Professional Documents
Culture Documents
First Snow Glacier and Groundwater Contribution Quantification in The Upper Mendoza River Basin Using Stable Water Isotopes PDF
First Snow Glacier and Groundwater Contribution Quantification in The Upper Mendoza River Basin Using Stable Water Isotopes PDF
To cite this article: Sebastián A. Crespo , Francisco Fernandoy , Leandro Cara , Sebastián Klarian
& Céline Lavergne (2020): First snow, glacier and groundwater contribution quantification in the
upper Mendoza River basin using stable water isotopes, Isotopes in Environmental and Health
Studies, DOI: 10.1080/10256016.2020.1797713
1. Introduction
Globally, glaciers are melting at unprecedented rates and in the Andes they display a wide-
spread retreat [1]. The water supply for the oasis irrigated by the Mendoza River depends
on the recharge of snow, which has been under extraordinary pressure following the
megadrought that affected the Central Andes since 2010 [1,2]. Even though the basin is
supplied mainly from snow contributions in years of normal to abundant loads [3], in very
dry years, hydrographs show a displacement of the maximum monthly flows from January
to February [4,5], which indicates higher contributions of glaciers to the Mendoza River
flow than during average or wet years [6]. According to the genetic classification of
solid feeding fluvial regimes, the regime of the Mendoza River could be classified in
normal years as ‘mitigated glacial’ [5,6]. However, for years of extreme drought (e.g.
1968, 2010), that maximum would be transferred to the month of February (Supplemen-
tary Figure S1, Supplementary Tables S4 and S5), becoming an ‘ultra-glacial’ regime which
would reinforce the supposed contribution of glaciers to the Mendoza River flow. In
addition, some authors mention that in dry years, the decreasing stream flows do not
follow the marked decrease of snowfall, which would be explained by a proportionally
greater contribution of ice bodies [4,7–10]. In accordance with previous hydrograph analy-
sis [4], when we analyze the Parde genetic classification of solid feeding fluvial regimes [6]
in very dry years, a displacement of the maximum monthly flows from January to February
occurs, which would reinforce the supposed contribution of glaciers to the Mendoza River
flow. Measurements were compared with the Cuevas River streamflow records, where the
ice bodies drain their water, establishing their proportion of contribution (Table 5).
The more widely used techniques for ice melt quantification in a glaciated basin (hydro-
logical and traditional or satellite imagery glacial mass balances) is generally not able to
detect the shift in the water contribution from snow to glacial melt. Those traditional
methodologies cannot detect or quantify the groundwater input in a river. Naturally occur-
ring tracers, such as ion and water stable isotopic (18O, 2H or deuterium excess) compo-
sition may facilitate that differentiation. Different contact time of water with air and
sediments is different for each water source, and results in distinct ions and stable
isotope composition [11]. These chemical properties provide natural tracers of flow
inputs along the melting season by different water sources and sub-basins to a river
basin [11]. In other similar glaciated basins from different geographical regions like the
Bhagirathi River in Indian Himalayas [12], the use of electric conductivity and 18O compo-
sition served as tracers to identify water from ice, snow, and rain water. Comparable
findings were published by Lambs et al. [13] for the Garonne Valley (France), where
runoff water from high altitudes was identified using stable oxygen isotopes and conduc-
tivity data from river water samples. Similarly, Pu et al. [14], using only 18O composition
from different water sources in the Baishui River catchment (China), built a hydrograph
separation between rain and melt water contributions to the river, but did not differentiate
snow, groundwater and ice melting contributions.
Panarello et al. have documented differences in the stable isotope composition of river
water and precipitation in the Central Andes, even between rivers in the same area such as
the Río Tunuyán and Río Mendoza [15]. Hoke et al. reported isotopic differences for pre-
cipitation in the Cordillera Frontal versus Cordillera Principal geological provinces [16].
At the same latitude as the Mendoza River basin but in the Juncal River basin, located at
the western slopes of the Andes (in the Chilean side), Ohlanders et al. [17] studied the
hydrological components of the Juncal river basin of Chile including groundwater,
glacial melt, and precipitation in the dry melt year of 2011/2012. Even when they could
not calculate exactly the water sources contribution, they assumed at least 62% of strea-
mflow water input from glacial sources (inferred from the minimum 15 December glacier
contribution) for the melting season. Continuing that work, in the same basin, Rodriguez
ISOTOPES IN ENVIRONMENTAL AND HEALTH STUDIES 3
et al. [18] presented a hydrograph separation, which aimed at identifying flow sources
among ungauged headwater sub-catchments through a combination of observed strea-
mflow and data on natural tracers including isotopes and dissolved solids along the
2011–2012 dry water year [18]. They combined isotopic and solute information for inform-
ing mixing models for hydrologic research in mountain environments, using similar PCA
and Bayesian mixing analysis as presented here. Even when they were able to provide
fairly reliable estimates of the contribution of a specific sub-catchment that is strongly
determined by the presence of a clean glacier covering a significant fraction of its
surface area, they were unable to discriminate between different types of water
sources, specially between seasonal snowpack and glacier melt, but some results are
similar the findings of this study (discussed in Section 3).
In a previous work, Crespo et al. [11] found clear differences in ionic and stable isotope
chemistry between the main mountain ranges of the Mendoza River basin, related to
different precipitation systems (Pacific and Atlantic moisture) and geological settings.
Differences between geological provinces (Cordillera Principal, Cordillera Frontal and Pre-
cordillera) were found in both ionic and isotopic compositions in a seasonal scale along
the dry period 2011–2012. Hydrological sources in the Cordillera Principal, such as ice
bodies, groundwater, snow catchments, and precipitation also showed significant differ-
ences in ionic and stable isotope chemistry. A seasonal variability in natural tracers was
also found in main streams and rivers of Cordillera Principal, suggesting changes in the
relative contribution input from different water sources, but was not possible to quantify
their contribution.
Continuing the Crespo et al. [11] work, the aim of this work was to identify and quantify
different water source contributions (snow, groundwater, glacial and periglacial environ-
ments) in a time series analysis along a melting season (November 2013–March 2014),
which represents half of the year hydrological contribution (Table S4) in a particular dry
period framed by the ‘megadrought’, with 30% snow water equivalent deficits and 0.5–1°
C warmer conditions compared with the past decade values [2]. We used two approaches:
gauging glacier flows and naturally occurring chemical tracers. Along the melting season, all
the water sources are disbursing water, overlapping their signals. To be able to separate the
different water sources, the chemical composition of the water sources was done in a weekly
time resolution, which was also recommended by Ohlanders et al. [17].
The present study constitutes the first water source quantification and hydrograph sep-
aration in the Mendoza River basin. It shows the contribution from snow, glaciers, rock gla-
ciers and groundwater and points to the strategic relevance of groundwater and water
melt from glacial and periglacial environments to maintain the river basal flow in a dry
period [1,2]. This is relevant as a basis to preparedness efforts in confronting the projected
dry, warm regional climate scenarios [2].
2. Methodology
2.1. Streamflow and environmental variables analysis
Two glacial outflows from representative cryoforms (glacier and rock glacier) of the
Mendoza River basin were measured: Horcones Inferior glacier and Mt. Tolosa rock glaciers
conglomerate, respectively. Meteorological variables were also measured at two stations
4 S. A. CRESPO ET AL.
in Laguna de Horcones at 3043 m a.s.l. (‘1’ in Figure 1) and in Plaza Francia (‘2’ in Figure 1),
at 4016 m a.s.l., on the banks of the Horcones Inferior glacier.
To estimate the melting water flow, ultrasonic limnimeter sensors were installed in the
pro-glacial rivers of the mentioned glaciers. These sensors measured the height of the
streams automatically and continuously. Periodical calibration curves of height–discharge
were developed in the field. The installed sensors were Campbell Scientifics, SR50a model,
powered by a solar panel with Campbell Scientifics data logger CR200x, with 0.25 mm res-
olution and ± 1 cm sensor accuracy.
Two different procedures for each analyzed site were applied for the construction of the
calibration curve, which links the height measured continuously by the sensor and the
volumetric flow. One was by constant saline flow [21] and the other by velocity–area
method [22].
Mt. Tolosa rock glaciers conglomerate is composed of four rock glaciers and is located
between the peaks Leñas del Tolosa and the Morro El Paso, at 32.80 °S – 70.01 °W
(Figure 1). The rock glaciers range from 3509 to 3749 m a.s.l., with an average altitude
of 3614 m a.s.l., a mean length of 378 m and a total covered area of 0.16 km2.
This site was chosen for the simplicity provided by the cryogenic origin of the rock
glaciers covering this sub-basin, accessibility throughout most of the year and because
it is near (∼ 6 km), and in the same geological province, to the Horcones Inferior glacier.
Furthermore, the similarity in orientation (south) and average altitude of both glaciated
basins reduces the environmental noise when comparing water dynamics of these
different types of ice bodies.
Figure 1. Map with the digital elevation model, sampling sites and ice bodies. Glacier shapes (marked
with contour) were taken from the national glacier inventory [19, 20].
ISOTOPES IN ENVIRONMENTAL AND HEALTH STUDIES 5
Because the Mt. Tolosa rock glaciers conglomerate has a low flow, saline tracer constant
flow measurements were performed periodically [21]. This method is especially useful in
turbulent mountain streamflow [21,23] and consists in introducing a chemical tracer sub-
stance (in this case sodium chloride) in the stream and then monitoring concentration
changes downstream. The salt solution was injected at a constant flow rate using a Mar-
iotte bottle built for that purpose [23]. Using a digital conductivity meter, calibration curves
were constructed with measurements of salinity downstream, after pouring the saline
solution.
The Horcones Inferior glacier, located in the Aconcagua Mt. Provincial Park (32.73 °S and
69.97 °W), drains to the Horcones River, affluent of the Cuevas River, which is one of the
tributaries of the Mendoza River together with Tupungato and Vacas rivers, when they
join in Punta de Vacas (Figure 1). The glacier is distributed from 3472 to 5460 m a.s.l.,
with an average altitude of 4151 m a.s.l. It presents southeast orientation, a total length
of 12.7 km and 6.95 km2 area (Figure 1). To calculate the glacier streamflow, speed–area
method gauges were performed [22].
The General Department of Irrigation (DGI) weather station (labelled as ‘1’ in Figure 1),
located at 3043 m a.s.l. in Laguna de Horcones (32.80°S–69.95°W), measured: air tempera-
ture, soil temperature, wind speed and direction, relative humidity, incident radiation and
snow water equivalent, hereinafter ‘station 1’. Air and soil temperature HOBO sensors were
also installed in the Horcones Inferior glacier (labelled as ‘2’ in Figure 1), at 4016 m a.s.l.
(32.69°S – 69.97°W), hereinafter ‘station 2’ (Table S2). To analyze each summer storm,
data from nearby weather stations, MODIS images and records from Aconcagua Park
rangers was collected (Table S3).
samples, Horcones River, Horcones Superior River, and the gauging of the Horcones
Inferior glacier the study period was from December 2013 to March 2014. Electric conduc-
tivity and temperature were measured in situ with a digital multiparametric meter (Thermo
Scientific Orion Star A329) and a conductivity electrode DuraProbe 4. Samples for isotope
analyses were taken in plastic tubes of 15 ml and sealed with a thermoplastic cohesive
(PARAFILM®) to prevent evaporation. Samples for ion concentration analysis were col-
lected in 1-L plastic bottles, refrigerated in the field and kept frozen until analysis.
315). For Cl−, electric conductivity, pH and temperature, a multiparametric sensor (Thermo
Scientific Orion Star A329) was used with specific electrodes for chlorides, electric conduc-
tivity and pH. The SO−2 4 ion was analyzed with a Hach spectrophotometer (DR2800) with
the 8051-SulfaVer 4 method, adapted from Standard Methods for the Examination of
Water and Wastewater [24], based on the precipitation of BaSO4.
A laser mass spectrometer Picarro L2130–I WS–CRDS (wavelength scanned cavity ring–
down spectroscopy) was used to analyze water stable isotopes at the Environmental
Studies Group (GEA) lab, San Luis Applied Mathematics Institute (IMASL–CONICET). The
δ 18O and δ 2H analytical uncertainty was 0.3 and 0.5 ‰, respectively. Standardization
was based on the Vienna Standard Mean Ocean Water (VSMOW).
different water sources (snow, rock glacier, glacier and groundwater) and requires the use
of two tracers (in this case: electric conductivity and 18O), based on Gaussian parsimony
and expressed in average values for the different water sources with credibility intervals
(95%), which means that the reported effect has 95% probability of falling within this
range [31]. The water sources selected were the more representative of the possible
water inputs to the main rivers along the melting season in the Cordillera Principal.
Groundwater represents the mountain spring samples, cryogenic rock glaciers are the
major geomorphological expression of the periglacial environment, glaciers from glacial
environment and snow for the solid precipitation type that usually fall in this high altitude
region.
With simmr analysis we quantified the contribution from different water sources
throughout the melting season. For validating their contribution, we compare the
results of three simmr analyses. In the first analysis we considered all the Cordillera Prin-
cipal rivers as replicates: Cuevas, Vacas, Tupungato and Mendoza rivers sampled in
Punta de Vacas, as well as the Horcones River and Cuevas River in Puente del Inca (analysis
A, Table 4). In the second analysis (B), were analyzed the main Cordillera Principal rivers
without considering the Cuevas River (both in Puente del Inca and in Punta de Vacas).
This was done to avoid the potential noise introduced by the Puente del Inca geothermal
water input. Finally, the resulting proportions of the Mendoza River sampled in Punta de
Vacas, downstream from the union of the three Mendoza River formers were analyzed
(analysis C). In this way, the proportions given by the different analyses were compared,
to contrast and validate the results (Table 4: D1, D2 and D3).
draining to the Mendoza River basin (Cordillera Principal, Cordillera Frontal and Precordil-
lera) and to analyze the different water sources (glaciers, rock glaciers, precipitation and
groundwater) and seasonal influence. They collected different water sources samples
throughout 2011–2012 in the different geological provinces with a seasonal frequency
and showed that water from the different geological provinces could be differentiated
by their ionic composition (due to different geological features) and isotope differences,
due to the effects of the prevalence of distinct moisture masses incoming from
different origins. The Cordillera Principal waters presented lower stable isotopic compo-
sitions due to winter precipitation incoming almost exclusively from the Pacific, while
the Precordillera were notably enriched by Atlantic Ocean summer precipitation. The Cor-
dillera Frontal, located between the other two mountain ranges, represented a moisture
sources mixture, since it was reached by both precipitation systems. Water source differ-
ences were observed and seasonal variability was also found in the composition of natural
tracers in the main streams and rivers of the Cordillera Principal, suggesting changes in
different water sources relative contribution. However, due to seasonality interference
in natural tracer composition (mainly winter snowfall) and low sampling density of
different water sources in summer, where glacier melt occurs, it was not possible to quan-
tify their contribution. With the aim of quantifying the contribution of different water
sources to rivers of the Cordillera Principal, we carried out this work. We sampled through-
out the 2013–2014 melting season to eliminate seasonal noise and, to provide a greater
statistical robustness, a weekly sampling of different water sources was designed, as
was also recommended by Ohlanders et al. [17].
those from any other water source, highlighting the significance levels of summer precipi-
tation and snow catchments (p < 0.001), rock glaciers (p < 0.005) and geothermal ground-
water (p < 0.05), (Table 1).
The model evaluating electric conductivity among sources presented an Akaike weight
of 1 (R 2 = 0.99). It indicates a significant difference between the intercept, Horcones
Inferior glacier, and geothermal and groundwater (p < 0.001), but not with other water
sources (rock glaciers, snow and summer precipitation) (Table 2).
Table 1. Mean, standard deviation (SD) and confidence interval (CI) of 18O composition for water source
and the intercept (Glacier).
Variable Mean (‰) SD CI (2.5–97.5%) p
Glacier –20.15 0.62 –21.37 & 18.93 <2e−16
Geothermal 1.82 0.87 0.11 & 3.53 0.0372*
Summer precipitation 10.37 0.93 8.55 & 12.19 <2e −16***
Rock glacier 2.88 1.01 0.91 & 4.86 0.0043**
Rivers and streams 1.39 0.77 –0.11 & 2.90 0.0692.
Snow 2.92 0.87 1.20 & 4.64 0.0008***
Groundwater 1.95 0.87 0.24 & 3.66 0.0253*
Note: The significant codes are: 0; ‘***’ 0.001; ‘**’ 0.01; ‘*’ 0.05; ‘.’ 0.1. The significant variables are marked in bold.
Table 2. Mean, standard deviation (SD) and confidence interval (CI) of electric conductivity for different
water sources in comparison with the intercept (Glacier).
Variable Mean (µS cm−1) SD CI (2.5–97.5%) p
Glacier 699.4 330.2 −52.14 & 1346.61 0.0342
Summer precipitation −55.9 551.8 −1,137.44 & 1025.69 0.9193
Rock glacier 136.5 545.3 −932.35 & 1205.32 0.8024
Rivers and streams 371.2 411.7 −435.79 & 1178.18 0.3673
Snow −439.8 476.6 −1,373.94 & 494.33 0.3561
Geothermal 21,982.3 470.5 −21,060.04 & 22,904.54 <2e−16***
Groundwater 1556.7 470.5 634.46 & 2478.95 0.0009***
Note: The significant codes are: 0; ‘***’ 0.001; ‘**’ 0.01; ‘*’ 0.05; ‘.’ 0.1. The significant variables are marked in bold.
Figure 3. Time-series showing the snow covered area (SCA), mean daily streamflow (MDS), mean daily
air temperature (MDT) and δ2H values of the Horcones Inferior Glacier proglacial stream (bold black
dots).
storms. These summer rain or light snow events (Table S7) increased the streamflow stable
isotope values up to −18.8 ‰ (usually around −20.5– −21.0 ‰), showing the sensitivity of
the measurements.
Mt. Tolosa rock glaciers conglomerate was sampled from 13 November 2013 until 20
May 2014 and captures the change in the isotope signatures of the stream during the tran-
sition from snow to ice melting. The first three stream samplings (13 and 26 November and
11 December) present lower deuterium excess d (−1.93 ‰) expressing more evaporated
conditions (Figure 4) than the other samples taken in the same site along the rest of the
melting season (8.60 ‰). At this early date, snowfall of 14, 19 and 21 November, and sub-
sequent snowmelt were recorded, according to the SCA satellite derived product for the
rock glaciers conglomerate area (Table S3 and Figure 4). The water from the proglacial
ISOTOPES IN ENVIRONMENTAL AND HEALTH STUDIES 11
Figure 4. Mean daily streamflow (MDS) of Mt. Tolosa rock glaciers conglomerate and snow covered
area percentage (SCA). The dots are the deuterium excess d (d ‰ = δ2H – 8∗δ18O [37]). The circle
marks the streamflow peak.
stream was 18O enriched (−15.5 ‰) and presented lower d values (<0 ‰) in those spring
samplings, indicating a probable major contribution from superficial snowmelt water
(Figure 4 and Table 3).
Finally, after 20 December, streamflow stable isotopic composition was steady,
depleted (δ 18O values around −17.3 ‰) and shifted from negative d values (average d
= −1.93 ‰) to positive ones, which indicated less evaporated composition (averaged d
values = 8.6 ‰). This water stable isotopic composition is highly relevant because it
allowed deciphering the provenance of the streamflow peak from the rock glaciers con-
glomerate (circle in Figure 4). The streamflow peak (circle in Figure 4) was fed by glacial
ice melt water instead of the presumed snowmelt contribution, as could be interpreted
if only satellite products were considered.
Main rivers isotopic composition are scattered, with low values similar to glaciers and
up to the lowest values of pure snow catchments (Figure 5). Although the rivers of the Cor-
dillera Principal would be fed by snowmelt in normal/abundant winter snow loads years
[3], for the studied dry period the stable isotopic composition of the Cordillera Principal
major rivers presented values closer to ice melting composition than snow signature
(Figure 5).
Table 3. Mean and standard deviation for stable isotopes and electric conductivity of the different
water sources analyzed in SIMMR analysis.
Water source Glacier Rock glacier Snow Groundwater
n 16 7 9 12
Mean δ18O (‰) −20.15 −17.27 −17.23 −18.19
SD 0.78 0.12 0.13 0.24
Mean δ2H (‰) −150.53 −129.56 −130.94 −136.51
SD 6.03 0.59 0.88 1.19
d (‰) 10.70 8.60 6.90 9.00
SD 1.36 0.51 0.64 1.00
EC (µS cm−1) 677 836 248 2256
SD 141 48 161 262
Note: Snow represents snow catchments samples obtained during the seasonal snow melt, in November and December
2013, when the snowy streams were active. For glaciers, rock glaciers and groundwater were considered the samples
obtained in January and February 2014, as representatives of the water sources.
occurred because of the Puente del Inca geothermal water input. With high electric con-
ductivity (23 mS cm−1), the geothermal waters drain into the Cuevas River and, although
this process does not represent a significant flow, it may increase ionic composition during
water transport to the Cuevas River sampling site. The Puente del Inca geothermal waters
were confined in a very narrow region of the scatter plot (Figure 5). The low dispersion of
these stable isotopes may indicate the isolation of water, compared to the surface waters,
without being affected by the fluctuations in precipitation or water inputs from snow or ice
melting. The deviation from the meteoric water lines indicates an enrichment in 18O, prob-
ably due to prolonged isotope exchange with rocks at temperatures between 25 and 100°
C [39,40]. The geothermal water presented stable temperatures of 33°C in all samples. Due
to the extreme electric conductivity values of geothermal waters, they could not be used
Figure 5. δ18O - δ2H scatter plot of the analyzed samples. The lower line represents the global meteo-
ric water line [38]. The upper line represents the local meteoric water line (δ2H = 8.1281*δ18O + 9.2755,
R 2 = 0.99).
ISOTOPES IN ENVIRONMENTAL AND HEALTH STUDIES 13
Figure 6. Isospace plot of different water sources contributing to the upper Mendoza River basin. RG:
refers to rock glacier. Mixtures refers to the main rivers of Cordillera Principal along melting season; 1:
December, 2: January, 3: February and 4: March.
Table 4. Proportional contribution of each water source to the upper Mendoza River basin according to
different simmr analysis.
Analysis Analysis
products products
Source/Month A SD A B SD B C SD C mean SD D1 D2 D3
December
Glacier 0.43 0.09 0.52 0.14 0.55 0.11 0.50 0.06 –0.09 0.12 0.03
Groundwater 0.13 0.07 0.09 0.06 0.06 0.03 0.09 0.04 0.05 –0.07 –0.03
Rock glacier 0.22 0.13 0.16 0.11 0.13 0.08 0.17 0.05 0.06 –0.09 –0.03
Snow 0.22 0.11 0.23 0.13 0.27 0.11 0.24 0.03 –0.01 0.05 0.04
January
Glacier 0.34 0.11 0.57 0.17 0.62 0.10 0.51 0.15 –0.23 0.28 0.05
Groundwater 0.30 0.12 0.10 0.07 0.11 0.05 0.17 0.11 0.20 –0.19 0.01
Rock glacier 0.20 0.12 0.16 0.12 0.14 0.09 0.16 0.03 0.04 –0.06 –0.02
Snow 0.16 0.10 0.18 0.13 0.13 0.08 0.16 0.02 –0.02 –0.03 –0.05
February
Glacier 0.42 0.06 0.51 0.06 0.43 0.10 0.45 0.05 –0.09 0.01 –0.08
Groundwater 0.29 0.09 0.17 0.06 0.22 0.07 0.23 0.06 0.12 –0.07 0.05
Rock glacier 0.16 0.09 0.17 0.09 0.19 0.11 0.17 0.01 –0.01 0.03 0.02
Snow 0.13 0.07 0.15 0.07 0.16 0.09 0.15 0.01 –0.02 0.03 0.01
March
Glacier 0.34 0.13 0.51 0.13 0.43 0.08 0.43 0.08 –0.17 0.09 –0.08
Groundwater 0.36 0.15 0.20 0.09 0.34 0.07 0.30 0.09 0.16 –0.02 0.14
Rock glacier 0.16 0.11 0.16 0.11 0.13 0.07 0.15 0.02 0.01 –0.03 –0.03
Snow 0.13 0.09 0.13 0.09 0.11 0.06 0.13 0.01 0 –0.02 –0.02
Note: A: refers to the simmr analysis of Cordillera Principal rivers sampled and considered as replicates: Cuevas, Vacas,
Tupungato and Mendoza rivers in Punta de Vacas, as well as the Horcones and Cuevas rivers in Puente del Inca. B: con-
siders all the previously described rivers without considering the Cuevas river in Puente del Inca or in Punta de Vacas. C:
analysis of the Mendoza River in Punta de Vacas. D1: A-B difference. D2: C–A difference. D3: C-B difference.
14 S. A. CRESPO ET AL.
as a water source in simmr modelling. However, this difference between analyses implies
that the groundwater contributions obtained in A are over-estimating groundwater con-
tribution and sub-estimating the remaining water sources contributions, specially the ionic
depleted ones, as glaciers. This process forces the simmr model to interpret this saline
signal as groundwater source in analysis A, which includes the Cuevas River in its calcu-
lation, growing from 13% to 30% in December and January, respectively. The high salinity
of the Cuevas River also impacts the total average values, increasing the groundwater con-
tribution from 9% to 17% for the same months, far from the 9 and 10% observed in
product B.
The contribution of rock glaciers represented an average of 17 and 16% for December
and January, respectively. The snow contribution showed response homogeneity in the
analysis, with a 67% drop from December (24%) to January (16%). This result is consistent
with the 60% recession in snow covered areas in the Mendoza river between December
(14.6%) and January (8.7%) (Tables 4 and S6).
In February the groundwater contribution continues to enhance its input, growing up
to 23% on average, with the share of glaciers decreasing to 45%. However, as explained
before, the increasing groundwater contribution is due to the Cuevas River water input,
salinized by the geothermal waters of Puente del Inca. This can be observed in analysis
B with a glacial contribution of 51% and decreasing the input of groundwater to 17%,
well below the 29% groundwater contribution in A. Likewise, snow continues to decrease
in its contribution (15%) and rock glaciers slightly increase their contribution to an average
of 17%, something that is also very consistent in all the analyses for both, snow and rock
glacier contributions.
The glacial contribution is widespread in March, ranging from 34 to 51% for A and B,
respectively. Once again, this difference is due to the greater proportional contribution
of groundwater presented by A (36%) with respect to product B (20%), which ends up
affecting the glacial contribution. In March the greatest influence of groundwater is
observed, due to a delayed thermal inertia in soil which affects all the basin, even for B,
which has grown steadily since December (9%), January (10%), February (17%) to reach
20% in March. Rock glaciers and snow contributions were uniform in all the analysis in
March, accounting 15 and 13% on average, respectively (Table 4).
The greatest source of water supply along the melting season in the Mendoza River
came from the melting of glaciers (51%), followed by groundwater (18%), snow (17%)
and finally, but also relevant, periglacial environment contribution represented by rock
glaciers (15%) (analysis C, Table 4).
Glaciers cover an area of 335 km2 in the Upper Mendoza River basin, which totalizes
3023 km2 [19,20]. Although glaciers represent 11% of the basin area (Table S8), during
the analyzed melting period, the simmr model shows a contribution around 55, 62, 43
and 43% in December, January, February and March, respectively (analysis C, Table 4).
Rock glaciers cover an area of 69.62 km2 in this basin and represent 2.3% of the basin
area and, according to simmr, contribute 13, 14, 19 and 13% in December, January, Feb-
ruary and March, respectively. When analyzing all the cryoforms together, they cover
13.4% of the total basin area representing 404.2 km2, and contributing to 68, 76, 62 and
56% of streamflow in December, January, February and March, respectively (Figure 7).
By January 2014, most of the snow had melted, leaving 8.7% of SCA in the upper
Mendoza River basin (Figure S2 and Table S6). The simmr model calculated an average
ISOTOPES IN ENVIRONMENTAL AND HEALTH STUDIES 15
Figure 7. Boxplot of different water source contribution along the melting season in the Mendoza
River, obtained by the simmr model.
snow contribution of 27% for December, but it drops almost by half in January (13%)
(Table 4). From December to January, a 59.7% drop is observed in the SCA, from 14.6–
8.7%. Subsequently, a small increase of up to 16% for the snow contribution in February
is observed (due to a couple of thin snowfalls with great coverage, Table S3) and coinci-
dent with the increase until 16% of SCA observed (Figure S2 and Table S6). In March, the
snow contribution estimated by simmr drops to 11%, but the snow coverage detected by
the SCA satellite derived product rises to 16.5%, due to a couple of thin snowfalls (Tables
S3 and S6).
3.2.1. Glaciers
Because the streamflow measurements were carried out in glaciers flowing to the Cuevas
River, after the modelling for all the Cordillera Principal main rivers (explained in section
2.4), we run a model including each of the Mendoza River headstreams (Cuevas, Vacas
and Tupungato rivers), with results available in Supplementary Data (Table S9). Glaciers
16 S. A. CRESPO ET AL.
Table 5. Glaciers streamflow contribution to the Cuevas River (December 2013–March 2014).
Mendoza River Cuevas River HIG Tolosa RGC
(hm3) (hm3) 3
(hm ) Cuevas% Mendoza% 3
(hm ) Cuevas% Mendoza%
Dec 191.6 19.3 8.94 46.3 4.67 0.08 0.41 0.042
Jan 233.5 23.6 6.61 28.0 2.83 0.03 0.13 0.013
Feb 161.1 12.4 4.22 34.1 2.62 0.01 0.08 0.006
Mar 109.3 11.0 2.50 22.8 2.29 0.02 0.18 0.018
Total 695.4 66.3 22.27 33.6 3.20 0.15 0.23 0.022
Average 173.9 16.6 5.57 32.8 3.10 0.04 0.20 0.020
Cuevas/Mendoza% refers to the proportion of streamflow contributed by the cryoform regarding the Cuevas/Mendoza
rivers streamflow measured at Punta de Vacas. Sources: Cuevas/Mendoza River at Punta de Vacas streamflow: Secretariat
of Water Resources, (32.86 °S and 69.77 °W), Horcones Inferior glacier (HIG) and Tolosa Rock glacier conglomerate (Tolosa
RGC) streamflow were measured in this study.
represent 6% of the Cuevas River basin area (Table S8) but, according to the simmr analy-
sis, contribute 26% to the Cuevas River streamflow in December, falling to 13 and 14% in
January and February, respectively, decreasing its contribution in comparison with
groundwater, until 5% in March (Table S9). The Horcones Inferior glacier, just one of the
190 cryoforms in the Cuevas River basin [19], measured streamflow represents 28 and
34% (for January and February, respectively) of the absolute Cuevas River streamflow
measured at Punta de Vacas (Table 5). In this case, contributions estimated with direct
streamflow measurements of one glacier are much larger than those estimated with
simmr, pointing to the increase in ionic composition in the Cuevas River due the geother-
mal water input at Puente del Inca. This process forces the simmr model to interpret this
saline signal as groundwater source in the analysis A of the simmr model (Table 4). As was
explained before, this geothermal influence is observable when analyzing the glacial con-
tribution for January and February, being 57 and 51%, respectively (analysis B). Total
summer streamflow (Dec–March) of the Horcones Inferior glacier (22.27 hm3) represents
3.2% of the entire streamflow of Mendoza River (695.39 hm3) for the same period, while
the area of the glacier (6.95 km2) represents 3.5% of the upper Mendoza River basin
(3023 km2) [19]. If we extrapolate the Horcones Inferior glacier contribution to the
entire glacier area of the Mendoza River basin (334.58 km2), glaciers should contribute
1072.10 hm3, 54.17% more than the Mendoza River actually measured streamflow,
showing an important aquifer recharge process. But this is not directly linkable, because
the Horcones Inferior glacier is one of the biggest Mendoza River glaciers and not all
the glaciers present an equivalent ice mass per area unit.
period (695.39 hm3). The area occupied by this rock glacier conglomerate (0.16 km2) rep-
resents just the 0.005% of the upper Mendoza River basin (3023 km2).
The simmr model shows 26, 18, 14, and 9% input from rock glacier sources for the same
months (Table S9). If we extrapolate linearly the contribution measured in the Tolosa rock
glacier conglomerate streamflow and its rock glacier area to the Cuevas basin, the
17.6 km2 occupied by rock glaciers in the Cuevas River should contribute 16.5 hm3 to
the Cuevas River streamflow from December to March, representing 25% of the total
Cuevas River streamflow for the summer season (66.28 hm3). The simmr model is lower
than that value for the Cuevas River (17%) (Table S9), and also for the analysis: A (19%),
B (16%) and C (15%) (Table S10). Summarizing, the streamflow value (25%) extrapolated
to the total Cuevas River basin rock glacier area, is 8% higher than simmr average calcu-
lations (17%) (Table S9), pointing to infiltration processes, and probably lower melting
rates in massive rock glaciers due to higher altitudes and other unknown processes, as
thicker rock isolation, shadow, etc.
Rodriguez et al. [18] studied the Juncal river basin (Chilean side of the Andes), located at
the same latitude as the Mendoza River along the hydrological year 2011–2012. They
analyze streamflows and ionic and isotopic compositions to develop a hydrograph separ-
ation. Based on geographical source analysis, they identify and quantify three different
sources of flow in the basin, namely: South, West and East; instead of different water
sources differentiation. They conclude that most of the Juncal river catchment can be
attributed to the Southern basin region, strongly related to glacier presence and low
ion concentrations, and supplying 50% of the average water contribution in summer
(26 November 2011–31 March 2012). In this work 51% glacial contribution is reported
(analysis C, Table 4 and S10), which is close to Rodriguez et al. [18], but lower than the pre-
vious 62% contribution inferred in Ohlanders et al. in the Juncal river basin [17]. The West
basin region, associated with quick-flowing snowmelt, results in 22% average water con-
tribution, while the Mendoza River catchment analyzed in this work presented 17% of the
average snow contribution. The East basin region, related to high ion concentrations
proper to high rock–water interaction and a depleted isotopic signal proper of high-
elevation accumulation zones, showed 29% average water contribution in summer. This
proportion is similar to the 33% average water input contribution from rock glaciers
and groundwater reported here together with 15% and 18%, respectively. Even when
this work cannot be linearly compared with Rodriguez et al. [18], due to the different
study zones and methodological approaches, there are similarities in the results. Even
when these results cannot be extrapolated to normal or abundant snowfall years, these
findings in both works demonstrate the capability of combined isotopic and solute infor-
mation for informing mixing models for hydrologic research in mountain environments in
a dry hydrological year.
Understanding the different contribution from different water sources is relevant in an
uninterrupted sequence of dry years that started in 2010 and was called megadrought.
The moderate intensity but protracted, spatially intensive and warm character of the
megadrought is consistent with the climate change signal expected for western subtropi-
cal South America [2]. This work offers information to develop hydrological adaptation and
mitigation strategies for future climate change scenarios in the region.
18 S. A. CRESPO ET AL.
Acknowledgements
To Julieta Aranibar, Claudia Feijoo, Chris Harrod, Victor A. Ramos, Eduardo Kruse, Álvaro González–
Reyes, Ezequiel Tourn, Mariano Masiokas and Ricardo Villalba for their valuable contributions. To
Esteban Jobbagy for his help with the isotope analysis in the Environmental Studies Group (GEA),
Applied Mathematics Institute (IMASL – Conicet). To DRNR of Mendoza province, for the permission
to deploy collectors, carry out sampling, and work with park rangers in protected areas (A. Zalazar,
V. Ottero, R. Massarelli, O. Aranibar and J. Gimenez). To A. Casteller, J.P. Scarpa, M. Castro, G. Costa,
L. Zalazar, I. Pecker, L. Gomez and Marcelo Quiroga for their fieldwork support. To the CR2 (FONDAP
1511009). To the editor and referees for their comments and suggestions which significantly
improved the manuscript.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Funding
This work was supported by Fondo Nacional de Desarrollo Científico y Tecnológico [grant number
3180682]; Agencia Nacional de Promoción Científica y Tecnológica [PICT 2011-2703]; Universidad
Nacional de Cuyo [PID (2011–2013 and 2013–2015)].
ISOTOPES IN ENVIRONMENTAL AND HEALTH STUDIES 19
ORCID
Sebastián A. Crespo http://orcid.org/0000-0003-3142-751X
Francisco Fernandoy http://orcid.org/0000-0003-2252-7746
Céline Lavergne http://orcid.org/0000-0002-2002-8655
References
[1] Dusaillant I, Berthier E, Brun F, et al. Two decades of glacier mass loss along the Andes.
Nat Geosci. 2019;12:802–808.
[2] Garreaud RD, Alvarez-Garreton C, Baricivich J, et al. The 2010–2015 mega-drought in
central Chile: impacts on regional hydroclimate and vegetation. Hydrol Earth Syst Sci.
2017;21:6307–6327.
[3] Masiokas M, Villalba R, Luckman B, et al. Variaciones de la precipitación nívea en los Andes cen-
trales de Argentina y Chile, 1951–2005: Influencias atmosféricas de gran escala e implicancias
para los recursos hídricos en la región. J Climate. 2006;19:6334–6352. Spanish.
[4] Boninsegna J. Impacto del cambio climático en los oasis del oeste argentino. Asociación
Argentina para el progreso de las ciencias. Cienc Investig. 2013;64:45–58. Spanish.
[5] Lascano M, Villalba R. Algunas precisiones sobre el rol de los glaciares en el escurrimiento
andino. XXI Congreso Nacional del Agua (CONAGUA2007); 2007 May 15–17, Tucumán,
Argentina. Spanish.
[6] Bruniard ED. Los regímenes fluviales de alimentación sólida en la República Argentina: ensayo
de elaboración de un modelo hidroclimático de la vertiente oriental de los Andes. Buenos Aires:
Academia Nacional de Geografía; 1994; (Publicación especial; 7). Spanish.
[7] Leiva JC. Recent fluctuations of the Argentinean glaciers. Glob Planet Change. 1999;22:169–177.
[8] Boninsegna J, Villalba R. Los condicionantes geográficos y climáticos. Documento marco sobre
la oferta hídrica en los oasis de riego de Mendoza y San Juan. First report. Mendoza: Secretaría
de Ambiente y Desarrollo Sustentable de la Nación; 2006; Spanish.
[9] Masiokas MH, Villalba R, Luckman BH, et al. Intra-to multidecadal variations of snowpack and
streamflow records in the Andes of Chile and Argentina between 30° and 37°S. J
Hydrometeor. 2010;11:822–831.
[10] Lauro C, Vich A, Moreiras SM. Streamflow variability of river basins from Cuyo region. Geoacta.
2016;40:28–51.
[11] Crespo S, Aranibar JN, Gomez L, et al. Ionic and stable isotope chemistry as indicators of water
sources to the Upper Mendoza River basin, Central Andes of Argentina. Hydrolog Sci J.
2016;62:588–605.
[12] Lambs L. Correlation of conductivity and stable isotope 18O for the assessment of water origin
in river system. Chem Geol. 2000;164:161–170.
[13] Lambs L, Loubiat M, Richardson W. The use of stable isotopes to evaluate water mixing and
water use by flood plain trees along the Garonne valley. Isot Environ Health Stud.
2010;39:301–310.
[14] Pu T, He Y, Zhu G, et al. Characteristics of water stable isotopes and hydrograph separation in
Baishui catchment during the wet season in Mt. Yulong region, south western China. Hydrol
Process. 2013;27:3641–3648.
[15] Panarello H, Dapeña C. Mecanismos de recarga y salinización en las cuencas de los ríos
Mendoza y Tunuyán, Mendoza, República Argentina: evidenciados por isótopos ambientales
(Volume II). XII Geological Survey Congres of Bolivia; 1996. Spanish.
[16] Hoke G, Aranibar JN, Viale M, et al. Seasonal moisture sources and the isotopic composition of
precipitation, rivers, and carbonates across the Andes at 32.5–35.5°S. Geochem Geophys
Geosyst. 2013;14:962–978.
[17] Ohlanders N, Rodriguez M, McPhee J. Stable water isotope variation in a Central
Andean watershed dominated by glacier and snowmelt. Hydrol Earth Syst Sci. 2013;17:1035–
1050.
20 S. A. CRESPO ET AL.
[18] Rodriguez M, Ohlanders N, Pellicciotti F, et al. Estimating runoff from a glacierized catchment
using natural tracers in the semi-arid Andes cordillera. Hydrol Process. 2016;30:3609–3626.
[19] IANIGLA-ING. Informe de las subcuencas de los ríos de las Cuevas y de las Vacas, cuenca del río
Mendoza, Inventario Nacional de Glaciares. Mendoza: Instituto Argentino de Nivología,
Glaciología y Ciencias Ambientales (IANIGLA-CONICET); 2018; Spanish.
[20] IANIGLA-ING. Informe de la subcuenca del río Tupungato, cuenca del río Mendoza, Inventario
Nacional de Glaciares. Mendoza: Instituto Argentino de Nivología, Glaciología y Ciencias
Ambientales (IANIGLA-CONICET); 2018; Spanish.
[21] Gordon N, McMahon T, Finlayson BL, et al. Stream hydrology: An introduction for ecologists.
2nd ed Chichester: John Wiley & Sons; 2013.
[22] Francoud B, Pouyaud B. Métodos de observación de glaciares en los Andes Tropicales. 1st ed
France: Unité de Recherche Great-Ice, Institut de Recherche pour le Développement (IRD); 2004;
Spanish.
[23] Moore D. Construction of a Mariotte bottle for constant-rate tracer injection into small stream.
Streamline. 2004;8:15–16.
[24] Clesceri L, Greenberg A, Eaton A, editors. Standard methods for the examination of water and
wastewater. 20th ed. Washington (DC): APHA, AWWA and WPCF; 1999.
[25] Husson F, Fraser J. Multivariate exploratory data analysis and data mining with R, R package
FactoMineR; 2014.
[26] Oksanen J, Blanchet FG, Kindt R, et al. Vegan: community ecology package. R Packag. version
2.0-7. 2013.
[27] Pinheiro J, Bates D, DebRoy S, et al. Nlme: linear and nonlinear mixed effects models, R package
version 3, 1-109, 2013.
[28] Crawley MJ. The R book. Chichester: John Wiley and Sons; 2007.
[29] Zuur AF, Leno EN, Walker NJ, et al. Mixed effects models and extensions in ecology. New York
(NY): Springer Science and Business Media; 2009.
[30] Parnell A. Simmr: a stable isotope mixing model. R package version 0.4.1; 2019.
[31] Makowski D, Ben-Shachar MS, Lüdecke D. BayestestR: describing effects and their uncertainty,
existence and significance within the Bayesian Framework. J Open Source Softw. 2019;4.
DOI:10.21105/joss.01541.
[32] Riggs G, Hall D, Salomonson V. MODIS snow products user guide to collection 5. Digit Media.
2006;6:1–80.
[33] Dietz A, Kuenzer C, Gessner U, et al. Remote sensing snow–a review of available methods. Int J
Remote Sens. 2012;33:4094–4134.
[34] Roy A, Royer A, Turcotte R. Improvement of springtime streamflow simulations in a boreal
environment by incorporating snow-covered area derived from remote sensing data. J
Hydrol. 2010;390:35–44.
[35] Bergeron J, Royer A, Turcotte R, et al. Snow cover estimation using blended MODIS and AMSR-E
data for improved watershed-scale spring streamflow simulation in Quebec, Canada. Hydrol
Process. 2014;28:4626–4639.
[36] Cara L, Masiokas M, Viale M, et al. Análisis de la cobertura nival de la cuenca superior del río
Mendoza a partir de imágenes MODIS. Meteorologica. 2016;41:21–36. Spanish.
[37] Dansgaard W. Stable isotopes in precipitation. Tellus. 1964;16:436–468.
[38] Craig H. Isotopic Variations in Meteoric Waters. Science. 1961;133(3465):1702–1703. DOI:10.
1126/science.133.3465.1702
[39] Craig H. The isotope geochemistry of water and carbon in geothermal areas. In: Tongiorgi E,
editor. Nuclear geology on geothermal areas. Pisa: Consiglio Nazionale delle Ricerche,
Laboratorio di Geologic Nucleare; 1963. p. 17–53.
[40] Kharaka YK, Mariner RH. Geothermal systems. In: Aggarwal PK, Gat JR, Froehlich KF, editor.
Isotopes in the water cycle. Dordrecht: Springer; 2005. p. 243–270.