Environ Monit Assess (2017) 189:440 DOI 10.1007/s10661-017-6149-2
Using synoptic tracer surveys to assess runoff sources in an Andean headwater catchment in central Chile A Nauditt & C Soulsby & C Birkel & A Rusman & C Schüth & L Ribbe & P Álvarez & N Kretschmer
Received: 18 January 2017 / Accepted: 27 July 2017 # Springer International Publishing AG 2017
Abstract Headwater catchments in the Andes provide critical sources of water for downstream areas with large agricultural communities dependent upon irrigation. Data from such remote headwater catchments are sparse, and there is limited understanding of their hydrological function to guide sustainable water management. Here, we present the findings of repeat synoptic tracer surveys as rapid appraisal tools to understand dominant hydrological flow paths in the semi-arid Rio Grande basin, a 572-km2 headwater tributary of the A. Nauditt (*) : L. Ribbe Institute for Technology and Resources Management in the Tropics and Subtropics, Technical University Cologne, Cologne, Germany e-mail:
[email protected] C. Soulsby School of Geosciences, University of Aberdeen, Aberdeen, Scotland, UK
11,696-km2 Limarí basin in central Chile. Stable isotopes in stream water show a typical altitudinal effect, with downstream enrichment in δ2H and δ18O ratios. Seasonal signals are displayed in the isotopic composition of the springtime melting season water line with a steeper gradient, whilst evaporative effects are represented by lower seasonal gradients for autumn and summer. Concentrations of solutes indexed by electrical conductivity indicate that there are limited contributions of deeper mineralised groundwater to streamflow and that weathering rates vary in the different sub-catchments. Although simplistic, the insights gained from the study could be used to inform the structure and parameterisation of rainfall runoff models to provide seasonal discharge predictions as an evidence base for decision making in local water management. Keywords Tracers . Stable isotopes . Mountainous runoff generation . Andes . Semi-arid central Chile . Steep elevation gradient
C. Birkel Department of Geography, University of Costa Rica, San Pedro, Costa Rica A. Rusman : C. Schüth Institute for Applied Geosciences, University of Darmstadt, Darmstadt, Germany P. Álvarez Department of Agricultural Engineering, University of La Serena, La Serena, Chile N. Kretschmer Department of Geology and Mining, University of La Serena, La Serena, Chile
Introduction Throughout the world, mountainous areas act as Bwater towers^ providing river flows and water resources to lowland areas downstream (Viviroli et al. 2007). Runoff from montane regions, often regulated by snow and glacier melt, sustains a range of ecosystem services, not least the provision of drinking water supplies and irrigation water to communities downstream (Price and Egan 2014). Hydrological regimes in mountainous areas
440
Environ Monit Assess (2017) 189:440
Page 2 of 17
are often strongly influenced by snow and glacier melt and are therefore vulnerable to the effects of climatic warming and variability (Barnett et al. 2005). This, in turn, has implications for the security of downstream water supplies for human consumption and health, maintenance of food security and industrial development (Döll and Siebert 2002). Globally, almost 20% of the world’s population is dependent on such melt regimes for their water resources. The Andes form one of the world’s most extensive mountain ranges and have a critical role in sustaining river flows and water supplies in many South American countries (Price and Egan 2014). Cities and agricultural areas in the western lowlands of the long, thin landmass of Chile are particularly dependent on river flows from the high Andes in the east. Despite this dependence, the headwaters of strategically important Chilean rivers draining the Andes are sparsely monitored, usually ungauged and remote (Ohlanders et al. 2013). This provides a limited evidence base for decision making for sustainable water management, which is exacerbated by fundamental uncertainty regarding precipitation inputs in the high mountain areas (Hublart et al. 2015). The current pressure on water resources in Chile, together with the potential for strong inter-annual variability in climate between wetter El Nino years and drier El Nina events, dictates a high level of vulnerability to water supplies and associated food insecurity (Nauditt et al. 2016). Furthermore, the likely deleterious implications of climate change dictate that there is an urgent need to better understand the hydrology of these Andean headwater catchments (Vicuña et al. 2011). Given the remoteness and logistical constraints, there is a need for rapid appraisal tools to identify the dominant hydrological sources sustaining river flow in these regions and evaluate how these change over the course of the hydrological year in order to aid climate change assessment. Whilst the development of data-rich, experimental catchments in the Andes is progressing (e.g. Ohlanders et al. 2013), most areas are effectively ungauged and need hydrological assessment. Environmental tracers provide such tools that can help identify the geographic source areas generating runoff and their temporal dynamics (Leibundgut et al. 2009). Such tracers include geochemicals that can establish the provenance of water source areas and isotopes which can additionally indicate the timescales of the runoff response (Kendall and Caldwell 1998; Soulsby et al. 2009; Soulsby et al. 2011). Synoptic sampling of river basins can show the spatial variability
in tracers which can infer the spatial interaction of hydrological processes (Soulsby et al. 2003; Zimmer et al. 2013; McGuire et al. 2014). Repeat surveys can provide insight into how these sources change with hydroclimatic conditions (Lessels et al. 2016). Whilst such studies are increasingly common in temperate catchments, applications in semi-arid Andean areas remain limited, though recent years have seen an increase in research efforts (e.g. Hoke et al. 2013; Ohlanders et al. 2013). Here, we report the results of synoptic surveys of the river network and groundwater sources in an Andean headwater in northern central Chile. We used repeat tracer surveys of isotopes and other hydrochemical variables on four occasions encompassing both the dry season and melt-dominated period in order to assess the dynamics of the hydrological sources which govern the generation of streamflow. The overall aim of this study was therefore to improve the understanding about hydrological processes in this remote Andean environment by identifying sources of water from tracer-based synoptic surveys. The specific objectives were (a) to assess spatial variability in runoff sources, (b) understand how these change over the year as a basis for developing conceptual models of hydrological function, which can inform numerical models and water-related decision making and (c) inform follow-up sampling strategies to be implemented by local stakeholders.
Materials and methods Study area The Rio Grande is a 572-km2 headwater tributary of the 11,696-km2 Limari river basin (Fig. 1). The Andean headwaters of the Rio Grande range in altitude between 1381 and 4375 m. The catchment area is 572 km2 and drains into the discharge station Las Ramadas. The catchment has a mean hillslope gradient of 23.5°, and the mean gradient of the river network is 9.6°. The drainage density derived from a 1:200,000 scale map is extremely low at 0.085 km−1 km2. The geology of the tectonically active area is highly heterogeneous and fractured as evidenced by thermal springs (Fig. 1a). Plutonic and volcanic formations comprise the higher elevation areas, with fractured metamorphic and sedimentary formations at lower altitudes. Alluvial sediments occupy the main river valleys, and a range of glacial deposits are also present (Fig. 1b). Soil
Environ Monit Assess (2017) 189:440
Page 3 of 17 440
Fig. 1 a The Rio Grande headwater catchment topography and sampling points, Las Ramadas weather and discharge stations at the outfall. Photographs of b the mainstream at Las Ramadas
gauging station at 1381 m of elevation and c the upper valley at 3100 m showing cushioned peatlands in the riparian zone
development is limited, with shallow leptosols dominating. Vegetation below 2000 m of elevation is predominantly sparse coverage of shrubs and cacti growing in the leptosols or over exposed rock outcrops. In the downstream alluvial valleys, riparian trees and shrubs predominate. At higher elevations of 2000 m, in the valley bottoms, areas of peat have formed (Fig. 1c) (Squeo et al. 2006).
The climate is semi-arid. Precipitation is highly seasonal and falls mainly in the winter months as snow at high altitudes between May and October, with very little in summer (Souvignet et al. 2012). At the outfall of the Rio Grande catchment at an altitude of 1381 m at Las Ramadas, long-term average annual precipitation (1970–2013) is 307 mm (Fig. 2). Recent modelling studies suggest at least double this precipitation at
440
Environ Monit Assess (2017) 189:440
Page 4 of 17
higher altitudes (Nauditt et al. 2016). Mean annual temperatures here are 5.1 °C, varying between a summer maximum of 39 °C and winter minimum of − 12 °C. The higher summer temperatures result in high annual potential evapotranspiration which averages 2542 mm (CNR, Ciren, Comisión Nacional de Riego 1997). The seasonality of precipitation and temperatures is reflected in the annual hydrological regime which typically shows a snow melt dominance, with flows increasing markedly from October, reaching a peak in November/ December, but declining soon after and remaining low for the remainder of the summer (Fig. 2). Average annual runoff is 259 mm at the Las Ramadas gauging station; however, inter-annual variability is marked depending on large-scale climatic influences, and this can vary between 20 mm in dry years and 897 mm in wet years which for this region might coincide with La Niña (dry) or El Niño (wet) years (Nauditt et al. 2016). Data The outfall of the Rio Grande headwater catchment has relatively good quality meteorological and hydrological data extending back to 1968 (Nauditt et al. 2016). Precipitation and temperatures have been recorded at the Las Ramadas climate station next to where streamflow has been measured at the discharge gauging station also called BLas Ramadas^. This provided the context for the
synoptic surveys which were carried out on four occasions in 2013 (January, May, September and December) to gain Bsnapshot^ coverage of the spatial variation in tracer concentrations over the hydrological year (Fig. 2). These sampling occasions encompassed a range of discharges; on 11 January 2013, the flow was 0.75 m3 s−1 which equated to runoff of 0.12 mm day−1. Very similar flows were sustained on 11 May 2013 (0.81 m3 s−1 and 0.13 mm day−1). By 5 September, flows had doubled to 1.73 m3 s−1 (0.27 mm day−1) and remained at a similar level on 2 December (at 1.61 m3 s−1 and 0.26 mm day−1). Unfortunately, continued snow cover precluded sampling some of the higher altitude sites during the September visit. Although increased flows were captured by the September/December sampling, these were much lower than it usually occurs during these months due to an extended dry period since 2008 which was mostly dominated by La Niña years (Nauditt et al. 2016). Figure 2 compares the hydrograph (mm day−1) of the sampling year 2013 at Las Ramadas station with the daily average discharge (1968–2013). Monthly precipitation of the sampling year is also shown in comparison to the monthly average (1969–2013). Methods Sampling encompassed 10 sites on the main stem of the Rio Grande, 14 tributary sites and 6 spring or borehole
Fig. 2 Long-term mean daily precipitation-discharge relationship at Las Ramadas gauging station in comparison to the sampling year 2013 indicating the sampling dates
2874
18
20
Rio Gordito
Rio Gordito
1600
32c
34
Río Grande
Río Grande outfall
15
23
28
29
Left tributary
Left tributary
Left tributary
Left tributary
32
32b
Right tributary
Right bank
9
12
13
16c
GW borehole right tributary
GW borehole right tributary
GW borehole right tributary
GW Virgen right tributary
Groundwater
27
31
Right tributary
Right tributary
2950
3237
3267
3313
1668
1759
1845
2133
2398 2254
26
Right tributary
2024
2107
Right tributary
Right tributary
3122
11 2553
3310
21
Las Cuevas
2556
1380
Left tributary
Left tributaries
1937
30
Río Grande
2543 2268
22
25
Rio Grande
Rio Grande
Rio Grande main river
2950
16a 2570
3147
14
Río Gordito
Rio Gordito thermal springs
Elev. (m)
3317
Site
10
Rio Gordito
Rio Gordito (upstream)
Río Grande
1
0
2.5
1.5
11
80
15
1
15
5
2
40
30
180
1500
800
400
380
306
170
150
78
17
Q (l s−1) 11.05.
10
35
300
40
4
25
1730
1600
1200
800
400
Q (l s−1) 05.09.
2
0
1
1.4
48.2
35
293
28.3
25
44
130.4
278.5
1550
1493
998
800
507
450
220
105.8
40
Q (l s−1) 02.12.
− 108.5 − 105.1 − 100.7 − 102.1
− 14.7 − 14.3 − 13.9
− 73.9
− 9.9
− 15.1
− 97.3 − 78.4
− 13.4 − 10.9
− 99.8 − 93.1
− 13.5 − 12.7
− 93.2
− 99.3 − 93.7
− 13.4 − 12.7 − 12.8
− 96.0 − 105.6
− 12.9
− 88.5
− 14.6
− 11.3
− 94.9
− 12.7
− 11.2 − 97.4
− 98.8 − 90.3
− 13.5
− 96.9
− 92.6
− 13.0
− 101.0
− 13.8 − 12.0
− 13.0
− 100.0
δ2H (‰)
− 13.9
δ18O (‰) 11.01.
Table 1 Stable isotope and discharge values for the four sampling campaigns in Rio Grande in 2013
− 98.7
− 13.6
− 14.4
− 12.4
− 13.8
− 14.3
− 11.3
− 9.8
− 13.8
− 13.5
− 102.2
− 96.3
− 103.8
− 104.1
− 78.1
− 76.2
− 98.7
− 95.3
− 101.4
− 95.9
− 13.5 − 13.0
− 95.0
− 98.8
− 96.5
− 98.1
− 91.1
− 97.6
− 98.2
− 99.9
− 101.2
− 97.7
− 95.0
− 88.0
− 98.7
δ2H (‰)
− 13.0
− 13.4
− 11.4
− 13.6
− 11.7
− 13.3
− 13.2
− 13.5
− 13.7
− 13.2
− 12.4
− 9.6
− 12.5
δ18O (‰) 11.05.
− 86.2 − 80.5
− 11.7
− 88.1
− 103.3
− 98.3
− 103.0
− 97.1
− 103.5
− 97.6
− 99.7
− 105.7
− 106.8
− 107.3
δ2H (‰)
− 12.4
− 12.6
− 13.8
− 13.6
− 13.6
− 12.9
− 13.8
− 13.5
− 13.9
− 14.6
− 14.4
− 14.0
− 14.4
δ18O (‰) 05.09.
− 13.9
− 14.6
− 11.0
− 13.2
− 13.3
− 13.3
− 14.1
− 12.9
− 11.9
− 12.1
− 13.3
− 13.2
− 13.6
− 13.3
− 13.4
− 13.8
δ18O (‰) 02.12.
− 104.6
− 108.2
− 82.0
− 99.5
− 99.5
− 104.1
− 105.7
− 98.2
− 92.8
− 94.0
− 101.4
− 102.0
− 101.9
− 100.2
− 102.6
− 102.6
δ2H (‰)
Environ Monit Assess (2017) 189:440 Page 5 of 17 440
Environ Monit Assess (2017) 189:440 − 108.2
− 108.0 − 14.4
− 14.1
− 106.6 − 14.6 − 105.5
− 94.6 − 11.7
− 104.8
− 104.8
− 14.3 5
2950
2948
2687
16b
17
19
7
1517 33
1
1.5
10
2.5
− 14.3
− 14.5
− 107.3 − 15.0 − 106.0 − 14.6
Q (l s−1) 05.09. Elev. (m) Site
Q (l s−1) 11.05.
− 14.5
− 75.7 − 11.3 − 77.8 − 11.5 − 73.4 − 10.0
δ18O (‰) 02.12. δ2H (‰) δ18O (‰) 05.09. δ2H (‰) Q (l s−1) 02.12.
δ18O (‰) 11.01.
δ2H (‰)
δ18O (‰) 11.05.
− 109.5
Page 6 of 17
δ2H (‰)
440
sites (Table 1, Fig. 1). We use the term Bright tributary^ for those streams which flow into the Rio Grande river from the right mountain ranges when walking from upstream to downstream. Accordingly, we call the streams coming from the left side Bleft tributaries^. Sampling of tributaries was carried out at locations before entering the mainstream. Samples for stable isotope analysis were collected in 5-ml bottles at each point during each sampling campaign from January to December 2013. Samples collected in January and May were analysed for stable isotopes of water (2H and 18O) at the Northern Rivers Institute at the University of Aberdeen, with a Los Gatos Research (LGR) DLT-100 laser diode water isotope analyser and transformed into the δ notion (‰) according to Vienna Standard Mean Ocean Water (VSMOW) standards. Analytical precision is 1.0 ‰ for δ2H and 0.2‰ for δ18O. Samples collected in September and early December 2013 were analysed to the same precision at the Technical University of Darmstadt Laboratory with a Picarro L2130-i cavity rind down spectrometer (CRDS) isotope analyser connected to a Picarro A0211 high precision vaporiser as described in Coplen (1996). For other tracers, the electric conductivity (EC) of stream waters (μS cm−1) was measured in situ with a Hach Lange HQ30D handheld device. The temperature (T in °C) and dissolved oxygen (DO in mg l−1) of samples were also recorded. In addition, at selected sites (as time allowed), flows were measured using dilution gauging from springs, tributaries and main stem river sites upstream of Las Ramadas. The discharge measurements were carried out with the salt dilution method using an EasyFlow measurement device. As a tracer, sodium chloride (NaCl; usual salt) is used. It facilitated a rapid, portable tool for flow gauging. The device records the salt cloud and gives the result in litre per second at the end of the operation (MADDTechnologies 2014).
Left tributary
Left tributary
Thermal spring
GW spring right tributary
Spatial variability of observed parameters Thermal springs
Río Grande
Table 1 (continued)
Results
Spatial variability in stream composition Table 1 shows the values for stable isotopes for the four sampling campaigns. Discharge was measured only in May, September and December. The marked spatial
Environ Monit Assess (2017) 189:440
Page 7 of 17 440
variation in isotope ratios was evident in stream waters especially in the synoptic surveys under the lowest flow conditions in January and May 2013. This primarily reflects a strong altitudinal control in terms of more depleted isotopes at high altitude sites, which presumably is caused by altitudinal effects on the isotopic composition of precipitation. These persistent patterns see the δ2H ratios increase in the main stem of the river from around − 101‰ at 3147 m on the main headwater of the Rio Gordito to − 88‰ at 1380 m at the Las Ramadas gauging station at 1386 m (Table 1). Similarly, the δ18O ratios increase from approximately − 15 to − 10‰ along the main stem of the river (Fig. 3a). Figure 3a shows the altitudinal gradient for the surface water samples in January 2013. Values for groundwater and thermal springs are plotted separately. In Fig. 3b, we present average altitudinal gradients calculated for other latitudes of the Andean mountain range by Windhorst et al. (2013) and Vogel et al. (1975). Electrical conductivity, used as a proxy for total dissolved solids, shows a similar general change downstream with levels increasing under low flow conditions in the January and May sampling periods from approximately 100 μS cm−1 in the headwaters of the Rio Gordito to approximately 200 μS cm−1 at Las Ramadas gauging station (Table 2). However, the highest values (> 400 μS cm−1) along the main stem of the river are found in the upper part of the catchment in the lower reaches of the Rio Gordito (Fig. 5). The latter values are impacted by the extremely high EC values of the thermal spring (3880 μS cm−1 in January) located at an
Generally, the isotopic composition of the tributaries sampled is also correlated with altitude like that of the main stem of the river (Fig. 4). The upper tributaries (above 2900 m) generally have δ2H and δ18O values below − 95 and − 12% respectively (Table 1). In contrast, the lower altitude tributaries that enter the Rio Grande below 1900 m generally have δ2H values > − 95‰ and δ18O values > − 12‰. Although in the September and December samples, some sites see values depressed during the main melt period. The relationship between δ2H and δ18O shows some scatter
Fig. 3 a January 2013 (summer) sampling data and b fitted altitudinal gradients for the surface waters sampled in January, September and December 2013 compared to the altitudinal
gradients for precipitation sampled in Ecuador (Windhorst et al. 2013) and for the Argentinian Andes by Vogel et al. (1975)
altitude of 2950 m next to the Gordito river and the left tributary source called Aguas Negras with high EC values (> 600 μS cm−1) suggesting geogenic origin. Likewise, with electrical conductivity, the values were generally lower throughout the river system under higher flows (Table 2). These ranged from 62 μS cm−1 in the upper Rio Gordito to 233 μS cm−1 at Las Ramadas (Table 2). However, the higher levels previously observed in the lower parts of the Rio Gordito (> 400 μS cm−1) were less pronounced under high flows and reduced to values (< 300 μS cm−1) (Table 2). Figure 4 gives an overview on the value range of samples for the different seasons. Unsurprisingly, lower δ18O values between − 13 and − 14‰ dominate for the colder snow melt season in September, whilst for the summer and autumn samples, they range between − 12 and − 13.5‰. Spatial variation in tributary inputs
2874 2570
20
Rio Gordito
32c 1600 34
Río Grande
Río Grande outfall
15 23 28 29
Left tributary
Left tributary
Left tributary
Left tributary
Left tributary
27 31 32 32b 1668
Right tributary Colorado
Right tributary
Right tributary Pangue
Right tributary 9 12 13
GW borehole right tributary
GW borehole right tributary
GW borehole right tributary
Groundwater
26
Right tributary
3237
3267
3313
1759
1845
2133
2254
24
Right tributary
2553
2024
2107
2553
3122
3310
2556
1380
1937
64.6
238
113.7
335
160
84.7
198
198.9
110.5
69.6
43.1
68.5
69
63.9
201.2
187
222
236
362
269
78.5
97.7
119.4
110.9
334
158
75.1
246
217.7
109.7
75.5
108.5
38.2
62.4
71.2
52
209
210.6
249
295
462
318
88.5
110.2
143.5
EC (μS cm−1) 11.05.
72.2
71.6
181.4
164.6
92.7
56.2
53.9
148.9
165.9
191.4
227
194.5
137.3
463
186
107
205
204.8
48
63
41
85
233
171.6
190.7
256
279
213
72
96
87
0.5
0.5
4.7
6.8
7.1
6.9
6.7
5.9
6.9
6.1
6.7
6.8
6.4
7.1
7.1
6.4
6.5
6.3
6.3
6.8
6.2
5.1
1.0
0.8
4.8
8.7
9.3
8.6
8.4
7.8
8.3
8.1
8.0
7.9
8.8
8.2
9.4
9.1
8.1
8.3
8.1
8.1
8.2
8.2
7.8
9.3
8.9
9.0
8.6
8
8.5
8.3
8.8
8.6
8.5
8.3
7.8
0.6
1.1
7.4
7.6
7.5
6.6
6.5
7.2
7.1
7.2
7.4
6.6
6.7
6.9
7.2
13.1
16.4
16
23
19.8
20.9
19.1
23.8
20
25.9
19.6
13.7
22.6
25
24.2
23.5
23.3
22.6
21.6
20.2
17.2
19.6
10.8
14.8
11.7
11.9
8.6
10.5
8.9
12.2
11.5
11.2
11.1
8.2
3.8
10.2
10
10.7
12.1
11.1
11.2
9.5
8.8
5.3
2.4
8.6
9.5
9.2
9
11.9
11
10.2
11
10.6
10.4
10.1
9.8
16.7
15.7
16.1
15.3
20.1
15.7
19.9
19.9
16.3
13.9
18.2
18.4
18
11.2
EC EC DO DO DO DO T (°C) T (°C) T (°C) T (°C) (μS cm−1) (μS cm−1) (mg l−1) (mg l−1) (mg l−1) (mg l−1) 11.01. 11.05. 05.09. 02.12. 05.09. 02.12. 11.01. 11.05. 05.09. 02.12.
Page 8 of 17
Right tributary
21 11
Las Cuevas
Left tributaries
30
Río Grande
2543
Rio Grande
2268
22 25
Rio Grande
Rio Grande main river
Rio Gordito
16a 2950 18
Rio Gordito
3147
10 14
Río Gordito
3317
Site Elevation EC (μS cm−1) 11.01.
Rio Gordito
Rio Gordito
Río Grande
Table 2 Values for conductivity, dissolved oxygen and water sample temperature in the Rio Grande catchment
440 Environ Monit Assess (2017) 189:440
13.1 19.5 8.6 6.4
7.7 8.1
7.6
5.3
7.4
41.4
26.9 1.2
5.1
5.3
13.3
13.4
12.9
25
10.3 14.2
12.2 11.9
14.6
6.6 6.9 6.1
Page 9 of 17 440
for tributary sites, presumably as the different altitudinal ranges of the sub-catchments is not fully reflected in the altitude of the sampling site alone (Fig. 4). The solute composition of the tributaries was reflected by the electrical conductivity which showed substantial variation under both low and high flow conditions (Table 2). The highest altitude tributaries (sites 11 and 15) exhibit consistently low conductivities below 80 μS cm−1. In the middle part of the catchment, the Vega Larga and Aguas Negras (sites 17 and 19) had much higher conductivities which were > 400 μS cm−1 for the former and > 600 μS cm−1 for the latter, though these were also depressed in the September samples when flows were higher. Inputs in this area seem to at least partly explain the highest concentrations observed along the main stem of the river. Most of the remaining tributaries in the lower catchment generally had lower (< 200 μS cm−1) conductivities, similar to that of the lower Rio Grande near to Las Ramadas (Table 2).
825 928 2687 19 Left tributary
819
415 424 2948
16b 2950 Thermal spring left bank
Thermal springs
Vega Larga left bank tributary 17
3.7 mS cm−1 3.9 mS cm−1
128 104.4 33
154.9 16c 2950
GW spring right
1517
159.3
638
552
5370
144
227
Spatial variation in groundwater inputs
GW Virgen right tributary
Río Grande
Table 2 (continued)
Site Elevation EC (μS cm−1) 11.01.
EC (μS cm−1) 11.05.
EC EC DO DO DO DO T (°C) T (°C) T (°C) T (°C) (μS cm−1) (μS cm−1) (mg l−1) (mg l−1) (mg l−1) (mg l−1) 11.01. 11.05. 05.09. 02.12. 05.09. 02.12. 11.01. 11.05. 05.09. 02.12.
Environ Monit Assess (2017) 189:440
Groundwater sampled from boreholes or discharging from the headwater springs of the Rio Gordito at altitudes above 2300 m has depleted isotopic signatures with δ2H and δ18O values generally remaining below − 100 and − 12% respectively. These also remain rather constant over the sampling periods (Table 1). The electrical conductivity however shows more marked spatial and temporal variation; site 9 (a stream bank borehole) consistently has a high electrical conductivity exceeding 330 μS cm−1, whilst site 10 (a headwater spring) ranged between 65 and 144 μS cm−1. The latter was more similar to stream water and two other boreholes (sites 12 and 13) further downstream (Table 2). Further downstream at an altitude of 2950 m, a thermal spring (site 16b) and groundwater discharge (site 16c) close to the confluence of the Rio Gordito and Aquas Negras show similar isotopic composition to the higher altitude groundwater sources, but higher solute concentrations reflected in their much higher electrical conductivities. The latter ranged between 3600 and 5400 μS cm−1 in the thermal spring (whose temperature was 40 °C), but the former were more comparable to the higher altitude sources (150–230 μS cm−1) at site 16c. A lower altitude spring at 1517 m (site 33) had more enriched isotopic ratios that were quite constant across the four sampling periods with δ2H and δ18O values remaining around − 75 and − 11 to − 10%
440
Environ Monit Assess (2017) 189:440
Page 10 of 17
Fig. 4 Boxplots (quartiles) of δ18O (‰) values for surface water samples (mainstream and tributaries) to compare seasonal differences in range values of each sampling campaign showing outliers in May and December
respectively. This site also had a lower electrical conductivity varying between 104 and 144 μS cm−1. Taken together, the isotopes and electrical conductivity of groundwater imply a common source of recharge in terms of snow melt which reflects the altitudinal variation in the isotopic composition of precipitation inputs, but different geological sources which have low, moderate or high solute composition reflected by the varying electric conductivity values. However, the stream water seems more dominated by sources of lower/moderate solute composition on the four sampling occasions here. Altitudinal gradient As previously described, there is a notable altitudinal effect displayed in the decrease of stable isotope values with elevation due to the marked Andean slope (Figs. 4 and 5). For the Rio Grande, a mean altitudinal gradient of − 0.14 was observed. For January and September, the observed gradient was − 0.14, for December − 0.12. For May, however, almost no altitudinal effect could be detected which might be attributed either to the strong evaporative effect in standing waters and peatland during the dry autumn which might also have their source at higher elevations (Table 1). Seasonal variability of isotopes and hydrochemistry: evaporative and moisture-related fractionation effects Sampling was carried out in summer (January), late autumn (May), snow melt-dominated spring (September)
and early summer (December 2013). In September, sampling points above 2900 m of elevation could not be reached due to snow-covered pathways. Dual isotope data from each sampling occasion were plotted against the Global Meteoric Water Line (GMWL) (Fig. 6). The data from the sampling campaigns exhibit lower slopes in January and May (6.42 and 5.17) respectively showing evidence of evaporative fractionation effects in the hot and dry summer period. This might be supported by the intensified evapotranspiration from standing water in the peatland Bbofedales^ valleys (Squeo et al. 2006). In contrast, the September and December samples have steeper gradients more like the GMWL and other neighbouring meteoric water lines (9.42 and 8.21 respectively). Groundwater samples are plotted separately and show a strong relationship with surface water origin.
Discussion Inferring a LMWL for the Andean Rio Grande catchment using stream water isotopes For most humid and continental regions, the correlation between δ2H and δ18O is equal to the GMWL. However, locally, the isotopic composition may strongly differ from the GMWL depending on the geographical, altitudinal and climatic conditions. Therefore, we need to define local meteoric water lines (LMWL) to be able to derive stream water origin from stable water isotopes
Environ Monit Assess (2017) 189:440
Fig. 5 Comparison of spatial variability of conductivity, discharge and δ18O values throughout the drainage network for the sampling dates 11 May and 2 December 2013 in the Rio Grande
Page 11 of 17 440
catchment. The black triangle highlights the location of the thermal springs which have a strong influence on the surrounding hydrochemistry
440
Page 12 of 17
Environ Monit Assess (2017) 189:440
Fig. 6 Local seasonal water lines (surface water samples) for January, May, September and December compared to the Global Meteoric Water Line (GMWL) indicating mainstream, tributary,
groundwater and thermal spring samples. Groundwater and thermal spring water samples were excluded from the regression line calculation
(Leibundgut et al. 2009). LMWLs are usually established from a time series of precipitation data at least through the hydrological year. For the semi-arid headwater catchments that we investigated, no stable isotope values for precipitation are available for the sampling dates and data collection was infeasible given the remoteness of the site and the infrequent visits. The utility of riverine-based data for the establishment of LMWLs in data-sparse areas was highlighted by Halder et al. (2015) who evaluated large-scale and long-term data from 235 stations of the Global Network of Isotopes in Rivers (GNIR). Precipitation and river samples all lie along the GMWL and show a strong global isotopic correlation between average rainfall and river discharge (R2 = 0.88). Despite this similarity when δ2H and δ18O data are plotted in dual isotope space, there is a marked difference in the seasonal amplitude of
isotopic values from streamflow and precipitation data, with the former being markedly damped. Looking at the GNIR data, the average seasonal δ18O amplitude is 2.5‰ for rivers compared to 7.5‰ for precipitation (Halder et al. 2015). This seasonal range for river water does also not increase with latitude, as it is usually observed for precipitation values. This could be explained by mixing and storage processes in surface and groundwaters, which may be more marked in mountainous regions (Jasechko et al. 2016). Based on these findings, there is increased confidence that streamflow sampling campaigns can be informative in the absence of precipitation data in such semi-arid environments of strong seasonality. The surface water samples (Table 1) do hence allow us to tentatively infer a LMWL for the Rio Grande of δ2H = 7.8 × δ18O + 5.94.
Environ Monit Assess (2017) 189:440
To put this into a regional context, we show precipitation-based neighbouring and downstream LMWLs that have been reported: The Chilean Meteoric Water Line is δ2H = 8.3δ18O + 9.8 (IAEA, International Atomic Energy Agency/World Meteorological Organization 2005), and a LMWL established by Oyarzún et al. (2016) in the Limarí basin downstream of Rio Grande catchment is δ2H = 8.5 δ18O + 16, the LMWL δ2H = 7.8 δ18O + 9.7 for the latitude of 21° (Aravena et al. 1999) and the LMWL δ2H = 7.8 δ18O + 10.3 for the Salares de Atacama at 23° (Fritz et al. 1981). All of these are close to the GMWL. Altitudinal gradient Knowledge about the seasonally varying local isotope altitudinal gradient enables altitudinal effects to be separated from other moisture-dependent fractionation signals to enhance our understanding of the origin of water sources. This is particularly important for mountainous catchments of steep topography with differing contributions from ice, snow and groundwater. However, altitudinal gradients strongly contrast not only in different studies but also among different elevation zones. For example, Gat (1987) observed altitudinal gradients in δ18O precipitation data from − 0.15 to − 0.40‰ per 100 m worldwide for midlatitudes and Windhorst et al. (2013) for the tropical Andes gradients ranging from − 0.1 to − 0.3‰ per 100 m (Fig. 4). Further, Siegenthaler and Oeschger (1980) reported a 0.32 decrease in δ18O per 100-m increase in elevation for the Swiss Alps. For the drier extra-tropical Andean region at higher latitudes, varying altitudinal gradients from 0.1 to 0.4 were discussed by Vogel et al. (1975), Fritz et al. (1981), Aravena et al. (1999), Rozanski and Araguás (1995) and Ohlanders et al. (2013). The strong variation among these reported gradients can be either attributed to (1) varying gradients for different elevation bands within the landscape, showing, for example, steeper gradients between 2000 and 3000 m of elevation (Fritz et al. 1981) and by inference to the varying ranges of elevation (e.g. 0–6000 m different to 4000–6000 m) (Siegenthaler and Oeschger 1980; Aravena et al. 1999); (2) seasonally (wind dependent) varying moisture sources, for example, from moisture-rich tropical Atlantic trade winds or cold dry Pacific winds (Hoke et al. 2013; Windhorst et al. 2013); and (3) differing water sampling strategies as, e.g. eventbased precipitation (Windhorst et al. 2013) or
Page 13 of 17 440
cummulative (e.g. accumulated over a week or month) precipitation, and integrated and mixed ground and surface water. Ohlanders et al. (2013) indicated that at higher altitudes, the isotopic signal of streamflow will be affected by the spatial distribution of the snowpack and by the temporal dynamics of snow melt flow paths at different elevations. Consequently, the isotopic composition of stream waters in such snow melt-driven headwaters depends on the timing of when meltwater leaves the snowpack and how long it was exposed to evaporation. For the Rio Grande, for δ18O, a mean altitudinal effect of − 0.14‰ per 100 m was observed. Figure 3b compares the seasonal gradients for Rio Grande, a high elevation catchment in Ecuador (Windhorst et al. 2013) and the neighbouring Argentinean Andes (Vogel et al. 1975). Although there are insufficient values to derive a statistically significant elevation lapse rate values for each section, the gradient does seems to be steeper between 2000 and 3000 m. This would be consistent with Fritz et al. (1981) who for the the latitude of 21° in the Andes found the highest altitudinal lapse between 2000 and 3000 m. It would also be in line with Vogel et al. (1975) who for the Argentinean cordillera at the latitude of ca. 32° reported an increasing gradient (− 0.3‰ per 100 m) above 2000 m of elevation compared to lower lying areas. For areas above 3000 m of elevation, Fritz et al. (1981) and Vogel et al. (1975) reported a decreasing altitudinal gradient. This might be explained by changing sources from tropical moisture-rich air masses which increasingly influence the Andes at higher elevations (Mook et al. 2001; Hoke et al. 2013; Rozanski and Araguás 1995). Seasonality The relationships of δ2H and δ18O to the GMWL in each of the four sampling periods do reflect the strong climatic seasonality impacting on moisture availability and evaporative processes (Fig. 6). The regression lines have a lower slope for the samples collected during the dry season in summer and autumn (December, January and May) starting after the melting season when the discharge of surface and groundwater declines and surface waters might be subject to evaporative fractionation at the high elevation peatlands in the valley bottoms. The end of the melt season varies mainly depending on winter rainfall and temperatures at high elevation. In our
440
Page 14 of 17
dry study year, by December, there was almost no snow left. Due to the low flows and potential exposure to evapotranspiration, the May (autumn) samples show the lowest slope (5.16) followed by the ones collected in January (summer) with a slope of 6.42. The September samples, in contrast, with 9.42 show the steepest slope due to snow melt and wetter conditions in early spring time (Fig. 6). Stream water sources and residence times The knowledge about the local isotopic fractioning processes (LWML) including particular altitudinal and seasonal effects as described in the previous sections is indispensable to identify stream water sources or estimate residence times.The four surveys encompassing both the dry season and melt-dominated period allowed a preliminary assessment of the temporal dynamics of the hydrological sources which govern the generation of streamflow along the stream network. The characteristics of these perennial headwaters are to a large extent influenced by the seasonally varying climate, the steep gradient and the Andean geology. Streamflow is generated mainly by contributions from higher elevations as snow melt and water exfiltrating from the fractured basalt and metamorphic formations (Aravena et al. 1999). This is also reflected in the isotopic and geochemical composition of streamflow in the Rio Grande and its tributaries upstream of the geothermally influenced section at an elevation of 2950 m. Low electrical conductivity values suggest that precipitation and melt water has limited geochemical interaction with the local geology. This is supported by the analysis of major ions and cations showing increased amounts of base cations and sulphate below the area of geothermal alteration (Rusman 2014) and the fact that groundwater sample values—as illustrated in Fig. 6—show a strong similarity with surface water values. However, although low solute concentrations indicate that residence times of dominant sources are likely to be younger (i.e. < 1 year) in the mountainous areas (e.g. Jasechko et al. 2016), the exact time scales involved in the hydrological pathways remain unclear. Due to the fact that the stream and most tributaries are perennial, we can infer that there is significant storage in the fractured basalt which is able to slowly release melt water throughout the hydrological year. Strong diurnal variation in radiation inputs and temperatures lead to night-time freezing and a decrease of melting velocities.
Environ Monit Assess (2017) 189:440
The very low slope of the May water line (Fig. 6) caused by evaporative fractionation (Sprenger et al. 2017) might be supported by the intensified evapotranspiration from standing water in the peatland bofedales valleys (Squeo et al. 2006) found at elevations higher than 2000 m (Fig. 1c). It stores water in its deep peat layers and exposes it to evapotranspiration. Thus, the tracer survey results have provided us with a useful preliminary overview of the general flow paths in the Rio Grande. However, due to the limited number of values in space and time and the dry nature of the study year, we are unable to provide a more quantitative analysis on the exact sources and residence times of the individual sample points. To what extent can the tracer survey results inform hydrological models and aid water management? The stable isotope and geochemical signatures strongly infer that streamflow is composed of rainfall or snow melt quickly moving through the strongly fractured geological environment. This, together with observations related to air and stream temperature, soils and vegetation and observed discharges at different elevations and tributaries is useful information for hydrological models and can help improve the conceptualisation, structure and parameterisation of rainfall-runoff models. For the dry season, relative depletion of δ2H and deviation from the GMWL show evaporative effects caused by lower discharge and the stream water being exposed to evaporative losses in the extended peatland areas. The tracer data suggest a simple conceptualisation of runoff generation from a fractured groundwater reservoir, mainly fed by snow melt which is slowly released to the channel network. Moreover, it is clear that marked altitudinal increases in precipitation volume in the very high Andean peaks at 6000 m of elevation need to be inferred to explain the isotope depletion in the upper catchment which might help adjust precipitation inputs to models to allow the water balance to be adequately reproduced. On this basis, Nauditt et al. (2016) used dry and wet sub-periods to calibrate the conceptual HBV light model to simulate discharge at the outlets of Rio Grande and the neighbouring Tascadero catchments. Good statistical efficiencies and plausible hydrograph simulations were obtained for the wet season; those for the dry season were poorer, but still informative. As water managers in this region need seasonal discharge forecasts based on winter precipitation, these models
Environ Monit Assess (2017) 189:440
have potential predictive value. They could be further enhanced by data on snow cover, depth and density which could be derived from remotely sensed data such as Modis snow-water equivalent data which could be used to help drive the models. Similarly, although the tracer surveys presented here provide valuable information to inform model calibration and water management, more spatially and temporally resolved data would also be advantageous (e.g. Birkel et al. 2014). Potential for future sampling strategies This study has provided seasonal reference LMWLs and an altitudinal gradient as a basis to further assess Andean streamflow generation processes, groundwater residence times and stream water composition. In parallel, a number of other tracer-based assessment initiatives have been carried out in recent years in central Chile due to increased awareness of the urgent need for an evidence base for seasonal water availability predictions for water management. Some of the results have been published by Ohlanders et al. (2013), Oyarzún et al. (2014, 2016) and Hoke et al. (2013); other studies are still ongoing also in similar environments in Argentina and Peru. In the context of joint research meetings and training courses, most of the works were presented and discussed with representatives of local water user associations and public institutions who are interested in seasonal runoff forecasts and are willing to support sampling during their monitoring and maintenance field work. We expect that tracer surveys will play a major role in informing future water management about water contributions from data-sparse high-elevation catchments where the effects of climate change are expected to be marked (Vuille et al. 2015). As priorities, we therefore suggest the following strategic priorities: (1) establishing a large network of spatially distributed precipitation collectors and streamflow sampling sites for stable isotopes in all seasons to determine the dynamics of dominant water sources across the main (feasible) altitudinal ranges; (2) high temporal resolution sampling at selected streamflow sites to assess diurnal variability in isotope composition (and runoff sources) during the melt season; (3) synoptic surveys of stream sampling sites, wells and springs for full geochemical analysis to establish the hydrogeological provenance and impact on streamflow; (4) improve streamflow monitoring at higher altitude sites and assess diurnal discharge variability to provide
Page 15 of 17 440
further insight into runoff generation processes and varying water sources especially in autumn and summer; (5) assess groundwater age and residence times by sampling boreholes and springs and analysing for 3H, CFC and 13C samples; (6) install continuous air temperature measurements with simple sensors and data loggers at several elevation points of the Rio Grande catchment to obtain an accurate temperature lapse rate; and (7) carry out parallel sampling campaigns in the neighbouring Tascadero catchment (Nauditt et al. 2016) where grazing has been prohibited since 2014 in order to evaluate the impacts of earlier human activities on headwater catchments in tracer data from 2012 and 2013.
Conclusion Synoptic surveys using stable isotope and hydrochemical tracers were carried out in a non-glacierised, highelevation headwater catchment of the Chilean Central Andes which supplies downstream irrigation schemes with critical water resources. The data-sparse and semiarid region was exposed to a multi-year drought from 2008 to 2015 leading to severe water scarcity with profound economic implications for local agricultural production. Four sampling campaigns in January, May, September and December were conducted at a catchment elevation between 1250 and 3500 m to analyse the isotopic and hydrochemical composition of stream and spring waters with the aim to understand the seasonal contribution of snow, ground and spring waters to streamflow. The resulting data sets for this remote region are consistent with theoretical concepts of altitudinal influence on isotopes, as well as with published data in the lower Limarí basin and some neighbouring Andean catchments. The isotopic composition of stream and groundwaters was found to be in the range of the other LMWLs reported for altitudinal effect of the steep Andean topography, but also showing downstream enrichment in δ2H and δ18O which under the lowest flows also provided evidence for evaporative fractionation. The effect of snow melt is evident in the isotopic composition for the springtime melting season water line with a steeper gradient, whilst fractionation effects are supported by lower seasonal gradients for autumn and summer. Concentrations of solutes indexed by electrical conductivity indicate that there are limited contributions of deeper
440
Page 16 of 17
mineralised groundwater to streamflow, though such waters are evident in thermal springs. Nevertheless, weathering rates vary in the different sub-catchments, though generally conductivity values are relatively low suggesting that shallow groundwater sources recharged by snow melt are the dominant sources of runoff. To further develop a tracer-based understanding of the catchment, more intensive and extensive sampling strategies need to be implemented by local water user and research institutions to obtain more comprehensive isotope samples of precipitation, groundwater and streamflow with a higher resolution in space and time. Acknowledgements Funding for field visits was provided by the BMBF (German Federal Ministry for Education and Research) in the scope of the research projects BWeb based drought information system^ and BIncreasing water use efficiency in irrigation management^ (2012–2014). We especially thank our local project partners from the University of La Serena: Pablo Álvarez, Fabián Reyes and Nicole Kretschmer from the Centre for Advanced Studies in Arid Regions (CEAZA). Their support and hospitality were of key importance to be able to carry out the sampling campaigns. We also thank Christoph Schüth, head of the Institute for Applied Geosciences of the University of Darmstadt, to let us use the laboratory and facilities.
References Aravena, R., Suzuki, O., Peña, H., Pollastri, A., Fuenzalida, H., & Grilli, A. (1999). Isotopic composition and origin of the precipitation in northern Chile. Applied Geochemistry, 14, 411–422. doi:10.1016/S0883-2927(98)00067-5. Barnett, T. P., Adam, J. C., & Lettenmaier, D. P. (2005). Potential impacts of a warming climate on water availability in snow-dominated regions. Nature, 438, 303– 309. doi:10.1038/nature04141. Birkel, C., Soulsby, C., & Tetzlaff, D. (2014). Developing a consistent process-based conceptualization of catchment functioning using measurements of internal state variables. Water Resources Research. doi:10.1002/2013WR014925. CNR, Ciren, Comisión Nacional de Riego. 1997. Calculo y cartografía de la evapotranspiración potencial en Chile, http://bibliotecadigital.ciren.cl/gsdlexterna/collect/estudios/ index/assoc/ HASH01a7/d06c2417.dir/CNR-0029_1.pdf, accessed on 11.08.2015. Coplen, T. B. (1996). New guidelines for the reporting of stable hydrogen, carbon, and oxygen isotope ratio data. Geochimica et Cosmochimica Acta, 60, 3359. Döll, P., & Siebert, S. (2002). Global modeling of irrigation water requirements. Water Resources Research, 38(4). Fritz, P., Silva, C., Suzuki, O., & Salati, E. (1981). Isotope hydrology of groundwaters in the Pampa del Tamarugal, Chile. Journal of Hydrology, 53, 161–184. doi:10.1016/0022-1694 (81)90043-3.
Environ Monit Assess (2017) 189:440 Gat, J. R. (1987). Variability (in time) of the isotopic composition of precipitation: consequences regarding the isotopic composition of hydrologic systems. Isotope techniques in water resource development, IAEA-SM-319/39 (pp. 551–563). Vienna: IAEA. Halder, J., Terzer, S., Wassenaar, L. I., Araguás-Araguás, L. J., & Aggarwal, P. K. (2015). The Global Network of Isotopes in Rivers (GNIR): integration of water isotopes in watershed observation and riverine research. Hydrology and Earth System Sciences, 19, 3419–3431. doi:10.5194/hess-193419-2015. Hoke, G. D., Aranibar, J. N., Viale, M., Araneo, D. C., & Llano, C. (2013). Seasonal moisture sources and the isotopic composition of precipitation, rivers, and carbonates across the Andes at 32.5–35.5_S. Geochemistry, Geophysics, Geosystems, 14. doi:10.1002/ggge.20045. Hublart, P., Ruelland, D., Dezetter, A., & Jourde, D. (2015). Reducing structural uncertainty in conceptual hydrological modelling in the semi-arid Andes. Hydrology and Earth System Sciences, 19, 2295–2314. doi:10.5194/hess-192295-2015. IAEA, International Atomic Energy Agency/World Meteorological Organization. 2005. Isotope hydrology information system; IAEA/WMO: Vienna, Austria, http://isohis.iaea.org, accessed in 2015. Jasechko, S., Kirchner, J. W., Welker, J. M., & McDonnell, J. (2016). Substantial proportion of global streamflow less than three months old. Nature Geoscience, 9, 126–129. doi:10.1038/ngeo2636. Kendall, C., & Caldwell, E. A. (1998). Fundamentals of isotope geochemistry. In C. Kendall & J. J. McDonnell (Eds.), Isotope Tracers in Catchment Hydrology (pp. 51–86). Amsterdam: Elsevier Science. Leibundgut, C., Maloszewski, P., & Külls, C. (2009). Tracers in hydrology. Chichester: John Wiley & Sons Ltd. Lessels, J., Tetzlaff, D., Birkel, C., Dick, J., & Soulsby, C. (2016). Water sources and mixing in riparian areas revealed by tracers and geospatial analysis. Water Resources Research, 52(1), 456–470. doi:10.1002/2015WR017519. MADDTechnologies. http://www.madd.ch/index.php?option= com_deeppockets&task=contShow&id=56& Itemid=318, assessed in 2014. McGuire, K., Torgersen, C., Likens, G., Buso, D., Lowe, W., & Bailey, S. W. (2014). Network analysis reveals multiscale controls on stream water chemistry. Proceedings of the National Academy of Sciences of the United States of America, 111, 7030–7035. Mook, W. G., Gat, J. R., & Meijer, H. A. J. (2001). Environmental isotopes in the hydrological cycle: principles and applications, v. IV: groundwater—saturated and unsaturated zone technical documents. Hydrology, 39 SC.2001/WS/37. Nauditt, A., Birkel, C., Soulsby, C., & Ribbe, L. (2016). Conceptual modelling to assess the influence of hydroclimatic variability on runoff processes in data scarce semi-arid Andean catchments. Hydrological Sciences Journal. doi:10.1080/02626667.2016.1240870. Ohlanders, N., Rodriguez, M., & McPhee, J. (2013). Stable water isotope variation in a central Andean watershed dominated by glacier and snowmelt. Hydrology and Earth System Sciences, 17, 1035–1050. doi:10.5194/hess-17-1035-2013.
Environ Monit Assess (2017) 189:440 Oyarzún, R., et al. (2014). Multi-method assessment of connectivity between surface water and shallow groundwater: the case of Limarí River basin, north-central Chile. Hydrogeology Journal, 22, 1857–1873. doi:10.1007 /s10040-014-1170-9. Oyarzún, R., Zambra, S., Maturana, H., Oyarzún, J., Aguirre, E., & Kretschmer, N. (2016). Chemical and isotopic assessment of surface water–shallow groundwater interaction in the arid Grande river basin, North-Central Chile. Hydrological Sciences Journal, 61(12), 2193–2204. doi:10.1080 /02626667.2015.1093635. Price, M. F., & Egan, P. A. (2014). Policy brief: our global water towers: ensuring ecosystem services from mountains under climate change. Paris: UNESCO. Rozanski, K., & Araguás, L. (1995). Spatial and temporal variability of stable isotope composition over the South American continent. Bulletin de l'Institut français d'études andines, 24, 379–390. Rusman, A. (2014). Hydrogeological assessment in the high Andean cordillera, North Central Chile. MSc Thesis at the Department of Geosciences at the University of Darmstadt. Siegenthaler, U., & Oeschger, H. (1980). Correlation of 18O in precipitation with temperature and altitudes. Nature, 285, 314–318. Soulsby, C., Rodgers, P., Smart, R., Dawson, J., & Dunn, S. (2003). A tracer-based assessment of hydrological pathways at different spatial scales in a mesoscale Scottish catchment. Hydrological Processes, 17, 759–777. doi:10.1002/hyp.1163. Soulsby, C., Tetzlaff, D., & Hrachowitz, M. (2009). Tracers and transit times: windows for viewing catchment scale storage? Hydrological Processes, 23, 3503–3507. doi:10.1002/hyp.7501. Soulsby, C., Birkel, C., Tetzlaff, D., & Dunn, S. M. (2011). Inferring groundwater influences on surface water in montane catchments from hydrochemical surveys of springs and streamwaters. Journal of Hydrology, 333, 199–213. doi:10.1016/j.jhydrol.2006.08.016. Souvignet, M., Oyarzún, R., Koen, M., Verbist, J., Gaese, H., & Heinrich, J. (2012). Hydro-meteorological trends in semi-arid
Page 17 of 17 440 North-Central Chile (29-32° S): water resources implications for a fragile Andean region. Hydrological Sciences Journal, 57(3), 479–495. doi:10.1080/02626667.2012.665607. Sprenger, M., Tetzlaff, D., Tunaley, C., Dick, J., Soulsby, C. (2017). Evaporation fractionation in a peatland drainage network affects stream water isotope composition, Water Resources Research, 53, 851–866. doi:10.1002/2016 WR019258. Squeo, F. A., Barry, G., Warner, R. A., & Espinoza, D. (2006). Bofedales: high altitude peatlands of the central Andes Bofedales: turberas de alta montaña de los Andes centrales. Revista Chilena de Historia Natural, 79(245–255), 2006. Vicuña, S., Garreaud, R. D., & McPhee, J. (2011). Climate change impacts on the hydrology of a snowmelt driven basin in semiarid Chile. Climatic Change, 105(3), 469–488. doi:10.1007/s10584-010-9888-4. Viviroli, D., Dürr, H. H., Messerli, B., Meybeck, M., & Weingartner, R. (2007). Mountains of the world, water towers for humanity: typology, mapping, and global significance. Water Resources Research, 43(7), W07447. doi:10.1029/2006WR005653. Vogel, J. C., Lerman, J. C., & Mook, W. G. (1975). Natural isotopes in surface and groundwater from Argentina. Hydrological Sciences Bulletin, 20, 203–221. Vuille, M., E. Franquist, R. Garreaud, W. Lavado, B. Caceres (2015). Impact of the global warming hiatus on Andean temperature. Journal of Geophysical Research, 120(9), 3745–3757. doi:10.1002/2015JD023126. Windhorst, D., Waltz, T., Timbe, E., Frede, H. G., & Breuer, L. (2013). Impact of elevation and weather patterns on the isotopic composition of precipitation in a tropical montane rainforest. Hydrology and Earth System Sciences, 17(1), 409–419. doi:10.5194/hess-17-409-2013. Zimmer, M. A., Bailey, S. W., McGuire, K. J., & Bullen, T. D. (2013). Fine scale variations of surface water chemistry in an ephemeral to perennial drainage network. Hydrological Processes, 27, 3438–3451. doi:10.1002/hyp.9449.
