Content uploaded by Skafti Brynjólfsson
All content in this area was uploaded by Skafti Brynjólfsson on Nov 24, 2017
Content may be subject to copyright.
© The Author(s) 2017
Reprints and permissions:
The variations in summer temperature (mean ablation season air
temperature, Ts) and winter accumulation exert a particularly
decisive influence on the dynamics of the debris-free glaciers
(Eythorsson, 1935; Liestøl, 1967; Ohmura et al., 1992). For
example, Caseldine (1985b) pointed out that the combined effect
of a low Ts and normal winter precipitation led to a glacier
advance over a 10-year period in Tröllaskagi peninsula. The
debris-free glaciers are especially sensitive to climate variations
and were extensively studied in the peninsula until the late 1980s
(Caseldine, 1983, 1985a, 1985b, 1987; Caseldine and Culling-
ford, 1981; Caseldine and Stötter, 1993).
For many of the glaciers in the central highlands of Iceland
and for the majority of those in the Tröllaskagi peninsula, the
maximum glacial advance in the second half of the Holocene was
reached during the ‘Little Ice Age’ (LIA; Flowers et al., 2007;
Kirkbride and Dugmore, 2001, 2006; Larsen et al., 2011;
Schomacker et al., 2003).
The Holocene temperature variation range is thought to be
around 3°C (Stötter et al., 1999): mean annual air temperature
(MAAT) during the Holocene thermal maximum (HTM) is esti-
mated to be 3°C higher than that for the period 1961–1990 (Casel-
dine et al., 2006; Geirsdóttir et al., 2009), and therefore comparable
to the warmest decades of the 20th century (Stötter et al., 1999),
while conditions during the post-Preboreal Holocene minimum
were similar to those during the second half of the 19th century
(Stötter et al., 1999). Precipitation is estimated to have doubled
between both climatic extremes (Stötter et al., 1999). Caseldine
and Stötter (1993) suggest that from the LIA maximum in the late
19th century to the 1980s, the Ts increased by 2°C and winter
precipitation by 600 mm (+41%, from 1450 mm) at the
equilibrium line altitude (ELA). Climate evolution during the last
High sensitivity of North Iceland
(Tröllaskagi) debris-free glaciers
to climatic change from the
‘Little Ice Age’ to the present
José María Fernández-Fernández,1 Nuria Andrés,1
Þorsteinn Sæmundsson,2 Skafti Brynjólfsson3
and David Palacios1
The Tröllaskagi peninsula is located in northern Iceland, between meridian 19°30′W and 18°10′W, jutting out into the North Atlantic to latitude 66°12′N.
The aim of this research is to study recent glacier changes in relation to climatic evolution of the Gljúfurárjökull and Tungnahryggsjökull debris-free valley
glaciers in Tröllaskagi. Glacier extent mapping and spatial analysis operations were performed with ArcGIS (ESRI), using analysis of aerial photographs
from 1946, 1985, 1994 and 2000, and a 2005 SPOT satellite image. The results show that these glaciers lost a quarter of their surface area between the
‘Little Ice Age’ and 2005. In this paper, the term ‘Little Ice Age’ follows Grove (2001) as the most recent period when glaciers extended globally between
the medieval period and the early 20th century. The abrupt climatic transition of the early 20th century and the 25-year warm period 1925–1950 triggered
the main retreat and volume loss of these glaciers since the end of the ‘Little Ice Age’. Meanwhile, cooling during the 1960s, 1970s and 1980s altered the
trend, with advances of the glacier snouts. Between the ‘Little Ice Age’ and the present day, the mean annual air temperature and mean ablation season
air temperature increased by 1.9°C and 1.5°C, respectively, leading to a 40–50 m rise in the equilibrium line altitude (ELA) of the glaciers during this
period. The response of these glaciers depends not only on the mean ablation season air temperature evolution but also on other factors such as winter
precipitation. The models applied show a precipitation increase of up to more than 700 mm since the ‘Little Ice Age’.
climatic change, deglaciation, equilibrium line altitude, Iceland, ‘Little Ice Age’, Tröllaskagi peninsula
Received 31 July 2016; revised manuscript accepted 16 November 2016
1 Research Group of High Mountain Physical Geography, Department of
Geography, Faculty of Geography and History, Complutense University
of Madrid, Spain
Faculty of Life and Environmental Sciences, University of Iceland, Iceland
3Icelandic Institute of Natural History, Iceland
José María Fernández-Fernández, Research Group of High Mountain
Physical Geography, Department of Geography, Faculty of Geography
and History, Complutense University of Madrid, Profesor Aranguren
Street, 28040 Madrid, Spain.
683262HOL0010.1177/0959683616683262The HoloceneFernández-Fernández et al.
2 The Holocene
millennium, and its relationship to sea ice expansion, is relatively
well known from the work of Koch (1945), Bergþórsson (1969),
Ogilvie (1984, 1996, 2005, 2010) and Ogilvie and Jónsson (2001).
Especially cold climatic episodes at the beginning and end of the
13th century, in much of the second half of the 14th century
(1350–1380) and during the later years of the 16th century have
been reported by Ogilvie (1984, 1991) and Ogilvie and Jónsson
(2001). Koch (1945) suggested that between AD 1600 and 1900,
the climate was particularly cold in Iceland and East Greenland,
but Ogilvie (1984, 2005, 2010) pointed out that there was great
variability with some mild periods and different levels of inci-
dence of the sea ice: a more temperate climate occurred during the
1640s, 1650s and early 18th century, while very cold climatic epi-
sodes in the late 17th century (1690s), mid-18th century (1740s in
the north and 1750s in the south) and during the 19th century
(1810s, 1830s and 1880s; Ogilvie, 2010; Ogilvie and Jónsdóttir,
2000; Ogilvie and Jónsson, 2001), coinciding with the maximum
sea ice extent. This is in contrast to the 20th century, which was
more temperate than the three preceding centuries (Ogilvie, 1984,
1996, 2010; Ogilvie and Jónsson, 2001).
The limits and definition of the LIA are complex issues as they
vary between authors, probably because of the differences in
regions and the approaches applied (Grove, 2001). The origins
and uses of the term ‘Little Ice Age’ are discussed in detail in
Ogilvie and Jónsson (2001). Grove (1988) considered it to have
begun earlier and extended from 1450 to 1900. However, it is sug-
gested this period should be expanded, as much evidence has
demonstrated that the LIA was under way in the early 14th cen-
tury along North Atlantic regions as synthesized by Grove (2001),
for example, glacial activity found in Iceland during the 13th and
14th centuries. Nevertheless, the period from 1600 to 1900 must
have been more important for the glacial activity in Iceland as it
was not interrupted by major warming (Guðmundsson, 1997),
except a mild period around 1640–1680 (Ogilvie, 2005, 2010).
We define the LIA in this paper according to Grove (2001) as the
most recent period when glaciers extended globally and remained
enlarged between the Medieval period and the warming begin-
ning in the early 20th century (Grove, 1988).
The Gljúfurárjökull and Western Tungnahryggsjökull glaciers
reached their maximum extent during the LIA, more precisely
during the second half of the 19th century (Caseldine, 1983,
1985b; Caseldine and Cullingford, 1981), coinciding with the
Holocene maximum advance of many Icelandic ice-cap outlet
glaciers (Geirsdóttir et al., 2009; Kirkbride and Dugmore, 2008).
Nevertheless, the maxima of Gljúfurárjökull and Tungnah-
ryggsjökull were not synchronous: Gljúfurárjökull reached maxi-
mum extent around AD 1898–1903, while the Western
Tungnahryggsjökull reached its maximum in AD 1868 (Casel-
dine, 1983, 1985b; Caseldine and Cullingford, 1981). Thus, the
LIA climate of the Tröllaskagi peninsula was characterized by a
Ts 2°C lower than at present and an average winter precipitation
of 1450 mm at the ELA (946 m; Caseldine and Stötter, 1993).
Rising temperatures from the end of the 19th century caused
the glaciers to retreat from their LIA positions. The retreat of the
Western Tungnahryggsjökull (Caseldine, 1985b) started decades
earlier than in Gljúfurárdalur, because of the steeper gradient of
the Vesturdalur valley and the reduced thickness of the glacier.
The retreat throughout the 19th century was interrupted by differ-
ent advance phases with moraine formation: 1876–1878, 1882–
1887 and 1898–1903 (Caseldine, 1985b).
Gljúfurárjökull retreated 250 m from its LIA position during
the first 20 years of the 20th century (Caseldine, 1983, 1987).
This retreat was interrupted with moraine formation in 1910 and
1913–1917, and later slowed down between the mid-1920s and
1930. During the early 1930s, the retreat accelerated again (ca.
200 m) and was interrupted by minor advances which enabled
moraine deposition around 1935. Once again, in the late 1940s
there was a short re-advance, concluding in 1950–1951 (Casel-
dine, 1983), which left a series of moraine arcs. Marginal mea-
surements of the Icelandic Glaciological Society (IGS) show that
Gljúfurárjökull continued to retreat 422 m until it reached its
minimum extent in the mid-1970s. Glacial retreat data since the
LIA obtained by Caseldine and Cullingford (1981) using photo-
grammetry and lichenometry show that the terminus retreated 265
m between 1939 and 1972 and 151 m between 1953 and 1960.
After reaching its minimum extension, the trend reversed in 1977.
The glacier commenced a new, more continuous re-advance of
greater scope than the advances which had occurred during degla-
ciation: the snout advanced 50 m in 1977–1979, slowing in the
following years to 30 m from 1979 to 1981 and 25 m from 1981
to 1983 (Caseldine, 1983, 1985a, 1985b, 1988; Caseldine and
The snout of Gljúfurárjökull was located at altitude 580 m
with an ELA ca. 960 m in 1985 (Caseldine, 1985b). The ELA for
48 glaciers in Tröllaskagi was higher, at 992 m (Caseldine and
Stötter, 1993). The last publication about Gljúfurárjökull in the
late 1980s (Caseldine, 1988) pointed out the end of its advance in
1986. On the other hand, the annual IGS marginal measurements
show that the Gljúfurárjökull advance ended in the late 1980s;
after that, it began to retreat again, by more than 160 m between
1989 and 2013.
The relationship between the climate and glacier response
(glacier termini, mass balance) was first studied in Iceland by
Björnsson (1971), who proposed 8°C (at Akureyri) as the Ts
threshold which would reflect the change in the mass balance sign
in the Tröllaskagi glaciers. The temperature evolution in Iceland
was studied by Einarsson (1991), who differentiated six thermal
phases between 1901 and 1990, depending on whether these were
cold (1901–1925, 1947–1952 and 1965–1971) or warm (1926–
1946, 1953–1964 and 1972–1990).
The study of the above debris-free glaciers in Tröllaskagi
enabled the impact of climate change in northern Iceland to be
monitored for decades and compared with the evolution of the
large ice-caps in central and southern Iceland. The aim of this
article is to extend the previous work and record the evolution
of the Gljúfurárjökull and Tungnahryggsjökull glaciers up to
the present, focusing on termini retreat, area/volume loss and
ELA variation in these glaciers. In addition, the trends and rela-
tionships of these parameters with climate evolution will be
The limits of the Tröllaskagi peninsula in north Iceland are
Skagafjörður to the west and Eyjafjarðardalur to the east (Figure
1) between meridian 19°30′W and 18°10′W, jutting out into the
North Atlantic to latitude 66°12′N and linked to the central high-
lands to the south. The peninsula consists of over 4000 km2 of
tertiary flat-summit highlands and crests at 1000–1400 m com-
posed of jointed basaltic lava flows often separated by 30–50 cm
lithified sedimentary horizons (Jóhannesson and Sæmundsson,
1989; Sæmundsson et al., 1980). The highlands are cut by deeply
entrenched valleys with steep slopes and sheer headwalls. These
headwall areas host 167 small cirque glaciers (Icelandic Meteoro-
logical Office, 2015), a few of which are debris-free and the most
sensitive to climatic fluctuations (Häberle, 1991; Kugelmann,
The Tröllaskagi corrie-glaciers are found in north-facing
cirques resulting from the leeward accumulation of snow blow-
ing from the plateau areas (Caseldine and Stötter, 1993) and the
solar radiation shadow. Most of the glaciers are debris-covered
and rock glaciers because of significant slope activity. The insu-
lating effect of the debris cover makes them static and less sensi-
tive to climate variations (Martin et al., 1991).
Fernández-Fernández et al. 3
The 1961–1990 weather data series show a MAAT of 2–4°C
on the Tröllaskagi coasts and −2° to −4°C on the summits (Etzel-
müller et al., 2007). At Akureyri (1901–1990), MAAT is 3.4°C,
while mean summer (June–August) and Ts reach 9.9°C and
8.4°C, respectively (Einarsson, 1991); in winter (January–March),
the mean value drops to −1.6°C (Einarsson, 1991), although it can
be higher if the October–April period is considered, with −0.3°C.
Precipitation in the Tröllaskagi area oscillates between 400 mm in
some lowland areas of Skagafjörður and Eyjafjörður and up to
2500 mm on the summits (1971–2000 data series; Crochet et al.,
2007). Two weather stations have been used for the analysis, one
of which is located in the capital town of northern Iceland,
Akureyri (65°41′N; 18°06′W; 23 m a.s.l.), in inner Eyjafjörður;
and the other on the mountain road at Öxnadalsheiði (65°28′Ν;
18°41′W; 540 m a.s.l.) in southern Tröllaskagi.
Gljúfurárjökull and Tungnahryggsjökull have been selected as
the largest glaciers with no superficial debris cover in the study
area (Figure 1), with areas between 4 and 9 km2 (Tables 2 and 4).
This feature makes them optimal for assessing their level of sus-
ceptibility to climatic variations.
Glacier monitoring and spatial analysis operations were per-
formed with ArcGIS (ESRI), using analysis of aerial photographs
(≈1:30,000 scale) from 1946, 1985, 1994 and 2000 (National
Land Survey of Iceland, 2015). A 2005 SPOT satellite image was
also used. The glaciers were delimited at different dates by photo-
interpretation and georeferencing (RMS error 3.1–5.9 m) of the
aerial photographs and the previously georeferenced satellite
ArcGIS was used to calculate the area and retreat of the gla-
ciers at different dates. The glacier extent during the LIA maxi-
mum was delimited over the position of the morainic ridges dated
to the late LIA (end of the 19th century) in the bibliographic
references cited in section ‘Introduction’ (Caseldine, 1983, 1985a,
1985b; Caseldine and Cullingford, 1981; Caseldine and Stötter,
1993). The glaciers and moraines were mapped in early publica-
tions after previous fieldwork aided by several techniques such as
triangulation, tachometry and photogrammetry (Caseldine and
Cullingford, 1981), which enabled the authors to make an accurate
mapping and contouring of the glacier surface. Regarding dating
methods, lichenometry was used in previous works to date the
most recent and LIA maximum moraines (Caseldine, 1983, 1985a,
1985b; Caseldine and Cullingford, 1981; Caseldine and Stötter,
1993; Kugelmann, 1991), in combination with radiocarbon
(Häberle, 1991), tephrochronology and Schmidt hammer (Casel-
dine, 1987), where lichenometry was not applicable. Lateral
moraines were used to reconstruct the glacier topography and ice
thickness during the LIA maximum and for each date analysed.
The ELA of the glaciers was calculated automatically for each
date available with the ArcGIS toolbox designed by Pellitero
et al. (2015), implementing the accumulation area ratio (AAR;
Brückner, 1886, 1887) and the area altitude balance ratio (AABR;
Osmaston, 2005) methods. For the AAR method, the ratio 0.67
was applied, previously used by Caseldine and Stötter (1993), as
the results obtained in Tröllaskagi were similar to the maximum
elevation of the lateral moraines (Stötter, 1990). For the AABR
method, the balance ratio (BR) of 1.5 ± 0.4 proposed by Rea
(2009) as representative for Norwegian glaciers was used.
The results obtained from the glacial remote studies (ELA
depressions) and the climatological data were used to estimate
the temperatures and paleotemperatures at the snouts and the
ELA, assuming a lapse rate of 0.66°C 100−1 m. To estimate the
current precipitation and the paleoprecipitation, glacio-climatic
models were used (Ballantyne, 1989; Braithwaite, 2008; Ohm-
ura et al., 1992). These relate variables such as MAAT or Ts
with precipitation or ablation, measured on an annual, seasonal
or daily scale. The first model used is based on an exponential
relationship existing between the mean ablation season
Figure 1. Location of the study area in the interior of the Tröllaskagi peninsula. This figure is available in colour in the online version.
4 The Holocene
temperature and winter accumulation at the ELA of Norwegian
glaciers (Liestøl, 1967; Sutherland, 1984), expressed by equa-
tion (1) established by Ballantyne (1989), and later applied by
Dahl and Nesje (1992) in southern Norway and by Caseldine
and Stötter (1993) in Tröllaskagi
==<0 915 0 989 0 0001
0 339 2
where A is the winter accumulation (October–April) in metres of
water equivalent, and Ts is the mean ablation season temperature
(May–September) in degree Celsius.
The second model used was proposed by Ohmura et al. (1992),
defined by equation (2), of the best fit polynomial curve obtained
through the regression analysis between the mean temperature of
the three summer months (June, July and August) and total annual
precipitation (winter balance plus summer precipitation) at the
ELA for a dataset of 70 glaciers worldwide:
T=+ +645 296 9 2 (2)
where P is the total annual precipitation (in mm water equivalent)
and T is the mean temperature of the three summer months (in
°C). Standard error is 200 mm.
The last method used was the ‘degree-day’ model (Braithwaite,
2008; Brugger, 2006), based on the existing proportionality between
snow or ice melt (ablation, expressed in mm of water equivalent)
and the sum of temperatures above freezing point (degree-day sum).
The quotient of the two variables obtains the degree-day factor (df)
ratio (expressed in mm day−1 °C−1). The value for df used was the
average 4.1 mm day−1 °C−1, obtained by Braithwaite (2008) from 66
of the 70 glaciers in the dataset included in Ohmura et al. (1992).
According to this approach, equation (3) shows that melt (Md) only
occurs with positive temperatures, so that:
where Md is the daily melt (in mm water equivalent) and Td is the
mean daily temperature (in °C), obtained from equation (4), based
on a sinusoidal distribution throughout the year around the mean
temperature so that:
where Ay is the amplitude (calculated as half of the annual tem-
perature range), d is the Julian day (1–365), λ is the period (365
is the phase angle (1.93 to reflect that January is the
coldest month). MAAT is expressed in °C.
The result of applying the model is the total annual ice or snow
melt (in mm of water equivalent), which is equivalent to the accu-
mulation, given that the mass balance at the ELA is 0.
Calculations of glacial volume were then performed following
the indications by Bahr et al. (1997), according to which the vol-
ume (V) of any glacier is related to its surface area (S) using expo-
nential equation (5) based on a dimensionless scaling exponent
(γ) which includes the morphometric characteristics of the glacier
(width, slope, side drag and mass balance):
V ∝ cSγ (5)
where c is an empirical power law coefficient of 0.2055 (expressed
in units of m3−2γ), derived from Chen and Ohmura (1990), and γ is
derived from equation (6):
where q = 0.6, m = 2 and f = 0 are obtained from empirical data,
and where r may be 0 for steep surface slopes, or r = (1 − m + n
− nf) / (2 (n + 1)) for gentle slopes; and n = 3. So γ applied is in
turn 1.375. Further details on the mathematical basis and its appli-
cation can be found in Bahr et al. (1997) and Radić et al. (2007).
In parallel to studying the glaciers, the climatological data
series from the two meteorological stations presented above
was statistically processed. The series from Akureyri were used
because it is the longest at the study area, with data recorded
since 1882 (Icelandic Meteorological Office, 2015), and from
Öxnadalsheiði because it is close to the study area and is located
at a higher altitude, with a useful period from 2000 to 2014
(Icelandic Road and Coastal Administration, 2016). The statis-
tical processing analysed the running-means (5 years) and cor-
relation between the two meteorological stations. The
correlation analysis was used to reconstruct the MAAT at the
Öxnadalsheiði meteorological station using the regression
Evolution of the glacier snouts
Studies of aerial photographs and satellite images show that the
glacier snouts have retreated by more than 1300 m on average
since the LIA maximum (considered to be AD 1898 in Gljúfurár-
jökull and AD 1868 in both Western and Eastern Tungnah-
ryggsjökull as explained in the publications presented in section
‘Introduction’; Figure 2), with an altitudinal rise of more than 100
m. The retreat accelerated rapidly (15.3 m yr−1) during the first
half of the 20th century (Figure 2). In the second half of the 20th
century, the retreat decelerated considerably, reflected in the low-
est values around 1985 (5.2 m yr−1) and a trend shift in 1994, with
an advance observed in Gljúfurárjökull. The trend then altered
again and Gljúfurárjökull retreated in the years 1994–2005.
During the period 1898–1946, the snout of Gljúfurárjökull
retreated 635 m, almost two-thirds of the total distance from the
LIA maximum (1898–1903) to 2005 (Figures 2 and 3), at an aver-
age rate of 13.2 m yr−1 (Table 1). The rise of the snout during that
period (46 m) was almost half of the total rise. By 1985, the retreat
and ascent since 1898 was almost the total for the 1898–2005
period. However, the velocity of the retreat in 1946–1985 was
lower than in 1898–1946. The 1994 aerial photograph reveals a
change in this trend, with a snout position 20 m more advanced
compared with 1985 (Figure 2). Nevertheless, from 2000
onwards, there was a slow but continuous retreat.
The trend in Western Tungnahryggsjökull during the first half
of the 20th century was a more rapid retreat, showing the highest
average rates of the whole period (19.5 m yr−1). By 1946, this
glacier had retreated almost 90% of the total recorded between the
LIA maximum (1868) and 2005 (Table 1). In the 1946 photo-
graph, this significant retreat of the ice reveals two large moraines
in the centre of the deglaciated area. The snout retreat slowed
down considerably during the second half of the century, espe-
cially in 1985 (1.5 m yr−1). By this date, the aerial photograph
shows a complex terminus covered with debris, with an uneven
retreat, from 60 m in the centre to 150–170 m on the margins, and
a vertical rise of more than 200 m since 1946. The 1994 aerial
photograph shows a similar snout, although with an advance in
the western sector of ≈40 m and a retreat in the eastern sector of
≈20 m (Figure 2). In 2000, the snout, still covered with debris,
retreated mainly in the centre. The glacier then continued to
retreat, although more slowly than Gljúfurárjökull (6.4 m yr−1)
preserving the debris-covered snout (Figures 2 and 3).
Just as in the glaciers described above, the retreat of the East-
ern Tungnahryggsjökull from its LIA position was more intense
during the first half of the 20th century (Table 1), and in 1946 its
snout was only 200 m from its current position. The snout then
Fernández-Fernández et al. 5
continued to retreat more slowly and by 1985 had already lost its
most westerly tongue (Figure 2), where the margin retreated
more than 400 m. The 2000 aerial photograph shows that an
advance of at least 41 m had taken place since 1985. Neverthe-
less, between 2000 and 2005, the snout retreated 17 m, even
more slowly than Western Tungnahryggsjökull.
Figure 2. Evolution of the glacier snouts. The greatest retreat took place between the LIA maximum and 1946 and was especially significant in
the Tungnahryggsjökull. This figure is available in colour in the online version.
LIA: ‘Little Ice Age’ maximum.
Figure 3. Snout and moraine positions from field observations. (a, b) The Gljúfurárjökull and (d) Eastern Tungnahryggsjökull LIA moraines
are easily recognized from their sharp-crested shape. The debris cover on the (c) Western Tungnahryggsjökull determines its complex snout
evolution. This figure is available in colour in the online version.
LIA: ‘Little Ice Age’ maximum.
6 The Holocene
Evolution of ice – area and volume of the glaciers
During the LIA maximum, the total surface area of the three gla-
ciers exceeded 18 km2, with almost half corresponding to the
Western Tungnahryggsjökull. From then until 2005, the glaciers
lost a quarter of their surface area (Table 2), with almost 20% lost
during the first half of the 20th century. In 1985, the loss rate was
considerably reduced, and slight increases in the surface area of
Gljúfurárjökull and Western Tungnahryggsjökull occurred in
1994 (Table 2; Figure 2). Since 2000, the surface loss of the gla-
ciers has not reached 2%.
A third of the LIA maximum ice volume had been lost by
2005. The greatest volume loss (25%) occurred between the LIA
maximum and 1946. The most intense volume loss rate was in
Western Tungnahryggsjökull, around 2.5 km3 yr−1 10−3 (Table 3;
Figure 4). During the second half of the 20th century, the losses
were lower, with maximum average 7.8% (Table 3), although in
1985 the Eastern Tungnahryggsjökull had lost around 15% com-
pared with 1946. In 1994, the Gljúfurárjökull and Western Tung-
nahryggsjökull volumes increased. However, the reduction in
volume continued from 2000 onwards; the loss rate intensified
during 2000–2005 with values similar to or even higher than
those of the first half of the 20th century (Table 3).
Evolution of the ELA
Applying the AAR and AABR methods obtains a mean ELA of
≈1010 m during the LIA maximum and a rise of 40–50 m in the
period analysed between LIA maximum and 2005 (Table 4).
Using the AAR method, the greatest rise in the ELA (29 m)
occurred between the LIA maximum and 1946, coinciding with
the most important snout retreat and the greatest surface area and
volume losses. From 1946 to 1985, there was a smaller rise (10
m), slightly more intense in Gljúfurárjökull and Western Tung-
nahryggsjökull. Although 1994 showed a trend shift advancing,
the ELAs for the two glaciers remained stagnant. Since 2000, the
ELA has remained practically stable around 1050 m, Neverthe-
less, a sharp intensification can be clearly seen in the ELA rise
ratio in the last period 2000–2005, with a mean rate higher than in
the period between the LIA maximum and 1946 (Table 4). The
results obtained using the AABR method were reasonably close
to those obtained using the AAR method, with maximum differ-
ences of ±10 m (Table 4).
The MAAT calculated for Akureyri (1882–2014) and Öxnad-
alsheiði (2000–2014) data series was 3.35°C and 0.97°C, respec-
tively. A lapse rate of 0.66°C 100 m−1 was obtained from the
MAAT data series for the common period 2000–2014. Regression
analysis of the two MAAT series for the period 2000–2014
showed strong correlation (r = 0.79; n = 15), which enabled a first
approximation of the Öxnadalsheiði series reconstruction for the
period 1882–2000. The least squares equation used was y =
0.9092x − 3.0223 (r2 = 0.63), with an overall average (MAAT =
0.02°C) only slightly different from the result obtained using
extrapolation of the lapse rate (MAAT = −0.05°C).
Using the Akureyri temperature series and the 5-year run-
ning-means deviation compared with the overall series average,
nine homogeneous periods were identified (Figure 5; Table 5).
Thus, four cold periods with negative deviations (1882–1924,
1951–1955, 1966–1973 and 1979–1986) and five warm periods
with positive deviations (1925–1950, 1956–1965, 1974–1978,
1987–1999 and 2000–2014). The MAAT was 2.5°C during the
period 1882–1924, coinciding with the end of the LIA. This
cold period ended in the early 1920s, with a sharp temperature
rise and MAAT ≈4°C, maintained until the mid-1960s (Figure
5). This warm period was interrupted by brief cooling, with
MAAT ca. 3.6°C, between 1950 and 1955. The temperature
increase between the late LIA (1882–1924) and the period
1925–1950 was 1.4°C. However, the MAAT fell to 2.9°C
between 1966 and 1986, marking the first cold period with
negative deviations since the end of the LIA, interrupted by
higher MAAT (3.6°C) between 1974 and 1978. Finally, the cli-
mate trend shifted again to warmer conditions during the period
1987–2014. Two sub-periods were identified; during the first
period (1987–2002), the MAAT was 3.8°C, while in 2003 an
abrupt warming of 0.6°C occurred, marking the onset of the
warmest period with an MAAT of 4.4°C from 2003 to 2014.
Table 1. Glacier advance/retreat and snout elevation shift from the LIA maximum. Values in bold represent glacier advances.
Distance from the LIA maximum position (m)
Glaciers LIA 1946 1985 1994 2000 2005 Total retreat
Gljúfurárjökull – 635 910 890 916 993 993
Tungnahryggsjökull (W) – 1524 1584 1610 1703 1735 1735
Tungnahryggsjökull (E) – 1027 1298 – 1257 1274 1274
Average – 1062 1264 – 1292 1334 1334
Advance/retreat rate (m yr−1)
Glaciers LIA LIA–1946 1946–1985 1985–1994 1994–2000 2000–2005 LIA–2005
Gljúfurárjökull – −13.2 −7.1 2.2 −4.3 −15.4 −9.3
Tungnahryggsjökull (W) – −19.5 −1.5 −2.9 −15.5 −6.4 −12.7
Tungnahryggsjökull (E) – −13.2 −6.9 – – −3.4 −9.3
Average – −15.3 −5.2 – – −8.4 −10.4
Glacier snouts elevation (m a.s.l.)
Glaciers LIA 1946 1985 1994 2000 2005 ↑LIA–2005
Gljúfurárjökull 512 558 594 591 593 622 110
Tungnahryggsjökull (W) 540 741 786 759 779 793 253
Tungnahryggsjökull (E) 597 679 718 – 705 711 114
Average 550 659 699 – 692 709 159
LIA: ‘Little Ice Age’.
Fernández-Fernández et al. 7
The Akureyri climatological log shows that increases in the
MAAT and Ts were 1.9°C and 1.5°C, respectively, between the
LIA and the present (Table 5). The average temperature calculated
from the reconstructed Öxnadalsheiði data series suggests that the
MAAT was below freezing level above 500 m a.s.l. in the interior
of the Tröllaskagi peninsula during the late LIA cold period,
1966–1973 and 1979–1986.
As regards precipitation, the mean annual value for the 1950–
2014 period at Akureyri is 515 mm, while in the winter/accumula-
tion season (October–April) it is 357 mm, that is, 69% of the total.
Running mean analysis (Figure 6c) showed below average values
from the late 1950s to the late 1960s (ca. 250 mm minima) and
from the late 1970s to the early 1980s (ca. 200 mm minima in
1980). From the mid-1980s onwards, winter precipitation has
been above the average, reaching maxima in the early 1990s (ca.
430 mm). Since then, winter precipitation has been relatively
regular and close to average, although an increasing trend started
in the mid-2000s. Individual values peaked in 1989 and 2014, at
over 550 mm.
The temperatures (mean annual, May–September, June–July–
August) of the above logs, averaged for the phases identified, and
later extrapolated to the successive ELAs through the lapse rate of
0.66°C 100 m−1 (Table 6), were input into the glacio-climatic
models (Ballantyne, 1989; Braithwaite, 2008; Ohmura et al.,
1992). The Ballantyne (1989) model predicted winter precipita-
tion of 2159 mm at the 2005 mean ELA which supposes an
increase of 19% 100 m−1 during the winter season if the Akureyri
mean winter precipitation over the 30-year period 1976–2005
Table 2. Ice surface evolution from the LIA maximum in the Gljúfurárjökull and Tungnahryggsjökull glaciers. Values in bold represent area gains.
Glaciers LIA 1946 1985 1994 2000 2005 ↓LIA–2005
Gljúfurárjökull 4.372 3.540 3.407 3.413 3.384 3.335 1.038
Tungnahryggsjökull (W) 8.735 6.939 6.689 6.700 6.604 6.512 2.222
Tungnahryggsjökull (E) 5.348 4.532 4.035 – 4.054 3.940 1.408
Total 18.455 15.011 14.131 – 14.041 13.787 4.668
Area gain/loss rate (km2 yr−1)
Glaciers LIA LIA–1946 1946–1985 1985–1994 1994–2000 2000–2005 ↓LIA–2005
Gljúfurárjökull – −0.017 −0.003 0.001 −0.005 −0.010 −0.010
Tungnahryggsjökull (W) – −0.023 −0.006 0.001 −0.016 −0.018 −0.016
Tungnahryggsjökull (E) – −0.010 −0.013 – – −0.023 −0.010
Area gain/loss (%)
Glaciers LIA LIA–1946 1946–1985 1985–1994 1994–2000 2000–2005 ↓LIA–2005
Gljúfurárjökull – −19.04 −3.75 0.18 −0.86 −1.46 −23.73
Tungnahryggsjökull (W) – −20.55 −3.61 0.16 −1.43 −1.39 −25.44
Tungnahryggsjökull (E) – −15.27 −10.96 – – −2.79 −26.33
Total – −18.67 −5.86 – – −1.81 −25.29
Table 3. Ice volume evolution from the LIA maximum in the Gljúfurárjökull and Tungnahryggsjökull glaciers. Values in bold represent volume
Glaciers LIA 1946 1985 1994 2000 2005 ↓LIA–2005
Gljúfurárjökull 0.278 0.208 0.197 0.198 0.195 0.191 0.086
Tungnahryggsjökull (W) 0.720 0.524 0.498 0.500 0.490 0.481 0.239
Tungnahryggsjökull (E) 0.367 0.292 0.249 – 0.250 0.241 0.126
Total 1.364 1.024 0.944 – 0.936 0.913 0.451
Volume gain/loss rate (km3 yr−1 10−3)
Glaciers LIA LIA–1946 1946–1985 1985–1994 1994–2000 2000–2005 ↓LIA–2005
Gljúfurárjökull – −1.459 −0.273 0.055 −0.387 −0.781 −0.808
Tungnahryggsjökull (W) – −2.502 −0.663 0.125 −1.632 −1.863 −1.744
Tungnahryggsjökull (E) – −0.958 −1.104 – – −1.912 −0.918
Volume gain/loss (%)
Glaciers LIA LIA–1946 1946–1985 1985–1994 1994–2000 2000–2005 ↓LIA–2005
Gljúfurárjökull – −25.21 −5.12 0.25 −1.17 −2.00 −31.10
Tungnahryggsjökull (W) – −27.12 −4.93 0.23 −1.96 −1.90 −33.22
Tungnahryggsjökull (E) – −20.38 −14.76 – – −3.82 −34.30
Total – −24.92 −7.77 – – −2.43 −33.08
LIA: ‘Little Ice Age’ maximum.
8 The Holocene
(357 mm) is taken into account. This result suggests an increase
of ca. 50% (714 mm) in the winter precipitation compared with
around 1445 mm during the late LIA (1882–1924). The results
indicate that when the Ts increased by 1.2°C since the LIA, the
winter precipitation increased by 714 mm at the ELA. In turn, the
Ohmura et al. (1992) model, with an estimated annual precipita-
tion of 2022 mm at the 2005 mean ELA, suggests a smaller
increase compared with the late LIA (1578 mm) of around 28%
(444 mm). If the estimated annual precipitation for 2005 is com-
pared with that recorded in Akureyri in the period 1976–2005
(517 mm), the pluviometric gradient obtained is lower, with an
increase of 14% 100 m−1. Finally, the degree-day model (Braith-
waite, 2008) obtained 1741 mm annual precipitation at the ELA
in 2005 (Table 6), an increase of 518 mm (42%) since the LIA
(1223 mm). Using the degree-day model, the vertical pluviomet-
ric gradient obtained was lower than with the other models, with
13% 100 m−1. This model also estimated precipitation much more
sensitive to temperature variations (Table 6). For example, the
degree-day model estimates a reduction of 565 mm in precipita-
tion from 1946 to 1985, because of a drop of 1.1°C in the MAAT
between these two dates.
From the LIA to the present day, MAAT and Ts at the ELA
increased by 1.7°C and 1.2°C, respectively, while in Akureyri the
rise was 1.9°C and 1.5°C, respectively. These values are much
higher than the temperature increase deduced from the rise of the
ELA (0.3°C). For most of the dates, the MAAT in the glacier
snouts remained close to freezing level. In Tungnahryggsjökull,
the MAAT of the snouts was below freezing level at all the dates
and fell below −1°C and −2°C in the coldest periods of the LIA
and 1985, respectively. In the Gljúfurárjökull snout, apart from
these cold periods, the MAAT was positive, around 0.5–0.6°C
Interpretation of the results
The results of this research show a gradual climate warming from
the end of the LIA, as well as a regressive trend for the northern
Iceland glaciers. This process was not uniform, with considerable
temperature variations in this region (Einarsson, 1991) which led
to important changes in the debris-free glaciers studied.
The most important retreat of the Tröllaskagi glaciers between
LIA maximum and the present occurred during the first half of the
20th century. The study of the three glaciers presented here shows
that most of the glacier snout retreat, area reduction and volume
loss had already occurred by 1946; a similar trend was observed
at southeast Vatnajökull outlet glaciers, whose volume loss before
1945 represented the half of the post-LIA total loss (Hannesdóttir
et al., 2015). This is reflected in the combination of the field mea-
surements carried out by Caseldine (1983) and the IGS at Gljú-
furárdalur. However, the figures are different (Table 8). Our
remote measurements on the aerial photographs show that Gljú-
furárjökull retreated 635 m in the period 1898–1946. On the other
hand, the retreat reported by Caseldine (1983) would be at least
450 m if the retreat from the LIA maximum to 1915–1917 (>250
m) and the retreat during the 1930s (200 m) are considered. The
key to this glacier response is found in four main factors: (1) the
sharp 1.4°C rise of the MAAT and 1.2°C rise in the Ts (Akureyri)
between the cold period at the end of the LIA (1882–1924) and
the warm period 1925–1950 (Figure 5). (2) Warm conditions with
MAAT ≈4°C and Ts = 9°C were maintained between 1925 and
1950 (Böðvarsson, 1955). (3) The predominant south-westerly
airflow after 1920 proposed by Kirkbride (2002), which kept
summers warm and caused increased ablation. (4) Other later cold
periods did not last longer than 10 years (Caseldine, 1985b). This
sharp increase in temperature triggered an ELA rise of ≈30 m
compared with the ELA during the LIA maximum. Increased win-
ter precipitation from the LIA maximum (Table 6) did not appear
to have a major impact on the termini variation at that moment,
but probably in further advances (e.g. mid-1970s to mid-1980s, or
early 1990s) by increasing the mass flux and reducing the termini
retreat rate (Kirkbride, 2002). In this context, the Western Tung-
nahryggsjökull glacier seems to be the most sensitive to the
increased temperature of the three glaciers, as it presents the high-
est values for retreat rates, area and volume losses, and the great-
est ELA rise.
Stötter et al. (1999) indicate that the coldest period after the
LIA was from the early 1960s to the mid-1970s, when tempera-
tures fell to levels equivalent to the warmest recorded in the 19th
century. This cooling is the reason given by Caseldine (1983,
1985a, 1985b, 1988) to explain the advance of the Gljúfurárjökull
between the mid-1970s and the mid-1980s, which can be clearly
seen in Figure 6. This would suggest a time response to Ts cooling
close to 10 years. The retreat from 1946 to 1985 calculated using
IGS field measurements (322 m) appears to be overestimated if
we consider our results of 275 m for the same period (Table 8).
This discrepancy can be explained by technical issues such as the
accuracy of the georeferencing (RMS error), the change in field
measurement procedures (estimates; see Sigurðsson et al., 2007)
and the vague and scarce (incomplete) data about termini varia-
tions prior to the 1950s provided by Caseldine and Cullingford
(1981) and Caseldine (1983). So the GIS measurements over a
photograph with snow-free termini (taken at the end of the abla-
tion season) and properly georeferenced can provide the best
results, avoiding estimates when fieldwork is not possible. In this
Figure 4. Evolution of retreat rates and area and volume loss
during the different periods analysed. From 2000 to 2005, the rates
are close to those recorded in the first half of the 20th century.
LIA: ‘Little Ice Age’ maximum; ND: no data.
Fernández-Fernández et al. 9
paper, two points are mentioned which may clarify the glacial
evolution after the 1980s: (1) the 1994 aerial photograph reveals
a more advanced position of the Gljúfurárjökull compared with
1985 and (2) between 1979 and 1986 another cold period is iden-
tified (with temperatures not as cold as in the previous one, but
separated from it by a brief warm 4-year period), characterized
especially by a fall in the Ts below 8.5°C and even 8°C in the
Akureyri station (Figures 5 and 6b). This cold period between
1979 and 1986 seems to have been the continuation of the cooling
which started in the early 1960s and the reason why Gljúfurár-
jökull continued to advance after 1985. However, by the year
1994 the advance had ended, because of (1) the low advance rate
of Gljúfurárjökull inferred from the positions of the snout in 1985
and 1994 (2.2 m yr−1) compared with the advance velocities
occurring in previous years (Caseldine, 1983, 1985a, 1988; Casel-
dine and Cullingford, 1981) and (2) the time lag (8 years) from
1994 to the end of the cooling in 1986. These findings confirm the
Ts value of 8–8.5°C at Akureyri proposed by Björnsson (1971)
and Caseldine (1985b) as the threshold for the trend shift in the
glacial mass balance and also suggest that less than 10 years with
cold summers may be required for the glacier advance. However,
the increase in winter precipitation (671 mm; see Supplementary
Material, available online) obtained in this study at the Gljúfurár-
jökull ELA between 1985 and 1994, and that obtained in Akureyri
also seem to explain the glacier advance during the 1990s when
the Ts in Akureyri reached over 8.5°C. It is reasonable to assume
that a greater increase in winter precipitation reduced the number
of cold summers required for the glacier snout advance. Such a
clear advance was not observed in 1994 in the Western Tungnah-
ryggsjökull, because of the difficulty in identifying precisely the
lateral margins of the glacier in snow-covered areas. However,
different sectors of the snout advanced or retreated compared with
1985. The explanation may be found in the uneven debris cover
and the insulating effect it exerted on the ice.
From the mid-1980s, there was a gradual rise in the Ts (Casel-
dine, 1988), which triggered the retreat of the glaciers in 2000
from their position in 1994. A sharp temperature rise occurred
around the year 2003, which intensified the reduction in glacial
volume and the ELA rise at rates comparable to those in the first
half of the 20th century. Nevertheless, the snout retreat did not
accelerate dramatically. The retreat rate intensified in the period
2000–2005 compared with 1994–2000, but did not reach the rates
recorded before 1946 (Table 1). The glacier evolution in recent
years is characterized by continuous retreat, which can be
explained by the high Ts above 9°C since 2003. According to
Caseldine and Stötter (1993), the effect of the climate warming
observed from the LIA to the mid-1980s was a 50-m ELA rise in
the glaciers in northern Iceland. This value is similar to the 40–50
m ELA rise obtained in this study for Gljúfurárjökull and Tung-
nahryggsjökull between the LIA maximum and 2005. The AAR
(0.67) and AABR (1.5) methods applied in this paper to calculate
the ELAs obtained homogeneous results, suggesting a good adap-
tation of the application of AABR = 1.5, representative of the
Norwegian glaciers (Rea, 2009), to the debris-free glaciers of
According to Caseldine and Stötter (1993), although the ELA is
the parameter which best expresses the relationship between gla-
ciers and climate, the use of its rise or fall to estimate Ts variations
may lead to significant underestimation in the results. This has
Table 4. ELAs and ELA changes over variable periods calculated by AAR and AABR methods for the Gljúfurárjökull and Tungnahryggsjökull
ELA-AAR (0.67; m a.s.l.)
Glaciers LIA 1946 1985 1994 2000 2005 ↑LIA–2005
Gljúfurárjökull 954 974 985 982 984 988 34
Tungnahryggsjökull (W) 1046 1082 1092 1090 1091 1094 48
Tungnahryggsjökull (E) 1029 1061 1069 – 1071 1073 44
Average 1010 1039 1049 – 1049 1052 42
ELA-AAR rise rate (m yr−1)
Glaciers LIA LIA–1946 1946–1985 1985–1994 1994–2000 2000–2005 ↑LIA–2005
Gljúfurárjökull – 0.42 0.28 −0.33 0.33 0.80 0.32
Tungnahryggsjökull (W) – 0.46 0.26 −0.22 0.17 0.60 0.35
Tungnahryggsjökull (E) – 0.38 0.21 – – 0.40 0.32
Average – 0.42 0.25 – – 0.60 0.33
ELA-AABR (1.5 ± 0.4; m a.s.l.)
Glaciers LIA 1946 1985 1994 2000 2005 ↑LIA–2005
Gljúfurárjökull 960 ± 20 975 ± 15 986 ± 20 988 ± 15 990 ± 15 994 +15/−10 34
Tungnahryggsjökull (W) 1047 + 20/−15 1093 ± 10 1103 ± 10 1101 ± 10 1102 +10/−5 1105 ± 10 58
Tungnahryggsjökull (E) 1020 + 20/−15 1062 ± 15 1075 ± 10 – 1072 +15/−10 1079 ± 10 59
Average 1009 ± 36 1043 ± 50 1055 ± 50 – 1055 ± 47 1059 ± 47 50
ELA-AABR rise rate (m yr−1)
Glaciers LIA LIA–1946 1946–1985 1985–1994 1994–2000 2000–2005 ↑LIA–2005
Gljúfurárjökull – 0.31 0.28 0.22 0.33 0.80 0.32
Tungnahryggsjökull (W) – 0.59 0.26 −0.22 0.17 0.60 0.42
Tungnahryggsjökull (E) – 0.54 0.33 – – 1.40 0.43
Average – 0.48 0.29 – – 0.93 0.39
ELA: equilibrium line altitude; AAR: accumulation area ratio; AABR: area altitude balance ratio; LIA: ‘Little Ice Age’ maximum.
10 The Holocene
been proved in this study, observing that if the maximum depres-
sion of the ELA (41 m) and the lapse rate of 0.66°C 100 m−1 are
taken into consideration, the rise in the ELA would indicate a
lower Ts increase, approximately 0.3°C, assuming the precipita-
tion remained constant. However, the Ts rise recorded in Akureyri
(1.5°C) or the rise extrapolated at the ELA (1.2°C) between the
LIA maximum and 2005 is considerably higher. This shows that
in addition to temperature, other factors may have been decisive
in the glacial evolution, such as precipitation and wind (Caseldine
and Stötter, 1993).
The model applied by Caseldine and Stötter (1993) and Stöt-
ter et al. (1999) suggests that the precipitation in northern Iceland
during the LIA was significantly lower than at the present day.
Applying the same model in this study shows a similar trend. The
model designed for Norwegian glaciers (Ballantyne, 1989) pre-
dicted winter precipitation of 1445 mm at the mean ELA (at
Figure 5. Evolution of (b) mean annual air temperature (MAAT), mean ablation season air temperature (Ts) at (a) Akureyri and MAAT
reconstruction at (c) Öxnadalsheiði. The coloured numbers in the middle of the periods are the mean value of MAAT/Ts for each period. This
figure is available in colour in the online version.
Table 5. Cold/warm periods at Akureyri and Öxnadalsheiði weather stations and seasonal values.
Period Type Akureyri Öxnadalsheiði
Air temperature (°C)
Annual Ablation season Three-month summer Annual range Annual
1882–1924 Coldest 2.52 7.86 9.41 15.59 −0.73
1925–1950 Warm 3.95 9.05 10.38 14.80 0.57
1951–1955 Cold 3.62 8.60 9.96 14.71 0.27
1956–1965 Warm 3.79 8.38 9.52 14.56 0.43
1966–1973 Cold 2.81 8.12 9.47 15.44 −0.46
1974–1978 Warm 3.61 8.73 10.31 15.75 0.38
1979–1986 Cold 2.93 7.91 9.85 14.84 −0.35
1987–2002 Warm 3.77 8.95 10.27 14.71 0.45
2003–2014 Warmest 4.44 9.32 10.92 13.89 0.96
Source: Icelandic Met Office (IMO) and Icelandic Road and Coastal Administration.
Temperatures at Öxnadalsheiði prior to 2000 were reconstructed through the least squares equation obtained from the regression analysis between
Akureyri and Öxnadalsheiði MAAT series for the common period 2000–2014.
Fernández-Fernández et al. 11
altitude 1010 m) during the LIA maximum, and an increase of
more than 714 mm (twice the modern winter precipitation at
Akureyri) from then to 2005. Caseldine and Stötter (1993) esti-
mated practically identical precipitation at the ELA during the
LIA (1450 mm) and an increase of around 600 mm until the mid-
1980s for the Tröllaskagi glaciers. Dahl and Nesje (1992) using
the same model calculated a relatively similar increase of 690
mm in Nordfjord (Western Norway) since the LIA, where the
Figure 6. (a) Relationship between the variations in Gljúfurárjökull snout (taken from the Iceland Glaciological Society, 2016), (b) ablation
season temperature, (c) winter precipitation at Akureyri and (d) winter NAO index since 1950. Black dotted lines show 5-year running mean
(temperature and precipitation) and LOESS regression in the NAO index (modified from Cropper etal., 2015). Red points in (a) are the years
with marginal measurements. There is a clear relationship between the 1980s advance and the previous cooling of the mean ablation season
air temperature (Ts). The winter precipitation evolution shows a curve parallel to that of the NAO index (high winter precipitation, positive
NAO index phase), suggesting a connection between the NAO mode and the precipitation, especially in the early 1980s and 1990s. This figure
is available in colour in the online version.
Table 6. Temperature and precipitation at the ELA calculated for each year: comparison between different models. All models agree on a
wetter climate at the present day than during the LIA maximum.
Period LIA 1946 1985 1994 2000 2005 ↑LIA–2005
Mean air temperature (°C)
Ablation season (May–September) 1.35 2.35 1.13 – 2.18 2.53 1.18
Three-month summer (June–July–August) 2.90 3.67 3.08 – 3.50 4.13 1.23
Mean annual −4.00 −2.75 −3.84 – −3.00 −2.35 1.65
Precipitation (mm water equivalent)
Winter (Ballantyne, 1989 model) 1445 2029 1344 – 1913 2159 714
Annual (Ohmura etal., 1992 model) 1578 1854 1641 – 1791 2022 444
Annual (Braithwaite, 2008 model) 1223 1713 1148 – 1552 1741 518
ELA: equilibrium line altitude; LIA: ‘Little Ice Age’.
12 The Holocene
current climate is wetter and milder (Olden, 78 m a.s.l., Ts =
12.2°C, winter precipitation = 812 mm; see Dahl and Nesje,
1992) because of the influence of the North Atlantic Drift and the
frequent frontal precipitation associated with the polar front
position. The maritime location of the Tröllaskagi glaciers and
those (Norwegian) used to devise the Ballantyne (1989) model,
and also the good adaptation of a Norwegian AABR for ELA
calculation, may postulate this model as the most suitable of the
three to infer temperature and precipitation changes in the Tröl-
laskagi glaciers. This model gives higher values of precipitation
than the other models (e.g. Ohmura et al., 1992) at warmer dates
(e.g. 1946, 2000 and 2005) because of the exponential nature of
its formula (see equation (5) in section ‘Methods’). This deter-
mines that equal values of temperature as input will give higher
output precipitation in the Norwegian model than the Ohmura
et al. (1992) one. Only at the coldest dates (e.g. LIA maximum
and 1985) were the results inverse with higher precipitation in
the Ohmura et al. (1992) model, when the Ts was far below the
mean 3-month summer temperature.
The precipitation pattern observed in this paper, lower during
the LIA cold period and higher during the warm periods, fully
coincides with the model proposed by Stötter et al. (1999) for
northern Iceland. This is also coherent with a lower ocean surface
temperature (Geirsdóttir et al., 2009) linked to the greater pres-
ence of Arctic sea ice (Ogilvie, 1984, 1996; Ogilvie and Jónsdót-
tir, 2000; Ogilvie and Jónsson, 2001) which weakened the
convective processes (Lehner et al., 2013). Nesje and Dahl (2003)
and Holmes et al. (2016) link the precipitation changes to the
variations in the North Atlantic Oscillation (NAO) phase (Hurrell,
1995) and in the position of the polar front. In this sense, the dates
at which high precipitation was obtained in this research (e.g.
1946, 2005) would correspond to a positive NAO phase (Figure
6d) which reinforced the zonal flow of the westerlies and the
W–SW winds, coinciding with a northwards displacement of the
low-pressure cells and the polar front (Jansen et al., 2016). This
situation facilitated the predominance of warm wet sub-tropical
masses responsible for warm wet winter weather in Iceland
(Holmes et al., 2016). On the contrary, the dates when the calcu-
lated precipitation was lowest (e.g. LIA, 1985) would have coin-
cided with the negative NAO phases (Figure 6d) in which Arctic
air masses predominated as a result of the southward displace-
ment of the polar front and prevailing N–NW winds (Holmes
et al., 2016). This atmospheric configuration would favour cold
dry summers (Jansen et al., 2016). Thus, it is reasonable to sup-
pose that the variations in precipitation between the cold and
warm periods may also be explained by conditions that either hin-
dered or facilitated convection, respectively (Burn et al., 2016).
However, extra accumulation from snow-blowing should also be
taken into account in the corrie glaciers studied. In a deeply
incised cirque surrounded by a plateau, the snow may deflate
from the plateau and accumulate in the cirque, either by direct
accumulation or avalanching from the cirque walls (Dahl and
Nesje, 1992; Sissons and Sutherland, 1976; Sutherland, 1984).
Although most Tröllaskagi glaciers are surrounded by sharp
peaks, ridges and summits, they receive snow blown from far out
on the plateau mountains. Caseldine and Stötter (1993) suggested
that up to 35% of the total winter accumulation could be attrib-
uted to the processes explained above (Tangborn, 1980). Based on
this relationship between accumulation and snow-blowing, previ-
ous authors (Caseldine and Stötter, 1993; Dahl and Nesje, 1992)
proposed that the changes in winter accumulation may also reflect
changes in the direction of the prevailing wind. According to this
reasoning, the increase in precipitation between the LIA and 2005
could be explained by a current predominance of the wind from
the plateau (with snow-blowing), coherent with the changes in
atmospheric circulation explained above. However, the results
from the wind data processing (see Supplementary Material,
available online) do not provide strong support for the changes in
wind directions based on differences of winter accumulation, at
least in the present day, with NW–NE (36%) and SW–SE (35%)
as the dominating wind directions above 10 m s−1 during winter
(October–April) at Grimsey. Further research on wind and snow-
fall is required to shed light on this issue.
The NAO exerts control over mass balance by influencing
temperature and precipitation anomalies (Marzeion and Nesje,
2012), and therefore, a link between NAO phases and termini
variations has been suggested on the literature. Bradwell et al.
(2006) found that Lambatungnajökull (north-eastern outlet of
Vatnajökull, South Iceland) advanced during negative NAO
phases and linked this to positive mass balances. On the contrary,
Nesje et al. (2000) linked negative mass balance with negative
NAO indices. Nevertheless, such relationships are not so clear, at
least in Gljúfurárjökull. The continuous retreat from 1950 until
the late 1970s in Figure 6a is mostly characterized by negative
NAO indices, so it is reasonable to think that, at least for that
period, there may have been a link between negative NAO index
and negative mass balance. However, this relationship during the
1980s advance is less clear as it coincides with a negative NAO
phase interrupted by 3 years with positive indices (1981, 1983
and 1984). We found that this advance started in the mid-1970s,
that is, 20 years after the major reversal of the NAO index mode
(negative to positive) occurred in 1955 (see Figure 6), in good
agreement with the reaction times at the decadal scale reported by
Kirkbride (2002) and the theoretical calculations of glacier
response times for small mountain glaciers (Jóhannesson et al.,
1989). The glacier termini/winter–NAO relationship at this gla-
cier is even fuzzier if we consider the advanced position in 1994
coinciding with positive indices (since 1989) and the retreat since
the late 1990s over two positive and negative NAO phases. Fur-
thermore, fuller and more detailed research on the past and future
behaviour of the glacier termini and mass balance is needed to
determine the future behaviour of the glaciers and elucidate the
relationship with the NAO.
This research has shown the high sensitivity of the debris-free
Gljúfurárjökull and Tungnahryggsjökull glaciers to climatic
Table 7. Mean annual air temperature (MAAT) extrapolated to the
Gljúfurárjökull and Tungnahryggsjökull snouts.
Mean annual air temperature (°C)
Glacier LIA 1946 1985 1994 2000 2005
Gljúfurárjökull −0.71 0.42 −0.83 0.02 0.01 0.49
Tungnahryggsjökull (W) −0.89 −0.78 −2.10 −1.09 −1.22 −0.64
Tungnahryggsjökull (E) −1.27 −0.38 −1.65 – −0.73 −0.10
Average −0.96 −0.25 −1.53 –−0.65 −0.08
LIA: ‘Little Ice Age’ maximum.
Table 8. Glacier termini variations of Gljúfurárjökull: comparison
between remote and field measurements.
Period Snout variation measurements (m)
Remote (GIS) Field (IGS)
1898–1946 −635 −450a
1946–1985 −275 −322
1985–1994 20 39
1994–2000 −26 −25
2000–2005 −77 −63
Total −993 −821
GIS: Geographical Information System; IGS: Icelandic Glaciological
IGS measurements have been summarized for the periods analysed.
aInferred from Caseldine and Cullingford (1981).
Fernández-Fernández et al. 13
fluctuations (Häberle, 1991; Kugelmann, 1991), especially to the
Ts (Eythorsson, 1935; Liestøl, 1967; Ohmura et al., 1992). Conse-
quently, they experienced an important retreat during the periods
characterized by warm summers and advanced during the short
periods with cold summers, even when the duration of these peri-
ods was shorter than the 10 years proposed by Caseldine (1985b).
The debris-free Gljúfurárjökull and Tungnahryggsjökull are
important indicators of climate change, as the absence of debris
and reduced dimensions mean they are highly sensitive to climate
fluctuations. As a result, the abrupt climatic transition of the early
20th century and the 25-year warm period 1925–1950 triggered
the most important glacier retreat and volume loss since the end
of the LIA; meanwhile, cooling during the 1960s, 1970s and
1980s altered the trend, with glacier snout advances.
Calculating the ELAs for the Tröllaskagi glaciers using the AAR
and AABR methods showed a good fit of the AABR = 1.5 proposed
for Norwegian glaciers. Analysis of the relationships between ELA
evolution and climatic data also revealed that the glacier response
depends not only on the Ts but also on other factors such as precipi-
tation. The models applied, especially the one obtained from Nor-
wegian glaciers, show a precipitation increase of more than 700 mm
since the LIA, compatible with an increase in the surface tempera-
ture of the North Atlantic and with a change in the direction of the
prevailing wind, currently from the plateau. Nevertheless, the evolu-
tion of the glaciers in the last 10 years shows an uncertain trend
because of the lack of updated data (except for Gljúfurárjökull),
which may become clearer with further monitoring of the glaciers
over the coming years. The relationship between glacier evolution
and atmospheric circulation patterns remains unclear.
We thank the Icelandic Institute of Natural History and Hólar
University College for their support in the field. We also thank
the anonymous referees, whose valuable comments improved the
quality of the earlier version of the manuscript.
This paper was funded by the projects CGL2012-35858 and
CGL2015-65813-R (Spanish Ministry of Economy and Competi-
tiveness) and Nils Mobility Program (EEA GRANTS), and with
the help of the Research Group of High Mountain Physical Geog-
raphy (Complutense University of Madrid). José María Fernández-
Fernández received a grant from the FPU programme (Spanish
Ministry of Education, Culture and Sport).
Bahr DB, Meier MF and Peckham SD (1997) The physical
basis of glacier volume-area scaling. Journal of Geophysical
Research 102: 20355–20362.
Ballantyne CK (1989) The Loch Lomond readvance on the Isle
of Skye, Scotland: Glacier reconstruction and palaeoclimatic
implications. Journal of Quaternary Science 4: 95–108.
Bergþórsson P (1969) An estimate of drift ice and temperature in
1000 years. Jökull 19: 94–101.
Björnsson H (1971) Bægisarjökull, North Iceland. Results of gla-
ciological investigations 1967–68. Part I. Mass balance and
general meteorology. Jökull 21: 111–118.
Böðvarsson G (1955) On the flow of ice-sheets and glaciers.
Jokull 5: 1–8.
Bradwell T, Dugmore AJ and Sudgen DE (2006) The Little Ice
Age glacier maximum in Iceland and the North Atlantic
Oscillation: Evidence from Lambatungnajökull, southeast
Iceland. Boreas 35: 61–80.
Braithwaite RJ (2008) Temperature and precipitation climate
at the equilibrium-line altitude of glaciers expressed by the
degree-day factor for melting snow. Journal of Glaciology 54:
Brückner E (1886) Die Hohern Tauern und ihre Eisbedeckung.
Zeitschrift des Deutsch-Österreichische Alpenvereins 17:
Brückner E (1887) Die Höhern der Schneelinie und ihre Bestim-
mung. Meteorologische Zeitschrift 4: 31–32.
Brugger KA (2006) Late Pleistocene climate inferred from the
reconstruction of the Taylor River glacier complex, southern
Sawatch Range, Colorado. Geomorphology 75: 318–329.
Burn MJ, Holmes J, Kenedy LM et al. (2016) A sediment-based
reconstruction of Caribbean effective precipitation during the
‘Little Ice Age’ from Freshwater Pond, Barbuda. The Holo-
cene 26: 1237–1247.
Caseldine CJ (1983) Resurvey of the margins of Gljúfurárjökull
and the chronology of recent deglaciation. Jökull 33: 111–118.
Caseldine CJ (1985a) Survey of Gljúfurárjökull and features
associated with a glacier burst in Gljúfurárdalur, Northern
Iceland. Jökull 35: 61–68.
Caseldine CJ (1985b) The extent of some glaciers in Northern
Iceland during the Little Ice Age and the nature of recent
deglaciation. The Geographical Journal 151(2): 215–227.
Caseldine CJ (1987) Neoglacial glacier variations in Northern
Iceland: Examples from the Eyjafjördur area. Arctic and
Alpine Research 19(3): 296–304.
Caseldine CJ (1988) Fluctuations of Gljúfurárjökull, Northern
Iceland 1983–1987. Jökull 38: 32–34.
Caseldine CJ and Cullingford RA (1981) Recent mapping of
Gljúfurárjökull and Gljúfurárdalur. Jökull 31: 11–22.
Caseldine CJ and Stötter J (1993) ‘Little Ice Age’ glaciation of
Tröllaskagi peninsula, northern Iceland: Climatic implica-
tions for reconstructed equilibrium line altitudes (ELAs). The
Holocene 3: 357–366.
Caseldine CJ, Langdon P and Holmes N (2006) Early Holocene
climate variability and the timing and extent of the Holocene
thermal maximum (HTM) in northern Iceland. Quaternary
Science Reviews 25: 2314–2331.
Chen J and Ohmura A (1990) Estimation of Alpine glacier water
resources and their change since the 1870s. Hydrology in
Mountain Regions 193: 127–135.
Crochet P, Jóhannesson T, Jónsson T et al. (2007) Estimating the
spatial distribution of precipitation in Iceland using a linear
model of orographic precipitation. Journal of Hydrometeorol-
ogy 8(6): 1285–1306.
Cropper T, Hanna E, Valente MA et al. (2015) A daily Azores–
Iceland North Atlantic Oscillation index back to 1850. Geo-
science Data Journal 2: 12–24.
Dahl SO and Nesje A (1992) Paleoclimatic implications based
on equilibrium-line altitude depressions of reconstructed
Younger Dryas and Holocene cirque glaciers in inner Nordf-
jord, western Norway. Palaeogeography, Palaeoclimatology,
Palaeoecology 94: 87–97.
Einarsson MÁ (1991) Temperature conditions in Iceland 1901–
1990. Jökull 41: 1–20.
Etzelmüller B, Farbrot H, Guðmundsson Á et al. (2007) The
regional distribution of mountain permafrost in Iceland. Per-
mafrost and Periglacial Processes 18: 185–199.
Eythorsson J (1935) On the variations of glaciers in Iceland. Some
studies made in 1931. Geografiska Annaler 17: 121–137.
Flowers GE, Björnsson H, Geirsdóttir Á et al. (2007) Glacier fluc-
tuation and inferred climatology of Langjokull ice cap through
the Little Ice Age. Quaternary Science Reviews 26: 2337–2353.
Geirsdóttir Á, Gifford HM, Axford Y et al. (2009) Holocene and
latest Pleistocene climate and glacier fluctuations in Iceland.
Quaternary Science Reviews 28: 2107–2118.
Grove JM (1988) The Little Ice Age. London: Methuen.
Grove JM (2001) The initiation of the ‘Little Ice Age’ in regions
round the North Atlantic. Climatic Change 48: 53–82.
14 The Holocene
Gudmundsson HJ (1997) A review of the Holocene environmen-
tal history of Iceland. Quaternary Science Reviews 16: 81–92.
Häberle T (1991) Holocene glacial history of the Hörgádalur area,
Tröllaskagi, northern Iceland. In: Maizels JK and Caseldine
C (eds) Environmental Change in Iceland: Past and Present.
Dordrecht: Kluwer Academic Publishers, pp. 193–202.
Hannesdóttir H, Björnsson H, Pálsson F et al. (2015) Changes in
the southeast Vatnajökull ice cap, Iceland, between ~1890 and
2010. The Cryosphere 9: 565–585.
Holmes N, Langdom PG, Caseldine CJ et al. (2016) Climatic
variability during the last millennium in Western Iceland from
lake sediment records. The Holocene 26(5): 756–771.
Hurrell JW (1995) Decadal trends in the North Atlantic Oscilla-
tion: Regional temperatures and precipitation. Science 269:
Icelandic Glaciological Society (2016) Glacier termini variations
data. Available at: http://spordakost.jorfi.is (accessed 15 July
Icelandic Meteorological Office (2015) Climatological data.
Available at: http://en.vedur.is/climatology/data/ (accessed 13
Icelandic Road and Coastal Administration (2016) Climatological
data. Available at: http://www.road.is/ (accessed 24 Septem-
Jansen HL, Simonsen JS, Dahl SO et al. (2016) Holocene glacier
and climate fluctuations of the maritime ice cap Høgtuvbreen,
northern Norway. The Holocene 26(5): 736–755.
Jóhannesson H and Sæmundsson K (1989) Geological map of
Iceland. 1:500.000. Bedrock. Reykjavik: Icelandic Institute
of Natural History.
Jóhannesson T, Raymond C and Waddington E (1989) Time-scale
for adjustment of glaciers to changes in mass balance. Journal
of Glaciology 35: 355–369.
Kirkbride MP (2002) Icelandic climate and glacier fluctuations
through the termination of the ‘Little Ice Age’. Polar Geog-
raphy 26(2): 116–133.
Kirkbride MP and Dugmore AJ (2001) Timing and significance
of mid-Holocene glacier advances in northern and central Ice-
land. Journal of Quaternary Science 16: 145–153.
Kirkbride MP and Dugmore AJ (2006) Responses of mountain lee
caps in central Iceland to Holocene climate change. Quater-
nary Science Reviews 25: 1692–1707.
Kirkbride MP and Dugmore AJ (2008) Two millennia of glacier
advances from southern Iceland dated by tephrochronology.
Quaternary Research 70: 398–411.
Koch L (1945) The East Greenland Ice. Medd. Grønland 130: 1–374.
Kugelmann O (1991) Dating recent glacier advances in the Svar-
faðardalur-Skíðadalur area of northern Iceland by means of a
new lichen curve. In: Maizels J and Caseldine C (eds) Envi-
ronmental Change in Iceland: Past and Present. Dordrecht:
Kluwer Academic Publishers, pp. 203–217.
Larsen DJ, Miller GH, Geirsdóttir Á et al. (2011) A 3000-year
varved record of glacier activity and climate change from
the proglacial lake Hvítárvatn, Iceland. Quaternary Science
Reviews 30: 2715–2731.
Lehner F, Born A, Raible CC et al. (2013) Amplified inception of
European Little Ice Age by sea ice–ocean–atmosphere feed-
backs. Journal of Climate 26: 7586–7602.
Liestøl O (1967) Storbreen glacier in Jotunheimen. Norsk Polar-
institutts Skrifter 141: 1–63.
Martin E, Whalley WB and Caseldine C (1991) Glacier fluctua-
tions and rock glaciers in Tröllaskagi, Northen Iceland, with
Special reference to 1946–1986. In: Maizels J and Caseldine
C (eds) Environmental Change in Iceland: Past and Present.
Dordrecht: Kluwer Academic Publishers, pp. 255–264.
Marzeion B and Nesje A (2012) Spatial patterns of North Atlantic
Oscillation influence on mass balance variability of European
glaciers. The Cryosphere 6: 661–673.
National Land Survey of Iceland (2015) Available at: http://www.
lmi.is/en/ (accessed 13 June 2015).
Nesje A and Dahl SO (2003) The ‘Little Ice Age’ – Only tempera-
ture? The Holocene 13(1): 139–145.
Nesje A, Lie Ø and Dahl SO (2000) Is the North Atlantic Oscilla-
tion reflected in Scandinavian glacier mass balance records?
Journal of Quaternary Science 15: 587–601.
Ogilvie AEJ (1984) The past climate and sea-ice record from Ice-
land, Part 1: Data to A.D. 1780. Climatic Change 6: 131–152.
Ogilvie AEJ (1991) Climatic changes in Iceland A.D. c. 865 to
1598. In the Norse of the North Atlantic (presented by G.F.
Bigelow). Acta Archaeologica 61: 233–251.
Ogilvie AEJ (1996) Sea-ice conditions off the coasts of Iceland
A.D. 1601–1850 with special reference to part of the Maun-
der Minimum period (1675–1715). In: Pedersen ED (ed.)
AmS-Varia 25. Stavanger: Archaeological Museum, pp. 9–12.
Ogilvie AEJ (2005) Local knowledge and travellers’ tales: A selec-
tion of climatic observations in Iceland. In: Caseldine C, Rus-
sell A, Hardardottir J et al. (eds) Iceland – Modern Processes
and Past Environments. Amsterdam: Elsevier, pp. 257–287.
Ogilvie AEJ (2010) Historical climatology, climatic change, and
implications for climate science in the twenty-first century.
Climatic Change 100: 33–47.
Ogilvie AEJ and Jónsdóttir I (2000) Sea ice, climate, and Icelan-
dic fisheries in the eighteenth and nineteenth centuries. Arctic
Ogilvie AEJ and Jónsson T (2001) ‘Little Ice Age’ research: A
perspective from Iceland. Climatic Change 48: 9–52.
Ohmura A, Kasser P and Funk M (1992) Climate at the equilib-
rium line of glaciers. Journal of Glaciology 38: 397–411.
Osmaston H (2005) Estimates of glacier equilibrium line altitudes
by the Area × Altitude, the Area × Altitude Balance Ratio and
the Area × Altitude Balance Index methods and their valida-
tion. Quaternary International 138–139: 22–31.
Pellitero R, Rea BR, Spagnolo M et al. (2015) A GIS tool for
automatic calculation of glacier equilibrium-line altitudes.
Computers & Geosciences 82: 55–62.
Radić V, Hock R and Oerlemans J (2007) Volume-area scaling vs
flowline modelling in glacier volume projections. Annals of
Glaciology 46: 234–240.
Rea BR (2009) Defining modern day Area-Altitude Balance
Ratios (AABRs) and their use in glacier-climate reconstruc-
tions. Quaternary Science Reviews 28: 237–248.
Sæmundsson K, Kristjansson L, McDougal I et al. (1980) K-Ar
dating, geological and paleomagnetic study of a 5-km lava
succession in Northern Iceland. Journal of Geophysical
Research 85(7): 3628–3646.
Schomacker A, Krüger J and Larsen G (2003) An extensive late-
Holocene glacier advance of Kötlujökull, central south Ice-
land. Quaternary Science Reviews 22: 1427–1434.
Sigurðsson O, Jónsson T and Jóhannesson T (2007) Relation
between glacier-termini variations and summer temperature
in Iceland since 1930. Annals of Glaciology 46: 170–176.
Sissons JB and Sutherland DG (1976) Climatic inferences from
former glaciers in the south-east Grampian Highlands, Scot-
land. Journal of Glaciology 17: 325–346.
Stötter J (1990) Geomorphologische und landschaftsgeschichtliche
Untersuchungen im Svarfaðardalur-Skiðadalur, Tröllaskagi,
N-Island. Münchener Geographische Abhandlungen 9: 1–166.
Stötter J, Wastl M, Caseldine C et al. (1999) Holocene palaeo-
climatic reconstruction in northern Iceland: Approaches and
results. Quaternary Science Reviews 18: 457–474.
Sutherland DG (1984) Modern glacier characteristics as a basis for
inferring former climates with particular reference to the Loch
Lomond Stadial. Quaternary Science Reviews 3: 291–309.
Tangborn W (1980) Two models for estimating climate-glacier
relationships in the North Cascades, Washington, U.S.A.
Journal of Glaciology 25: 3–21.