligible (0.2%). In 2010/11 the total amount of refreezing is less due to decreased surface melt, whereas the proportion of refrozen melt water is much larger than in 2009/10 with 24% compared to 12% in 2009/10.

Fig. 2.17: Glacier-wide monthly (a) SEB components (see section 2.3.1 for abbreviations) and (b) MB compo-nents from May 2009 to May 2012 at Zhadang glacier.