Potential controls on net accumulation distribution

Non-parametric correlation of the potential control variables with net accumulation data tested to see which variables (if any) could account for observed spatial variability (Table 2.3). Relationships with elevation were weak at TG, but statistically significant at FJG in 2008 and 2009. Due to the importance placed on the elevation-accumulation relationship in snow accumulation modelling, closer inspection of the

elevation-accumulation relationship is made (Figure 2.5). On both glaciers, the elevation-accumulation area covers a small elevation range, compared to the elevation range over the entire glacier.

However, it is in the accumulation area where the relationship between elevation and accumulation is found to weaken, demonstrated by the significantly lower regression coefficients. There are no mass balance data for 2008 on FJG between 500-1800 m a.s.l.

as this section of the glacier is characterised by ice falls and crevassing. Previous measurements at ~900 m a.s.l. (Anderson et al., 2006) indicate a strong relationship between elevation and mass balance on the lower glacier trunk.

Table 2.3: Results of Spearman's Rank correlation between net snow accumulation and variables that potentially influence net accumulation distribution. Results that are statistically significant at 0.01 are in bold, and those significant at 0.05 are italicised.

Variables FJG 2007 (n=14)

1 Temperature depends largely on elevation and influences surface melting through sensible heat fluxes.

Decreasing temperature with increasing elevation also influences the position of the snow/rain temperature threshold.

2 The rate at which an air mass cools and condenses as it rises up and over an orographic barrier influences precipitation onset and intensity. In this case net accumulation is related to the distance from the Alpine Fault or alternatively the Main Divide.

3 Aspect affects melting by influencing the duration and intensity of sun exposure. Aspect also influences the orientation of the snow surface to varying wind directions.

4 Slope angle influences gravitational processes (i.e. avalanche). Slope angle also influences wind redistribution processes through its effect on wind dynamics.

5 Insolation is a direct source of energy for melting.

6 The exposure or sheltering of the snow surface to different wind directions influences the susceptibility of the snow surface to wind redistribution.

Figure 2.5: Relationship between elevation and mass balance for the 2008 mass balance year.

A. Mass balance data from both the accumulation and ablation area (measured by stakes this study). B. Accumulation area only. An outlier at TG below 1000 m a.s.l. is due to insulating effects of surface moraine. The r² of the regression between elevation and net accumulation is shown.

Aside from elevation, other variables that demonstrated statistical significance with net accumulation include the wind exposure indices, aspect, distance from the Alpine Fault, distance from the Main Divide and insolation. On FJG correlation coefficients show that higher net accumulation is associated with measurement points that were exposed to NW (and NE) winds, yet sheltered from SE winds. Results for aspect showed that higher (lower) accumulation was associated with points on northerly and easterly (southerly and westerly) aspects. Interestingly, measurement points with higher ablation

season insolation had higher net accumulation. Only in 2008 was a significant

relationship found between net accumulation and distance from the Main Divide, with increased net accumulation associated with decreased distance from the Main Divide (further distance from the Alpine Fault).

PCA analysis followed by multiple-regression of PC factor scores produced statistically significant results for FJG in 2008 and 2009 (Table 2.4). In 2008 PC1 had maximum loadings on elevation and N-S aspect, while PC2 was dominated by the NE/SW wind exposure indices. In 2009 PC1 loadings were on elevation, E-W aspect and insolation, PC2 was dominated by wind exposure indices (NE/SW) and PC3 by distance from the Alpine Fault and Main Divide, the orographic indicators.

Table 2.4: Principle Component Analysis of variables influencing net accumulation distribution on Franz Josef Glacier for the 2008 and 2009 mass balance years. Refer to footnotes in Table 3 for additional information about the processes associated with each of the variables.

Variables 2008 2009

On TG less statistical significance was gained, but in 2008 and 2009 increased net accumulation was associated with reduced distance to the Alpine Fault. Distance from the Main Divide was not a good predictor at TG, due to the range curving around the top of the accumulation area, resulting in most points having similar proximity to the Main Divide (Figure 2.1). Measurements of net accumulation that were lee of SW winds, and/or had southerly aspects, had higher accumulation in 2008. In 2009 there is an unusual positive correlation between slope and net accumulation, possibly influenced

by the small data-set. Multiple-regression of PC factors did not produce statistically significant results for TG in any year.

As well as a statistically significant difference in net accumulation between the three FJG snowfields, there was also a significant difference in some of the control variables between the snowfields, with the exception of SW and NE wind exposure indices and slope. In particular largest differences were found in elevation, distance from Main Divide and ablation season insolation (Table 2.3). The Geikie Snowfield (which had lowest net accumulation) has the lowest elevation, is the greatest distance from the Main Divide, has lowest insolation, and is most exposed to SE winds (sheltered from NW). Conversely Davis Snowfield (which recorded the highest net accumulation) has the highest elevation, is closest to the Main Divide, has highest insolation and is most exposed to the NW winds (sheltered from SE).

The fact that the Davis Snowfield recorded highest net accumulation and highest insolation is surprising and warrants further investigation. Using a positive degree-day approach, we would expect summer ablation at the Geikie Snowfield to be ~0.6 m w.e.

more than at Davis Snowfield, based on an elevation difference of 250 m a.s.l. Ablation from net radiation is implicitly included in a degree-day model, but we can estimate the proportion of melting attributable to net radiation using the energy balance for a melting snow surface (Neale and Fitzharris, 1997). The net radiation difference between

measurement points in the Davis and Geikie Snowfields was 178 MJ m^-2 (assuming albedo at 0.65), which can only account for a ~0.2 m w.e difference in ablation.

Therefore the elevation-temperature difference outweighs the net radiation effect, resulting in the Davis Snowfield experiencing lower ablation (and higher net accumulation) compared to the Geikie Snowfield.