Journal of Hydrology

Surface hydrology on the highest volcano of the high Dry Andes, the Ojos del Salado;
interannual fluctuations and moisture sources
--Manuscript Draft--

Manuscript Number: HYDROL58640

Article Type: Research paper

Keywords: Andes, permafrost, hydrology, tritium, stable isotope

Corresponding Author: Adam Igneczi, Ph.D

University of Bristol
Bristol, UNITED KINGDOM

First Author: Balázs Nagy

Order of Authors: Balázs Nagy

Sebastián RUIZ-PEREIRA

Adam Igneczi, Ph.D

József KOVÁCS

Kaveh GHAHRAMAN

Gábor MIHAJLIK

Marianna TÚRI

Zoltán Kern

Abstract: The Atacama Plateau in the Central Andes (28-22°S) is characterised by a dry and
cold periglacial tundra due to the high altitude, low precipitation, and high evaporation.
Endogenous freshwater sources – e.g.: seasonal streams and lakes, subsurface
reservoirs, surface snow/ice patches – are available, though they are highly sensitive
to climatic changes. The near surface hydrological network is highly modified by the
distribution and seasonal evolution of perennial frozen ground, i.e. permafrost, which is
also expected to change in the future. The interplay between permafrost and
hydrology, especially in relation to future climate change, is poorly explored. To
address this issue, we carry out long-term ground temperature measurement and
modelling, snow coverage survey, tritium- and stable isotope analysis of surface
waters on the Ojos del Salado Massif, which is representative of high altitude
mountains on the Atacama Plateau. According to our results, a highly transient surface
hydrological network – lakes, springs and streams – forms during each summer where
permafrost is widespread and ground thawing (i.e. active layer) is present (~4900-6500
m a.s.l.). In this system, the water is of meteoric origin and relatively young (<10
years). The development of the network is strongly influenced by the active layer,
which plays a crucial role in storing, seeping, and discharging groundwater. However,
future permafrost degradation is expected to reduce the seasonal presence of shallow
water, and hence, modify groundwater recharge patterns.

Suggested Reviewers: Darío Trombotto

CONICET Mendoza
[email protected]
Decades of expertise in Andean geocryology.

Gino Casassa
Direccion General de Agua, Chile
[email protected]
Expertise in Andean glaciology, geocryology

Kenji Yoshikawa
University of Alaska
[email protected]
Expert in Arctic and Andean geocryology.

Shelley MCDonnell

Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
 University of Canterbury
[email protected]
Expertise in Andean glaciology, hydrology

Lukas Arenson
BGC Engineering Inc
[email protected]
Expertise in Andean hydrology and permafrost

Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Manuscript (double-spaced and continuously LINE and PAGE Click here to view linked References
numbered)

1 Surface hydrology on the highest volcano of the high Dry Andes, the Ojos del
2 Salado; interannual fluctuations and moisture sources
3 Balázs NAGY1, Sebastián RUIZ-PEREIRA2, 3, Ádám IGNÉCZI4*, József KOVÁCS5, Kaveh
4 GHAHRAMAN1, Gábor MIHAJLIK6, Marianna TÚRI7, Zoltán KERN8

5 1: Department of Physical Geography, Eötvös Loránd University, Budapest, Hungary

6 2: Escuela de Ingeniería, Pontificia Universidad Católica de Chile, Santiago, Chile
7 3: PermaChile Network, Santiago, Chile
8 4: School of Geographical Sciences, University of Bristol, Bristol, UK
9 5: Department of Physical and Applied Geology, Eötvös Loránd University, Budapest, Hungary
10 6: Department of Atomic Physics, Budapest University of Technology and Economics, Budapest, Hungary
11 7: Isotope Climatology and Environmental Research Centre, Institute for Nuclear Research, Eötvös Loránd Research
12 Network, Debrecen, Hungary
13 8: Institute for Geological and Geochemical Research, Research Centre for Astronomy and Earth Sciences, Eötvös
14 Loránd Research Network, Budapest, Hungary

15 * Correspondance: Ádám Ignéczi <[email protected]>

16 Abstract
17 The Atacama Plateau in the Central Andes (28-22°S) is characterised by a dry and cold
18 periglacial tundra due to the high altitude, low precipitation, and high evaporation. Endogenous
19 freshwater sources – e.g.: seasonal streams and lakes, subsurface reservoirs, surface snow/ice
20 patches – are available, though they are highly sensitive to climatic changes. The near surface
21 hydrological network is highly modified by the distribution and seasonal evolution of perennial
22 frozen ground, i.e. permafrost, which is also expected to change in the future. The interplay
23 between permafrost and hydrology, especially in relation to future climate change, is poorly
24 explored. To address this issue, we carry out long-term ground temperature measurement and
25 modelling, snow coverage survey, tritium- and stable isotope analysis of surface waters on the
26 Ojos del Salado Massif, which is representative of high altitude mountains on the Atacama
27 Plateau. According to our results, a highly transient surface hydrological network – lakes, springs
28 and streams – forms during each summer where permafrost is widespread and ground thawing
29 (i.e. active layer) is present (~4900-6500 m a.s.l.). In this system, the water is of meteoric origin
30 and relatively young (<10 years). The development of the network is strongly influenced by the
31 active layer, which plays a crucial role in storing, seeping, and discharging groundwater.
32 However, future permafrost degradation is expected to reduce the seasonal presence of shallow
33 water, and hence, modify groundwater recharge patterns.

1
34 Keywords: Andes, permafrost, hydrology, tritium, stable isotope

35 1. Introduction
36 Mountains could be considered freshwater reservoirs for lowlands (Viviroli &
37 Weingartner, 2008) with major recharge from the cryosphere (Barnett et al., 2005). However,
38 throughout the Central and Southern Andes, atmospheric warming and precipitation loss is
39 expected to significantly reduce snow cover by 2100 (Boisier et al., 2018; Bozkurt et al., 2018),
40 thereby accelerating glacier and ice bearing permafrost retreat (Baraer et al., 2012). Thus
41 recharge at high elevations and consequently water supply in lowlands (Cheng & Jin, 2013;
42 Oliva & Fritz, 2018) is also expected to decrease in the region, significantly deteriorating water
43 security (Boisier et al., 2018; Bozkurt et al., 2018). Current trends corroborate these projections, e.g.
44 annual snow cover persistence declined by about 2-5 days between 2000-2016 in the Southern
45 Andes (29-39° S) (Saavedra et al., 2018). This shortage of mountain snow has already caused an
46 extended, severe drought that is unprecedented in the hydrological and climatological records of
47 this region (Masiokas et al., 2020).
48 On the Atacama Plateau (Puna de Atacama) in the Central Andes (between 28-22°S)
49 dry and cold periglacial tundra conditions are predominant due to the high altitude (> 5000 m
50 a.s.l.), low precipitation, and high evaporation (Lobos-Roco et al., 2021). Despite the
51 arid/hyperarid periglacial environment, endogenous freshwater sources such as surface streams
52 and surface/subsurface reservoirs are available (Nicholson, 1998). However, these water sources
53 are highly sensitive to the expected future decline in precipitation across the Dry Andes (Bozkurt
54 et al., 2018) that will diminish recharge in the preferential zones (Carroll et al., 2019) and
55 exacerbate future droughts (Alvarez-Garreton et al., 2021). Furthermore, the degradation of
56 permafrost and subsurface ice in a warming climate (Masson-Delmotte et al., 2021) is also
57 expected to modify infiltration patterns and snowmelt partitioning (Hammond et al., 2019);
58 groundwater pathways, flux, and confinement (Cochand et al., 2020); and groundwater discharge
59 to surface streams (Somers & McKenzie, 2020).
60 Hence, interstitial ice thaw within the arid periglacial zone of the Dry Andes – e.g. the
61 Ojos del Salado near the latitude of 27°S (Obu et al., 2019, Nagy et al., 2019) – due to future
62 warming and drying will have direct implications for aquifer recharge and also for the water
63 budget of the wider region including nearby lowlands. Despite, the expected increase in the role

2
64 of permafrost – and its degradation – in the hydrological system of the Dry Andes, and other
65 regions with similar conditions, only limited research is available (Arenson et al., 2022). To
66 address this shortcoming, we deploy a wide variety of methods – including long-term ground
67 temperature measurement and modelling, satellite remote sensing of snow coverage, radioactive
68 and stable isotope analysis of water samples – on the Ojos del Salado to gain a comprehensive
69 view of the complex hydrological system of the mountain. Based on these insights, we also
70 develop a conceptual model of this system, which is directly applicable to the numerous other
71 high altitude massifs of the Dry Andes, and also to similar high-altitude cold and dry tundra
72 environments.

73 2. Study site
74 Several major massifs rising above 6000 m a.s.l. – mostly dormant and inactive
75 volcanoes – can be found along the 1000 km long Dry Andes (part of the Central Andes), and
76 also within the even more arid Andean Dry Diagonal (between 27° and 30°S). The highest
77 massifs of the Andean Dry Diagonal have a topographic prominence of about 2000-2500 m from
78 the surrounding Puna Plateau (Figure 1). Their climate is extremely arid and cold, with
79 precipitation dominated by winter snow which usually forms a thin cover that is present for only
80 a few weeks due to strong sublimation (Kull et al., 2002). Thus active glaciers are arguably
81 absent, though ground ice (i.e. ice bearing permafrost) can be found above 5000 m a.s.l. (Gjorup
82 et al., 2019). Due to the presence of this subsurface ice deposits/reservoirs, these cold high-
83 altitude deserts could be considered the frozen water towers of the Dry Andes.
84 The highest of these peaks is the Ojos del Salado (6893 m a.s.l., 27°07′34″S
85 68°32′26″W), the prominent peak of the corresponding Ojos del Salado massif (OSM) which
86 also includes the immediate vicinity/foreground of the volcano (Figure 1, 2). The OSM is a
87 representative example of this chain of volcanic giants. Measurements from the foreground of
88 the OSM – near the Laguna Verde Station at 4914 m a.sl. (Figure 1, 2) – between 2012 and 2015
89 indicate extremely low annual mean precipitation (181 ±25.5 mm). Furthermore, only two major
90 snow events were detected in the same period: one in 2012 lasting ~5 days with a maximum
91 snow depth of 62 mm, and another in 2015 lasting ~32 days with a maximum snow depth of 243
92 mm (Enguita et al., 2016). Accordingly, the presence of surface ice is very limited as the climatic
93 snowline (~7000 m a.s.l.) is estimated to be just above the Ojos del Salado summit (Houston and

3
 94 Hartley, 2003). However, permafrost and ground ice has been detected above 5000-5500 m a.s.l.
95 on the OSM (Nagy et al., 2019). Surface freshwater such as small and shallow ponds (Aszalós et
96 al. 2016, Nagy et al., 2019, 2020), and short streams (Kereszturi et al., 2022) has also been
97 reported from the OSM recently, though only from high elevations. Larger, saline lakes – e.g. the
98 Lagune Verde with a corresponding saturated aquifer of 128.7 km2 (Vargas Paysen, 2013) –
99 located in the lower foreground (below ~5000 m a.s.l.) of the OSM are well-known and studied.
100 These saline lakes are fed predominantly by warm (up to 36 °C) saline springs and considered to
101 be permanent. However, the temporal persistence, water sources, and hydrological connectivity
102 of high elevation surface ponds and streams are largely unknown (Aszalós et al. 2016, 2020;
103 Nagy et al. 2020).

104 3. Methods

105 3.1. Ground temperature measurements

106 Ground thermal regimes, permafrost distribution, and active layer dynamics have been
107 investigated on the OSM since 2012 (Nagy et al., 2019) using measurements from a shallow
108 ground temperature monitoring network (Figure 1, 2) that we established on the OSM "from-toe-
109 to-top" (4550-6893 m a.s.l.). The influence of local bias on our temperature measurements (e.g.
110 Ishikawa, 2003; Brenning et al; 2005) is suppressed by careful site selection, i.e. on relatively
111 flat areas covered by debris representative of the wider area, away from cliff faces, topographic
112 depressions, and perennial snow patches (Nagy et al., 2019). Hourly ground temperatures at
113 shallow depths (10 to 60 cm) – capturing the active layer – are measured by HOBO Pro v2
114 temperature loggers (operation range: −40°C to 70°C, accuracy: ±0.21°C) that buried at
115 representative sites of all the major vertical environmental zones of the OSM (Nagy et al., 2019).
116 The study period extends from March, 2012 to January, 2020; except for one depth level (-10
117 cm) at the Atacama Camp, where the first two years of data is unavailable due to instrument
118 error.
119 Permafrost, ground with temperatures below 0°C for at least two consecutive years,
120 presence can be established directly from ground temperatures (van Everdingen et al., 2005).
121 However, where the permafrost table – i.e. the upper boundary surface of permafrost – is below
122 the deepest temperature logger more indirect methods are necessary (e.g. Ishikawa, 2003). We
123 build upon a detailed investigation of ground thermal regimes – i.e. ground temperature

4
124 distribution, heat flows, and their time-dependence (van Everdingen et al., 2005) – on the Ojos
125 del Salado that was carried out by Nagy et al. (2019) for the period of 2012-2016. Conclusions
126 presented in this study provide the basis for determining the representative vertical profile of
127 permafrost on the OSM. This aids our assessment of how permafrost and the hydrological
128 system interacts on the OSM, e.g. how the duration and course of thawing in the active layer
129 predetermine and modulate infiltration and recharge. Linear regression was also applied on the
130 monthly average temperature records to determine long-term trends in ground temperature.

131 3.2. Ground temperature modelling

132 A major shortcoming of any ground temperature monitoring network is the incapability
133 to directly obtain information about the time-dependent liquid water and ice ratio of the regolith.
134 The presence or absence of water can be inferred by investigating temperature periodicity during
135 freezing-thawing cycles, i.e. zero-curtain method (Outcalt et al., 1990; Nagy et al., 2019).
136 However, with this indirect method, it is not possible to estimate the ratio of water and ice in the
137 ground. Thus, we have developed a 1D thermal model – presented in Nagy et al. (2020) – that
138 can be used to estimate the range of realistic water content in the ground by tuning modelled
139 ground temperatures to observations, i.e. analysing the performance of model runs using
140 different sets of parameters (e.g. water content).
141 Our model simulates temperature distribution and water/ice phase changes within the
142 active layer to clarify interannual properties and determine any existing trends. The ground
143 material is assumed to be a composite of liquid water, ice, rock, and air. The volumetric ratio of
144 the fractions is time-independent, only the ratio of liquid water and ice is allowed to evolve with
145 time. A continuous local thermal equilibrium is assumed between the ice, water, rock and air
146 components. Hence, weighted average quantities – calculated by using the volume fractions of
147 the different components – are used to represent the density, heat capacity and heat conduction
148 coefficient of the composite ground material. The model is using a transient conduction equation
149 with the latent heat of fusion, thus includes the effects of ice- water phase changes on the heat
150 transfer and temperature variations (Nagy et al., 2020).
151 We restricted our modelling exercise to two study sites: the Atacama Camp (5260 m
152 a.s.l.) and the Tejos Camp (5830 m a.s.l.), as these are the sites where permafrost, ice-bearing
153 permafrost and well-developed activate layer are all present (Nagy et al., 2019). Temperature

5
154 records from three data loggers (buried at 10 cm, 35 cm, and 60 cm below the surface) are used
155 for both sites. The logger closest to the surface (i.e. at -10 cm) provides input values to the
156 model, while the other two provides independent data streams, which are used to evaluate the
157 performance of the model (Nagy et al., 2020).
158 Main assumptions of the thermal model: (1) one-dimensional geometry; (2) time
159 independent air, water (liquid plus solid), and rock ratio; (3) constant -2.5°C temperature at the
160 bottom of the model domain, 5 m from the surface; the temperature was based on the observed
161 annual mean ground temperatures at -60 cm, while the depth was based on the performance of
162 multiple model runs with different domain depths (1-10 m, with 1 m interval); (4) the volumetric
163 changes of water during freezing/thawing are neglected.
164 The key issue with this model (Nagy et al., 2020) is the difficulty of empirically
165 determining the volume ratio of the fractions (i.e. air, rock, and water/ice) as functions of depth
166 which is crucial for accurately simulating phase-changes and the thermal behaviour of the
167 regolith. Nagy et al. (2019) analysed the physical properties of several regolith samples taken
168 from multiple depths at the Atacama and Tejos Camps. As the samples exhibited weak spatial
169 variations, an overall 60% (v/v) porosity and rock fraction density (2700-3000 kg/m3) was used
170 in the model of Nagy et al., (2020). Thermal properties of the rock fraction was assumed to be
171 similar of andesite-basalt material (Eppelbaum et al., 2014). Water content (liquid and solid) was
172 not measured on the field by Nagy et al. (2019), thus a depth dependent but temporally fixed
173 water content was prescribed by manually tuning modelled temperatures to the measurements
174 (Nagy et al., 2020).
175 This method is labour intensive and might not arrived at the best solution due to the
176 limited number of possible tuning experiments. Thus, we have upgraded the model to provide a
177 more rigorous automated solution for determining the appropriate depth functions of fraction
178 volume ratios, that provide the best match between modelled and measured temperatures. As a
179 compromise, we describe the aforementioned depth dependence by a sparse set of parameters
180 and employ a couple of simple assumptions: (1) the volume ratio of air is monotonically
181 decreasing as function of depth; (2) similarly, volume ratio of rock is monotonically increasing.
182 These simplifications enable us to automatically optimise the depth functions of fraction volume
183 ratios by matching modelled and observed temperatures within gradient descent framework
184 (Ruder, 2016), which is in essence the basis of deep learning (LeCun et al., 2015) and artificial

6
185 intelligence schemes. Besides the parameterized depth dependent fraction volume ratios (Figure
186 3), the melting point (i.e. temperature) of water is also included in the optimisation scheme as a
187 control. This was found to be very close to 0°C, which demonstrates that the optimisation
188 technique did employ unphysical corrections. This also indirectly proves the presence of water.

189 3.3. Satellite survey of snow coverage and lake presence

190 As sporadic snow cover on the OSM could play a significant role in determining surface
191 water presence and groundwater recharge, we assessed the presence of two high-elevation small
192 lakes at 5900 m and 5890 m a.s.l. (Figure 4) in relation to snow coverage between mid-winter of
193 2017 to late summer of 2020. This timeframe was chosen as we had complex field activity every
194 summer during this period, thus we have continuous field observations to aid the interpretation
195 of remotely sensed data.
196 Snow coverage and lake presence were surveyed – within the catchments of the two
197 lakes – by using Sentinel-2 optical satellite imagery with a maximum of 10% cloud coverage,
198 between 2/Aug/2017 and 28/Feb/2020 (n=88). Sentinel-2 images have been widely used to carry
199 out high resolution surveys of surface water bodies (Kaplan & Avdan, 2017; Pena-Regueiro et
200 al., 2020; Yang et al., 2017). Thanks to the dual satellite system, Sentinel-2 has approximately 5
201 days revisit frequency which makes it a useful tool for land cover mapping (Drusch et al., 2012).
202 Both visual inspection of true colour images (using bands 2, 3, 4) and normalised
203 difference snow index (NDSI) products were utilised to survey snow coverage within the lake
204 catchments. Lake presence was assessed similarly, using the same set of input satellite imagery
205 and true colour images. However, instead of NDSI, normalised difference water index (NDWI)
206 products (McFeeters, 1996) were calculated to aid the detection of the surface lakes. All of the
207 aforementioned data products were created by using the Sentinel Application Platform (SNAP).

208 3.4. Isotope analyses of water samples

209 Fresh snow, and water samples were collected from the “Upper Lake” and a stream near
210 the Tejos Camp on 25/Feb/2018. Water samples were also taken from the “Upper Lake” and
211 “Lower Lake” on 17/Feb/2020. Besides the stream sampled near Tejos Camp, a lower elevation
212 stream near the Atacama Camp was also sampled on 19/Feb/2020 (Table 1, Figure 4). Water
213 samples were collected in 500 mL HDPE containers leaving only minimal headspace and were

7
214 closed tightly to avoid leakage or evaporation. Isotope analyses were carried out on the water
215 samples to estimate the potential contribution of different water sources such as present-day
216 precipitation, ‘old’ or ‘young’ water from the thawing permafrost (Gibson et al., 2016, Wan et
217 al., 2019), or other potential sources (Stewart et al., 2017).

Code Site Date Lat Lon Alt (m

a.s.l.)
OS_2018_1 snow 25/Feb/2018 27°05'19.20" S 68°32'04.50" W 5910
OS_2018_2 "Upper Lake" 25/Feb/2018 27°05'18.70" S 68°32'03.40" W 5825
OS_2018_3 stream (Tejos) 25/Feb/2018 27°05'16.90" S 68°32'13.50" W 5825
OS-2020-1 "Upper Lake" 17/Feb/2020 27°05'18.70" S 68°32'03.40" W 5900
OS-2020-2 "Lower Lake" 17/Feb/2020 27°05'18.40" S 68°31'51.53" W 5890
OS-2020-3 stream (Tejos) 19/Feb/2020 27°05'16.90" S 68°32'13.50" W 5825
OS-2020-4 stream (Atacama) 19/Feb/2020 27°03'31.20" S 68°32'58.48" W 5240

218 Table 1. Date and location of sampling around the OSM in 2018 and 2020 collected for isotope
219 hydrological analysis

220 In the lab, a few mL water was sub-sampled from each container for stable isotope
221 analyses. The rest of each sample was distilled and filled into metal canisters. After degassing,
222 the metal canisters were closed and stored for several months to allow 3He ingrowth from tritium
223 decay. Finally, the canisters were connected to the inlet line of a noble gas mass spectrometer
224 (Helix SFT) and the He fraction was allowed to enter the dual collector mass spectrometer. 3He
225 and 4He were measured simultaneously, applying the peak height method. During the sample
226 intake, an ultrapure 4He spike was added to each sample (Palcsu et al., 2010). Tritium activity is
227 expressed in tritium units (1 TU equals to the 3H/H ratio of 10-18, and corresponds to an activity
228 concentration of 0.119 Bq/L). The overall precision of the tritium measurements is usually better
229 than 2.0% above 1 TU, if sample size is larger than 500 mL.
230 Measured tritium activities are compared to decay corrected reconstructed tritium
231 activities of precipitation to help determine the origin of contemporary surface water on the
232 OSM. Reconstructed annual mean regional precipitation 3H activities have been retrieved from a
233 set of regional records for the temperate zone of South America (Basaldúa et al., 2022). The
234 region situated closest to the OSM was selected from this dataset. These reconstructions,
235 available from 1960 to 2022 (Basaldúa et al., 2022), were corrected for radioactive decay using
236 the estimated 3H half-life of 12.32 years (Lucas and Unterweger, 2000) before comparisons with
237 the measured tritium activities of our samples.
238

8
239
240 The measured tritium activities of the samples were compared to gridded mean annual
3
241 H in precipitation, which is estimated by inverse distance weighting from 1955 to 2008
242 (Jasechko and Taylor, 2015) and by a robust regression model – based on environmental and
243 geospatial covariates for the multi-year mean 3H of contemporary (2008–2018) precipitation –
244 from 2008 to 2018 (Terzer-Wassmuth et al., 2022). The decay-corrected tritium activity were
245 calculated for 2018 using a half-life of 12.32 years (Lucas and Unterweger, 2000).
246 The stable hydrogen and oxygen isotope composition of the samples was determined by
247 laser spectroscopy. Samples collected in 2018 were analysed by a Liquid Water Isotope Analyzer
248 (LWIA-24d, Los Gatos Research) at the Institute for Geological and Geochemical Research,
249 Budapest, Hungary. Samples and laboratory standards were pipetted into 2 mL vials, 1 μL
250 specimens from these vials were injected into the vaporizer of the laser analyser where they were
251 evaporated at 80 °C in low vacuum. Although 6 injections were made from each vial, only the
252 latest four measurements were used to determine the isotope composition of the samples to
253 minimize memory effect (Czuppon et al., 2021). All samples were measured at least two times.
254 The laboratory standards, calibrated to international standards BWS1, BWS2 and BWS, have the
255 following composition: δ²H = –9.0‰; –74.9‰; –147.7‰; and δ18O = –0.53‰; –10.41‰; –
256 19.95‰, respectively. Samples collected in 2020 were analysed by a Liquid Water Isotope
257 Analyzer (LGR LWIA-24i, ABB-Los Gatos Research) at the Institute for Nuclear Research,
258 Debrecen, Hungary. Stable hydrogen and oxygen isotope composition of the water samples are
259 expressed as δ²H and δ18O in ‰ relative to V-SMOW (Vienna Standard Mean Ocean Water)
260 (Coplen 1994). The reproducibility considering both datasets is better than 1.0‰ and 0.15‰ for
261 δ²H and δ18O, respectively.

262 3.5. Developing a conceptual hydrological model

263 Previous hydrological datasets and analyses were combined with new information
264 obtained in this study and theoretical considerations to develop a conceptual model that
265 characterises the current hydrological configuration – i.e. water sources and their connectivity –
266 of the OSM headwaters. We utilised a simplified hydrological budget equation, developed for
267 permafrost affected glacial forelands (Cooper et al., 2011), as a guide for the conceptual
268 framework:

9
269 𝑊𝐹 = 𝑊𝑃 + 𝑊𝑅 − 𝑊𝐸 − 𝑊𝑆𝑆𝑆 − 𝑊𝑆𝑅 ± 𝑊𝛥𝑊𝑠 (Eq. 2)

270 where the main channel net water flux (WF ) consists of; precipitation (WP ), channel recharge
271 from the active layer ( 𝑊𝑅 ), evaporation ( 𝑊𝐸 ), sub-surface seepage (active layer discharge
272 outside channel) (𝑊𝑆𝑆𝑆 ), surface runoff (𝑊𝑆𝑅 ) and the change in the water storage (WΔWs ).
273 Information about surface hydrology in the region was obtained from previous field
274 reports of the local Water Directory: D.G.A. Chile (Vargas Paysen, 2013; Enguita et al., 2016)
275 and the PermaChile network (Földgömb Foundation, Hungary). Ground thermal regime datasets
276 from field studies (Nagy et al., 2020) were also used to evaluate surface conditions that could
277 modify infiltration and subsurface hydrological dynamics at preferential recharge sites. Our
278 conceptual hydrological model can be used to evaluate current hydrological organisation and the
279 possible future pathways of its evolution.

280 4. Results

281 4.1. Timing and length of ground thawing

282 The ground temperature model – using continuous, long-term hourly temperature
283 measurements and machine learning optimization (Section 3.2) – allowed us to investigate the
284 vertical profile of thawing/freezing in the active layer and the associated phase changes of water.
285 As we have no measurements from depths larger than -60 cm which could be used to constrain
286 our model, we focus on the interpretation of model outputs for the uppermost 60 cm of the
287 ground. This almost completely covers the actual thickness of the active layer at the Tejos Camp,
288 while at the Atacama Camp the coverage is partial (Nagy et al., 2019) (Figure 5).
289 Results of the temperature modeling are presented graphically (Figure 5). Accuracy of
290 the model was assessed by comparing the presence of daily ground temperature periodicity –
291 which is a key indicator of thermal and phase-change processes in the active layer (Nagy et al.,
292 2019; 2020) – in the observed and modeled temperature signals (Text S1, Table S1). Observed
293 and modelled daily mean temperatures are also compared directly to aid this evaluation (Text S1,
294 Table S2). Both tests – and the visual comparisons – suggest that our model performs well and
295 captures the thermal evolution of the ground with sufficient accuracy.
296 Our modelling for the Tejos Camp indicated that positive ground temperatures and
297 phase changes are common at shallow depths, though the active layer is typically thinner than 1

10
298 m (Figure 5). The thickness of the active layer exhibits a notable (>20%) interannual oscillation,
299 reaching only about 55-60 cm in La Niña years (e.g. 2012/13, 2016/17, 2017/18), and >70 cm in
300 El Niño years (Figure 5). Although, confidence levels are lower in the latter case as the active
301 layer extends deeper than our model constraints, the scale of this extrapolation is small (< 20
302 cm). Hence, we propose that the ground thermal model can provide a full picture about the
303 evolution of the active layer at the Tejos Camp.
304 Ground thaw at 35 cm depth usually starts in the second half of December, or the
305 beginning of January, while at 60 cm depth it starts about 23-39 days later (Figure 5, Table S3).
306 Thawed ground usually persists for about 105-120 days per year (about 24% of the investigated
307 period) at 35 cm depth, although this can be considerably shorter in some years, e.g. in 2017/18
308 it lasted for only 80 days. Ground stays thawed for shorter periods with increasing depth, e.g. at
309 60 cm it lasts for about 90-112 days per year (Figure 5, Table S3).
310 The active layer is thicker and ground thawing lasts longer at the Atacama Camp.
311 According to the model results, the ground is thawed at 35 cm depth for about 45.8% of the
312 investigated period. At 35 cm depth, thawing usually starts in November, though this varies
313 significantly between different years (Figure 5, Table S3). The start of thawing only lags about
314 5-12 days behind at the depth of 60 cm. The ground stays thawed for about 160-180 days at 35
315 cm depth. At 60 cm depth, this period is only about 5-8 days shorter, though in many summer
316 seasons there is virtually no difference (Figure 5, Table S3). Hence, thawing propagates towards
317 larger depths quickly, while the length of the thawed period does not vary significantly close to
318 the surface at the Atacama Camp, in contrast to the Tejos Camp. The complete refreezing of the
319 ground is usually achieved by April and May at both sites (Figure 5, Table S3). As the modelled
320 thickness of the active layer extends well beyond the model constrains during each summer
321 (Figure 5), we cannot estimate the actual thickness of the active layer at the Atacama Camp with
322 sufficient confidence.

323 4.2. Presence of surface snow and water

324 High-altitude lakes on the OSM are formed in depressions on the flat tongues of former
325 lava flows, in basins between lavalobes, within tributary craters, and behind parasitic cones. We
326 investigated two regularly recurring lakes on the OSM – the "Upper Lake" (UL) and "Lower
327 Lake" (LL) – in the 5890-6150 m zone (Figure 6). The area around the two larger lakes (UL and

11
328 LL) has been field surveyed five times between 2014 and 2020. Apart from 14/Jan/2019 the
329 presence of these lakes were confirmed on the field (Figure 6h). Although field surveys are
330 useful for ground validation, they only provide a short snapshot about the temporal evolution of
331 the lakes; e.g. in 2019 satellite images indicate that the lakes appeared in February, similarly to
332 other years, a few weeks after our field visit (Figure 6). Satellite image surveys between
333 Aug/2017 and Feb/2020 reveal that open water in the lake basins usually appears in the second
334 half of the summer – i.e. from February onwards (Figure 6). This is several months later than the
335 start of ground thawing within the 5800-6000 m a.s.l. elevational zone, starting in November
336 (Figure 5). Open water within the lake basins was usually present for about 6-8 weeks during
337 each summer, after this time the water either drained away and/or a thin ice layer formed in the
338 shallow ponds for the first half of the winter.
339 The timing of lake appearance in this elevational zone suggests an inverse connection
340 with the seasonal evolution of active layer thickness. Early in the melt season, until January, the
341 active layer is very thin, below 20 cm (Figure 5). Although the water storage and seepage
342 capacity of this thin active layer is small, meltwater infiltrates and seeps away instead of
343 accumulating in meltwater ponds. Lakes only appear within poorly drained areas around mid-
344 summer, when the active layer reaches the thickness of ~50 cm and its growth rate slows down
345 (Figure 5, 6). Later during the summer, as water saturation in the active layer becomes more
346 widespread, groundwater seepage overflows the lava plateaus and small, short-lived springs form
347 on the slopes of lava lobes (Figure 4). in parallel with this, small surface streams also appear as
348 the melt season progresses. These streams are fed either by the aforementioned springs (5800-
349 6000 m a.s.l.) or by melting perennial snow patches (Figure 4), especially at higher elevations
350 (6200-6300 m a.s.l.). The discharge of these streams exhibits a very strong diurnal variability.
351 Observations from 2018/19 indicate, that the presence of significant snow coverage is not a
352 strong prerequisite of lake formation. Even though only light snow coverage was detected in the
353 preceding winter, lakes do appear as usual in February (Figure 6). However, major snowfall
354 events in the summer strongly promote the formation and growth of lakes and meltwater ponds
355 due to the quick saturation of the active layer by the melting snowpack.

356 4.3. Isotope compositions of the surface waters

12
357 Stable hydrogen and oxygen isotope composition of fresh snow samples collected in
358 late February, 2018 were -69.5‰ for δ²H and -10.33‰ for δ18O, the 3H activity was 5.36±0.10
359 TU. Stable isotope composition of surface waters collected around the OSM in 2018 and 2020
360 ranged from -114.3 to -31.4‰ for δ²H, and from -15.32 to -2.0‰ for δ18O (Table 2). Tritium
361 activities of the samples ranged from 4.7 to 8.7 TU (Table 2). 3H activities of fresh snow samples
362 collected by us in 2018 (Table 2) and by an independent study in 1990 (6.22±0.38 TU, Grosjean
3
363 at al., 1995) agree within uncertainty with the annual mean precipitation H activity
364 reconstructions of Basaldúa et al. (2022), which were obtained for a nearby region. This supports
365 using this reconstruction as reference for the evaluation of the 3H activities of surface waters and
366 groundwater from the OSM region.
Sampling site 3
H δ²HVSMOW δ18OVSMOW
Sample code (TU) (‰) (‰)
OS_2018_1 snow 5.36±0.10 -69.5 -10.33
OS_2018_2 "Upper Lake" 5.54±0.11 -72.2 -10.15
OS_2018_3 stream (Tejos) 5.61±0.11 -90.2 -12.70
OS-2020-1 "Upper Lake" 8.71±0.33 -31.4 -2.01
OS-2020-2 "Lower Lake" 6.93±0.36 -36.9 -4.14
OS-2020-3 stream (Tejos) 6.11±0.36 -60.5 -7.22
OS-2020-4 stream (Atacama) 4.72±0.36 -114.3 -15.23
367 Table 2. Tritium activity and stable isotope composition of snow and surface waters collected
368 around the OSM in 2018 and 2020.

369 5. Discussion

370 5.1. Contemporary configuration of surface hydrology

371 Snow coverage on the OSM above 5000 m a.s.l. exhibits strong annual and seasonal
372 variations, while rain is rare. Thus, snow presence and melting determines the supply of liquid
373 water. Permafrost occurrence becomes widespread above ~ 5200 m a.s.l. on the OSM; this also
374 needs to be considered in the hydrological budget as infiltration and recharge is confined by the
375 permafrost table, while storage and seepage can also occur at shallow depths in the active layer
376 (Figure 7). The evolution of the snowpack also impacts the timing and length of ground thawing
377 due to its insulating effect (Figure 5) (Nagy et al.,. 2019). This in turn affects the thickness of the
378 active layer and its capacity for infiltration and groundwater seepage, and has consequences for
379 the depth and horizontal coverage of the permafrost table. Hence, at high elevations on the OSM
380 infiltration is restricted to preferential recharge zones that are controlled by the downslope

13
381 convergent topography (Carroll et al., 2019), and the spatiotemporal distribution of permafrost
382 (Figure 7), active layer, and snow coverage. This further complicates the already significantly
383 restricted recharge due to the strong evaporation and sublimation in this arid environment; i.e.
384 recharge is only about 5-10% of the precipitation in lower altitude arid areas near the OSM, e.g.
385 the alluvial basins at 27°S (Enguita, 2016).
386 Based on our ground temperature measurements and thermal model, freeze-thaw cycles
387 occur in the ground at both the Tejos and Atacama Camps. Strong early and late summer
388 thawing-freezing cycles without significant zero curtain periods are typical at shallow depths of
389 the Atacama Camp. However, similar cycles also occur at the Tejos Camp, especially in the early
390 summer (e.g.: Dec/2012, Dec/2016, Dec/2018). Snow coverage affects these cycles at both
391 locations. These transitional periods are characterised by large and abrupt changes in the
392 conductivity and storage capacity of the active layer. As the melt season progresses, and snow
393 coverage becomes insignificant, the ground remains thawed at shallow depths. Thawed ground
394 (i.e. active layer with liquid water) is present for about 32% and 39% of the observation period at
395 the Tejos and Atacama sites respectively. Hence, the active layer can be considered a seasonal
396 shallow unconfined aquifer (Figure 7), which gets thicker during the melting season – reaching
397 about 60 cm – and thus provides a deepening confinement table for infiltration. Infiltration and
398 shallow groundwater seepage can also be disrupted by short-term (lasting 1-2 weeks) refreezing
399 events, e.g. during late summer at the Tejos site, before ceasing completely at the end of
400 summer. The change in subsurface water storage (WΔWs) at the end of the melt season, when
401 main channel water flow is paused due to freezing (i.e. W F = 0), equals with the recharge (WR)
402 minus the active layer discharge (WSSS), according to the site specific hydrological budget (Eq.
403 2). The active layer is thicker/deeper and it is present for a longer time (+7% of the observed
404 period) at the Atacama Camp than at the Tejos Camp. Thus, considering similar magnitudes and
405 timing of precipitation, evaporation, and recharge – which is a reasonable assumption given the
406 close proximity of the two sites – change in active layer storage at the end of the melt season is
407 expected to be significantly more negative at the Atacama Camp than at the Tejos Camp due to
408 larger active layer discharge.
Altitude
Feature Attribute Reference
m a.s.l.
Transient seasonal Small and shallow with about Aszalós et al., 2016;
5200-6500
high-altitude lakes 0.8 - 1.4 m depth Nagy et al., 2020

14
 Lavalobes connecting plateaus
Seepage slopes 5830-6000 Nagy et al., 2020
with shallow lakes
Nagy et al., 2020;
Thaw slumping
5200 Slumps are 1-3 m in diameter Kereszturi et al.
(thermokarst)
2022
Retrogressive thaw Due to melting of firn buried
5200 Nagy et al., 2020
slumps within steep sand headwalls
Currently ephemeral with
Surface streams 4900-6300 strong diurnal and seasonal This study
variations; relict fluvial valleys
Perennial snow Associated with preferential
>5900 Nagy et al., 2020
patches recharge areas
409 Table 3. Summary of main hydrological structures of the Ojos del Salado massif

410 Although meltwater form the snowpack (Table 3) – and perennial firn patches in
411 sheltered locations – can infiltrate to the active layer and towards larger depths through
412 permafrost discontinuities and rock fractures, this is strongly inhibited where the active layer is
413 shallow and permafrost is widespread/continuous, especially above ~5600 m a.s.l. (Nagy et al.,
414 2019; 2020). Hence, a highly transient surface hydrological network consisting of meltwater
415 ponds and streams can form can form within the wider permafrost zone – i.e. continuous and
416 discontinuous permafrost between ~ 5000-6500 m a.s.l. (Nagy et al., 2019; 2020) – during the
417 summer (Table 4). Shallow seasonal meltwater ponds also form in this zone, i.e. within the
418 elevational range of ~5800-6300 m a.s.l. These ponds can appear even in the absence of
419 significant snow coverage, thus active layer discharge (WSSS in Eq. 2) likely plays an important
420 role in their formation. This water could be sourced from melting ground ice and stored surface
421 meltwater (i.e. from the melting snowpack and firn patches). Landforms generated by thermal
422 erosion and thaw slumping (Nagy et al., 2020; Kereszturi et al. 2022) – e.g. near the Atacama
423 Camp at 5200 m a.s.l. – and transient meltwater ponds (Aszalós et al. 2020) corroborate the
424 presence and degradation of ground ice, suggested by the analysis of ground temperature records
425 (Nagy et al., 2019). These features could also indicate preferential recharge sites due to the
426 presence of a highly porous mixture of regolith and ground ice (Nagy et al., 2020), and a
427 disturbed permafrost table. Seasonal surface streams within the wider permafrost zone have
428 highly variable discharge depending on diurnal temperature variations, the spatiotemporal
429 distribution of precipitation, and the distribution and volume of perennial firn patches. Liquid

15
430 surface water is unlikely to appear at higher elevations – above ~6500 m a.s.l. – as temperatures
431 only rise above 0°C occasionally (Nagy et al., 2019, 2020) and sublimation is strong.

432 5.2. Hydrological connectivity

433 Hydrological connectivity can be inferred from the isotopic signatures of our samples.
434 Stable isotope compositions of the surface water samples from the OSM are within the range of
435 compositions measured for precipitation samples in northern Chile (Jordan et al., 2015; Boschetti
436 et al., 2019). Snow and stream water samples exhibit a relatively more depleted composition and
437 fit well to the Northern Chile meteoric waterline (Boschetti et al., 2019), while lake water
438 samples usually show relatively more enriched composition and are scattered below it (Figure
439 8a) indicating some evaporation effect (Craig et al., 1963). The source composition of lake water
440 is estimated by a back extrapolation to the meteoric waterline along the empirical evaporation
441 line (Gat 1995). According to this, the estimated source composition of lake water is consistent
442 with the stable isotope composition of snow and stream waters (Figure 8a). These results suggest
443 that 1) high altitude summer lakes on the OSM are recharged by snowmelt and streams, and 2)
444 lake water is isotopically modified at the air-water interface due to evaporation, as expected in
445 such a dry environment (Gat 1995). In conclusion, the stable isotope composition of all our
446 samples from streams and lakes within the permafrost belt of the OSM suggest meteoric origin.
447 For the period between 2018 and 2020, tritium (3H) activities from ephemeral streams
448 and lakes between 5800-5900 m a.s.l. align with measurements taken from the fresh snow
449 samples (Table 3, Figure 8b) – collected near the Upper lake in 2018. The tritium signal also
450 broadly aligns with independent measurements made at another high elevation site in the North
451 Chilean Andes (Kinnard et al., 2020) and the Andean background level (Houston 2007).
452 Considering reconstructions of the annual mean tritium activity in precipitation for the OSM
453 region (Basaldúa et al., 2022), the decay-corrected activity is steadily below 4 TU before 2012
454 (Figure 8b). The tritium activity of our samples collected from surface waters was above 4 TU
455 without exception. Thus, we propose that if water from the thawing permafrost is a major source
456 of surface water (i.e., lakes and streams), then this water in the period of 2018-2020 could not
457 have been replenished before 2012. This means that the main fraction of the active layer and
458 surface water is young, with multiannual storage under 8 years pertaining to ‘perennial’ snow

16
459 cover and snowmelt sources as well. Recent reports about snowpack burial (Nagy et al., 2020)
460 and thawing in subsequent melt seasons (Kereszturi et al., 2022) support this interpretation.

461 5.3. Future evolution of surface hydrology

462 The spatiotemporal distribution of permafrost and the state of the active layer
463 (thickness, water content, seasonal evolution, etc.) are among the main controls of the
464 hydrological system in periglacial regions (Wang et al., 2009). The extension of the sporadic (i.e.
465 low probability) permafrost zone on the OSM under future RCP atmospheric warming scenarios
466 could lead to widespread permafrost degradation (Ruiz Pereira et al., 2021), modifying recharge
467 and infiltration patterns in the permafrost zone (Hiyama et al., 2013; Sjöberg et al., 2021).
468 Therefore, evaluations of the changing hydrological regime of cold-dry environments (Arenson
469 et al., 2022), such as the OSM, should consider changes in active layer storage (Evans & Ge,
470 2017) and groundwater recharge (Kuchment et al., 2000) due to permafrost degradation.
471 Nevertheless, the sensitivity of shallow interstitial ice to changes in moisture and temperature
472 forcing is uncertain at high altitudes (>5,000 m a.s.l.) on the OSM, and similar cold deserts.
473 These uncertainties are propagated when assessing the responses of groundwater flow, and the
474 response time, to surface climatic changes in cold deserts (Cuthbert et al., 2019) especially
475 beyond the next century (Bense et al., 2009).
476 Finally, surface, and subsurface hydrological processes and their interplay on the OSM
477 are likely to be influenced by hydrothermal circulation. For example, hydrothermal influx under
478 the lower boundary of permafrost (Figure 7) could imply a secondary recharge from stored
479 subsurface relict ice. This interaction is largely unknown on the OSM, with inherent uncertainty
480 stemming from the unknown thickness of the permafrost layer. This is heavily controlled by the
481 geothermal gradient and ranges broadly even on active volcanoes, i.e. from hundreds of metres
482 where the gradient is low (Abramov et al., 2008), to just a few metres where the gradient is high,
483 e.g.: Chajnantor Volcano (22°59′S, 67°44′W) where a 200 °C/km geothermal gradient limits
484 permafrost table to 5 m depth at 5000 m a.s.l. (Mena et al., 2021).

485 6. Conclusions

486 A transient surface water network – consisting of ephemeral lakes and streams – is
487 present between ~4900 and 6500 m a.s.l. on the OSM. Seasonal lakes – depressions and behind

17
488 dams of volcanic debris – are most abundant in the elevation range of 5800 to 6100 m a.s.l.
489 which is also associated with the high probability presence of both permafrost and ground ice,
490 which constitute a subsurface layer with low permeability. Surface water originates from local
491 snowmelt and active layer discharge – supplied by infiltration and seepage and/or in-situ thawing
492 of ground ice. During most of the year lake basins are dry or hosting small patches of snow and
493 ice. However, these basins fill up with meltwater (0.5-2 m deep) during the first half of the
494 summer, and the resulting lakes persist for about 6 to 8 weeks, through late summer and the
495 autumn transition period.
496 Seasonal surface streams can occur in the elevation range between 4900 and 6500 m
497 a.s.l., though they are the most abundant above 5500 m a.s.l. where permafrost hinders deep
498 infiltration and seepage and there are plenty of water sources. Like lakes, streams are fed by
499 melting seasonal and perennial surface snow patches and also by active layer discharge. The
500 latter is especially important in the zone of 5800 to 6100 m a.s.l., where lakes are also the most
501 abundant. Here, the active layer is relatively thick so it can store a significant amount of water,
502 the permafrost table is mostly continuous so it hinders deep infiltration and seepage, and there
503 are ample water sources so the active layer can become saturated during the melt season. Streams
504 usually have a shorter life span than lakes, lasting around 2 to 4 weeks during the summer. Stable
505 isotope compositions and tritium activities both suggest that the water in ephemeral streams and
506 lakes originates from recent (< 10 years old) precipitation.
507 Above 6500 m a.s.l. the formation of a surface hydrological network is strongly
508 inhibited by the cold temperatures, which limit thawing to a few hours during the summer and
509 make the active layer virtually absent. Here, sublimation is the dominant process that reduces the
510 thickness of the snowpack and causes the drying out of the top layers of the regolith. Thus, even
511 if temperatures allow some melting, this meltwater will quickly infiltrate and refreeze in the
512 ground. The formation of a surface hydrological network is also inhibited below 4900 m a.s.l.
513 due to the low probability of permafrost (i.e. sporadic permafrost), which do pose any
514 considerable hindrance to infiltration. Water sources are also scarce in this zone due to lower
515 precipitation and the scarcity of ground ice. Thus, meltwater produced locally or transported via
516 surface streams infiltrates quickly towards deeper layers, possibly contributing to groundwater
517 recharge, though the volumes of this water is likely low. Hence, we suggest that the thermal
518 regime of the ground – which determines the distribution and seasonal evolution of the

18
519 permafrost and the active layer – plays a preponderant role in the characteristics and evolution of
520 surface hydrology and recharge on the OSM and similar high altitude massifs in the Dry Andes.
521 The permafrost zone is expected to drift towards higher elevations and the active layer
522 is expected to become thicker at a given altitude in the future, due to the predicted warming.
523 Thus infiltration towards deeper layers will become less restricted, and the storage capacity of
524 the active layer will increase. Unless significant surplus meltwater becomes available from the
525 region above 6000 m a.s.l. – e.g. from ground ice or buried and surface snow patches – this
526 process will fundamentally reduce the seasonal presence of surface water. Thus, future climatic
527 changes could significantly change recharge patterns through their effect on permafrost. We
528 suggest that this interplay should be considered when evaluating future hydrological pathways
529 and water security in the region.
530

531

19
532 References

533 Abramov, A., Gruber, S., Gilichinsky, D. (2008). Mountain permafrost on active volcanoes: field
534 data and statistical mapping, Klyuchevskaya volc DGA-Arcadis, 2015ano group,
535 Kamchatka, Russia. Permafrost and Periglacial Processes 19(3). 261-277.

536 Arenson, L. U., Harrington, J. S., Koenig, C. E., & Wainstein, P. A. (2022). Mountain
537 permafrost hydrology—a practical review following studies from the andes. Geosciences,
538 12(2), 48.Aszalós, J. M., Krett, G., Anda, D., Márialigeti, K., Nagy, B., & Borsodi, A. K.
539 (2016). Diversity of extremophilic bacteria in the sediment of high-altitude lakes located in
540 the mountain desert of Ojos del Salado volcano, Dry-Andes. Extremophiles, 20(5), 603-
541 620.

542 Baraer, M., Mark, B. G., McKenzie, J. M., Condom, T., Bury, J., Huh, K.-I., Portocarrero, C.,
543 Gómez, J., & Rathay, S. (2012). Glacier recession and water resources in Peru’s Cordillera
544 Blanca. Journal of Glaciology, 58(207), 134–150.

545 Barnett, T. P., Adam, J. C., & Lettenmaier, D. P. (2005). Potential impacts of a warming climate
546 on water availability in snow-dominated regions. Nature, 438(7066), 303–309.

547 Barry, R., & Gan, T. Y. (2011). The global cryosphere: past, present and future. Cambridge
548 University Press.

549 Basaldúa, A., Alcaraz, E., Quiroz-Londoño, M., Dapeña, C., Ibarra, E., Vélez-Agudelo, C.,
550 Copia, L., & Martínez, D. (2022). Reconstruction of the record of tritium in precipitation in the
551 temperate zone of South America. Hydrological Processes, 36(9), e14691.
552 https://doi.org/10.1002/hyp.14691

553 Bense, V.F., Ferguson, G., Kooi, H. (2009). Evolution of shallow groundwater flow systems in
554 areas of degrading permafrost. Geophysical Research Letters 36 (22).
555 https://doi.org/10.1029/2009GL039225

556 Bockheim, J. G. (2015). Global distribution of cryosols with mountain permafrost: an overview.
557 Permafrost and Periglacial Processes, 26(1), 1–12.

558 Boisier, J. P., Alvarez-Garretón, C., Cordero, R. R., Damiani, A., Gallardo, L., Garreaud, R. D.,
559 Lambert, F., Ramallo, C., Rojas, M., & Rondanelli, R. (2018). Anthropogenic drying in
560 central-southern Chile evidenced by long-term observations and climate model simulations.
561 Elemementa Sci Anth, 6(1). 74. https://doi.org/10.1525/elementa.328

562 Boschetti, T., Cifuentes, J., Iacumin, P., & Selmo, E. (2019). Local meteoric water line of
563 Northern Chile (18 S–30 S): An application of error-in-variables regression to the oxygen
564 and hydrogen stable isotope ratio of precipitation. Water 11(4), 791.

565 Bozkurt, D., Rojas, M., Boisier, J. P., & Valdivieso, J. (2018). Projected hydroclimate changes
566 over Andean basins in central Chile from downscaled CMIP5 models under the low and
567 high emission scenarios. Climatic Change, 150(3–4), 131–147.
568 https://doi.org/10.1007/s10584-018-2246-7

20
569 Brenning A, Gruber S, Hoelzle, M. (2005). Sampling and statistical analyses of BTS
570 measurements. Permafrost and Periglacial Processes, 16, 383-393.
571 https://doi.org/10.1002/ppp.541

572 Bresciani, E., Cranswick, R. H., Banks, E. W., Batlle-Aguilar, J., Cook, P. G., & Batelaan, O.
573 (2018). Using hydraulic head, chloride and electrical conductivity data to distinguish
574 between mountain-front and mountain-block recharge to basin aquifers. Hydrology and
575 Earth System Sciences, 22(2), 1629–1648.

576 Carroll, R. W. H., Deems, J. S., Niswonger, R., Schumer, R., & Williams, K. H. (2019). The
577 importance of interflow to groundwater recharge in a snowmelt‐dominated headwater
578 basin. Geophysical Research Letters, 46(11), 5899–5908.

579 Cheng, G. D., & Jin, H. J. (2013). Groundwater in the permafrost regions on the Qinghai-Tibet
580 Plateau and it changes. Hydrogeology & Engineering Geology, 40(1), 1–10.

581 Cochand, M., Molson, J., Barth, J. A. C., van Geldern, R., Lemieux, J. M., Fortier, R., &
582 Therrien, R. (2020). Rapid groundwater recharge dynamics determined from
583 hydrogeochemical and isotope data in a small permafrost watershed near Umiujaq
584 (Nunavik, Canada). Hydrogeology Journal, 28(3). https://doi.org/10.1007/s10040-020-
585 02109-x

586 Cooper, R., Hodgkins, R., Wadham, J., & Tranter, M. (2011). The hydrology of the proglacial
587 zone of a high-Arctic glacier (Finsterwalderbreen, Svalbard): Sub-surface water fluxes and
588 complete water budget. Journal of Hydrology, 406(1–2), 88–96.

589 Coplen, T. B. (1994). Reporting of stable hydrogen, carbon, and oxygen isotopic abundances
590 (technical report). Pure and applied chemistry, 66(2), 273-276.

591 Craig, H., Gordon, L. I., & Horibe, Y. (1963). Isotopic exchange effects in the evaporation of
592 water: 1. Low‐temperature experimental results. Journal of Geophysical Research, 68(17),
593 5079-5087.

594 Cuthbert, M. O., Gleeson, T., Moosdorf, N., Befus, K. M., Schneider, A., Hartmann, J., &
595 Lehner, B. (2019). Global patterns and dynamics of climate–groundwater interactions.
596 Nature Climate Change, 9(2), 137–141.

597 Czuppon, G., Bottyán, E., Kristóf, E., Weidinger, T., Haszpra, L., & Kármán, K. (2021). Stable
598 isotope data of daily precipitation during the period of 2013–2017 from K-puszta (regional
599 background monitoring station), Hungary. Data in Brief, 36, 106962.

600 DGA-Arcadis. (2015). Diagnóstico de disponibilidad hídrica en cuencas Alto-Andinas de la

601 Región de Atacama, fase 2.

602 Drusch, M., Del Bello, U., Carlier, S., Colin, O., Fernandez, V., Gascon, F., Hoersch, B., Isola,
603 C., Laberinti, P., & Martimort, P. (2012). Sentinel-2: ESA's optical high-resolution mission
604 for GMES operational services. Remote sensing of Environment, 120, 25-36.

21
605 Enguita, P. (2016). Diagnóstico de disponibilidad hídrica en cuencas Alto-Andinas de la Región
606 de Atacama, fase 3, AMPHOS Chile.

607 Eppelbaum, L. Kutasov, I., Pilchin, A. (2014). Applied Geothermics. Lecture Notes in Earth
608 System Sciences, Springer-Verlag Berlin Heidelberg.

609 Evans, S. G., & Ge, S. (2017). Contrasting hydrogeologic responses to warming in permafrost
610 and seasonally frozen ground hillslopes. Geophysical Research Letters, 44(4).
611 https://doi.org/10.1002/2016GL072009

612 Gat, J.R. (1995). Stable Isotopes of Fresh and Saline Lakes. In: Lerman, A., Imboden, D.M., Gat,
613 J.R. (eds) Physics and Chemistry of Lakes. Springer, Berlin, Heidelberg.
614 https://doi.org/10.1007/978-3-642-85132-2_5

615 Gibson, J. J., Birks, S. J., & Yi, Y. (2016). Higher tritium concentrations measured in permafrost
616 thaw lakes in northern Alberta. Hydrological Processes, 30(2), 245-249.

617 Gjorup, D.F., Francelino, M.R., Michel, R.F.M., Senra, E.O., Schaefer, C.E.G.R. (2019).
618 Pedoclimate monitoring in the periglacial high mountain soils of the Atacama Desert,
619 northern Chile. Permafrost and Periglacial Processes 30. pp. 310-329.

620 Goyanes, G., Vieira, G., Caselli, A., Cardoso, M., Marmy, A., Santos, F., Bernardo, I., & Hauck,
621 C. (2014). Local influences of geothermal anomalies on permafrost distribution in an active
622 volcanic island (Deception Island, Antarctica). Geomorphology, 225, 57–68.

623 Hammond, J. C., Harpold, A. A., Weiss, S., & Kampf, S. K. (2019). Partitioning snowmelt and
624 rainfall in the critical zone: effects of climate type and soil properties. Hydrol. Earth Syst.
625 Sci., 23(9), 3553–3570. https://doi.org/10.5194/hess-23-3553-2019

626 Hiyama, T., Asai, K., Kolesnikov, A. B., Gagarin, L. A., & Shepelev, V. V. (2013). Estimation
627 of the residence time of permafrost groundwater in the middle of the Lena River basin,
628 eastern Siberia. Environmental Research Letters, 8(3), 35040.

629 Houston, J., Hartley, A. J. (2003). The central Andean west‐slope rainshadow and its potential
630 contribution to the origin of hyper‐aridity in the Atacama Desert. International Journal of
631 Climatology 23(12). pp. 1453‐1464.

632 Houston, J. (2007). Recharge to groundwater in the Turi Basin, northern Chile: An evaluation
633 based on tritium and chloride mass balance techniques. Journal of Hydrology 334(3-4), 534-
634 544.

635 Ishikawa, M. (2003). Thermal regimes at the snow‐ ground interface and their implications for
636 permafrost investigation. Geomorphology 52(1‐ 2), 105‐ 120. https://doi.org/10.1016/S0169-
637 555X(02)00251-9

22
638 Jordan, T. E., Herrera, C., Godfrey, L. V., Colucci, S. J., Gamboa, C., Urrutia, J., ... & Paul, J. F.
639 (2019). Isotopic characteristics and paleoclimate implications of the extreme precipitation
640 event of March 2015 in northern Chile. Andean Geology 46(1), 1-31.

641 Kaplan, G., & Avdan, U. (2017). Object-based water body extraction model using Sentinel-2
642 satellite imagery. European Journal of Remote Sensing, 50(1), 137-143.

643 Kereszturi, Á., Aszalos, J.M., Heiling, Zs., Igneczi, Á., Kapui, Zs., Király, Cs., Leél-Őssy, Sz.,
644 Szalai, Z., Nemerkényi Zs., Pál, B., Skultéti, A., & Nagy, B. (2022). Wind-snow
645 interactions at the Ojos del Salado region as a potential Mars analogue site in the Altiplano -
646 Atacama desert region. Icarus 378, 15.

647 Kinnard, C., Ginot, P., Surazakov, A., MacDonell, S., Nicholson, L., Patris, N., ... & Squeo, F.
648 A. (2020). Mass balance and climate history of a high-altitude glacier, Desert Andes of
649 Chile. Frontiers in Earth Science, 8, 40.

650 Klene, A. E., Nelson, F. E., & Shiklomanov, N. I. (2001). The n-factor as a tool in
651 geocryological mapping: Seasonal thaw in the Kuparuk River basin, Alaska. Physical
652 Geography, 22(6), 449–466.

653 Kull, C., Grosjean, M., Veit, H. (2002). Modelling modern and Late Pleistocene glacio‐
654 climatological conditions in the North Chilean Andes (29‐30°S). Climate Change. 52(2)
655 359‐381.

656 LeCun, Y., Bengio, Y., Hinton, G. (2015). Deep Learning. Nature, 521(7553), 436–444.

657 Levy, J. S., Fountain, A. G., Gooseff, M. N., Welch, K. A., & Lyons, W. B. (2011). Water tracks
658 and permafrost in Taylor Valley, Antarctica: Extensive and shallow groundwater
659 connectivity in a cold desert ecosystem. Bulletin, 123(11–12), 2295–2311.

660 Lobos-Roco, F., Hartogensis, O., Vilà-Guerau de Arellano, J., de la Fuente, A., Muñoz, R.,
661 Rutllant, J. and Suárez, F. (2021). Local evaporation controlled by regional atmospheric
662 circulation in the Altiplano of the Atacama Desert. Atmos. Chem. Phys., 21 (11).
663 doi:10.5194/acp-21-9125-2021

664 Markovich, K. H., Manning, A. H., Condon, L. E., & McIntosh, J. C. (2019). Mountain‐block
665 recharge: A review of current understanding. Water Resources Research, 55(11), 8278–
666 8304.

667 Masiokas MH, Rabatel A, Rivera A, Ruiz L, Pitte P, Ceballos JL, Barcaza G, Soruco A, Bown F,
668 Berthier E, Dussaillant I and MacDonell S (2020) A Review of the Current State and Recent
669 Changes of the Andean Cryosphere. Front. Earth Sci. 8:99. doi: 10.3389/feart.2020.00099

670 Masson-Delmotte, V., P. Zhai, A. Pirani, S. L. Connors, C. Péan, S. Berger, N. Caud, Y. Chen,
671 L. Goldfarb, M. I. Gomis, M., & Huang, K. Leitzell, E. Lonnoy, J. B. R. Matthews, T. K.
672 Maycock, T. Waterfield, O. Yelekçi, R. Y. and B. Z. (2021). Summary for Policymakers.

23
673 in: Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to
674 the Sixth Assessment Report.

675 McFeeters, S. K. (1996). The use of the Normalized Difference Water Index (NDWI) in the
676 delineation of open water features. International journal of remote sensing, 17(7), 1425-
677 1432.

678 Mena, G., Yoshikawa, K., Schorghofer, N.,Pastén, C., Ochoa, F.A., Yoshii, Y., Doi, M., Miyata,
679 T., Takahashi, H., Casassa, G., & Sone, T. (2021). Freeze–thaw cycles and snow impact at
680 arid permafrost region in Chajnantor Volcano, Atacama, northern Chile. Arctic, Antarctic
681 and Alpine Research 53 (1). 60-66. https://doi.org/10.1080/15230430.2021.1878739

682 Nagy, B., Ignéczi, Á., Kovács, J., Szalai, Z., and Mari, L. (2019). Shallow ground temperature
683 measurements on the highest volcano on Earth, Mt. Ojos del Salado, Arid Andes, Chile.
684 Permafrost and Periglacial Processes 30. 3-18.

685 Nagy, B., Kovács, J., Ignéczi, Á., Beleznai, S., Mari, L., Kereszturi, Á., & Szalai, Z. (2020). The
686 thermal behavior of ice-bearing ground: the highest cold, dry desert on Earth as an analog
687 for conditions on Mars, at Ojos del Salado, Puna de Atacama-Altiplano region.
688 Astrobiology, 20(6), 701–722.

689 Nakayama, T., Wang, Q., & Okadera, T. (2021). Evaluation of spatio-temporal variations in
690 water availability using a process-based eco-hydrology model in arid and semi-arid regions
691 of Mongolia. Ecological Modelling, 440, 109404.

692 Nelson, F. E., & Outcalt, S. I. (1987). A computational method for prediction and regionalization
693 of permafrost. Arctic and Alpine Research, 19(3), 279–288.

694 Nicholson, S. (1998). Desert hydrology. In Encyclopedia of Hydrology and Water Resources
695 (pp. 176–183). Springer.

696 Obu, J., Westermann, S., Kaab, A., & Bartsch, A. (2019). Ground Temperature Map, 2000-2016,
697 Andes, New Zealand and East African Plateau Permafrost. In University of Oslo.
698 PANGAEA. https://doi.org/10.1594/PANGAEA.905512

699 Oliva, M., & Fritz, M. (2018). Permafrost degradation on a warmer Earth: Challenges and
700 perspectives. Current Opinion in Environmental Science & Health, 5, 14–18.

701 Outcalt, S.I., Nelson, F.E., Hinkel, K.M. (1990) The zero-curtain effect: Heat and mass transfer
702 across an isothermal region in freezing soil. Water Resources Research, 26(7), 1509-1516.
703 https://doi.org/10.1029/WR026i007p01509

704 Pena-Regueiro, J., Sebastia-Frasquet, M.-T., Estornell, J., & Aguilar-Maldonado, J. A. (2020).
705 Sentinel-2 application to the surface characterization of small water bodies in wetlands.
706 Water, 12(5), 1487.

24
707 Palcsu, L., Major, Z., Köllő, Z. and Papp, L. (2010), Using an ultrapure 4He spike in tritium
708 measurements of environmental water samples by the 3He-ingrowth method. Rapid
709 Commun. Mass Spectrom., 24: 698-704. https://doi.org/10.1002/rcm.4431

710 Rozanski, K., Gonfiantini, R., and Araguas-Araguas, L.: Tritium in the global atmosphere:
711 Distribution patterns and recent trends, J. Phys. G. Nucl. Partic., 17, S523–S536, 1991.

712 Ruiz Pereira, S., Marquardt, C., Beriain, E., & Lambert, F. (2021). Permafrost evolution in a
713 mountain catchment near Santiago de Chile. Journal of South American Earth Sciences,
714 109, 103293.

715 Saavedra, F. A., Kampf, S. K., Fassnacht, S. R., & Sibold, J. S. (2018). Changes in Andes snow
716 cover from MODIS data, 2000–2016. The Cryosphere, 12(3), 1027–1046.

717 Sjöberg, Y., Jan, A., Painter, S. L., Coon, E. T., Carey, M. P., O'Donnell, J. A., & Koch, J. C.
718 (2021). Permafrost promotes shallow groundwater flow and warmer headwater streams.
719 Water Resources Research, 57(2), e2020WR027463.

720 Somers, L. D., & McKenzie, J. M. (2020). A review of groundwater in high mountain
721 environments. Wiley Interdisciplinary Reviews: Water, 7(6), e1475.

722 Stewart, M. K., Morgenstern, U., Gusyev, M. A., & Małoszewski, P. (2017). Aggregation effects
723 on tritium-based mean transit times and young water fractions in spatially heterogeneous
724 catchments and groundwater systems. Hydrology and Earth System Sciences, 21(9), 4615-
725 4627.

726 Stimson, J., Rudolph, D., Frape, S., & Drimmie, R. (1996). Interpretation of groundwater flow
727 patterns through a reconstruction of the tritium precipitation record in the Cochabamba
728 Valley, Bolivia. Journal of hydrology, 180(1-4), 155-172.

729 Terzer-Wassmuth, S., Araguás-Araguás, L. J., Copia, L., & Wassenaar, L. I. (2022). High spatial
730 resolution prediction of tritium (3H) in contemporary global precipitation. Scientific
731 Reports, 12:10271 https://doi.org/10.1038/s41598-022-14227-5

732 Vargas Paysen, J. (2013). Diagnóstico de disponibilidad hídrica en cuencas Alto-Andinas de la

733 Región de Atacama, Estudios Dirección General de Aguas, Chile.

734 Viviroli, D., & Weingartner, R. (2008). “Water towers”—A global view of the hydrological
735 importance of mountains. In Mountains: Sources of water, sources of knowledge (pp. 15–
736 20). Springer.

737 Wan, C., Gibson, J. J., Shen, S., Yi, Y., Yi, P., & Yu, Z. (2019). Using stable isotopes paired
738 with tritium analysis to assess thermokarst lake water balances in the Source Area of the
739 Yellow River, northeastern Qinghai-Tibet Plateau, China. Science of The Total
740 Environment, 689, 1276-1292.

741 Wang, G., Hu, H., & Li, T. (2009). The influence of freeze–thaw cycles of active soil layer on
742 surface runoff in a permafrost watershed. Journal of Hydrology, 375(3-4), 438-449.

25
743 Woodside, W., & Messmer, J.H. (1961). Thermal Conductivity of Porous Media. Journal of
744 Applied Physics 32., 1688. https://doi.org/10.1063/1.1728419

745 Wu, W.-Y., Lo, M.-H., Wada, Y., Famiglietti, J. S., Reager, J. T., Yeh, P. J.-F., Ducharne, A., &
746 Yang, Z.-L. (2020). Divergent effects of climate change on future groundwater availability
747 in key mid-latitude aquifers. Nature Communications, 11(1), 1–9.

748 Yang, X., Zhao, S., Qin, X., Zhao, N., & Liang, L. (2017). Mapping of urban surface water
749 bodies from Sentinel-2 MSI imagery at 10 m resolution via NDWI-based image sharpening.
750 Remote Sensing, 9(6), 596.

751 Ruder, S. (2016). An overview of gradient descent optimization algorithms. arXiv, 2016;
752 arXiv:1609.04747.

753 van Everdingen, R. (2005). Multi‐ language glossary of permafrost and related ground‐ ice
754 terms. International Permafrost Association

755

26
756 Figure captions:

757 Figure 1. Location of the Ojos del Salado Massif (OSM) on the Puna de Atacama. Stars indicate
758 the sites of the monitoring system.
759
760 Figure 2. The Ojos del Salado Massif, viewed from the North. Major peaks within the OSM, our
761 monitoring sites (stars), and key meltwater flow directions are indicated.
762
763 Figure 3. Depth dependent fraction volume ratios of air, rock, and water/ice. These values are
764 obtained using gradient descent optimisation.
765
766 Figure 4. An inset satellite image showing lavalobes, lake basins, and surface streams on the
767 northern slope of the Ojos del Salado summit; UL: Upper Lake, LL: Lower Lake
768
769 Figure 5. (a) Measured and modelled temperature at the Tejos Camp, 5830 m a.s.l., (b) and at
770 the Atacama Camp, 5260 m a.s.l. Phase change in the ground is graded from 0 (completely
771 liquid) to 1 (completely solid).
772
773 Figure 6. (a) Generalised presence of the two regularly occurring lakes (UL, LL) in the 5890-
774 6150 m zone (blue: open water, white: lake-ice) between 02/Aug/2017 and 28/Feb/2020. (b)
775 Snow coverage within the catchment area of the lakes. (c) Daily mean air temperatures (2 m
776 above the surface) measured at 6000 m a.s.l. near the lake sites. (d) Daily mean ground
777 temperatures (blue: -10 cm; orange: -35 cm; grey: -60 cm) measured at the nearby Tejos Camp at
778 5830 m a.s.l. (e) Dates when field observations and photos are available for the UL; see panel
779 (h). (f) The binary mask indicates the modelled evolution of the active layer at the nearby Tejos
780 Camp (Figure 3). (h) Photographs showing the site of the Upper Lake (UL) on different dates.
781
782 Figure 7. The conceptual model of surface hydrology on the OSM.
783
784 Figure 8. Isotope hydrological characteristics of surface waters of the OSM. a) Stable hydrogen
785 and oxygen isotopes and b) 3H activities of the samples collected from fresh snow and the range
786 for stream waters (blue) and lake waters (orange) in 2018 and 2020. The Northern Chile
787 Meteoric Waterline (N Chile MWL, Boschetti et al., 2019) and the empirical evaporation line
788 fitted to the lake water data are shown with solid and dashed lines, respectively. The regional
789 estimates of mean annual 3H in precipitation (Basaldúa et al., 2022) is shown as a thin black
790 curve. The decay corrected activities calculated for 2018 (light gray) and 2020 (dark grey) must
791 be considered as references when assessing 3H activities of our samples.
792

27
Highlights (3 to 5 bullet points (maximum 85 characters including
spaces per bullet point)

Surface hydrology on the highest volcano of the high Dry Andes, the Ojos
del Salado; interannual fluctuations and moisture sources

Highlights:

- Surface hydrology is influenced by seasonal ground thawing at high elevations

- Surface waters originate from meteoric sources

- The contribution of pre-1960s precipitation to contemporary surface waters is minor

Abstract Click here to
access/download;Abstract;Ojos_hydro_abstract_v3.docx

Surface hydrology on the highest volcano of the high Dry Andes, the Ojos
del Salado; interannual fluctuations and moisture sources
Abstract
The Atacama Plateau in the Central Andes (28-22°S) is characterised by a dry and cold
periglacial tundra due to the high altitude, low precipitation, and high evaporation. Endogenous
freshwater sources – e.g.: seasonal streams and lakes, subsurface reservoirs, surface snow/ice
patches – are available, though they are highly sensitive to climatic changes. The near surface
hydrological network is highly modified by the distribution and seasonal evolution of perennial
frozen ground, i.e. permafrost, which is also expected to change in the future. The interplay
between permafrost and hydrology, especially in relation to future climate change, is poorly
explored. To address this issue, we carry out long-term ground temperature measurement and
modelling, snow coverage survey, tritium- and stable isotope analysis of surface waters on the
Ojos del Salado Massif, which is representative of high altitude mountains on the Atacama
Plateau. According to our results, a highly transient surface hydrological network – lakes,
springs and streams – forms during each summer where permafrost is widespread and ground
thawing (i.e. active layer) is present (~4900-6500 m a.s.l.). In this system, the water is of
meteoric origin and relatively young (<10 years). The development of the network is strongly
influenced by the active layer, which plays a crucial role in storing, seeping, and discharging
groundwater. However, future permafrost degradation is expected to reduce the seasonal
presence of shallow water, and hence, modify groundwater recharge patterns.
Declaration of Interest Statement

Declaration of interests

☒The authors declare that they have no known competing financial interests or personal relationships
that could have appeared to influence the work reported in this paper.

☐The authors declare the following financial interests/personal relationships which may be considered
as potential competing interests:
Cover Letter

Resubmission of
“Surface hydrology on the highest volcano of the high Dry Andes, the Ojos del Salado;
interannual fluctuations and moisture sources” HYDROL52913

Dear Dr Igneczi,

I very much regret to have to tell you that publication entitled, "Surface hydrology on the
highest volcano of the high Dry Andes, the Ojos del Salado; interannual fluctuations and
moisture sources" (Dr Adam Igneczi) in our journal is not recommended.

We would, however, consider as a new submission for review a substantially revised version
of this paper that addresses all of the reviewers' comments. Should you choose to submit such
a revised manuscript please refer to the present manuscript number, provide a detailed point-
by-point reply to all of the reviewers' comments, and state how the revised manuscript
addresses these.

An explanation for this decision is given in the attached review reports (and
on https://www.editorialmanager.com/hydrol/). I hope that the comments contained therein
will be of use to you.

Thank you for your interest in our journal.

With kind regards,

Marco Borga
Editor
Journal of Hydrology

AE: Dear authors, I have finally received two reviews for your manuscript and can now now
formulate my recommendation.

Reviews are strikingly different, with one recommending a minor revision and the other one
recommending rejection. Nonetheless, both acknowledge the value of your work, particularly
as it focuses on a very remote and undocumented region of the world. For this reason, I am
returning the manuscript to you to give you an opportunity to revise and resubmit. I chose the
“reject & resubmit” option as this will give you time to work on your revision, with no
constraint by a deadline.

Please consider comments by both reviewers very carefully, as I will send your manuscript to
both of them in case you choose to resubmit.
REPLY
Dear Editor,

Thank your for the chance to resubmit our manuscript (original submission ID:
HYDROL52913). We have carefully considered the recommendations of both reviewers and
revised our manuscript accordingly. Attached you can find detailed responses to all the
comments provided by both reviewers, with references to changes in the manuscript.

Key changes to the manuscript:

 - In response to concerns raised by Reviewer 3, we have revised Section 3.1 to provide a
better outline of how shallow ground temperature measurements can unravel
permafrost conditions within a high altitude mountain environment (Reviewer 3). Most
prominently, we cover (among other issues) the limitations of this approach, how site
selection is crucial, and what steps were taken to address these factors. We also provide
detailed responses to the relevant comments raised by the reviewer. We feature more
prominently (both in the manuscript and our responses) that further details can be found
in our previous paper (Nagy et al., 2019) which exclusively deals with such issues.
- In response to comments from both reviewers, we have extensively rewritten the section
that describes our numerical modelling scheme (Section 3.2) to provide more clarity.
Most importantly we provide a more detailed explanation about the new elements of
the modelling scheme presented in this paper, in comparison to the previous version of
this model, presented in Nagy et al. (2020). Chief among these is the new automated
approach (based on machine learning) that aides the estimation of volume ratio
fractions of air, rock and water.
- In response to comments by Reviewer 3, we extensively rewritten sections pertaining
to tritium measurements and the interpretation of these results to provide more clarity,
especially about the need to decay-correct reconstructions of historical tritium content
of precipitation (which we believed caused confusions). We also carried out additional
data analysis as a new study about the historical tritium content of precipitation within
the Ojos del Salado region has came to our attention (Basaldúa et al., 2022). This dataset
is now used as reference when evaluating our results. Figure 8 has been revised
accordingly.
- We believe that figure captions were not included in our previous submission (due to
our oversight when using the online submission system), we now ensure that these are
included.
- The highlights have been completely revised.

We hope that both you and the reviewers will find these changes satisfactory and will
reconsider our manuscript for publication.

Yours sincerely,

Dr Adam Igneczi (corresponding author, on behalf of the authors)

Bristol Glaciology Centre
School of Geographical Sciences
University of Bristol
UK
email: [email protected]
Revision Notes Reviewer2

Reviewer #2:

The interaction between permafrost and hydrology is very complex as a result of the special
thermal and physical properties under the influence of ground ice. In arid regions with little
precipitation, the permafrost tends to act as an important freshwater reservoir and tower for
supplying water resources of the industry and local residents. Many efforts had been made to
study the influences of permafrost dynamics on the hydrological cycles over the past. While
the interplay between permafrost and hydrology in Atacama Plateau in the high and dry Central
Andes (28-22°S) had not been sufficiently studied, which is indeed worthy of study as it is
characterized by high altitude, low precipitation, and high evaporation. Endogenous freshwater
sources, e.g.: seasonal streams and lakes, subsurface reservoirs, surface snow/ice patches are
closely related to the existence and dynamics of permafrost. The near
surface hydrological network is highly modified by the distribution and seasonal evolution of
perennially frozen ground, i.e., permafrost, which is also expected to change in the future. In
this aspect, this study is really interesting and meaningful. In general, this manuscript is well-
organized. In this study, long-term observations (from March 2012 to January 2020) and
numerical simulations of the hydrothermal processes in near-surface shallow soil layers (0~60
cm) in relation to the surface hydrological networks (lakes, streams, and ponds) had been
carried out for permafrost in the highest volcano across the world, the Ojos del Salado in the
Andes. The relationship between seasonal thawing and water storage had been
comprehensively analyzed. Systematic sampling and field investigations of the isotopes had
been collected to source the water and to analyze the spatiotemporal variations of the
permafrost hydrology. The satellite survey of snow cover and lake presence was also
investigated.
REPLY
Thank you for this favourable summary of our work.

1. Figure 1 and 2, a legend showing the meanings of different symbols should be presented;
REPLY
We made an error during the submission process, which caused figure captions to go missing.
We have now rectified this and provide proper captions which also contain explanations of the
symbols (in case of vary simple symbology, such as on Figure 1). In addition, we provide
graphical legend for Figure 2 as it has more complex symbology.

2. Line 43, delete "Boissier et al., 2018;";

REPLY
Thank you for noticing this mistake in our citation. We in fact wanted to cite Buzkurt et al.
(2018) here. This has been corrected.

3. Why do you set a constant temperature at the lower boundary of the model at the depth of 5
m? As far as I know, the temperature kept stably with not too much change usually appears at
the depth of zero annual amplitude of ground temperature. While temperature at the depth of 5
m, sometimes even experiences freezing and thawing cycles within a year. Moreover, could
you please provide the value of the zero annual amplitude of ground temperature?
REPLY
- Thank you for raising this issue. While it is possible to introduce an observed temporally
varying temperature signal as the lower boundary condition of the model (i.e. at the
maximum depth of the model domain), measurements deeper than 60 cm are
unavailable for this region. Using a time-varying signal at the lower boundary would
also significantly raise computational complexity (given that the model is optimised to
constrain other unknown parameters). If we were to chose -60 cm as the lower boundary
of our domain, we would be restricted to just one independent data stream for model
evaluation (i.e. at -35 cm), which would limit our ability to optimize model performance
for difference depth gradients of water/air/rock volume ratios.
- In order to overcome this issue we carried out multiple experiments that assumed zero
annual amplitude of the ground temperature, i.e. constant temperature, at different
depths (from 1 to 10 m, with 1 m interval). This constant ground temperature was
chosen to coincide with the interannual average ground temperature at -60 cm (as these
logger provide the closest measured approximation), i.e. -2.5 °C. Using our independent
data streams, we evaluated these experiments, and found that the 5 m experiment
provided the best match with our temperature measurements (at -60 and -35 cm).

We added the following explanation/clarification about this procedure (and the value of
the zero annual amplitude ground temperature) in Section 3.2
“…(3) constant -2.5°C temperature at the bottom of the model domain, 5 m from the
surface; the temperature was based on the observed annual mean ground temperatures
at -60 cm, while the depth was based on the performance of multiple model runs with
different domain depths (1-10 m, with 1 m interval)…;”

- Although we acknowledge that this is an approximation, it is also important to note that

our aim was not to estimate ground temperature evolution at depths larger than our
measurements, or to estimate the complete thickness of the active layer. Instead we
aimed to estimate water content and freezing/thawing cycles in the uppermost layers,
i.e. > -60 cm (where the model is properly constrained by observations). As we assumed
no geothermal heat flux, all the heat flux is coming from the surface. The uppermost
layers are also constrained by observations (which we mimic with out experiments),
thus the domain depth will mostly influence the attenuation rate (with depth) of the
annual ground temperature amplitude (below our temperature loggers), which we do
not aim to resolve as it would not realistic with the data we have.

We provide additional explanations in Section 4.1 to further convey the aim of the
modelling experiments and their limitations:
“…As we have no measurements from depths larger than -60 cm which could be used
to constrain our model, we focus on the interpretation of model outputs for the
uppermost 60 cm of the ground. This almost completely covers the actual thickness of
the active layer at the Tejos Camp, while at the Atacama Camp the coverage is partial
(Nagy et al., 2019) (Figure 5)…”

“…The thickness of the active layer exhibits a notable (>20%) interannual oscillation,
reaching only about 55-60 cm in La Niña years (e.g. 2012/13, 2016/17, 2017/18), and
>70 cm in El Niño years (Figure 5). Although, confidence levels are lower in the latter
case as the active layer extends deeper than our model constraints, the scale of this
extrapolation is small (< 20 cm). Hence, we propose that the ground thermal model
can provide a full picture about the evolution of the active layer at the Tejos Camp…”
“…As the modelled thickness of the active layer extends well beyond the model
constrains during each summer (Figure 5), we cannot estimate the actual thickness of
the active layer at the Atacama Camp with sufficient confidence...”
Revision Notes Reviewer3

Reviewer #3:

The area is a very remote and interesting place on earth. However, none of the "highlights" are
proven by the results of the field work.
REPLY
- Thank you for bringing this issue to our attention. We acknowledge that the highlights
are too generic and do not reflect the core findings accurately. We made several changes
to better reflect the outcomes of our work.
- We have completely removed original Highlight 1): “Transient surface hydrology forms
where permafrost and active layer is widespread” as we realise a more comprehensive
survey is required to confidently make this statement. Also we acknowledge that this
statement do not take into account surface hydrology at lower elevations (e.g. large
saline lakes on the Puna de Atacama).
- Instead we realign focus on the outcomes of the isotope survey, which we hope will
prove more convincing after the changes made in response to other comments. Original
Highlight 3 “Surface water, sourced from snow and ground ice, is meteoric and young
(< 20 yr)” was broken into two and reworded to provide a more accurate picture of the
outcomes:
o “Surface waters originate from meteoric sources”
o “The contribution of pre-1960s precipitation to contemporary surface waters is
minor“
- We also reworded original highlight (2) “Development of surface hydrology is strongly
influenced by seasonal ground thawing” to make it more specific to our area of interest,
and also to be more permissive of the influence of other factors. We hope that the
reviewer will agree that there is sufficient evidence base to make this realigned
statement, especially when taking into account the temporal correlation between
meltwater pond presence (observed on the field and on satellite imagery) and seasonal
ground thawing.
o “Surface hydrology is influenced by seasonal ground thawing at high
elevations”

Near surface ground temperature data (fig.5 and maybe fig.6 not clear) didn't show the presence
of permafrost. Also, the distribution of permafrost was never investigated in the field, only by
speculating on 3 locations of the surface temperature data. Also this thermal model is hard to
understand how to calculate and estimate the thaw depth without any measurement of soil
properties, ice content, etc. Latent heat flux (text says many ice-rich layers that affect thawing
model) must affect a lot of this model.
REPLY
We agree with the reviewer that these issues need to be carefully considered, and are often in
the focus of debates, especially when studying permafrost in a remote region. We grouped our
responses around the key issues raised in this comment:

Lack of direct permafrost observation:

- Indeed, direct observations of permafrost (e.g. extensive borehole data with
measurements of interstitial ice content) were not made by us due to logistical and
financial limitations. However, the Dirección General de Aguas, Chile has deep
 borehole measurements and stratigraphic surveys from comparable environments in the
Chilean Andes that corroborate our findings:
https://snia.mop.gob.cl/repositoriodga/handle/20.500.13000/7391

Using ground temperatures to indirectly survey permafrost:

- Indirect methods, relying on air/surface/ground temperatures, are widely used to
ascertain ground thermal regimes and the associated type of permafrost presence (or
absence) (e.g. Ishikawa, 2003). These indirect methods are especially important where
establishing deep boreholes is not possible either due to logistical/financial limitations
or environmental concerns (e.g. extensively disturbing the ground in a protected
environment). Ground temperature records can also be used to infer ground ice
thawing/freezing due to the zero curtain effect (Outcalt et al., 1990; van Everdingen et
al., 2005).

- In agreement with the reviewer, the most straightforward way to claim permafrost
presence is for a temperature logger to continuously (i.e. at least for 2 years) measure
ground temperatures below 0°C. However, this is not a necessary pre-requisite to infer
permafrost presence, as the active layer might be too thick to place a logger directly in
the permafrost. In such circumstances, the thermal regime of the ground can be
investigated. As pointed out by the Multi-language Glossary of Permafrost and Related
Ground Ice Terms (van Everdingen et al., 2005), published by International Permafrost
Association (in term 553: thermal regime of the ground): “Permafrost exists if the mean
annual surface temperature is perennially below 0°C.”.
Near surface mean annual ground temperatures are between -0.4 and -0.7 at the
Atacama Camp and -3.3 and -3.7 at the Tejos Camp. Furthermore, ground temperatures
at the Tejos Camp, at 60 cm depth, exceed 0°C only for brief periods (Fig 5). Thus, as
it sits on the fringe of the definition, permafrost is likely to be restricted to preferential
locations at the Atacama Camp (i.e. discontinuous/sporadic permafrost) while at the
Tejos Camp it is expected to be widespread (continuous permafrost). Global statistical
modelling by Gruber et al. (2012) also corroborates these conclusions, according to
their calculations the permafrost zonation index is 0.54 at the Atacama Camp and 0.94
at the Tejos Camp.

- We acknowledge that these complicated issues need to be covered in more detail, thus
we provide clarification in Section 3.1 and point the readers directly to our previous
publication where these issues are discussed in detail:
“…Permafrost, ground with temperatures below 0°C for at least two consecutive years,
presence can be established directly from ground temperatures (van Everdingen et al.,
2005). However, where the permafrost table – i.e. the upper boundary surface of
permafrost – is below the deepest temperature logger more indirect methods are
necessary (e.g. Ishikawa, 2003). We build upon a detailed investigation of ground
thermal regimes – i.e. ground temperature distribution, heat flows, and their time-
dependence (van Everdingen et al., 2005) – on the Ojos del Salado that was carried out
by Nagy et al. (2019) for the period of 2012-2016. Conclusions presented in this study
provide the basis for determining the representative vertical profile of permafrost on
the OSM…”

Thermal model
- Thank you for raising this issue to our attention. We agree that the description of the
thermal model needs more details. Below is a summary of the changes we have made.

- We reworded the first sentence to clarify that the primary goal is to estimate the
changing ratio of liquid water and ice, along with the temperature of the regolith:
“A major shortcoming of any ground temperature monitoring network is the
incapability to directly obtain information about the time-dependent liquid water and
ice ratio of the regolith…”

- We add the following section to describe the type of thermal modelling that was
employed and to clarify that latent heat is included in the conductive heat transfer
equations:
“…Our model simulates temperature distribution and water/ice phase changes within
the active layer to clarify interannual properties and determine any existing trends. The
ground material is assumed to be a composite of liquid water, ice, rock, and air; only
the ratio of liquid water and ice is allowed to evolve with time. A continuous local
thermal equilibrium is assumed between the ice, water, rock and air components.
Hence, weighted average quantities – calculated by using the volume fractions of the
different components – are used to represent the density, heat capacity and heat
conduction coefficient of the composite ground material. The model is using a transient
conduction equation with the latent heat of fusion, thus includes the effects of ice- water
phase changes on the heat transfer and temperature variations (Nagy et al., 2020)…”

- We acknowledge that it is hard to follow how we arrived at fraction volume ratios for
the model. Thus we include an additional section that explains the previous approach
followed by Nagy et al., (2020) that was based on field measurements of porosity, rock
fraction density etc. This approach then manually tuned a time-independent water
content to provide the best match between modelled and observed temperatures.
In this paper, instead of manually tuning the water volume fraction (to achieve an
optimal match between modelled and observed temperature), we use the gradient
descent method (Ruder, 2016) to automatically tune fraction volume ratios (water, rock,
air). This method, similar to a deep learning, carries out many experiments by changing
certain parameters to arrive at an optimal solution (that matches with temperature
observations). We included the melting point of ice in the optimisation (i.e. the gradient
descent method was free to change it) as a control, to check whether the optimisation
employed any unphysical corrections.
We added the following text to Section 3.2:
“…The key issue with this model (Nagy et al., 2020) is the difficulty of empirically
determining the volume ratio of the fractions (i.e. air, rock, and water/ice) as functions
of depth which is crucial for accurately simulating phase-changes and the thermal
behaviour of the regolith. Nagy et al. (2019) analysed the physical properties of several
regolith samples taken from multiple depths at the Atacama and Tejos Camps. As the
samples exhibited weak spatial variations, an overall 60% (v/v) porosity and rock
fraction density (2700-3000 kg/m3) was used in the model of Nagy et al., (2020).
Thermal properties of the rock fraction was assumed to be similar of andesite-basalt
material (Eppelbaum et al., 2014). Water content (liquid and solid) was not measured
on the field by Nagy et al. (2019), thus a depth dependent but temporally fixed water
content was prescribed by manually tuning modelled temperatures to the measurements
(Nagy et al., 2020).This method is labour intensive and might not arrived at the best
solution due to the limited number of possible tuning experiments. Thus, we have
 upgraded the model to provide a more rigorous automated solution for determining the
appropriate depth functions of fraction volume ratios, that provide the best match
between modelled and measured temperatures...”

- Apart from these extra paragraphs we have also extensively edited Section 3.2 to better
explain our method.

References:
- Ishikawa M. Thermal regimes at the snow‐ ground interface and their implications for
permafrost investigation. Geomorphology. 2003;52(1‐ 2):105‐ 120.
- van Everdingen R. Multi‐ language glossary of permafrost and related ground‐ ice terms.
International Permafrost Association; 2005.
- Outcalt IS, Nelson EN, Hinkel MK. The zero‐ curtain effect: heat and mass transfer across
an isothermal region in freezing soil. Water Resour Res. 1990;27(7):1509‐ 1516.
- Eppelbaum, L. Kutasov, I. and Pilchin, A. (2014) Applied Geothermics, Lecture Notes in
Earth System Sciences, Springer-Verlag Berlin Heidelberg.

I can't find the figure captions, but I assume the monitoring sites are the same as Nagy et al.
2019. Monitoring near-surface temperature at one location is never representative of the entire
area. This is especially the case in subtropical high mountains, where albedo and soil properties
strongly influence permafrost distribution and active layer thickness.
REPLY
- We must apologize the reviewers because, unfortunately, the figure captions have been
“lost” from the system generated PDF and this mistake has been avoided our attention.

- Yes the monitoring sites have been in operation at the same locations since 2012, as
reported in Nagy et al. (2019). Altogether we have 6 sites on the Ojos del Salado and
its vicinity, all placed within representative vertical environmental zones of the
mountain. Though in this study we concentrate on mid-range elevations (between
~5200-6000 m a.s.l.), where we have two monitoring sites.

- Although we agree with the assessment that measurements at single points can be
biased by local characteristics (e.g. albedo, aspect, slope, topographical situation, etc.),
we think that these biases can be minimized by the careful selection of the monitoring
sites. We aimed to chose sites on relatively flat areas covered by debris, that is
representative of the wider vicinity (e.g. bright sand patches were avoided), as far as
possible from cliff faces, local topographical depressions – which are preferential sites
for snow accumulation – and perennial snow patches.
We have added a clarification about this to Section 3.1:
“The influence of local bias on our temperature measurements (e.g. Ishikawa, 2003;
Brenning et al; 2005) is suppressed by careful site selection, i.e. on relatively flat areas
covered by debris representative of the wider area, away from cliff faces, topographic
depressions, and perennial snow patches (Nagy et al., 2019).”

Isotope data are difficult to understand. I understand that lake water is enriched in stable
isotopes by evaporation (of course), but how to conclude to a recent (<20 yr) water source?
REPLY
This conclusion is based on tritium data. We hope that the revised and extended arguments and
the responses given below will be convincing.

And especially using the tritium data. what is fig 8b? the 1963 peak so low TU (should be 1000
times higher) and recent increasing TU in the figure? I have never seen this type of pattern and
TU numbers.
REPLY
- We must apologize the reviewers because, unfortunately, the figure captions have been
“lost” from the system generated PDF and this mistake has been avoided our attention.
A rewritten caption is now provided for Figure 8.

- The missed caption could have caused some confusion. Reference tritium levels
(represented by the blue line on Fig8b) are corrected for radioactive decay, hence older
peaks are not that pronounced.
- In the revised study we have replaced the previously used reference data with a recently
published reconstructed annual mean regional (i.e., corresponding to a region situating
closest to the OSM) precipitation 3H from the set of regional records reconstructed for
the temperate zone of South America (Basaldúa et al., 2022). We are convinced that this
new data could help to clarify the interpretation.
The following section is added to Section 3.4:
“Measured tritium activities are compared to decay corrected reconstructed tritium
activities of precipitation to help determine the origin of contemporary surface water
on the OSM. Reconstructed annual mean regional precipitation 3H activities have been
retrieved from a set of regional records for the temperate zone of South America
(Basaldúa et al., 2022). The region situated closest to the OSM was selected from this
dataset. These reconstructions, available from 1960 to 2022 (Basaldúa et al., 2022),
were corrected for radioactive decay using the estimated 3H half-life of 12.32 years
(Lucas and Unterweger, 2000) before comparisons with the measured tritium activities
of our samples. ”
- As the new reconstructions of precipitation tritium content by Basaldúa et al. (2022)
align with tritium measurements from fresh snow both in 2018 (undertaken in this
study) and 1990 (Grosjean et al., 1995), we could simplify both our Results and
Discussion. In the first version of this manuscript a our results were bias corrected as
only lower quality/less relevant reconstructions were available.
Section 4.3 was extensively rewritten.
Section 5.2 was also rewritten and restructured

References:
Basaldúa, A., Alcaraz, E., Quiroz-Londoño, M., Dapeña, C., Ibarra, E., Vélez-Agudelo, C.,
Copia, L., & Martínez, D. (2022). Reconstruction of the record of tritium in precipitation in the
temperate zone of South America. Hydrological Processes, 36(9), e14691.
https://doi.org/10.1002/hyp.14691

What is the source of the blue line in the figure? and do not understand these explanations in
the text?
REPLY
Both the text and the figure has been improved. We hope the revised version is convincing to
the reviewer. Also we believe that missing captions contributed to the confusion, these are now
included.

I feel that many interpretations are not based on field work and question how to understand
the hydrological process of the area. Manuscripts should include evidence to support the data.
REPLY
We hope that the additional explanations, clarifications and the revisions to the results and
some figures will help to alleviate the concerns expressed by the reviewer. Also we would like
to point out that besides field observations, the manuscript relies heavily on laboratory
analyses, numerical modelling, and remote sensing.
Figure 1

a) b) 69°W Laguna Verde N

CHILE 4350 m
Salar de
Maricunga
4550 m

3770 m

27°S

ARGENTINA
Cerro
El Muerto
6488 m
Nevado
Tres Cruces
6749 m Ojos del Salado
6893 m

Cerro Vicuna
6067 m

Volcán Copiapó
6052 m Cerro El Muerto
6488 m
5260 m

5830 m
5900 m (lake)

Laguna 6750 m 6000 m

Negro Francisco
6893 m

4200 m

0 200 km 0 10 km 0 5 km
Figure 2

monitoring site Ojos del Salado map of the northern lavalobes,

6893 m streams and lake basins
meltwater flow direction
Cerro El Muerto Cerro Vicunas Cerro Solo
6488 m 6067 m 6200 m
Cerro Fraile
6062 m
Cerro
Camp Tejos Barrancas Blancas
Cerro 5830 m 6119 m
Mulas Muertas Camp Atacama
5897 m 5260 m

4550 m
fossil alluvial plain (fan)
Puna plateau
Laguna Verde
CHILE
Figure 3
Figure 4
546000 547500

N
rocky lava
surfaces lava plateau
water sample

7003500
5830 m
LL
5890 m
5900 m
UL
5890 m
5830 m
Camp Tejos 5900 m

7003500
6000 m
6000 m

rocky lava
surfaces

7002000
debris slope
7002000

6750 m
6750 m
snow
glacier snow
glacier
ice
ice
Ojos del Salado
6893
6893 mm rocky lava perennial
Ojos del Salado surfaces
snow
snow

0 1 km

monitoring site seasonal stream lateral moraine

lake basin springs of the trimline
seepage water
remnant
lava lobe main crater of a rock glacier
solifluction lobe outwash plain
Figure 5
a) snow coverage b) snow coverage

-35 cm: daily temperature cycles (observed) -35 cm: daily temperature cycles (observed)

-35 cm: daily temperature cycles (modelled) -35 cm: daily temperature cycles (modelled)

-35 cm: phase change -35 cm: phase change

1 1
0 0

-35 cm: observed and modelled temperature (°C) -35 cm: observed and modelled temperature (°C)
10 15
5 10
0 5
-5 0
-10 -5
-15 -10

02.02.2012 02.02.2013 02.02.2014 02.02.2015 02.02.2016 02.02.2017 02.02.2018 02.02.2019 02.02.2020 11.02.2014 11.02.2015 11.02.2016 11.02.2017 11.02.2018 11.02.2019 11.02.2020

observed temperature modelled temperature observed temperature modelled temperature

snow coverage snow coverage

-60 cm: daily temperature cycles (observed) -60 cm: daily temperature cycles (observed)

-60 cm: daily temperature cycles (modelled) -60 cm: daily temperature cycles (modelled)

-60 cm: phase change -60 cm: phase change

1 1
0 0
-60 cm: observed and modelled temperature (°C)
2 -60 cm: observed and modelled temperature (°C)
8
0
6
-2 4
-4 2
-6 0
-2
-8
-4
-10 -6
-8
02.02.2012 02.02.2013 02.02.2014 02.02.2015 02.02.2016 02.02.2017 02.02.2018 02.02.2019 02.02.2020
11.02.2014 11.02.2015 11.02.2016 11.02.2017 11.02.2018 11.02.2019 11.02.2020

observed temperature modelled temperature

observed temperature modelled temperature
depth of active layer thawing depth of active layer thawing
-10 cm -10 cm

-100 cm -100 cm
Figure 6

h)
a) A
100
b) %
0
5
c) 0
-5
B
-10
℃ -15
-20
-25
-30
C
10
d) 5
0
℃ -5
-10
-15 D
-20
e) C D E
10
f)
40
cm
E
70

100

A
Figure 7

continuous permafrost discontinuous permafrost no permafrost

surface level
6893 m negligible active-layer
a.s.l. sublimation dry permafrost
active-layer bottom
recharge
permafrost bottom
runoff

active-layer
recharge
5200 m
a.s.l.

evaporation
recharge
runoff
hydrothermal
circulation
4300 m
a.s.l.
mountain block recharge mountain front recharge alluvial basin
Figure 8
Figure captions

Figure 1. Location of the Ojos del Salado Massif (OSM) on the Puna de Atacama. Stars
indicate the sites of the monitoring system.

Figure 2. The Ojos del Salado Massif, viewed from the North. Major peaks within the OSM,
our monitoring sites (stars), and key meltwater flow directions are indicated.

Figure 3. Depth dependent fraction volume ratios of air, rock, and water/ice. These values
are obtained using gradient descent optimisation.

Figure 4. An inset satellite image showing lavalobes, lake basins, and surface streams on the
northern slope of the Ojos del Salado summit; UL: Upper Lake, LL: Lower Lake

Figure 5. (a) Measured and modelled temperature at the Tejos Camp, 5830 m a.s.l., (b) and at
the Atacama Camp, 5260 m a.s.l. Phase change in the ground is graded from 0 (completely
liquid) to 1 (completely solid).

Figure 6. (a) Generalised presence of the two regularly occurring lakes (UL, LL) in the 5890-
6150 m zone (blue: open water, white: lake-ice) between 02/Aug/2017 and 28/Feb/2020. (b)
Snow coverage within the catchment area of the lakes. (c) Daily mean air temperatures (2 m
above the surface) measured at 6000 m a.s.l. near the lake sites. (d) Daily mean ground
temperatures (blue: -10 cm; orange: -35 cm; grey: -60 cm) measured at the nearby Tejos
Camp at 5830 m a.s.l. (e) Dates when field observations and photos are available for the UL;
see panel (h). (f) The binary mask indicates the modelled evolution of the active layer at the
nearby Tejos Camp (Figure 3). (h) Photographs showing the site of the Upper Lake (UL) on
different dates.

Figure 7. The conceptual model of surface hydrology on the OSM.

Figure 8. Isotope hydrological characteristics of surface waters of the OSM. a) Stable

hydrogen and oxygen isotopes and b) 3H activities of the samples collected from fresh snow
and the range for stream waters (blue) and lake waters (orange) in 2018 and 2020. The
Northern Chile Meteoric Waterline (N Chile MWL, Boschetti et al., 2019) and the empirical
evaporation line fitted to the lake water data are shown with solid and dashed lines,
respectively. The regional estimates of mean annual 3H in precipitation (Basaldúa et al.,
2022) is shown as a thin black curve. The decay corrected activities calculated for 2018 (light
gray) and 2020 (dark grey) must be considered as references when assessing 3H activities of
our samples.
Supplementary material for on-line publication only

Click here to access/download

Supplementary material for on-line publication only
Ojos_hydro_sup_v3.docx