Understanding the hydrological and hydrochemical functioning of glacierized
catchments requires the knowledge of the different controlling factors and
their mutual interplay. For this purpose, the present study was carried out
in two sub-catchments of the glacierized Sulden River catchment
(130 km
At the monthly sampling scale, complex spatial and temporal dynamics for
different spatial scales (0.05–130 km
Runoff from glacierized catchments is an important fresh water resource to downstream areas (Kaser et al., 2010; Viviroli et al., 2011). High-elevation environments face rapid and extensive changes through retreating glaciers, reduced snow cover and permafrost thawing (Harris et al., 2001; Dye, 2002; Beniston, 2003; Galos et al., 2015). This will have impacts on runoff seasonality, water quantity and water quality (Beniston, 2006; Ragettli et al., 2016; Gruber et al., 2017; Kumar et al., 2019). Therefore, better understanding the behavior of high-elevation catchments and their hydrological and hydrochemical responses at different spatial and temporal scales is of uttermost importance in view of water management, water quality, hydropower and ecosystem services under the current phase of climate change (Beniston, 2003; Viviroli et al., 2011; Beniston and Stoffel, 2014).
In general, the hydrological response of catchments (i.e., runoff dynamics) is controlled by heterogeneous catchment properties (Kirchner, 2009), which become more diverse in catchments with large complexity in various landscape features like, in the case of mountainous, high-elevation glacierized catchments (Cook and Swift, 2012). In fact, those catchments are deemed to be highly dynamic geomorphological, hydrological and biogeochemical environments (Rutter et al., 2011). The advances in tracer and isotope hydrology made during the last decades can substantially contribute to gain more insight into the variability in different runoff components of high-elevation catchments (Vaughn and Fountain, 2005; Maurya et al., 2011; Xing et al., 2015; Penna et al., 2017b), catchment conceptualization (Baraer et al., 2015; Penna et al., 2017a) and sensitivity to climate change (Kong and Pang, 2012).
The main controls on hydrological and hydrochemical catchment responses are represented by climate, bedrock geology, surficial geology, soil, vegetation, topography, drainage network (Devito et al., 2005; Williams et al., 2015) and catchment shape (Sivapalan, 2003). These catchment properties may affect the partitioning of incoming water and energy fluxes (Carrillo et al., 2011).
First, a major role is attributed to the global and regional climate, having strong impacts on mountain glaciers and permafrost, streamflow amount and timing, water quality, water temperature, and suspended sediment yield (Milner et al., 2009; Moore et al., 2009; IPCC, 2013). The impact of climate is difficult to assess because it requires long time windows (e.g., decades), whereas meteorological drivers interact at a smaller temporal scales and thus are easier to quantify. Among different meteorological drivers, radiation fluxes at the daily timescale were identified as the main energy source driving melting processes in glacierized catchments in different climates (Sicart et al., 2008). Beside radiation, air temperature variations generally correlate well with streamflow under the presence of snow cover (Swift et al., 2005) and may affect the daily streamflow range (Penna et al., 2016; Zuecco et al., 2018) and streamflow seasonality (Hock et al., 1999; Cortés et al., 2011) only after an air temperature threshold has been reached.
Geology sets the initial conditions for catchment properties (Carrillo et al., 2011). The geological setting strongly controls catchment connectivity, drainage, groundwater discharge (Farvolden, 1963), runoff response (Onda et al., 2001), residence time (Katsuyama et al., 2010), hydrochemistry during baseflow conditions (Soulsby et al., 2006a) and melting periods (Hindshaw et al., 2011), and subglacial weathering (Brown and Fuge, 1998). Also geomorphological features such as talus fields may affect streamflow and water quality, resulting from different flow sources and flow pathways (Liu et al., 2004). Catchment storage, as determined by both geology and topography, was found to impact the stream hydrochemistry as well (Rinaldo et al., 2015).
The catchment hydrological conditions, commonly referring to the antecedent soil moisture, are also a relevant driver of the hydrological response (Uhlenbrook and Hoeg, 2003; von Freyberg et al., 2017). Specifically in high-elevation and high-latitude catchments, permafrost thawing also affects the hydrological connectivity (Rogger et al., 2017), leading to a strong control on catchment functioning as it drives the partitioning, storage and release of water (Tetzlaff et al., 2014). In more detail, retreating permafrost may also result in distinct geochemical signatures (Clark et al., 2001; Lamhonwah et al., 2017) and the release of heavy metals being previously stored in the ice (Thies et al., 2007; Krainer et al., 2015). As those contaminants do not only affect the water quality but also the aquatic biota such as macroinvertebrate communities in high-elevation and high-latitude environments (Milner et al., 2009), the hydrochemical characterization of permafrost thawing (i.e., from rock glaciers as a specific form of permafrost) and its impact on stream hydrology deserves further investigation (e.g., Williams et al., 2006; Carturan et al., 2016; Nickus et al., 2015; Colombo et al., 2017).
Although the effect of catchment characteristics and environmental conditions on stream hydrochemistry at different spatial and temporal scales has been studied well in lowland and midland catchments (e.g., Wolock et al., 1997; McGuire et al., 2005; Tetzlaff et al., 2009), only few studies have focused on this aspect in glacierized or permafrost-dominated catchments (Wolfe and English, 1995; Hodgkins, 2001; Carey and Quinton, 2005; Lewis et al., 2012; Kumar et al., 2018). In fact, investigating the geological, meteorological and topographic controls on catchment response and stream water hydrochemistry in high-elevation catchments is essential when analyzing the origin of hydrochemical responses in larger catchments (Chiogna et al., 2016; Natali et al., 2016), calibrating hydrological models (Weiler et al., 2017) and analyzing catchment storages (Staudinger et al., 2017).
In this paper, we aim to fill this knowledge gap by analyzing hydrochemical data from a 2-year monitoring campaign in two nearby glacierized catchments in the eastern Italian Alps, characterized by a similar size and climate but contrasting geological setting. We hypothesize that the markedly different geological properties affect the geochemistry and the hydrological response of both catchments. We test this hypothesis by sampling different water sources (precipitation, stream water, groundwater, snowmelt and glacier melt) for the electrical conductivity (EC), turbidity, and major, minor and trace element analysis.
Within the present study, we specifically aim to answer the following
research questions:
Does the temporal pattern of the hydrochemical stream signature in the
two catchments reflect the dominant rock substratum? Do nivo-meteorological indicators (precipitation, air temperature, solar
radiation and snow depth) impact the stream hydrochemical response during the
melting period? What is the temporal relationship of discharge and tracer characteristics
in the stream?
The study was carried out in the Sulden (Solda) River catchment, located in
the upper Vinschgau (Venosta) Valley (eastern Italian Alps; Fig. 1). The size
of the study area is about 130 km
Overview of the Sulden catchment with
The study area had a glacier extent of about 16.9 km
Meteorological characteristics of the weather station Madritsch (Madriccio) at 2.825 m a.s.l. in 2014 and 2015.
Precipitation, air temperature, humidity and snow depth are measured by an
ultrasonic sensor at a 10 min measuring interval at the automatic weather
station (AWS) Madritsch (Madriccio) at 2825 m a.s.l., run by the
Hydrographic Office, Autonomous Province of Bozen-Bolzano (Fig. 1). We take
data from this station to be representative for the glacier in the catchment
at a similar elevation. At the catchment outlet at Stilfserbrücke (Ponte
Stelvio), water stages are continuously measured by an ultrasonic sensor
(Hach Lange GmbH, Germany) at the 10 min measuring interval and converted to discharge via a flow
rating curve using salt dilution or photometric measurements (measurement
range: 1.2–23.2 m
Topographical data (such as catchment area and 50 m elevation bands) were derived from a 2.5 m digital elevation model.
Stream water sampling at the outlet was performed by an automatic sampling
approach using an Isco 6712 system (Teledyne Technologies, USA). Daily water
sampling took place from mid-May to mid-October 2014 and 2015 (on 331 d,
mainly during meltwater conditions) at 23:00 (all times reported in this
study refer to solar time) to ensure consistent water sampling close to the
discharge peak. In addition, grab samples were taken from different stream
locations, tributaries and springs in the Sulden and Trafoi sub-catchments
and the outlet, following the sampling scheme of Penna et al. (2014) to
account for spatial variability in the hydrochemistry at the catchment scale.
Sampling took place monthly from February 2014 to November 2015 (Table 2).
Samples were collected approximately at the same time (within less than an
hour of difference) on all occasions. In winter, however, a different
sampling time had to be chosen for logistical constraints (up to 4 h of
difference between both sampling times). However, this did not produce a bias
in the results due to the very limited variability in the hydrochemical
signature of water sources (related to nearly constant discharge) during
winter baseflow conditions (Immerzeel et al., 2012). Three outflows from two
active rock glaciers were selected to represent meltwater from permafrost
because rock glaciers are considered to be long-term creeping ice-rock
mixtures under permafrost conditions (Humlum, 2000). Located on quartz
phyllite bedrock in the upper Sulden sub-catchment, three springs at the base
of the steep rock glacier front at about 2600 m a.s.l. were sampled monthly
from July to September 2014 and July to October 2015. Snowmelt water was
collected as dripping water from snow patches from April to September 2014
and March to October 2015 (
Topographical characteristics of sub-catchments defined by sampling points.
All samples were stored in 50 ml PVC bottles with a double cap and no
headspace. The samples were kept in the dark at 4
The
The analysis of major, minor and trace elements (Li, B, Na, Mg, Al, K, Ca, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Rb, Sr, Mo, Ba, Pb and U) was carried out by inductively coupled plasma mass spectroscopy (ICP-MS iCAP Q, Thermo Fischer) at the laboratory of Eco-Research S.r.l (Bolzano).
In order to better understand the effect of meteorological controls at different timescales, different nivo-meteorological indicators derived from precipitation, air temperature, solar radiation and snow depth data from AWS Madritsch were calculated (Table 3).
Nivo-meteorological indicators derived from the weather station Madritsch (Madriccio) at 2825 m a.s.l.
We performed a temporal sensitivity analysis to better understand at which
temporal scale these nivo-meteorological indicators affect the hydrometric
and hydrochemical stream response at the outlet. For that purpose, we
calculated the indicators for each day of stream water sampling and included
in the calculations a period of time of up to 30 d prior to the sampling day
by using a 1 d incremental time step. As precipitation indicators, we
considered the cumulated precipitation
The temporal sensitivities of agreement between nivo-meteorological
indicators and hydrochemical signatures were expressed as Spearman's rank
correlation coefficients (
In order to understand the link among water sources and their hydrochemical composition, a principle component analysis (PCA), using data centered to zero and scaled to variance 1 (R core team, 2016), was performed. Data below detection limit were excluded from the analysis.
To assess the dampening effect of meltwater on stream water chemistry during
baseflow conditions and the melting period, the variability coefficient (VC)
was calculated following Sprenger et al. (2016; Eq. 1):
We applied a two-component mixing model based on EC and
Element concentrations of stream and rock glacier spring water are presented
in Table S1 and S2 in the Supplement. It is worth highlighting that heavy
metal concentrations (such as Al, V, Cr, Ni, Zn, Cd and Pb) showed the
highest concentrations during intense melting in July 2015 at all six
locations (partly exceeding concentration thresholds for drinking water –
see European Union (Drinking Water) Regulations, 2014). Element
concentrations were clearly higher at the most upstream sampling locations.
Relatively low variability coefficients (VC
In contrast, other element concentrations (such as As, Sr, K and Sb) generally
revealed higher concentrations during baseflow conditions and lower
concentrations during the melting period. This observation was corroborated
by relatively high variability coefficients for As (VC: 2–2.9) and Sb (VC:
2–2.2) at S1, S2 and T1. For example, while the highest Sr concentrations were
measured at S6, As was highest at the downstream locations T1, S2 and S1.
Regarding the rock glacier springs, their hydrochemistry showed a gradual
decrease in As and Sr concentration from July to September 2015. The observed
geochemical patterns are confirmed by PCA results (Fig. 2) and the
correlation matrix (Fig. 3), revealing that geochemical dynamics are driven
by temporal (PC1) and spatial controls (PC2) and a typical clustering of
elements, respectively. PC1 showed high loadings for heavy metal
concentrations (such as Al, V, Cr, Ni, Zn, Cd and Pb), supporting the clear
temporal dependency for the entire catchment (baseflow conditions vs. melting
period; Fig. 2a). PC2 was instead mostly characterized by high loadings of
Principle component analysis of element concentrations of stream
water and springs draining a rock glacier sampled in the Sulden and Trafoi
sub-catchments from March to October 2015. Data based on
Spearman's rank correlation matrix of hydrochemical variables. Values
are shown for a level of significance
The temporal and spatial variability in EC in the Sulden and Trafoi River
along the different sections, their tributaries and springs is illustrated in
Fig. 4. During baseflow conditions, from late autumn to early spring prior to
the onset of the melting period in May–June, water enriched with solutes had
an important impact on stream hydrochemistry, as stream and tributary
locations showed the most increased conductivity, ranging from 132.5 to
927
Spatial and temporal variability in EC (
Hereinafter, the hydrochemistry of the Sulden and Trafoi sub-catchment is
analyzed in terms of hydrochemical patterns of the main stream, tributaries,
springs and runoff contributions at the most downstream sampling location
above the confluence. At T1 and S2, hydrochemistry was statistically
different in its isotopic composition (Mann–Whitney rank sum test:
By the aid of both tracers, catchment-specific hydrochemical characteristics such as contrasting EC gradients along the stream were revealed (Figs. 4 and 5). EC in the Trafoi River showed a linear increase with increasing catchment area (from T3 to T1) during baseflow and melting periods (EC enrichment gradient).
Spatial variability in electrical conductivity along the Trafoi and Sulden River against catchment area. Electrical conductivity is averaged for sampling days during baseflow conditions (21 January, 26 February and 18 March 2015) and melting period (12 June, 18 July, 11 August and 9 September 2014).
In contrast, the Sulden River revealed relatively high EC
(926
Variability coefficient (VC) for selected locations along the Sulden and Trafoi River in 2014 and 2015.
Regarding the hydrochemical characterization of the tributaries in both
sub-catchments (Fig. 4), Sulden tributaries were characterized by a
relatively low EC variability (68.2–192.3
To identify the effect of meteorological controls at high elevations on the hydrometric and hydrochemical stream response at the outlet, we first present the relationship between meteorological parameters against snow depth differences (Fig. 6). Then, we show snow depth differences compared with discharge, EC and isotopic data (Fig. 7).
Box plots of environmental variables
Box plots of snowmelt expressed as snow depth differences at AWS
Madritsch on the variability in
Among the nivo-meteorological indicators listed in Table 3, daily maximum air
temperature
With respect to
As a consequence, high-elevation snowmelt played an important role in
explaining both the hydrometric and hydrochemical response at the outlet
Stilfserbrücke (Fig. 7). During the snowmelt period, discharge at the
outlet clearly increased with increasing snowmelt due to snow depth losses at
high elevation. For example, median discharges of 6.25 and
7.5 m
Moreover, the increasing amount of snowmelt resulted in decreasing EC and
lower
The temporal variability in the hydrochemical variables observed at the
catchment outlet and of the meteorological drivers is illustrated in Fig. 8.
Controlled by increasing radiation inputs and air temperatures above about
5
Time series from 2014 and 2015 of
Water turbidity was highly variable at the outlet and mirrored the discharge
fluctuations induced by meltwater or storm events. Winter low flows were
characterized by very low turbidity (
Furthermore, the interannual variability in meteorological conditions with
respect to the occurrence of days exceeding the 6.5 or 15
The daily variations in air temperature, discharge, turbidity and EC showed marked differences in the peak timing. Daily maximum air temperature generally occurred between 12:00 and 15:00, resulting in discharge peaks at about 22:00 to 01:00 in early summer and at about 16:00 to 19:00 during late summer. Turbidity peaks were measured at 22:00 to 23:00 in May to June and distinctively earlier at 16:00 to 19:00 in July and August. In contrast, the EC maximum occurred shortly after the discharge peak between 00:00 to 01:00 in early summer and at 11:00 to 15:00, clearly anticipating the discharge peaks.
It is interesting to highlight complex hydrochemical dynamics during the
baseflow period in November 2015, which were interrupted only by a
precipitation event on 28 and 29 October 2015. This event was characterized
by more liquid (12.9 mm) than solid precipitation (6.6 mm) falling on a
snowpack of about 10 cm (at 2825 m a.s.l.). While stream discharge showed
a typical receding hydrograph confirmed by EC being close to the background
value of about 350
To better characterize the temporal dynamics of hydrochemical variables,
Fig. 9 shows the different relationships of discharge, EC,
Monthly relationships between
During these periods, the
Finally, we evaluated the hysteretic pattern of discharge and EC in more
detail by comparing it against
Monthly relationships between discharge and electrical conductivity
(EC) at the outlet Stilfserbrücke with respect to
Hydrochemical dynamics were driven by a pronounced release of heavy metals (such as Al, V, Cr, Ni, Zn, Cd and Pb) shown for the entire catchment and, in contrast, by a specific release of As and Sr in the upper and lower Sulden sub-catchment (Fig. 2). Yet, as the explained variance was only at about 53 %, further controls may be present. In this context, PC3 explained 11.8 % of additional variance and may characterize the hydrochemistry of surface and subsurface flows resulting from different residence times within the different soils and rocks.
With respect to PC1, several sources of heavy metals could be addressed: these elements may be released by rock weathering on freshly exposed mineral surfaces and sulfide oxidation, typically produced in metamorphic environments (Nordstrom, 2011). Proglacial stream hydrochemistry may also strongly depend on the seasonal evolution of the subglacial drainage system that contributes to the release of specific elements (Brown and Fuge, 1998). In this context, rock glacier thawing may play an important role for the release of Ni (Thies et al., 2007; Mair et al., 2011; Krainer et al., 2015) and Al and Mn (Thies et al., 2013). However, high Ni concentrations were not observed in this study. Moreover, high heavy metal concentrations were measured during the melting period in mid-summer, which would be generally too early to derive from permafrost thawing (Williams et al., 2006; Krainer et al., 2015). Also bedrock weathering as major origin probably needs to be excluded because low concentrations of heavy metals occurred in winter when the hydrological connectivity at higher elevations was still present (according to running stream water at the most upstream locations).
It is therefore more likely that heavy metals derive from meltwater itself,
as the spatial and temporal dynamics indicated. The element release is
strongly coupled with melting and infiltration processes, when hydrological
connectivity within the catchment is expected to be highest during the
snowmelt period. To support this explanation, supplementary element analysis
of selected snowmelt (
In contrast, a clear geological source can be attributed to the origin of As and Sr, indicating bedrock-specific geochemical signatures. In the lower Sulden catchment (at locations S1, S2 and T1), As could mainly originate from As-containing bedrock. As-rich lenses are present in the cataclastic carbonatic rocks (realgar-bearing) and in the mineralized, arsenopyrite-bearing bands of quartz phyllites, mica schists and paragneisses of the crystalline basement. Different outcrops and several historical mining sites are known and described in the literature (Mair, 1996, Mair et al., 2002, 2009; Stingl and Mair, 2005). In the upper Sulden catchment, the presence of As is supported by the hydrochemistry of rock glacier outflows in the Zay sub-catchment (corresponding to the drainage area of ST2; Engel et al., 2018) but was not reported in other studies (Thies et al., 2007; Mair et al., 2011; Krainer et al., 2015; Thies et al., 2013). Also high-elevation spring waters in the Matsch Valley corroborated that As and Sr concentrations may originate from paragneisses and mica schists (Engel et al., 2017). However, the gradual decrease in As and Sr concentrations from rock glacier springs clearly disagrees with the observations from other studies that rock glacier thawing in late summer leads to increasing element releases (Williams et al., 2006; Thies et al., 2007; Krainer et al., 2015; Nickus et al., 2015). We suggest a controlling mechanism as follows: As and Sr originate from the quartz phyllite rocks, which form the bedrock of the rock glaciers (see Andreatta, 1952; Montrasio et al., 2012). Weathering and former subglacial abrasion facilitate this release (Brown, 2002). As- and Sr-rich waters may form during winter when few quantities of water percolate in bedrock faults and then are released due to meltwater infiltration during summer (Volkmar Mair, personal communication, 2018). As a clear delayed response of heavy metal concentrations in rock glacier outflow is revealed, the infiltration and outflow processes along flow paths in the bedrock near the rock glaciers may take up to 2 months to hydrochemically respond to snowmelt contamination (Hood and Hayashi, 2015).
As a consequence, a clear hydrochemical signature of permafrost thawing is difficult to find, and results may lack the transferability to other catchments, as not all rock glaciers contain specific elements to trace (Colombo et al., 2017). In this context, as precipitation and snowmelt affect the water budget of rock glaciers (Krainer and Mostler, 2002; Krainer et al., 2007), potential impacts of atmospheric inputs on rock glacier hydrochemistry can be assumed and deserve more attention in future (Colombo et al., 2017).
Furthermore, export of elements in fluvial systems is complex and may strongly be affected by the pH (Nickus et al., 2015) or interaction with solids in suspension (Brown et al., 1996), which could not be addressed in this study. Further insight on catchment processes might be gained, considering also element analysis of the solid fraction, to investigate whether water and suspended sediment share the same provenance.
Superimposing the impact of the geological origin, melting processes were
controlled by meteorological conditions, affecting stream hydrochemistry
during summer, as shown by isotope dynamics (Figs. 4 and 8) and hydrochemical
relationships (Fig. 9). It is well known that snowmelt is mainly driven by
radiation and temperature. Generally, radiation is the main energy source
driving melt processes in glacierized catchments of different climates
(Sicart et al., 2008; Vincent and Six, 2013) and may integrate the effect of
cloud coverage (Anslow et al., 2008). Moreover, a high correlation between
snow or glacier melt and maximum air temperature exists (U.S. Army Corps of
Engineers, 1956; Braithwaite, 1981), thus controlling daily meltwater
contributions to streamflow (Mutzner et al., 2015; Engel et al., 2016).
In this study, we show that a
Of course, further nivo-meteorological indicators such as the extent of snow cover (Singh et al., 2005), vapor pressure, net radiation and wind (Zuzel and Cox, 1975), or turbulent heat fluxes and longwave radiation (Sicart et al., 2006) may exist but were not included in the present study due to the lack of observations.
Moreover, with respect to spatial representativeness,
The temporal sensitivity analysis and the relatively large variability
related to snow depth losses (Figs. 6 and 7) are generally difficult to
compare due to the lack of suitable studies. Moreover, we considered a
The contrasting variabilities of discharge, EC and
Tracer dynamics of EC and stable isotopes associated with monthly discharge variations generally followed the conceptual model of the seasonal evolution of streamflow contributions (for example, isotopic depletion and low EC during snowmelt period in June, less isotopic depletion and low EC during glacier melt period), as described for catchments with a glacierized area of 17 % (Penna et al., 2017a) and 30 % (Schmieder et al., 2017). However, isotopic dynamics were less pronounced compared to these studies, likely resulting from the impact of the relative meltwater contribution related to different catchment sizes, the proportion of the glacierized area (Baraer et al., 2015) or the sampling year.
In addition, the hydrometric and geochemical dynamics analyzed in this study were
controlled by interplay of meteorological conditions, the heterogeneity of
geology and topography. Such interplay is highlighted by EC dynamics
further controlled by the contributing catchment area (i.e., EC gradients
along the Sulden and Trafoi River; Wolock et al., 1997; Peralta-Tapia et al., 2015; Wu, 2018). As EC was highly correlated to Ca concentration
(Spearman's rank correlation: 0.6,
The additional effect of topographical characteristics is underlined by the
findings that the Sulden River hydrochemistry at S2 was significantly more
depleted in
Meteorological conditions, geology and topography explain specific
hydrometric and hydrochemical relationships at the catchment outlet. For
example, the hysteretic relationship between discharge and EC (Fig. 8b)
corresponds well with the hysteresis observed in the nearby Saldur and Alta
Val de la Mare catchment (Engel et al., 2016; Zuecco et al., 2016), although
these studies focused on the runoff event scale. The initial phase of this
hysteresis in early summer was clearly snowmelt-induced, with snowmelt likely
originating from lower elevations as
Moreover, this relationship helps to identify the conditions with maximum
discharge and EC; during baseflow conditions, the Sulden River showed
the highest EC of about 350
As more extreme weather conditions (such as heat waves and less solid winter precipitation) are expected in future (Beniston, 2003; Viviroli et al., 2011; Beniston and Stoffel 2014), glacierized catchments may exhibit more pronounced hydrochemical responses such as shifted or broader ranges of hydrochemical relationships (Kumar et al., 2019) and increased heavy metal concentrations both during melting periods and baseflow conditions. However, identifying these relationships with changing meteorological conditions would deserve more attention and is strongly limited by our current understanding of underlying hydrological processes (Schaefli et al., 2007). In a changing cryosphere, more complex processes such as non-stationarity processes may emerge under changing climate, which was found to be a major cause of non-stationarity (Milly et al., 2008). In this context, explaining apparently ambiguous processes like the one we observed during the baseflow period in November 2015 (Fig. 8) will deserve further attention.
Finally, our results underline that long-term controls such as geology and topography govern hydrochemical spatial responses (such as bedrock-specific geochemical signatures, EC gradients and relative snowmelt contribution). In contrast, short-term controls such as daily maximum solar radiation, air temperature and snow depth differences drive short-term responses (such as discharge variability and EC dilution). Both statements are in general agreement with the findings of Heidbüchel et al. (2013). However, as the catchment response strongly depended on the melting period vs. baseflow conditions, controls at longer temporal scales interact as well. Thus, our findings suggest that glacierized catchments react in a much more complex way compared to non-glacierized catchments and that catchment responses cannot be attributed to one specific scale, justified by either short-term or long-term controls alone.
In this context, the present study provides novel insights into geological, meteorological and topographic controls of stream water hydrochemistry rarely addressed for glacierized catchments so far. Moreover, this study strongly capitalizes on an important dataset that combines nivo-meteorological indicators and different tracers (stable isotopes of water, EC, and major, minor and trace elements). This aspect finally underlines the need for conducting multi-tracer studies in glacierized catchments with different geological complexity in order to evaluate whether our findings (obtained in sedimentary and metamorphic substratum) are transferable to different geological settings.
The sampling approach combined a monthly spatial sampling with daily sampling at the outlet, which is in good agreement methodologically with other sampling approaches, accounting for increasing distance of sampling points to the glacier (Zhou et al., 2014; Baraer et al., 2015), intense spatial and temporal sampling (Penna et al., 2014; Fischer et al., 2015), synoptic sampling (Carey et al., 2013; Gordon et al., 2015), and different catchment structures such as nested catchments (Soulsby et al., 2006b). Sampling covered a variety of days with typical snowmelt, glacier melt and baseflow conditions during 2014 and 2015, confirming the representativeness of tracer dynamics within 2 years with contrasting meteorological characteristics (Table 1). However, short-term catchment responses (such as storm-induced peak flows and related changes in hydrochemistry) were difficult to capture by this sampling approach and would require a higher temporal sampling resolution. In this context, the representativeness of the outlet sampling time with respect to the peak discharge time at that location may also play an important role. In fact, the peak of hydrochemical response may not be synchronized with the hydrometric one and therefore may lead to stronger or weaker relationships.
Furthermore, 2 years of field data are probably not sufficient for capturing all hydrological dynamics, the hydrological catchment status and catchment responses to specific meteorological conditions. In this regard, long-term studies may have better chances of capturing the temporal variability in hydrochemical responses (Thies et al., 2007). Although time-,energy- and money-consuming, more complex and long sampling approaches should be developed to further unravel process understanding of glacierized catchments.
Our results highlight the complex hydrochemical responses of mountain glacierized catchments at different temporal and spatial scales controlled by meteorological conditions, topography and geological heterogeneity. To our knowledge, only few studies investigated the impact of controlling factors on stream water hydrochemistry by using nivo-meteorological indicators and multi-tracer data, which we recommend establishing as a prerequisite for studies in other glacierized catchments.
The main results of this study can be summarized as follows:
Hydrometric and geochemical dynamics were controlled by interplay of
meteorological conditions and the geological heterogeneity. The majority of
the variance (PC1: 36.3 %) was explained by heavy metal concentrations
(such as Al, V, Cr, Ni, Zn, Cd and Pb), associated with atmospheric deposition
on the snowpack and release through snowmelt. Remaining variance (PC2:
16.3 %) resulted both from the presence of a bedrock-specific geochemical
signature (As and Sr concentrations) and the role of snowmelt contribution. The isotopic composition of rock glacier outflow was relatively similar to
the composition of glacier melt, whereas high concentrations of As and Sr may
more likely result from bedrock weathering. Therefore, as the underlying
geology may prevails over a thawing permafrost characteristics, a specific
hydrochemical signature of rock glacier springs was difficult to obtain. At the monthly scale, for different sub-catchments (spatial scale: 0.05–130 km For the entire study area (spatial scale: 130 km Daily maximum air temperature
Finally, this study may support future classifications of glacierized catchments according to their hydrochemical response under different catchment conditions or the prediction of appropriate end-member signatures for tracer-based hydrograph separation being valid at longer timescales.
Hydrometeorological data are available upon request at the Hydrographic Office of the Autonomous Province of Bozen-Bolzano. Tracer data used in this study are freely available by contacting the authors.
The supplement related to this article is available online at:
ME, DP, GB and FC carried out water sampling; WT supported element analysis; GV measured and maintained instrumentation; ME and DP analyzed the data; and ME, DP, GB and FC wrote the paper.
The authors declare that they have no conflict of interest.
This research is part of the GLACIALRUN project and funded by the foundation of the Free University of Bozen-Bolzano and supported by the project “Parco Tecnologico – Tecnologie ambientali”.
The authors thank Andrea Rücker, Ana Lucia, Alex Boninsegna, Raffaele Foffa and Michiel Blok for their field assistance. Giulia Zuecco and Luisa Pianezzola are thanked for the isotopic analysis at TESAF at the University of Padua, and Christian Ceccon is thanked for the isotopic analysis in the laboratory of the Free University of Bozen-Bolzano. We also thank Giulio Voto at Eco-Research S.r.l. (Bolzano) for the element analysis. We appreciate the helpful support for the geological interpretation by Volkmar Mair. We acknowledge the project AQUASED, whose instrumentation infrastructure we could use. Furthermore, we thank the Hydrographic Office and the Department of Hydraulic Engineering of the Autonomous Province of Bozen-Bolzano for providing meteorological and hydrometric data. We acknowledge the Forestry Commission Office Prat, the National Park Stilfserjoch (Passo Stelvio), and the cable car Sulden GmbH for their logistical support and helpful advice. This work was supported by the Open Access Publishing Fund provided by the Free University of Bozen-Bolzano.
This paper was edited by Carlo De Michele and reviewed by three anonymous referees.