for spring, correlations being negative, i.e. the less (more) frequent dry-warm days
in spring, the higher (lower) yearly snow accumulation. Thus, this pattern of snow
evolution, which is the most frequent, responds inversely to the conditions that
hamper snow accumulation and favour snow melting. There are also significant
correlations (though less in number) with the number of dry-cold days in spring,
thus conditions that favour the conservation of the snowpack. In general, years in
which cold days (regardless precipitation) prevail over warm days, are years with
higher snow accumulation. It may seem strange that conditions that favour snow
accumulation (wet-cold days) do not show a significant explanatory capacity for
snow evolution (except for some cases, as can be seen by the dots of the boxplot,
which correspond to the 95th and 99th percentiles of the distribution); but this shall
be interpreted in terms of limiting conditions: wet-cold days is not a limiting factor
for inter-annual snow evolution, meaning that there will still be snowfalls in days
that are not extremely cold and particularly wet. In contrary dry-warm days is a
strong limiting factor for snowpack consolidation.
PC2 which explains less proportion of variance, and thus represents fewer snow
poles, shows a different evolution, and different explanatory variables. The index
that best correlates with PC2 is DC in spring, with higher correlations than with
PC1. Thus, in this case, the conditions favourable for snowpack conservation are a
limiting factor. But we also found significant correlations with wet-cold days in
spring, i.e. with conditions that favour snowfalls and snow accumulation. Contrary
to PC1, where conditions that can favour snowfalls were not a limiting factor, in
this case, the occurrence of wet-cold days determines whether the year will by more
or less snowy.
To find a geographical explanation for the differentiation of PC1 and PC2 and
their differing limiting factors, we have performed a Mann–Whitney–Wilcoxon
Test, for the following dependent variables: elevation, latitude and longitude.
Previously, we separated the snow poles that best correlated with PC1 from the
snow poles that best correlated with PC2, and the test was performed over these two
populations. Results of the test inform that elevation of snow poles from PC1 is
higher (mean elevation = 2300 m) and differs significantly (p-value = 0.015) from
elevation of snow poles from PC2 (mean elevation = 2100). Latitude and longitude
did not show significant differences between snow poles of PC1 and PC2 (p values = 0.656 and 0.657 respectively). This is physically consistent with the fact that
cold days were not a limiting factor for snow evolution of PC1, but they were for
snow evolution of PC2: the higher is the elevation, the less dependent is snow
accumulation to conditions of extreme cold, because the zero degrees isotherm
(which marks the limit for snowpack to consolidate) will be reached more frequently. Thus at high elevations, the occurrence of warm days, especially in spring
is observed to be the most important limiting factor for the consolidation of the
snowpack. Our results suggest that precipitation (absence or particularly high
values) is not a factor controlling snowpack variability. This result may be, however, a shortcoming of the joint-quantile indices used. Several studies demonstrated
that precipitation can be a good predictor of snowpack variability when used as a
single covariate. However, its predictive power depends greatly, on the elevation:
314
E. Morán-Tejeda et al.
in spring, the higher (lower) yearly snow accumulation. Thus, this pattern of snow
evolution, which is the most frequent, responds inversely to the conditions that
hamper snow accumulation and favour snow melting. There are also significant
correlations (though less in number) with the number of dry-cold days in spring,
thus conditions that favour the conservation of the snowpack. In general, years in
which cold days (regardless precipitation) prevail over warm days, are years with
higher snow accumulation. It may seem strange that conditions that favour snow
accumulation (wet-cold days) do not show a significant explanatory capacity for
snow evolution (except for some cases, as can be seen by the dots of the boxplot,
which correspond to the 95th and 99th percentiles of the distribution); but this shall
be interpreted in terms of limiting conditions: wet-cold days is not a limiting factor
for inter-annual snow evolution, meaning that there will still be snowfalls in days
that are not extremely cold and particularly wet. In contrary dry-warm days is a
strong limiting factor for snowpack consolidation.
PC2 which explains less proportion of variance, and thus represents fewer snow
poles, shows a different evolution, and different explanatory variables. The index
that best correlates with PC2 is DC in spring, with higher correlations than with
PC1. Thus, in this case, the conditions favourable for snowpack conservation are a
limiting factor. But we also found significant correlations with wet-cold days in
spring, i.e. with conditions that favour snowfalls and snow accumulation. Contrary
to PC1, where conditions that can favour snowfalls were not a limiting factor, in
this case, the occurrence of wet-cold days determines whether the year will by more
or less snowy.
To find a geographical explanation for the differentiation of PC1 and PC2 and
their differing limiting factors, we have performed a Mann–Whitney–Wilcoxon
Test, for the following dependent variables: elevation, latitude and longitude.
Previously, we separated the snow poles that best correlated with PC1 from the
snow poles that best correlated with PC2, and the test was performed over these two
populations. Results of the test inform that elevation of snow poles from PC1 is
higher (mean elevation = 2300 m) and differs significantly (p-value = 0.015) from
elevation of snow poles from PC2 (mean elevation = 2100). Latitude and longitude
did not show significant differences between snow poles of PC1 and PC2 (p values = 0.656 and 0.657 respectively). This is physically consistent with the fact that
cold days were not a limiting factor for snow evolution of PC1, but they were for
snow evolution of PC2: the higher is the elevation, the less dependent is snow
accumulation to conditions of extreme cold, because the zero degrees isotherm
(which marks the limit for snowpack to consolidate) will be reached more frequently. Thus at high elevations, the occurrence of warm days, especially in spring
is observed to be the most important limiting factor for the consolidation of the
snowpack. Our results suggest that precipitation (absence or particularly high
values) is not a factor controlling snowpack variability. This result may be, however, a shortcoming of the joint-quantile indices used. Several studies demonstrated
that precipitation can be a good predictor of snowpack variability when used as a
single covariate. However, its predictive power depends greatly, on the elevation:
314
E. Morán-Tejeda et al.