Rapid decline of snow and ice in the tropical Andes – Impacts, uncertainties
and challenges ahead
Mathias Vuille
a,⁎
, Mark Carey
b
, Christian Huggel
c
, Wouter Buytaert
d
, Antoine Rabatel
e
,
Dean Jacobsen
f
, Alvaro Soruco
g
, Marcos Villacis
h
, Christian Yarleque
a
, Oliver Elison Timm
a
,
Thomas Condom
e
, Nadine Salzmann
c,i
, Jean-Emmanuel Sicart
e
a
Dept. of Atmospheric & Environmental Sciences, Univ. at Albany, Albany, NY, USA
b
Robert D. Clark Honors College, University of Oregon, Eugene, OR, USA
c
Dept. of Geography, Univ. of Zurich, Switzerland
d
Dept. of Civil and Environmental Engineering, Imperial College London, London, UK
e
Univ. Grenoble Alpes, CNRS, IRD, Institut des Géosciences de l'Environnement (IGE), Grenoble, France
f
Freshwater Biological Laboratory, Dept. of Biology, University of Copenhagen, Copenhagen, Denmark
g
Instituto de Investigaciones Geológicas y del Medio Ambiente, Universidad Mayor de San Andres, La Paz, Bolivia
h
Depto. de Ingenieria Civil y Ambiental, Escuela Politecnica Nacional, Quito, Ecuador
i
Dept. of Geosciences, Univ. of Fribourg, Switzerland
Glaciers in the tropical Andes have been retreating for the past several decades, leading to a temporary increase
in dry season water supply downstream. Projected future glacier shrinkage, however, will lead to a long-term
reduction in dry season river discharge from glacierized catchments. This glacier retreat is closely related to the
observed increase in high-elevation, surface air temperature in the region. Future projections using a simple
freezing level height- equilibrium-line altitude scaling approach suggest that glaciers in the inner tropics, such as
Antizana in Ecuador, may be most vulnerable to future warming while glaciers in the more arid outer tropics,
such as Zongo in Bolivia, may persist, albeit in a smaller size, throughout the 21st century regardless of emission
scenario. Nonetheless many uncertainties persist, most notably problems with accurate snowfall measurements
in the glacier accumulation zone, uncertainties in establishing accurate thickness measurements on glaciers,
unknown future changes associated with local-scale circulation and cloud cover affecting glacier energy balance,
the role of aerosols and in particular black carbon deposition on Andean glaciers, and the role of groundwater
and aquifers interacting with glacier meltwater.
The reduction in water supply for export-oriented agriculture, mining, hydropower production and human
consumption are the most commonly discussed concerns associated with glacier retreat, but many other aspects
including glacial hazards, tourism and recreation, and ecosystem integrity are also affected by glacier retreat.
Social and political problems surrounding water allocation for subsistence farming have led to conflicts due to
lack of adequate water governance. Local water management practices in many regions reflect cultural belief
systems, perceptions and spiritual values and glacier retreat in some places is seen as a threat to these local
livelihoods.
Comprehensive adaptation strategies, if they are to be successful, therefore need to consider science, policy,
culture and practice, and involve local populations. Planning needs to be based not only on future scenarios
derived from physically-based numerical models, but must also consider societal needs, economic agendas,
political conflicts, socioeconomic inequality and cultural values. This review elaborates on the need for adap-
tation as well as the challenges and constraints many adaptation projects are faced with, and lays out future
directions where opportunities exist to develop successful, culturally acceptable and sustainable adaptation
strategies.
⁎
Corresponding author at: Department of Atmospheric and Environmental Sciences, University at Albany, SUNY, 1400 Washington Ave., Albany, NY 12222, USA.
E-mail address: mvuille@albany.edu (M. Vuille).
1
http://doc.rero.ch
Published in "Earth-Science Reviews 176 (): 195–213, 2018"
which should be cited to refer to this work.
1. Introduction
Glaciers in the tropical Andes have been retreating rapidly during
the past decades, thereby temporarily increasing water supply in dry
regions downstream. Yet this increase is not sustainable, as glaciers
continue to shrink. Hence glacier retreat poses a significant challenge
for adaptation of a variety of natural and human systems throughout
the region. Almost all (> 99%, Kaser, 1999) of the remaining glaciers
in the tropics are located in the Andes; hence the situation in this region
is of particular relevance. Glaciers are also very helpful visual indicators
of our rapidly changing environment and can serve as sentinels of rapid
climate change. This is especially the case for tropical glaciers, which
are considered highly sensitive to changes in climate (Kaser and
Osmaston, 2002).
Glacier retreat in the tropical Andes is of particular concern for a
variety of reasons. First, unlike mid- and high-latitude mountain re-
gions, there is no seasonal snow cover outside the glacierized areas,
which could provide for an additional buffer and contribution to sea-
sonal river discharge. Because of the high solar radiation all year long,
snow falling outside of glacierized areas melts within a matter of a few
days (Lejeune et al., 2007; Wagnon et al., 2009). While this snowmelt
may contribute to immediate river discharge or potentially to ground-
water recharge, it prevents the build-up of a seasonal snowpack that can
provide water at the beginning of the dry season. Secondly, the
southern tropical Andes (central and southern Peru and Bolivia) ex-
perience a long and sustained dry season that lasts between five and six
months from April until September (Garreaud et al., 2003). During this
period rainfall is virtually absent and the glacier meltwater contribution
is fundamentally important for socioeconomic activities and environ-
mental services, especially those that take place in close proximity to
the glacierized catchments (Barnett et al., 2005; Bradley et al., 2006;
Kaser et al., 2010). This is why glacier retreat is much more of a concern
in the semiarid southern tropical Andes, at least from a water supply
perspective.
The northern tropical Andes (Colombia, Venezuela and Ecuador)
experience a much more humid climate with shorter dry periods and
some rainfall throughout the year. In addition, glaciers are generally
smaller in this region, and wetland ecosystems called páramos, can
provide for additional water storage, as their soils have a very high
water retention capacity (Buytaert et al., 2006, 2011). Nonetheless,
even in Ecuador glacier melt can temporarily provide for a significant
base flow that can become relevant during extreme drought periods
(Sicart et al., 2015).
The spatial extent and pace of Andean glacier retreat have been
documented in numerous studies, including two fairly recent in-depth
reviews (Vuille et al., 2008a; Rabatel et al., 2013). Similarly, the large-
scale drivers of this retreat, such as increasing temperature (Vuille and
Bradley, 2000; Bradley et al., 2009; Schauwecker et al., 2014; Vuille
et al., 2015) or potential changes in the spatiotemporal characteristics
of snowfall (Vuille and Ammann, 1997; Mernild et al., 2016; Saavedra
et al., 2016), thereby altering the glacier's energy and mass balance,
have been widely studied and discussed. Hence the goal of this paper is
not to provide yet another detailed review of these observed changes,
but rather to look ahead as to what impacts the region will face over the
coming decades as glaciers continue to retreat and some eventually
disappear.
While much emphasis in recent years has been put on better un-
derstanding the supply side of water resources through glacio-hydro-
logic modeling under a variety of future scenarios, there is an in-
creasing recognition that much of the future water scarcity and
struggles over allocation and access to clean water may be driven pri-
marily by the demand side (Carey et al., 2017). Indeed in wide regions
of the tropical Andes water scarcity will increase regardless of climate
change scenario, speed and magnitude of glacier retreat (Buytaert and
de Bievre, 2012). It is therefore indispensable to include socioeconomic,
cultural, legal and political aspects of water use and regulations at all
stages of future socio-hydrologic modeling attempts (e.g.
Carey et al.,
2014).
In addition glacier retreat increases the potential for natural
hazards, affects ecosystem composition in rivers downstream and may
impact economic sectors such as tourism or mining. Social values, local
perceptions and cultural beliefs are also closely intertwined with water
use and the existence of glaciers in the region. The major aim of this
review is thus to identify how glacier retreat affects Andean natural and
human systems today and to identify major gaps and roadblocks for
current adaptation efforts. A special emphasis will be put on the chal-
lenges that loom going forward, as Andean nations start implementing
adaption measures to address these impacts, in the light of large un-
certainties associated with future climate projections. Thereby we hope
to contribute to the discussion surrounding the major challenges but
also potential opportunities for climate change adaptation in the region
today.
2. Current state of glaciers and climate change in the tropical
Andes
Two recent comprehensive reviews on the current state of Andean
glaciers and their response to climate change were published by Vuille
et al. (2008a) and Rabatel et al. (2013). Both studies picture a situation
characterized by a rapidly shrinking surface area, length and volume of
glaciers due to an almost continuously negative mass balance, parti-
cularly since the late 1970s. Superimposed on the negative trends are
variations on interannual and decadal time scales, that primarily reflect
the state of tropical Pacific sea surface temperatures (SST), with often
(but not always) significant mass loss during warm episodes associated
with El Niño events, while cold phases (La Niña) tend to lead to less
negative or in some cases even balanced or slightly positive mass bal-
ance (e.g., Francou et al., 2003, 2004; Vuille et al., 2008b; Favier et al.,
2004; Lopez-Moreno et al., 2014; Veettil et al., 2014). Wagnon et al.
(2001) and more recently Maussion et al. (2015) examined the role
played by the El Niño - Southern Oscillation (ENSO) phenomenon in
more detail and quantified the various energy fluxes that contribute to
ENSO-related mass balance anomalies on glaciers in the Cordillera Real,
Bolivia, and Cordillera Blanca, Peru (see Fig. 1), respectively. They
confirmed that the influence of El Niño is transmitted primarily through
reduced and delayed snow accumulation and increased air temperature,
causing an elevated rain-snow line, lowered albedo and therefore in-
creased absorption of short-wave radiation. The higher temperature
also leads to an increased sensible heat flux, while the precipitation
deficit during most El Niño events contributes directly to a more ne-
gative mass balance (reduced snow accumulation and albedo effect
(Wagnon et al., 2001; Sicart et al., 2011)).
While ENSO events lead to significant interannual variability of
Andean glaciers' energy and mass balance, the long-term negative trend
cannot be pinned on tropical Paci fic SST's (Vuille et al., 2015). In fact
paleoclimatic studies show that throughout the Andes, glaciers have
been retreating for a long period of time. They reached a relative
maximum extent sometime during the Little Ice Age (LIA) between the
middle of the 17th and the beginning of the 18th century, likely asso-
ciated with cooler and wetter conditions. The exact timing of this ad-
vance however, varies somewhat depending on the mountain range
considered (
Rabatel et al., 2006, 2008; Solomina et al., 2007; Jomelli
et al., 2009).
But
in most locations glaciers have been receding ever
since, except for a few intermittent periods of glacier stabilization or
even mass gain, such as occurred for example in the mid 1960s to early
1970s in the Cordillera Real. According to Rabatel et al. (2013), the
retreat has accelerated over the past three decades, with an average
mass loss in the tropical Andes estimated at − 0.76 m water equivalent
(w.e.) over the period 1976–2010. This suggests that glaciers in the
tropical Andes are characterized by a more negative mass balance than
the average glacier monitored in the rest of the world (see Fig. 7 in
Rabatel et al., 2013). Glaciers located entirely below 5400 m above sea
level (a.s.l.) are generally small (< 1 km
2
) compared to high-altitude
2
http://doc.rero.ch
glaciers and experience an even more negative mass balance and more
rapid decline. Indeed these glaciers in their majority appear to be very
unbalanced (negative mass balance of − 1.2 m w.e. over the past 3
decades, compared to − 0.6 m w.e. for glaciers that extend above
5400 m a.s.l.), with many of them likely to disappear in the coming
decades (Rabatel et al., 2013). It is noteworthy that these glaciers re-
present the majority of all glaciers in the tropical Andes (e.g. 80% in the
Bolivian Cordillera Real in 1975), corresponding to half of the gla-
cierized surface area (Soruco et al., 2009). Overall the current rate of
glacier loss appears to be unprecedented for the period over which
reasonably accurate reconstructions exist (since the early 18th century).
Since the publication of Rabatel et al. (2013) a few new studies
focusing on specific mountain ranges have been published, which all
essentially confirm the results described above. Braun and Bezada
(2013) for example have documented in great detail the rapid shrinkage
and looming disappearance of tropical glaciers from the Andes in Ve-
nezuela. Hanshaw and Bookhagen (2014) and Lopez-Moreno et al.
(2014), based on a long series of satellite images, highlighted the rapid
shrinkage of glaciers in the Cordillera Vilcanota and Huaytapallana,
respectively. Both studies identified significant and fast glacier retreat
with a > 50% surface area loss in the Cordillera Huaytapallana be-
tween 1984 and 2011 and a roughly 25% area loss in the Cordillera
Vilcanota between 1962 and 2009.
These observed changes in glacier extent in the tropical Andes are
consistent with observations of temperature and precipitation changes
in the Andes during this same time period. A number of studies have
presented evidence for a rapidly changing climate in the tropical Andes.
Vuille and Bradley (2000) first documented a clear warming trend over
the Andes in the 20th century (based on a large number of in-situ sta-
tion data), covering the tropical Andes from Ecuador to northern Chile.
They also identified, for the first time, that the observed warming ap-
peared to be dependent on elevation and slope. Follow-up studies by
Vuille et al. (2003, 2008a) documented a warming of roughly 0.1 °C per
decade since the 1950s, when averaged over the entire tropical Andes
south of the equator. These results are consistent with more regional
temperature trend analyses performed later by Racoviteanu et al.
(2008), Poveda and Pineda (2009), Bradley et al. (2009), Gilbert et al.
(2010), Lavado Casimiro et al. (2013), Salzmann et al. (2013), Seiler
et al. (2013a) and Lopez-Moreno et al. (2014, 2016) in various locations
of the Andes. More recently Schauwecker et al. (2014) and Vuille et al.
(2015) analyzed the apparent slowdown of the warming, which may
have been linked with the cooling observed over the eastern tropical
Pacific and the cool phase of the Pacific Decadal Oscillation (PDO) since
the early 2000s. According to these studies, however, the Pacific
cooling aff ected
primarily
low-level coastal locations along the Pacific
shore, while the highest elevations in the tropical Andes have continued
to warm during this ‘hiatus’ period. In fact, as documented by Vuille
et al. (2015), the warming at higher elevation in the tropical Andes now
appears to have emerged outside the range of natural variability, ef-
fectively decoupling Andean temperature from the SST forcing in the
Pacific, which in previous decades served as a strong predictor for cold
or warm periods in the Andes.
Precipitation trends on the other hand are much more subtle and
difficult to identify, partly due to the paucity of long, high-quality ob-
servational records and partly due to the strong modulation of pre-
cipitation by topography, introducing significant east-west gradients
between the arid Pacific coast and the humid Amazon basin, thereby
rendering precipitation trends inherently more heterogeneous. Vuille
et al. (2003) initially suggested a tendency toward slightly increased
precipitation in the inner tropics (north of 11°S) and a decrease in the
outer tropical Andes further south, although they emphasized that
trends at individual stations were mostly insignificant. These results
were later confirmed in a follow-up study by Haylock et al. (2006),
identifying the same north-south contrast in precipitation trends. More
recently Salzmann et al. (2013) and Lavado Casimiro et al. (2013)
analyzed rainfall trends in the southern Peruvian Andes with newer and
denser data sets and were unable to find significant trends over the past
40 years. Instead most of the variability in the data appears to be as-
sociated with interannual variability driven by the ENSO phenomenon
(Rau et al., 2016).
In summary all these studies strongly link the observed glacier re-
treat to an increase in temperature — as the diverging and mostly in-
significant trends in precipitation seem inconsistent with the coherent
and strong glacier retreat observed throughout the tropical Andes re-
gion. We wish to stress, however, that estimating snowfall totals in the
accumulation zone of Andean glaciers is associated with large un-
certainties (see Section 4.1) and that past changes in snow accumula-
tion in the highest reaches of the tropical Andes are virtually unknown,
except for a few point-estimates based on ice-core studies (see review in
Vimeux et al., 2009). In addition, tropical glacier energy balance of
course strongly depends on the seasonal distribution of precipitation.
This distribution can change without being reflected in annual snowfall
totals and may also affect seasonal temperature trends, for example
through a delay of the wet season and reduced cloud cover.
3. Challenges for future projections – where are we headed?
3.1. Climate projections
Studies invoking future temperature projections in the Andes are
quite limited and have focused primarily on free-tropospheric tem-
perature. Bradley et al. (2004, 2006) were the first to consider future
changes in free-tropospheric temperature along the Andes Cordillera by
employing a multi-model ensemble from the Coupled Model Inter-
comparison Project Phase 2+ (CMIP2 +) and 3 (CMIP3) under a
2×CO
2
and the Intergovernmental Panel on Climate Change (IPCC)
Special Report on Emission Scenarios (SRES) A2 scenario, respectively.
According to their results, by the end of the 21st century the peak
elevations of the tropical Andes might warm by up to 4°–5 °C when
compared to the base period (1990–1999) under a high emission sce-
nario. The authors in these studies also pointed out that the models
consistently simulated an amplified warming at higher elevations,
Fig. 1. Map of the tropical Andes, with locations of glaciers and mountain ranges dis-
cussed in the text. White box (2°N–18°S/65°–82°W) indicates area over which CMIP 5
models were averaged to derive simulated historical and projected temperature change
over the Andes.
3
http://doc.rero.ch
although they did not diagnose or quantify the feedbacks involved in
this elevation-dependent warming. Urrutia and Vuille (2009) subse-
quently employed a regional climate model to simulate both near-sur-
face and free-tropospheric temperature change by the end of the 21st
century, compared to the 1961–90 average, based on the SRES B2 and
A2 scenarios. Their results also indicated that the warming would be
amplified at higher elevations going forward, reaching up to 5–6°Cat
the highest elevations in the A2 scenario and about half that amount in
the lower emissions scenario B2. Urrutia and Vuille (2009) further
pointed out that not only will temperature increase, but the interannual
variability will also be considerably enhanced, especially in the A2
scenario, leading to a higher probability of extremely warm years. Their
study, however, was based on only one climate model, which, as shown
by Buytaert et al. (2010), can lead to large uncertainties in the inter-
pretation, as ideally an ensemble of regional climate models should be
used for such numerical downscaling applications. Thibeault et al.
(2010) took a different approach by focusing on future projections of
temperature extremes, such as heat waves, warm nights or frost days on
the Bolivian Altiplano as simulated by an ensemble of CMIP3 models in
the SRES B1, A1B and A2 scenarios. According to their results, frost
days are projected to decrease by 2–4 standard deviations in all sce-
narios by the end of the 21st century when compared to the reference
year 2000, while warm nights and heat waves will become more fre-
quent by 3–7 standard deviations (warm nights) and 1–4 standard de-
viations (heat waves) respectively, by the year 2100. More recently
Seiler et al. (2013b) analyzed both CMIP3 and CMIP5 models over
Bolivia to estimate future temperature changes under a range of emis-
sion scenarios. According to their results, temperature in Bolivia will
increase anywhere from 2.5 °C to 6 °C by the end of the century com-
pared to the 1961–90 period, depending on model and emission sce-
nario considered.
Here we use seven CMIP5 models (HadGEM2-AO, HadGEM2-CC,
HadGEM2-ES, MIROC-ESM-CHEM, MPI-ESM-LR, MPI-ESM-MR and
NorESM1-ME, see Table 1) to analyze how surface temperature might
evolve throughout the 21st century in the tropical Andes. We calculate
anomalies with respect to the 1961–1990 reference period and analyze
results from a moderate (Representative Concentration Pathway, RCP
4.5) and a high-emission (RCP 8.5) scenario to probe future outcomes.
In order to compare future projections with past temperature trends we
average the model data over a box (2°N-18°S/65°-82°W, see Fig. 1) that
approximates the station coverage used for the recent Andean tem-
perature re-assessment in Vuille et al. (2015) and we also analyze
temperature trends in historical CMIP5 simulations of the same seven
models since 1950. For each simulation we select one ensemble
member per model.
Results of our analyses are shown in Fig. 2. They suggest that
temperature in the tropical Andes might continue to increase by 1°–5°C
(mean values of 2° –3.5 °C) beyond the 1961–90 values. This warming is
less than what was reported previously in some of the studies discussed
above, but it should be kept in mind that these are surface and not free-
tropospheric trends and that they represent a large-scale average over a
model topography that is significantly lower than reality. Hence surface
warming is likely to be larger than these values reported here at high
elevations where glaciers are located (see Vuille et al., 2015). Tem-
perature trends associated with rising freezing levels and implications
for the rise of the equilibrium-line altitude (ELA) of tropical Andean
glaciers are analyzed separately in
Section 3.2.It
is
noteworthy that the
trend in the simulated warming over the historical period (1950–2005)
is consistent with the observed warming by Vuille et al. (2015). But of
course averaging over multiple models tends to reduce the interannual
variability when compared with the observational record, as inter-
annual signals related to internal variability (e.g. ENSO) are not in
phase between models.
Several studies have tried to constrain future precipitation
Table 1
CMIP5 Models used in analysis of surface temperature and/or freezing level height.
Modeling center Institute ID Model name
Beijing Climate Center, China Meteorological Administration BCC BCC- CSM1.1(m)
College of Global Change and Earth System Science, Beijing Normal University GCESS BNU-ESM
Canadian Centre for Climate Modeling and Analysis CCCMA CanESM2
National Center for Atmospheric Research NCAR CCSM4
Community Earth System Model Contributors NSF-DOE-NCAR CESM1(BGC)
Commonwealth Scientific and Industrial Research Organization in collaboration with Queensland Climate Change Centre of Excellence CSIRO-QCCCE CSIRO-Mk3.6.0
The First Institute of Oceanography, SOA, China FIO FIO-ESM
National Institute of Meteorological Research/Korea Meteorological Administration NIMR/KMA HadGEM2-AO
Met Office Hadley Centre MOHC HadGEM2-CC
HadGEM2-ES
Japan Agency for Marine - Earth Science and Technology, Atmosphere and Ocean Research Institute (The University of Tokyo), and
National Institute for Environmental Studies
MIROC MIROC-ESM-CHEM
Max Planck Institute for Meteorology MPI-M MPI-ESM-LR
MPI-ESM-MR
Norwegian Climate Centre NCC NorESM1-ME
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
1950
1970
1990 2010 2030
2050 2070
2090
Year
Surface air temperature anomlay (°C)
observations
historical
RCP 4.5
RCP 8.5
Fig. 2. Simulated and observed annual mean air temperature anomalies in the tropical
Andes (departures from 1961 to 1990 mean) derived from station data (Vuille et al.,
2015, green, 1950–2010), historical CMIP5 (gray, 1950–2005), and future CMIP5 sce-
narios (RCP 4.5 in light blue and RCP 8.5 in red, 2006–2100). CMIP5 data averaged over
box 2°N-18°S, 65–82°W to match distribution of station data in Vuille et al. (2015).
Shading indicates ± 1.64 std. dev. (5–95% range across models). (For interpretation of
the references to colour in this figure legend, the reader is referred to the web version of
this article.)
4
http://doc.rero.ch
projections for the Andes, using a variety of different approaches. Most
recently Neukom et al. (2015) were able to show that precipitation
during the austral summer wet season (DJF) might drop below the
range of natural variability in the central Andes of Bolivia and southern
Peru within a matter of a few decades. Their analysis was based on a
combination of proxy evidence from tree rings and ice cores since the
year 1000 CE with historical model runs and future Coupled Model
Intercomparison Project Phase 5 (CMIP5)-based projections out to the
year 2100 CE. These results are consistent with earlier findings by
Urrutia and Vuille (2009), Minvielle and Garreaud (2011) and
Thibeault et al. (2010, 2012) who used dynamical and statistical
downscaling techniques, respectively, to determine future changes in
precipitation, based on the older IPCC-SRES scenarios. While these
projections of a roughly 10–30% decrease in precipitation by the end of
the century appear to be quite robust, given that they have been re-
produced by multiple studies employing different methodologies, the
uncertainty regarding future precipitation changes is much larger in the
more humid inner tropics. Buytaert et al. (2010) and more recently
Zulkafli et al. (2016), for example, showed that in the Andes of
Ecuador, models have difficulties in even agreeing on the sign of future
changes in precipitation. Nevertheless, some studies suggest that the
moisture flux from the Amazon basin to the Andes may weaken under a
future warming-scenario (e.g. Marengo and Espinoza, 2016).
3.2. Glaciologic projections
For longer-term projections of future watershed hydrology a quan-
titative prediction as to how the glacier area and volume will change
over time is essential (e.g. Frans et al., 2015). Ideally this requires a
realistic simulation of dynamic glacier flow, which is still a challenge in
hydro-glaciologic modeling. In addition the coupling between large-
scale climatic information, derived from historical simulations or future
projections of climate change, and glaciologic and hydrologic models
requires downscaling of the climate information to the catchment scale
to bridge this scale mismatch. Some studies have circumvented this
problem by general approximations of temperature focusing on the
close coupling between the ELA of tropical glaciers and the 0 °C iso-
therm (the freezing level) (Condom et al., 2007; Rabatel et al., 2012;
Sagredo et al., 2014), while others have exploited the relationship be-
tween freezing level height (FLH) and glacier extent ( Schauwecker
et al., 2017). As shown in Fig. 3, there is indeed a close relationship
between FLH and glacier ELA in the tropical Andes on interannual time
scales, consistent with previous reports (Vuille et al., 2008a; Bradley
et al., 2009; Rabatel et al., 2013). In our analysis we focus on glaciers
Antizana (0°29′S, 78°09′W, 4780–5760 m, Ecuador), Artesonraju
(8°57′S, 77°27′W, 4685–5979 m, Peru) and Zongo (16°15′S, 68°10′W,
4900–6000 m, Bolivia), which have the longest mass balance time
series available, and which allow us to probe a set of glaciers re-
presentative of inner and outer tropical sites (
Fig. 1).
The ELA data are
based
on measurements first published (and subsequently updated) in
Rabatel et al. (2012), Loarte et al. (2015) and Basantes-Serrano et al.
(2016).
Both ELA and FLH values were calculated based on averages over
the hydrologic year. The Antizana mass balance value from 2003 was
omitted from the analyses, as it was a large outlier, inconsistent with
the relationship shown in Fig. 3. Indeed glacier mass balance estimates
on Antizana are known to suffer from significant uncertainties due to
difficulties in obtaining accurate measurements from the accumulation
zone (Basantes-Serrano et al., 2016 ).
Schauwecker et al. (2014) have shown that significant differences
exist between reanalysis products, when assessing free tropospheric
temperature trends over the Andes, with NCEP/NCAR showing stronger
warming at mid-tropospheric levels than ERA-Interim. Here we there-
fore probe these same two products in assessing trends in FLH and its
relationships with the ELA in di fferent mountain regions. The FLH was
calculated by linearly interpolating temperature and geopotential
height between 500 and 600 hPa and then extracting geopotential
height at the interpolated level where temperature equals 0 °C. This was
done separately for both reanalysis products and each location (grid cell
covering glacier site).
Fig. 3 shows that the ELA is significantly correlated (p < 0.05) with
FLH at all sites. Depending on reanalysis product, changes in FLH ex-
plain between roughly half and two thirds of the total variance in the
ELA on Zongo, Artesonraju and Antizana, respectively. The slope of the
relationship tends to be steeper in the inner tropics (Antizana) and
become flatter toward the outer tropics (Zongo), consistent with the
tendency of inner tropical glaciers to be more sensitive to direct tem-
perature variations, while outer tropical glaciers, where the ELA is
several hundred meters above the FLH, tend to be more strongly tied to
changes in hydrologic variables (Favier et al., 2004). ERA slopes are
steeper than those derived from NCEP/NCAR, but all slopes are sig-
nificantly different from zero (F-test, p < 0.05).
We next employed the relationships examined above to probe future
changes in the ELA at these three glacier locations. We first calculated
4700
4800
4900
5000
5100
5200
5300
5400
4700 4800 4900 5000 5100
Antizana (1996-2002 / 2004-2012)
r = 0.74 (p<0.005)
ELA = 1386.5 + 0.76 FLH
Artesonraju (2000-2010)
r = 0.81 (p<0.01)
ELA = 1656.1 + 0.66 FLH
Zongo (1991-2015)
r = 0.72 (p<0.005)
ELA = 2103.3 + 0.62 FLH
a)
NCEP/NCAR Freezing Level Height [m]
Equilibrium Line Altitude [m]
Zongo and Arteson
4900
5000
5100
5200
5300
5400
4700 4800 4900 5000 5100
Equilibrium Line Altitude [m]
Antizana
ERA-Interim Freezing Level Height [m]
Zongo (1991-2015)
r = 0.67 (p<0.01)
ELA = 1986.5 + 0.64 FLH
Artesonraju (2000-2010)
r = 0.75 (p<0.05)
ELA = 407.1 + 0.92 FLH
Antizana (1996-2002 / 2004-2012)
r = 0.74 (p<0.005)
ELA = 694.4 + 0.90 FLH
b)
Fig. 3. Scatterplot and ordinary least squares
linear regression of hydrologic year ELA vs.
FLH on glaciers Zongo (Bolivia), Artesonraju
(Peru) and Antizana (Ecuador). FLH is derived
from a) NCEP/NCAR (Kalnay et al., 1996) and
b) ERA-Interim (Dee et al., 2011). ELA data is
from Rabatel et al. (2012), Loarte et al. (2015)
and Basantes Serrano et al. (2016).
5
http://doc.rero.ch
