Content uploaded by Ursula Enzenhofer
Author content
All content in this area was uploaded by Ursula Enzenhofer on May 31, 2021
Content may be subject to copyright.
Page I
Universität für Bodenkultur Wien
University of Natural Resources and Life Sciences, Vienna
Department of Civil Engineering and Natural Hazards,
Institute of Applied Geology (IAG)
An observational study of the
snow and firn regime
at four arctic glaciers, Svalbard
Masterarbeit / Master thesis
Angestrebter akademischer Grad / In partial fulfilment of the requirements
for the degree of
Diplomingenieur / graduate engineer
eingereicht von / submitted by:
ENZENHOFER, URSULA
Supervisor: Priv.-Doz. Dr. Mergili, Martin
Co-Supervisor: Professor Hodson, Andy
Student number 01140999 09.04.2020
Page II
Page III
Acknowledgements
This master thesis would not have been possible without the financial support from the Research
Council of Norway for the Arctic Field Grant (project Snow & Firn, RiS-ID11152)
Further a huge appreciation for the University Centre in Svalbard, which is an excellent university
which not only teaches magnificent science but also field skills and leadership skills. Special
thanks to logistic department for all the preparation meetings which made the fieldwork safe.
Thanks to Martin Mergili who is a wonderful mentor, an inspiring person with a huge interest and
curiosity for new things in science. Hard working and inspiring to learn from, but always with an
open heart for all the students.
Thanks to Andrew Hodson, for co supervising this thesis. Thanks for always having an open door
to everyone no matter how limited the time might be. Special thanks for paving the path to
extensive fieldwork skills which I could learn from and the endless motivation and energy you
bring into the fieldwork and into research.
Thanks to the polar bear and her cub who showed us that what you think is important is a whole
other thing for someone else. The fieldwork which was interrupted could be finished successfully
another day without any bear in sight but beautiful views for sure.
A huge thanks goes to all the good vibes from all field assistants. It is unimaginable how much
support, help and motivation can come from strangers. This world is full of them and in the end,
I am very proud to call them all friends now. All of them helped me collecting this exceptional
huge amount of data and much more. Especially Molly Peek, Erik Holmlund, Lukas Hillisch but
also Anna R. D. Bøgh, Alexandra Meyer and many more.
I am deeply honoured to be able to experience the pure joy of uncountable amazing moments
with the exceptional nature of Svalbard. I left this island with my heart full of crazy moments I
could experience in this wild island and I will cherish them forever. One of the last places on earth
where you instantly realize that men is nothing and nature’s will is everything. If you have the
chance to, let it move you and change your life like it changed mine.
With the ending of my study I want to express thanks to all the beautiful souls which walked by
my side from one lesson to the next. Especially, Ulrich, Jonas, Fruchti from my masters and
Moritz, Max, Anna, Lena and so many more from my bachelors.
Finally, I am grateful for my parents and siblings, who always supported me, not only financially
on my way but no matter what I do or where I go.
Page IV
Abstract (English)
Research has shown that glaciers in Svalbard are very sensitive to the persistence of snow and
firn and considerable knowledge gaps concerning the consequences of a changing climate and
the effects on glacier accumulation processes are existent. Especially, empirical studies are
necessary to update the knowledge by providing much-needed field data for monitoring and
model validation. Therefore, this study aims to (i) identify the amount, distribution and properties
of snow and firn in the accumulation zone and (ii) to estimate the amount of melt or rainwater
which refreezes within the snow/firn. The field campaign for data collection was carried out in
spring and summer 2019 in the accumulation zone of four arctic glaciers, Foxfonna,
Nordenskiöldbreen, Drønbreen and Von Postbreen all located on the Svalbard archipelago.
The snow and firn regime in the glacier accumulation zone were investigated using snow profiles,
ground penetrating radar surveys, firn cores and snow depth probing’s. Additionally, an
installation with iButtons (temperature loggers) was set up on Nordenskiöldbreen and a station
with two highly sensitive thermistor strings were installed on Foxfonna. The temperature datasets
were used to model the subsurface temperature evolution and the amount of melt or rainwater
refreezing with MATLAB by using a thermodynamic equation and the density evolution modelled
through meteorological parameters.
Analysis of the obtained GPR data revealed similar accumulation characteristics throughout all
collected radargrams. Further a trend of a densifying and thinning firn can be seen. Only the
height of the Equilibrium line altitude and therefore also the elevation where the accumulation
zone can be found varied. The total amount of refreezing on Foxfonna was modelled to be 593
mm for the first thermistor and 221 mm for the second thermistor string. On Nordenskiöldbreen,
the refreezing amount was modelled to be 389 mm using the iButton data. The results of the
refreezing modelling indicate that a lot of rainwater percolated into the snowpack and led to
refreezing seen in the calculated datasets.
Based on the results of this study, further investigations are recommended, especially towards
long-time monitoring of the accumulation zone dynamics, to be able to identify trends.
Additionally, to understand the effect of refreezing on the glacier mass balance, further research
is needed with a broader focus on all melt enhancing processes on the glaciers surface and their
mutual influence.
Page V
Abstract (Deutsch)
Forschungen haben gezeigt, dass die Gletscher in Spitzbergen sehr empfindlich auf das weitere
bestehen von Schnee und Firn reagieren. Außerdem sind erhebliche Wissenslücken hinsichtlich
der Folgen eines sich ändernden Klimas und die Auswirkungen auf die Akkumulationszone von
Gletschern vorhanden. Besonders, empirische Studien sind erforderlich, um den Wissenstand zu
erhöhen vor allem zur Validierung von Modellen.
Diese Arbeit hat das Ziel, (i) die Menge, Verteilung und Eigenschaften von Schnee und Firn in
der Akkumulationszone zu identifizieren und (ii) die Menge an Schmelz oder Regenwasser
abzuschätzen, die im Schnee / Firn wiedergefriert. Die Daten wurden im Frühjahr und Sommer
2019 in der Akkumulationszone der vier arktischen Gletscher Foxfonna, Nordenskiöldbreen,
Drønbreen und Von Postbreen gesammelt.
Der Schnee und Firn in der Akkumulationszone wurde mithilfe von Schneeprofilen, Bodenradar
Untersuchungen, Firn Bohrkernen und Schneehöhen Messungen untersucht. Außerdem wurde
am Nordenskiöldbreen ein Versuchsaufbau mit iButtons (Temperaturlogger) und auf Foxfonna
eine Station mit zwei hochempfindlichen Thermistor Strings installiert. Die Temperaturdaten
wurden verwendet, um die Menge des Wiedergefrorenen Wasser zu quantifizieren. Um das zu
erreichen wurde eine thermodynamische Gleichung verwendet, die auch eine Modellierung der
Dichte benötigte, was mithilfe von meteorologischen Daten gelang.
Die Analyse der gesammelten GPR-Daten ergab ähnliche Bilder in der
Akkumulationscharakteristik in allen gesammelten Radardatensätzen. Nur die Schneehöhe und
die Lage der Gleichgewichtslinie und damit auch der Höhenbereich, in dem sich die
Akkumulationszone befindet, variierten. Auch ein Trend wurde sichtbar, der Firn wird dünner und
dichter. Die Gesamtmenge an Wiedergefrorenen Wasser wurde modelliert und erzielte auf dem
Foxfonna Gletscher 593 mm für den ersten und 221 mm für den zweiten Thermistor String. Auf
Nordenskiöldbreen wurde die Menge des wiedergefrorenen Wassers unter Verwendung der
iButton-Daten mit 389 mm modelliert. Die Ergebnisse der Modellierung zeigen, dass viel
Regenwasser in die Schneedecke sickerte und wiedergefroren ist.
Weitere Forschung wird empfohlen, vor allem auf Basis der Ergebnisse in dieser Studie sind
Langzeitdaten über die Dynamik der Akkumulationszone wichtig, um Trends zu erkennen.
Um die Auswirkung des Wiedergefrierenden Wassers auf die Gletschermassenbilanz vollständig
zu verstehen, sind weitere Forschungen erforderlich. Wobei ein größerer Schwerpunkt auf allen
Prozessen liegen sollte, die Allgemein zur Schmelze auf der Gletscheroberfläche beitragen und
deren gegenseitiger Einfluss.
Page VI
Table of Content
1. Introduction ______________________________________________________ 1
2. Objectives and research questions ___________________________________ 2
3. Theoretical foundations ____________________________________________ 2
3.1. Glacier mass and energy balance ______________________________________ 2
3.1.1. The glacier mass balance terms ________________________________________ 3
3.1.2. Internal accumulation _________________________________________________ 3
3.1.3. Equilibrium line altitude _______________________________________________ 4
3.1.4. Surface energy balance _______________________________________________ 4
3.1.5. Turbulent heat fluxes _________________________________________________ 5
3.1.6. Albedo ________________________________________________________ 5
3.1.7. Heat sinks available due to rain _________________________________________ 6
3.2. Snow ____________________________________________________________ 6
3.2.1. Snow metamorphisms ________________________________________________ 6
3.2.2. Snow properties _____________________________________________________ 7
3.2.3. Snow water equivalent (SWE) __________________________________________ 8
3.2.4. Density ________________________________________________________ 8
3.2.5. Heat conduction _____________________________________________________ 8
3.2.6. Cold content (CC)____________________________________________________ 8
3.3. Firn ____________________________________________________________ 9
3.3.1. Densification ________________________________________________________ 9
3.3.2. Warm and cold firn ___________________________________________________ 9
3.4. Melt processes ____________________________________________________ 10
3.4.1. Water percolation and flow through snow and firn __________________________ 10
3.4.2. Refreezing _______________________________________________________ 11
4. Geography and climate in Svalbard__________________________________ 12
4.1. Geography _______________________________________________________ 12
4.2. Climate _________________________________________________________ 13
4.3. Glaciers in Svalbard ________________________________________________ 16
4.4. Field locations ____________________________________________________ 17
4.4.1. Foxfonna _______________________________________________________ 17
4.4.2. Nordenskiöldbreen __________________________________________________ 18
4.4.3. Von Postbreen _____________________________________________________ 19
4.4.4. Drønbreen _______________________________________________________ 20
5. Methods ________________________________________________________ 21
5.1. GPR (Ground penetrating radar) ______________________________________ 22
5.1.1. Characteristics _____________________________________________________ 23
5.1.2. Survey design______________________________________________________ 23
5.1.3. Processing _______________________________________________________ 24
5.2. Coring __________________________________________________________ 24
5.3. Snow profiles and snow depth survey __________________________________ 25
5.3.1. Experiment Setup ___________________________________________________ 25
5.4. Temperature loggers _______________________________________________ 26
Page VII
5.4.1. Measuring temperature with iButtons____________________________________ 26
5.4.2. Measuring temperature with thermistor strings ____________________________ 28
5.5. Density evolution modelling __________________________________________ 30
5.5.1. Model choice ______________________________________________________ 30
5.5.2. Model description ___________________________________________________ 31
5.5.3. Model workflow_____________________________________________________ 32
5.6. Calculating the amount of refreezing ___________________________________ 34
5.6.1. Model background __________________________________________________ 35
5.6.2. Model approach ____________________________________________________ 35
6. Results _________________________________________________________ 36
6.1. GPR ___________________________________________________________ 36
6.1.1. General results _____________________________________________________ 37
6.1.2. Wet firn _______________________________________________________ 37
6.1.3. Glacier hydrology ___________________________________________________ 38
6.1.4. Foxfonna _______________________________________________________ 38
6.1.5. Nordenskiöldbreen __________________________________________________ 39
6.1.6. Potpeschniggbreen (Von Postbreen) ____________________________________ 40
6.1.7. Phillipbreen (Von Postbreen) __________________________________________ 40
6.1.8. Drønbreen _______________________________________________________ 42
6.2. Snow profiles _____________________________________________________ 44
6.2.1. Foxfonna _______________________________________________________ 45
6.2.2. Nordenskiöldbreen __________________________________________________ 46
6.2.3. Von Postbreen _____________________________________________________ 47
6.2.4. Drønbreen _______________________________________________________ 49
6.3. Snow depth surveys ________________________________________________ 50
6.3.1. Foxfonna _______________________________________________________ 50
6.3.2. Nordenskiöldbreen __________________________________________________ 51
6.3.3. Drønbreen _______________________________________________________ 53
6.4. Firn cores ________________________________________________________ 54
6.4.1. Nordenskiöldbreen __________________________________________________ 54
6.4.2. Potpeschniggbreen (Von Postbreen) ____________________________________ 54
6.4.3. Drønbreen _______________________________________________________ 55
6.5. Thermistor string and iButton temperature measurements ___________________ 57
6.5.1. Subsurface temperature evolution from the thermistor strings ________________ 57
6.5.2. Subsurface temperature evolution from the iButtons ________________________ 60
6.6. Density evolution modelling __________________________________________ 61
6.6.1. Foxfonna _______________________________________________________ 61
6.6.2. Nordenskiöldbreen __________________________________________________ 63
6.7. Refreezing _______________________________________________________ 64
6.7.1. Foxfonna _______________________________________________________ 64
6.7.2. Nordenskiöldbreen __________________________________________________ 66
7. Discussion ______________________________________________________ 68
7.1. GPR surveys _____________________________________________________ 68
7.2. Snow profiles and firn cores __________________________________________ 69
Page VIII
7.3. Refreezing _______________________________________________________ 70
7.3.1. Density modelling ___________________________________________________ 70
7.3.2. Refreezing modelling ________________________________________________ 71
7.3.3. Glacier thermal switch from temperate to cold glacier in Svalbard _____________ 72
7.3.4. Influence of refreezing on the mass balance ______________________________ 73
8. Conclusion and outlook ___________________________________________ 73
8.1. Conclusions ______________________________________________________ 73
8.2. Outlook _________________________________________________________ 74
9. References ______________________________________________________ 76
10. Appendix _______________________________________________________ 80
11. Erklärung / Affirmation ____________________________________________ 92
Page IX
Abbreviations
SMB
ELA
GPR
SWE
RES
m a.s.l.
Si
CC
Surface Mass Balance
Equilibrium Line Altitude
Ground-penetrating radar
Snow water Equivalent
Radio-echo sounding. (Same as ground-penetrating radar)
metre above sea level
superimposed ice
Cold content
1
1. Introduction
With a changing climate, especially in Svalbard (Bintanja and Andry, 2017) the understanding of
the glacier SMB (surface mass balance) and runoff is becoming more crucial than ever.
Precipitation and temperature changes are leading to changes in the timing and duration of the
melt period, the type of ice undergoing accumulation (firn versus frozen liquid precipitation or
snowmelt) and the overall mass balance of the glacier. Several important modelling experiments
from Marchenko et al. (2017) , van Pelt et al. (2016) and others have shown that understanding
future mass balance and thermal state of Svalbard glaciers are very sensitive to the persistence
of snow and firn cover at high elevation. To quantify the glacier changes, a detailed understanding
of the processes happening on the glaciers surface and especially in the accumulation zones is
crucial. Since almost all currently available studies have in common that they largely rely upon
the heuristic application of state of the science models, there is a critical need for empirical studies
to update the knowledge by providing much-needed field data for monitoring and model validation.
Further the process of meltwater percolation and refreezing is lacking understanding, especially
when it comes to the amount of water which refreezes and the effect on the glacier SMB. (van As
et al., 2016) Several open questions are existent, starting at how much snow and firn can be
found on Svalbard glaciers accumulation zones and what is the local distribution of it? Further the
current processes of the SMB in a changing climate needs to be observed carefully. Especially
questionable is the role of refreezing on Svalbard glaciers, because either it is acting as a buffer
for mass loss or it is increasing the temperature of the surrounding snow structure and therefore
even enhances melt.
Therefore, this thesis project aims to provide a dataset including detailed observations of the snow
and firn regime of four arctic glaciers in Svalbard. Especially to (i) understand the current state of
arctic accumulation zones and (ii) to quantify meltwater retention in form of refreezing in snow
and firn for better understanding of the SMB.
2
2. Objectives and research questions
Considerable knowledge gaps concerning the complex interactions between the glacier surface,
the snow/firn and the atmosphere are existent. Especially in situ data, as outlined by previous
studies (van As et al., 2016) is missing to validate models and understand changes. Hence the
main objective of this thesis is to increase the general understanding of snow and firn processes
of glaciers in Svalbard via extensive in situ data collection and melt season observations. Four
glaciers are selected as study sites, because the snow and firn patterns vary locally and the aim
is a representative data collection. One focus of this study is to aim at a general increase of
understanding of the current accumulation zone state at different glaciers in Svalbard. The second
focus is the process of melt or rainwater refreezing within the snow and firn. Therefore, the
measuring of refreezing and further the quantification of the collected data is aimed at. Thus,
following research questions are outlined below.
Q1: How much Snow and Firn is present in the accumulation zone and what are their properties?
Q2: Can the amount of melt or rainwater, which refreezes within the snow and firn, be quantified?
3. Theoretical foundations
This chapter presents the theoretical background of the snow and ice processes on glaciers and
the ongoing complex interactions between the cryosphere, hydrosphere and atmosphere, which
are relevant for this thesis.
3.1. Glacier mass and energy balance
To understand the development of glaciers, they need to be observed as a system with complex
interactions of inputs and outputs. The glacier is determined by the gain and loss of ice mass,
which is referred to as glacier mass balance. As seen in Figure 1, the accumulation of mass at
the surface is driven by direct precipitation (snow and rain), riming, meltwater refreezing,
avalanches, drifted and blown snow. Mass loss is governed by, melt, evaporation but also
breaking off ice blocks or calving. Glaciers are direct responders to the climatic settings as they
recede when the ablation exceeds the accumulation and vice versa (Armstrong and Brun, 2008).
The mass balance outcome is highly dependent on local and regional climate, surface mass and
energy exchanges, specific balances and calving (Cuffey and Paterson, 2010).
3
3.1.1. The glacier mass balance terms
The overall mass balance of a glacier is referring to a change in mass over a specific time. The
term overall or total mass balance, implies the sum of all the mass balance components and
specific balances (Cogley et al., 2011). Usually the mass balance is monitored over a year and is
defined to as the time between first of October to end of September. The overall net mass balance
is a sum of the total accumulation and ablation. However, the winter and summer mass balances
can be calculated from the mean specific mass balance, which is the absolute change in mass
divided by the size of the glacier area (Benn and Evans, 2010). The term climatic mass balance
is the sum of the two components, SMB and internal accumulation. Therefore, the SMB is only
referring to all accumulation and ablation on the glacier surface.
Further the energy balance can also be used to describe glacial processes. The energy balance
describes the energy existent and changing in a defied volume (Cogley et al., 2011).
The measurement of the glacier mass balance can be performed by direct measurements such
as snow depth and point measurements, but also by hydrological methods, geodetic methods or
gravimetric methods (Benn and Evans, 2010).
3.1.2. Internal accumulation
As seen in Figure 1, the Internal Accumulation is defined as the sum of meltwater and rain which
enters the system and is undetected by surface measurements. The water enters the snow/firn
or ice through percolation, through efficient drainage into crevasses and moulins and through
water injected into fractures at the glacier bottom. The term refreezing, which describes the
process, is referred to internal accumulation and refreezing of meltwater within the snow.
The amount of internal accumulation can be measured within a small point scale, but the spatial
heterogeneity of the internal accumulation and refreezing varies greatly and therefore results in
Figure 1: Glacier-wide mass balance components, the arrows do not account for amount of mass
transfer, illustrated by Cogley et al. (2011).
4
major uncertainties within the mass balance measurements (Pfeffer et al., 1991; Pfeffer and
Humphrey, 1996). Summarizing, only the refreezing below the summer surface is referred to as
internal accumulation, because refreezing at the snow/firn or snow/glacier interface can be picked
up with density measurements at the annual summer mass balance measurements. Refreezing
at the interface of the snow and glacier ice is also named Si, superimposed ice (Cogley et al.,
2011).
3.1.3. Equilibrium line altitude
Generally, glaciers can be divided into two zones, the higher part in elevation, where the snow
survives the summer is called the accumulation zone and the lower part where warm summer
temperatures lead to melt of the surface winter accumulation, which is referred to as the ablation
zone. In between these two zones is the ELA, the Equilibrium Line Altitude which marks the border
of accumulation and ablation (Armstrong and Brun, 2008). Despite this apparently simple
definition, reality is more complex, which means that the equilibrium line is rarely a straight line
but an approximation (Oerlemans, 2001). Nevertheless, the equilibrium line is still a valid concept
to determine the transient climate conditions of any given year as it is giving an idea about the
glaciers mass balance (Armstrong and Brun, 2008).
3.1.4. Surface energy balance
The fate of a glacier depends on the moisture source, the local precipitation regime and the
physical interactions on the glacier surface. The energy balance is therefore referring to the SMB
terms like melt, sublimation and the radiative and non-radiative fluxes and together all the terms
are solved by SMB modelling. Energy modelling requires more data but also represents energy
exchanges more accurate than temperature-index models and is therefore an important tool to
determine the amount of energy exchanges (Cogley et al., 2011). Radiative fluxes are the
shortwave radiation (incoming and reflected shortwave) and the longwave radiation (incoming,
reflected) which leads to the net radiation flux (( 1 ); Benn & Evans, 2010)
!"#$%&' (%)*+,-.&' (.)*+,///////01/2345
( 1 )
!"
is defined as net radiation.
%&'
represents the incoming shortwave radiation, whereas
%)*+/
is
the reflected shortwave radiation.
.&'
is the incoming longwave radiation and
.)*+
is the outgoing
longwave radiation. The highest percentage of radiation is coming from the incoming shortwave
flux and can make up to 99 % of the total radiation. Nevertheless, this radiation can become less
due to clouds, particles or humidity in the air. On a cloudy day, the shortwave radiation is
decreasing underneath the clouds, but the terrestrial longwave radiation is reflected from the
clouds and can therefore cause a significant energy surplus even without direct sunlight available
(Benn and Evans, 2010). However, the incoming short-wave radiation plays a significant role in
the energy transport into the snow cover (Armstrong and Brun, 2008). The available energy for
melt of the snow is the radiation which is left after the reflected part of the shortwave radiation is
gone (Benn and Evans, 2010).
5
The whole energy balance on any given surface, where no energy can be destroyed or lost, is
equal zero. The energy balance equation for a surface is given by following variables (Equation 2;
Benn & Evans, 2010):
%1/-/.1/-/!6/- /!7/ -/!8/-/!9/ - /:/ # /;///////01/2345
( 2 )
%1
is expressed as net short-wave radiative flux,
.1
is the net long wave radiation flux.
!6
is
the sensible heat transfer and
!7
is the latent heat transfer.
!8
is the energy supply of rain and
:
is the energy used to melt ice or freeze water. Finally,
!9
is the energy used for temperature
change in the ice (Benn and Evans, 2010).
3.1.5. Turbulent heat fluxes
Turbulent Heat fluxes is a term for both sensible and latent heat flux. Sensible heat flux, QH, is
thermal energy exchange, happening directly between two materials. On a glacier this sensible
energy transfer can happen between the atmosphere and the ice surface (Benn and Evans,
2010). Since air is a decent insulator, the sensible heat flux is dependent on the near surface air
regime and is most effective when the surface is rough, and a temperature gradient is existent
(Paterson, 1994).
The latent heat flux expressed as
!7
in the energy balance equation, refers to convective water
vapor exchanges between snow and the atmosphere through evaporation, condensation,
sublimation and deposition. The amount of energy transfer depends on the vapor pressure
gradient, wind speed, surface roughness and stability of air. Latent heat exchanges within the
snow and ice matrix, associated with melt and refreezing are expressed in different terms by the
latent heat of fusion (Benn and Evans, 2010).
3.1.6. Albedo
The albedo determines how much of the incoming short-wave radiation is reflected from the
surface. It is given in percent of shortwave reflectance. Albedo is crucial, because it determines
how much energy will be available for melt or warming of the snowpack.
However, the distinctive albedo parameterization is influenced by high spatial variabilities
governed by surface parameters such as grain size, black carbon, dust, microorganisms, grain
shape, angle and wavelength of incoming solar radiation (Hock, 2005). Sensor accuracy is also
an important factor (Winther, 1993). The high reflectivity of snow is leading to positive feedbacks
to surface air temperature through the so-called “snow–albedo feedback”, which can lead to
increased warming. Generally, the values for albedo of dry snow range between 0.80 - 0.97.
However, melting snow can only reflect less shortwave radiation and therefore has an albedo
value of 0.66 - 0.88 (Paterson, 1994). Generally, Albedo values ranges from 0.1 for dirty ice to
more than 0.9 for fresh snow. Even though it is hard to parameterize it is responsible whether
melt is happening or not. For example, when a fresh snowfall is covering the glacier in summer,
the albedo is enhanced and can reduce melt and runoff immediately. Since albedo is also
determined by crystal grain size, the albedo of fresh snow can drop by 0.3 within a short time due
to metamorphisms. Further, overcast weather can increase the snow albedo due to a spectral
6
shift of the solar radiation reaching the snow surface (Warren, 1982). Simplified the equation for
the albedo is stated by the following parameters (Equation 3; Benn and Evans, 2010):
< #/%1)*+
%1&' ///////0=5
( 3 )
<
represents the Albedo in percent and is expressed through
%1)*+
, the reflected shortwave
radiation and
%1&'
, the incoming shortwave radiation (Benn and Evans, 2010).
3.1.7. Heat sinks available due to rain
Rain supplies energy by percolating into the snowpack, whereas the rain has a warmer
temperature than the surrounding snow, which causes an energy difference and leads to great
energy transfer in form of a heat flux. According to this effect, the biggest energy increase is
observed when the water is percolating into the snow and not running off the surface. The
equation is defined as following (Equation 4; Bamber and Payne, 2003):
!8 # >?@A@BC? @$9D( 9E,///////01/2345
( 4 )
!8
is the energy available from rain,
>?
is the density of water (1000 kg m-3),
A
is expressing the
rainfall rate in (m s-1) and
BC?/
is the specific heat capacity of water (4200 J kg-1 K-1).
9D
is the
temperature of rain in °C and
9E
is the surface-temperature of snow in °C (Bamber and Payne,
2003).
3.2. Snow
In this chapter the specific properties of snow are explained in detail. A snowpack is defined as
mixture of ice, air and water and due to the unique shape of snow crystals and its metamorphic
changes, snow creates several different layers within the snowpack (Fierz et al., 2008). The most
relevant snow characteristics (for glacier mass balance) are elucidated in this chapter.
3.2.1. Snow metamorphisms
Due to the fact that snow temperature, is mostly close to its melting point, snow is in a constant
state of transformation, referred to as snow metamorphism.
Once the snow is on the ground, it forms different layers in the snowpack depending on wind
velocity, precipitation, wind direction and snow metamorphisms. Each single stratigraphic layer is
unique and different from adjacent layers below and above. The layers differ at least in one
characteristic, ranging from density, temperature stratigraphic snow microstructure to impurities
within the snow (Fierz et al., 2008).
Several different post-depositional metamorphisms can be classified which can lead to changes
in density and crystal type. The wind can also lead to increased sublimation and changes of the
energy balance at the snow- atmosphere interchange (Armstrong and Brun, 2008). As snow has
its melting point temperature close to 0 °C, it has a high vapor pressure depending on the
prevailing thermal and meteorological settings. Therefore, ice is constantly in a phase change to
vapor without the liquid phase (sublimation). The metamorphisms within the snow pack often
7
come with a changing the aggregate state of the water molecules (Fierz et al., 2008). The
following metamorphisms can be subdivided:
Constructive metamorphisms
When the snowpack shows a temperature or vapor pressure gradient, and water vapor is
available in the pore space between snow grains, or simply sublimates, the vapor is moving
through the snowpack, usually upwards from high to low pressure. When a layer is too dense to
pass through for the vapor or it is cool, the vapor just deposits onto already existing snow crystals,
which can lead to building up of these crystals as depth hoar or faceting (Fierz et al., 2008).
Destructive metamorphisms
Typical destructive metamorphisms are the curvature effect, melting, gravitational settling and
wind. The curvature effect happens because the surface tension is at its maximum at the convex
edge of the snow crystal and therefore the water sublimates and then deposits again on the
concave part of the crystal, leading to rounding (Fierz et al., 2008). Melting due to temperature
increase and radiation, leads to liquid water filling the pore space between the crystals, and
because water transfers heat rather quickly, this rounding and melting process can be quicker
than other metamorphisms leading to big rounded grains (Armstrong and Brun, 2008). Wind is an
efficient destructive metamorphism, as it breaks the snow crystal and leads to broken, fragmented
needle like crystals and it can also increase the density because of high pressure on the slope of
luv facing mountain sides. The density of snow falling in wind increases around 20 kg m−3 for each
m s−1 of wind speed (Jordan et al., 1999). If a very intense snowdrift event is happening, a fresh
fallen snow layer with a density less than 100 kg m−3 can easily be transform into a drifted snow
layer with densities up to 300 kg m−3 within a short period of time. This densification process plays
a big role in polar regions where prevailing low temperatures slow down usual metamorphisms
(Dang et al., 2017).
3.2.2. Snow properties
Modern guidelines of snow properties classification can be found in two main publications
(Colbeck et al., 1990; Fierz et al., 2008). Snow depth is usually very tricky to measure as it can
vary immensely over short distances when strong wind drift happens. Therefore, many sample
points are taken, or snow-drift models are used obtain a decently resolved information on the
spatial distribution of the snow cover. Common physical quantities measured in snow are the
snow depth d, density
>
, snow water equivalent (SWE), porosity
F
, liquid water content
G?
,
permeability
H
, thermal conductivity
I
, and the temperature
9
(Kinar and Pomeroy, 2015). Crystal
stratigraphy can be determined in situ by performing a snow profile, which can give valuable inside
into the metamorphism and in terms of avalanches, the weakness of the snowpack can be
derived. Snow temperature is either measured directly or with automatic installed monitoring
systems. It can be measured per stratigraphic layer or every 10 centimetre, depending on the
objectives (Cogley et al., 2011).
8
3.2.3. Snow water equivalent (SWE)
The snow water equivalent (SWE) can be defined as the depth of water in millimetre, created by
melting of the snow. It is based on the fact that water, which spreads over 1 m2, has a depth of
1 mm and a mass unit of 1 kg, with a density of 1000 kg m−3. SWE is the mathematical product
of snow depth in millimetres
$JE,/
and the ratio of the density of snow and the density of water
(
KL
/KM/,
, both represented in kilograms per cubic metre. Therefore, the equation is stated as follows
(( 5 ); Pomeroy and Gray, 1995):
%17 #/ >E
>?@JE///////0225
( 5 )
3.2.4. Density
Density is the mass unit per unit volume expressed in kilograms per cubic metre. To determine
snow density, the snow weight of a known volume is measured. Snow density is a product of
density of air, liquid water and ice (Cuffey and Paterson, 2010). Alternative density measurements
can be obtained due to the dielectric properties of snow (Mätzler, 1996).
The density is measured in kilograms per cubic metres (kg m−3) which is the ratio of mass to
volume for a given snow sample. The density value can also vary immensely from fresh snow to
melt forms. For a snowpack, which is highly influenced by wind and refreezing meltwater events,
the density can reach up to 700 – 800 kg m−3 (Armstrong and Brun, 2008).
3.2.5. Heat conduction
Snow has an extremely low thermal conductivity (a measure of the ability to conduct heat) and
acts as thermal insulator, trapping heat in the ground. Snow has a high amount of air trapped,
compared to ice, and air is not good at conducting heat. Therefore, once the snow starts to settle
and increase density the heat conductivity is increasing too. The thermal conductivity of fresh
snow is N0.1 W m−1 K−1 which is 10 – 20 times lower than values for ice or wet soil. This has
major implications for the hydrological cycle, from insulation to lag of snowmelt. The meltwater
and rain, which is entering the snowpack and refreezes, does not only bring additional energy but
also leads to increased heat conductivity because of densification (Armstrong and Brun, 2008).
3.2.6. Cold content (CC)
The Cold Content (CC) describes the negative heat storage of the snowpack and is the amount
of energy needed to heat up the snowpack to isothermal zero degrees without a phase change
(Armstrong and Brun, 2008). The cold content
O
of a defined layer between the surface and the
depth
P
is expressed per unit area (Equation 6; Cogley et al., 2011):
O # BC@Q>$P,@09R(9$P,
S
T5/UP///////0V/2345
( 6 )
9
The cold content
O
does not lead to any mass change in this definition.
BC
is expressed as the
heat capacity of ice and
>
and
9
are density and temperature at a certain depth
P
.
9R
is the freezing
point of water or the temperature equal to 0 °C with a pressure of 1013.25 hPa, at which water
freezes and therefore releasing energy (latent heat of fusion). However, sometimes the
BB
is
expressed as the amount of water from melt or rainfall, that is percolating into the snowpack and
further must be refrozen in a subfreezing snowpack to warm the snowpack to isothermal 0 °C,
due to this change in density happens (( 7 )):
BB #O
>?@.R///////025
( 7 )
.R
is expressed as the latent heat of fusion, of refreezing water (where the temperature is equal
to the freezing point) and
>?
is the density of water (Cogley et al., 2011).
3.3. Firn
Snow grains which sustain over at least one melt season are called firn. A firn layer can only exist
if the SMB is positive and accumulation exceeds ablation (Sommerfeld and LaChapelle, 1970).
Freshly fallen snow has a low density due to the amount of air between the snow crystals, but
after one melt season, the grains usually have reached a density of 580 – 600 kg m−3 and are at
the metamorphic stage between snow and ice. Initially the main metamorphism to densify the
grains is melt-freeze but later the firn still evolves to higher densities through gravitational and
pressure packing (Cuffey and Paterson, 2010).
3.3.1. Densification
Firn density further increases with time (firn densification) because of refreezing of meltwater
which percolated into the available pore space within the firn grains. Once, the density of
917 kg m−3 is reached, the firn transformed into glacier ice. Firn density increases until around
830 kg m−3 where the connecting air and water filled passageways between the single grains are
sealed (pore close off). Further densification is happening through compression until the firn has
converted into glacier ice at a density of about 913 kg m−3 (Cuffey and Paterson, 2010).The
velocity of densification depends on melt. When rarely any ablation processes are present,
densification of firn hardly proceeds over 400 kg m−3 on average in a mid- latitude winter.
However, if plenty of meltwater percolates into the firn and refreezes within, bulk densities close
to pure ice can develop in short time, sometimes even within a day. Hence, depending on
accumulation, the glacier ice can develop in vast time ranges from a few years to a few centuries
(Cogley et al., 2011).
3.3.2. Warm and cold firn
Firn undergoes different stages per year depending on where it is located. Firn in the high
accumulation zone differs from firn in the low accumulation zone due to the processes happening.
One of the zones which can be distinguished from others is the percolation zone. This zone is
located where the water from surface melt or rain drains into the subsurface. In the upper
percolation zone, the water only infiltrates the snow, whereas in the lower percolation zone, the
10
water can reach the firn even below the summer surface and is also called wet snow zone.
Nevertheless, if a glacier reaches up very high in accumulation or the climate is dominated by
cold temperatures it is called the dry snow zone. In this dry snow zone, no surface melt or rainfall
occurs throughout the year (Cogley et al., 2011).
Depending on which zone the firn layer is located on the glacier, it can be warm or cold.
Cold firn is existent when the temperatures exceed the pressure melting point of water. Even
though meltwater is entering the firn in summer, the
BB
of the firn is high enough to refreeze the
water and keep temperatures at sub-zero. Another scenario is that firn stays cold throughout the
year because no rain or meltwater enters the firn due to a location in high elevation (Cogley et
al., 2011).
Warm firn can exist under various scenarios elucidated here. One reason for warm firn is that
melt or rainwater percolates deep into the firn and refreezes. Due to the process of refreezing,
the water releases latent heat which heats up the subsurface firn close to the melting point. This
process can occur seasonally and the firn can warm in spring and summer and cool in winter and
autumn (Cogley et al., 2011).
3.4. Melt processes
When the surface ice or snow reaches 0 °C temperature any additional intake of energy results
in melting. In contrast a low temperature near the surface can cool the surface and even lead to
accumulation through condensation of vapor or freezing of water molecules. Therefore, the
snowmelt period is defined by a positive energy balance. The snowmelt period can be divided
into three phases, the warming phase, the ripening and at last, the output phase. To be able to
directly measure the evaporation, sublimation and melt is complex, therefore the parameters are
mostly modelled using the energy inputs and outputs described in the SMB equation.
Nevertheless, melt of ice takes a high amount of energy, which leads to a delayed runoff and
therefore it is a crucial parameter for the mass balance year (Armstrong and Brun, 2008).
3.4.1. Water percolation and flow through snow and firn
Water flow through snow is similar to other granular material but affected by changes in the ice
matrix. Water moves rather quickly through because snow is a highly permeable and also
gravitational forces are higher than capillary forces. Low viscosity wetting fluids, like water, can
lead to concentrated flow paths in preferential channels, ahead of the wetting front. Similar to
thermal conductivity relating to the density structure, flow properties relate to the general pore
structure and spatially distributed homogeneity within the snowpack (Armstrong and Brun, 2008).
Therefore, water infiltration can happen homogenously or heterogeneously (preferential flow
paths in pipes) depending on the snow/firn structure (Pfeffer and Humphrey, 1996).
Most models use equations for homogenous flow through the snow/firn. A detailed understanding
in water flow within the snow/firn is currently missing due to the vast spatial variability of snow/firn
properties and ice layer formations. Concerning the quantification of refreezing, not only locally,
but interpolated on a glacier surface, the use of homogenous wetting front assumption results in
imprecise model outcomes. Therefore, the amount of glacier-wide refreezing can lead to overall
uncertainties in modelling water flow and refreezing quantification (van As et al., 2016).
11
3.4.2. Refreezing
Refreezing is part of the glacier accumulation and also referred to as internal accumulation. The
process is simple, rain or meltwater percolates into the snow/firn and with the cool ice crystals
around, the water is refreezing. When water is refreezing, it releases latent heat of fusion as
energy excess to the surroundings, causing a warming effect within the snow matrix. Therefore,
rain or meltwater is significant in transporting energy into the snow/firn and increase the
temperature by latent heat release (Benn and Evans, 2010).
Latent heat of fusion is the energy released or taken up by a molecule, when a phase change
happens from liquid to solid or vice versa. The amount of energy required for a phase change of
water is 333.5 kJ kg−1 (Cogley et al., 2011).
The amount of refreezing water is limited by the availability of pore space to percolate, energy
and amount of meltwater (Reijmer et al., 2012). Refreezing can be divided into two components
depending on when it occurs. The first is refreezing of meltwater in cool snow or firn in the spring
when first melt enters the snow/firn. The second refreezing component is in autumn when the
liquid water is held by capillary forces and further refreezes when the first negative temperatures
enter the snow or firn (Reijmer et al., 2012).
Refreezing results in an increased snow/firn density and delayed runoff in the ablation zone.
Further, it has mayor implications for the whole mass balance, because of enhanced mass
accumulation around the equilibrium line and the percolation zone (Reijmer et al., 2012).
In detail refreezing at first reduces the runoff, but later the melt can increase on the surface
through a positive conductive heat flux towards the surface. The overall temperatures of the
snow/firn pack can be raised faster to the melting point through the latent heat release (van Pelt
et al., 2016). Furthermore, refreezing contributes significantly to subsurface heating of the firn
layers (Marchenko et al., 2017). Another effect to consider is that the refreezing water can create
a solid ice layer which, when exposed to the surface, enhances melt through a decreased albedo
value. Finally, refreezing probably also extends the seasonal snow survival in summer and
therefore reduces the bare ice-free time of the glacier and thereby leads to decreased melt (van
Pelt et al., 2016).
To be able to determine the amount of refrozen water within the snow/firn pack is where it gets
complex. Not only because homogenous infiltration is assumed for most equations, but also
because many variables need to be considered in an appropriate way to be able to understand
the fate of the percolating water (Reijmer et al., 2012).
An expert summary showed that the most unknown processes in refreezing are lacking in situ
data to test the model outputs, the timing when liquid water is available and the spatial
heterogeneity of water percolation.
Further important factors, determining the model success are the lack of knowledge of ice layers
within the snow/firn pack, which prevents further percolation and the surface albedo, which also
can vary greatly within few metres (van As et al., 2016).
Recent literature about refreezing in Svalbard, has shown that the fate of a glacier is highly
impacted by the current and future state of the persistent firn layer (van Pelt et al., 2016). A
modelling study of the firn from Nordenskiöldbreen coupled a surface energy balance and firn
model. The results showed a scenario with densification of the firn and a significant loss of firn
depth due to increased melt which will outperform the positive runoff buffer of refreezing (van Pelt
et al., 2016). Nevertheless, the effect of refreezing is still significant thus it can lead to a climate
tipping point, where the runoff increases tremendously in short period of time, due to the
12
densification of the firn and therefore is expected to play a crucial role in the development of
glaciers in a changing climate context, especially on big ice masses, which contribute significant
to sea level rise (Charalampidis et al., 2016; van As et al., 2016). Until such a tipping point is
reached, refreezing will continue to influence the subsurface density and act as a buffer against
mass loss in Arctic Glaciers (Wright et al., 2007).
4. Geography and climate in Svalbard
This chapter provides an introduction and overview about the geographical and climatic
characteristics of the study area. The focus is on the climate of the archipelago which is highly
relevant for glacier accumulation.
4.1. Geography
The Svalbard archipelago is located around 1000 km north of the coast of mainland Norway and
1300 km south of the North Pole. Svalbard has four big islands, named Spitsbergen,
Nordaustlandet, Barentsrøya and Edgerøya. Around 59 % of the total land mass is covered by
2100 different glaciers. The total land area of Svalbard is 62.250 km² (Hagen, 1993). The following
overview map shown in Figure 2, gives an overview of the archipelago and its location. Further,
the map indicates the locations relevant for this thesis, the town Longyearbyen and the place of
the weather stations one and two.
Figure 2: Overview Map of Svalbard. In the right corner, the location of Svalbard in the Arctic ocean is indicated with
a circle. (OpenStreetMap.n.d.) and the two other maps indicate the locations relevant for this thesis (Norwegian Polar
Institute ,2018).
13
4.2. Climate
Considering Svalbard’s location, in the Arctic circle, the temperatures during winter can be higher
than expected, which means that temperatures above zero are well known to happen during mid-
winter. These warm spells are caused by warm ocean currents and frequent circulation of warm
air from lower latitudes. The end of the Gulf Stream encounters the western part of the island and
creates open water paths, even during the cold arctic winter. All these fairly unique climatic
settings lead to large fluctuations in winter temperatures and can be observed in snow profiles as
ice layers or as ice layer directly on the ground. Inside the fjord systems a slightly more continental
climate is present, with higher temperatures in summer and lower in winter (Adakudlu, Andresen,
Bakke, et al., 2019).
Arctic climate fluctuates enormously at a daily to annual scale. From the first weather station data
until now, no complete linear temperature development is visible, but when the data is examined
in depth, a significant trend in sub- periods and fluctuations can be observed. The results show a
slight warming period from 1910 to 1930, afterwards a downward trend until the 1960s. And finally
a clear warming trend is visible from 1960s until now (Førland et al., 2011). However, on a larger
scale, combined in situ recordings and reconstructions of the Arctic air temperature showed that
the temperature in the 20th century was higher compared to the last 400 years (Serreze and Barry,
2011).
In detail, the mean annual temperature is around - 6.7 °C. Despite large variations of climate
models due to the variability of the climate, the recordings show an increase of vegetation days
and a decrease of frozen soil days. Further a rise of winter temperature exceeds increases of the
summer temperature, and the length of the snow season has decreased by more than 20 days in
the period between 1958 and 2017 (Adakudlu, Andresen, Bakke, et al., 2019).
Due to the strong variations of weather during the past, predicting the future with modelling is not
straight forward and one needs to include inhomogeneity’s carefully (Førland and Hanssen-
Bauer, 2003).
The latest climate report, SCROCC (special report on the ocean and the cryosphere in a changing
climate) which was published at the IPCC Session in September 2019, supports the hypothesis
that the Arctic surface air temperature has increased at more than double the global average
(Pörtner et al., 2019). Further it is displayed that the climate warming is increasing since the
anthropogenic greenhouse gas emissions started to rise and that the climate will further warm
with high confidence (Overland and Wang, 2018; Pörtner et al., 2019).
Further extreme climate variations also proved this statement, for example the winter near surface
temperature of + 6 °C (relative to 1981-2010) have been measured in the winter 2016 and 2018
which is twice the temperature variation measured in previous years (Overland and Wang, 2018).
These warm temperatures were caused by a change in air convection, in particular a change of
the polar vortex which led to advection of warm air and further decreased the sea ice. (Overland
and Wang, 2018)Further the Winter sea ice extent in the arctic circle was at an unprecedented
low level compared to 1979 – 2014 (Overland and Wang, 2018).
In the Arctic, this general warming effect might be amplified due to several feedback mechanisms
like lacking sea ice extent, permafrost thaw, snow albedo changes and many more. Whereas,
when climate in Svalbard is discussed the term Arctic amplification should not be missed. This
term implies the feedback loops of greater rising of temperatures at arctic zones compared to the
rest of the world. A climate feedback response acts as an internal climate process which changes
14
the initial response of climate forcing. An example for feedback is the lack of polar sea ice which
enhances a rise of ocean water temperatures immediately due to a complete changed albedo
and it also changes the near surface temperatures due to the relatively warm ocean water. The
albedo is the fraction of radiation that is reflected, which varies tremendously from an ice or snow
surface to a rough ocean surface (Serreze & Barry, 2011). In general, the detailed processes of
Arctic amplification are still discussed by various study’s, but the outcomes are suggesting
intrinsic reinforced feedbacks in all Arctic research areas (Serreze and Barry, 2011; Goosse et
al., 2018; Stuecker et al., 2018).
Precipitation is generally about 400mm per year on the west coast of Svalbard and less inside
the fjord systems which is visible in Figure 3. On glaciers or highly elevated peaks, the
precipitation is higher but generally never more than 2 – 4 m of snow accumulation. The most
precipitation at the east coast is coming from low pressure systems and winds with an east wind
direction. Therefore, the precipitation is varying a lot not only north south, east west but also
locally, which is also shown in Figure 3 (Hagen, 1993).
The amount of precipitation was observed to increase significantly (15 – 25 %) at all stations in
the Arctic during the last 90 years. However, one has to be careful when interpreting this
statement, because it is known that the precipitation measurement of rain is way more accurate
compared to snow. Therefore, an overall increased measurement of precipitation can be a result
of more rain and less snow precipitation (Førland and Hanssen-Bauer, 2003).
Furthermore, trends are visible for the increase of precipitation around autumn and winter for all
Arctic islands, but the uncertainty is still considerably large. What has increased intrinsically is the
amount of heavy rainfalls during the year, which can make up to 25 % of the annual precipitation.
It is estimated that the amount of heavy rainfall and its frequency will increase on Svalbard.
(Adakudlu, Andresen, Bakke, et al., 2019).
Figure 3: Precipitation variations which are mostly
based on indirect measurements and observations
from GMB, AWS and observations of the ELA on
glaciers (Hagen et al ,1993).
15
Generally problematic is the scarce network of automatic weather stations and the precipitation
catchment errors due to dry snow and strong winds (Førland and Hanssen-Bauer, 2003). A study
with a great amount of data was performed by Bintanja and Andry (2017). and showed results of
37 different climate models in a projected future scenario. The outcome1 suggests an increase in
precipitation, but mostly rain not snow as the atmospheric warming will result in melting of
snowflakes before they reach the ground, especially over the North Atlantic and Barents Sea
(Bintanja and Andry, 2017).
The snow climate, at least in the close vicinity of the town Longyearbyen, is classified as ‘‘High
Arctic maritime snow climate’’. For this climate classification, it is common to have a rather thin
and cold snowpack and a basal layer of depth hoar with wind pressed snow and ice layers on
top. The annual snowpack duration is about 8 – 10 months of the year, even though precipitation
amounts are little. The thin snow cover usually has a high density, due to wind exposure during
and after snowfall events (Eckerstorfer and Christiansen, 2011).
Finally, wind also plays a crucial role in the redistribution of the precipitated snow. Especially with
Svalbard’s exposed topography, with no vegetation but steep mountain ridges. Therefore, local
wind fields can vary a lot within short distances. However, one can see large accumulation on the
leeward side of mountain ridges and of glaciers or channelled in small valleys (Eckerstorfer and
Christiansen, 2011). Most winds in the vicinity of Longyearbyen come from the easterly and north-
easterly direction as seen in Figure 4, resulting in most accumulation on the mountain slopes
facing west.
Figure 4: Wind rose from Svalbard Airport
Weather station showing the hours of winds
for each direction, indicating prevailing
easterly wind direction. (meteoblue)
16
4.3. Glaciers in Svalbard
Around 60 % of the landmass in Svalbard is covered by ice. Since the central part of the
archipelago has the mildest climate on Svalbard, only 18 % of the Adventdalen area is covered
by glaciers (Hagen, 1993).
In the early Holocene (approximately 9700 years from now) most glaciers disappeared. In the late
Holocene the glaciers built up again and formed the maximum glacier extent at the Little Ice Age
in the 19th century. Nowadays the glaciers in Svalbard lose more ice than they accumulate. A
reduction of 7 % of glacier area at all glaciers in Svalbard was estimated from 1961 – 2000. The
future state of Svalbard’s glaciers will be primarily depending on summer melt but given the
climate projections of increased warming, this will result in glacier mass loss (Adakudlu,
Andresen, Bakke, et al., 2019).
The first glaciological studies at Svalbard started in the 19th century, which were carried out as
basic explorations of the land but soon also included some ice-core studies, meteorological
observations. Thus, the first data available from observations is from Dutch and English whalers
who shipped in the vicinity around the islands in the 17th century. Later the Norwegian Polar
Institute initiated mass balance observations starting in the 1950s (Hagen, 1993).
The annual net specific mass balance is defined as the average net mass which is left on the
glacier surface or removed from the surface, after a full year of accumulation and ablation.
Moreover, this balance can act as a climate signal response to changes on Svalbard and
correlates highly with summer melt and thereby with the temperatures. The glaciers on Svalbard
show a steady decrease in mass, correlating with the steady increase of the temperature’s sine
the last century (Hagen and Liestøl, 1990). Nevertheless, this trend varies locally on Svalbard
and because of the observation of elevation changes, it could be observed that glaciers with high
accumulation zones are still accumulating and therefore gaining mass today (Nuth et al., 2010).
However, the glacier surface variations can only be studied profoundly since the first satellite
images were available, before only terrestrial photographs showed some areas of glaciers
(Hagen, 1993). An estimation of the total ice volume of Svalbard resulted in a value of 7000 m3
of ice. This ice mass is contributing 0.01 mm annually towards global sea level rise, calculated as
an average value over the past 30 years (Hagen et al., 2003).
Another indicator to see changes on glaciers is the ELA, which is expected to rise by up to 400 m
until 2100 (Adakudlu, Andresen, Bakke, et al., 2019).Thus, at the equilibrium line the amount of
accumulation is exactly the same as the amount of the ablation and therefore the net mass
balance is zero (Hagen, 1993). Especially with the current climate variations, the changes in ELA
will show a response. GMB is determined by winter precipitation and summer temperature. In
Svalbard mainly, south-easterly winds bring moist air and precipitation and therefore the ELA is
much lower on the south-eastern side compared to the western and central-northern part. On the
inner part of Wijdefjorden, the longest fjord in north of Svalbard, the ELA is even at 800 m above
sea level due to the continental climate (Hagen and Liestøl, 1990).
Besides that, all morphologically classified glacier types can be found in Svalbard. Most prominent
are the valley and cirque glaciers but also ice caps and piedmont glaciers can be found.
Nevertheless, sometimes the classification seems to be difficult as the glacier systems are flowing
together and covering large landmasses, divided only by nunataqs and mountain ridges.
Geophysically classified, most glaciers in Svalbard are “polythermal”, meaning the accumulation
zone is at the pressure melting point and the ablation zone is below zero. In the accumulation
area the meltwater is draining into the firn column and refreezing within, and thereby releasing a
17
lot of latent heat which is trapped in the glacier and can not only warm the whole glacier but also
form subglacial taliks (unfrozen zones) in the permafrost. The temperature regime within a glacier
determines its flow activity and affects the glacial hydrology and is therefore crucial to understand
the glacier system (Hagen and Liestøl, 1990).
Due to the fact that most of the glaciers in Svalbard are surge type glaciers, the climate change
effects cannot just be deduced from the surface length change (Sevestre et al., 2015).
Especially surge type glaciers are assumed to be sensitive towards temperature changes
concerning the speed and initiation of surging. Further, increasing surge activity could lead to
more glacier mass loss in a short amount of time, which can contribute significantly to sea level
rise (Adakudlu, Andresen, Bakke, et al., 2019).
Generally, surveys of climatic mass balance need to be conducted over several years to detect
trends in glacier mass loss or gain (Hagen and Liestøl, 1990).
4.4. Field locations
This chapter provides detailed information about the glacier systems and their former and current
state, where the data collection took place. In the overview map of Svalbard, seen in Figure
5 Figure 6, the four red stars indicate the glaciers relevant for this thesis. The topographic base
maps used in this thesis are derived from the Norwegian Polar Institute, (2018).
4.4.1. Foxfonna
Foxfonna is s a cold-based glacier, located to the south-east of Longyearbyen and situated in
Nordenskiöld Land, Svalbard. Foxfonna, which reaches up to 800 m a.s.l. (metres above sea
level) in elevation has no observed surge history. However, a radio-echo sounding thickness
survey from 1972 indicated ice depths ranging from 25 metres up to 60 metres deep with
completely cold temperatures measured in all boreholes (Liestøl, 1974). Basal temperatures were
found to be colder than the melting point, throughout the whole shallow plateau ice cap, Foxfonna,
where layers of impermeable Si may keep the ice cap from being heated up by percolating melt
Figure 6: Svalbard Overview with fieldwork
locations (Norwegian Polar Institute ,2018).
Figure 5: Locations of the fieldwork sites in central Svalbard (Norwegian
Polar Institute ,2018).
18
water (Liestøl, 1974). The glacier can be divided into the Foxfonna ice cap on top of the Breinosa
plateau which flows in two outlet glaciers, Foxbreen and Rieperbreen as seen in Figure 7 (Liestøl,
1974). A coal seed mine (Gruve-7), which is still active, is running underneath the bedrock and
glacier. Above the Gruve-7 mine a maximum ice depth of 100 metres was measured in 2005
(Christiansen et al., 2005). Within 2011 and 2012 the net glacier mass balance was assumed to
be at - 0.89 and - 0.42 m s-1 and is very likely to continue at this mass loss rate with the current
climate. Further no Si accumulation was observed at the end of both summers. There was a small
firn area (< 0.3 km2) observed on the higher end side where the glacier ends in steep slopes
(Koziol et al., 2019). Albeit negative mass balance observations in 2017, local mass increases
because of Si and firn patches were visible above the ELA. Runoff from the accumulation zone
was observed to be routed along low gradient, hydraulically inefficient supraglacial discharge
channels, which later joined into steeper, highly efficient supraglacial and englacial/subglacial
channels downstream (Rutter et al., 2011).
4.4.2. Nordenskiöldbreen
Nordenskiöldbreen is a vast glacier located inside a fjord system called Billefjorden and opposite
of Pyramiden, an abandoned Russian coal mining town. Because of its location deep within a
fjord system, it has a more continental precipitation pattern, which means up to 45 – 60 % less
accumulation compared to the west coast of Svalbard (Winther et al., 1998). The glacier has a
high accumulation zone reaching up to more than 1000 m in elevation towards Lomonosovfonna.
Figure 7: Foxfonna Location Map, A shows the glacier with the topographic map and B indicates the
loaction. C shows the glacier from a sattelite image (Norwegian Polar Institute ,2018).
19
In the middle part of the glacier, two rock formations stand out, which are named Terrier fjellet
and De Geer fjellet as seen in Figure 8.
Nordenskiöldbreen has an ELA of approximately 719 m a.s.l.. Further the glacier is characterized
by a cold ablation ice area and warm firn and ice in the accumulation zone (van Pelt et al., 2012).
Since the end of the Little Ice Age, the glacier retreated with a mean rate of 35 m which is also
stands for a loss of 5.3 % in glacier area (Rachlewicz et al., 2007).
Besides the thinning, Nordenskiöldbreen appears to have a dynamic accumulation zone above
720 m a.s.l. with enhanced mass gains due to refreezing of 25 % of all melt and rain water within
the snow and firn column. The refreezing contributes 0.27 m w.e. a-1 to the net mass balance (van
Pelt et al., 2012). The firn layer depth was estimated to be around 20 metres, derived from cores
in 2009 (Wendl, 2014). Figure 63 in the Appendix shows Nordenskiöldbreen from the south facing
Pyramiden and Ferrier fjellet and Terrier fjellet in the middle, on an old Aerial image made in 1936
by the North Polar Institute. This side shows the not surging, retreating ice stream from top to
bottom, where the fieldwork for this thesis was conducted.
4.4.3. Von Postbreen
Von Postbreen is a large land-terminating glacier, the biggest glacier of Brünsow Land together
with Nordenskiöldbreen. The glacier stretches over a length of 31 km and an area of 1 km2 (König
et al., 2014). The glacier has two outlets which are terminating on land since the last surge of the
glacier in 1870 (DeGeer, 1912). A large moraine, which was deposited during the surge, is
separating the glacier from the neighbouring Tunabreen glacier (Forwick et al., 2010). North of
Von Postbreen is another glacier, Bogebreen, which surged in 1980 (Dowdeswell, 1984). At the
moment, Von Postbreen is not active.
Figure 8: Nordenskiöldbreen Location Map, A shows the glacier with the topographic map and B indicates
the loaction. C shows the glacier from a sattelite image (Norwegian Polar Institute, 2018).
20
A radar survey performed by Sevestre et al. (2015) with a 100 MHz antenna revealed the same
structure that was found by Dowdeswell (1984). A two layered thermal regime, with a cold layer
in the ablation zone with a thickness of 70 – 110 m and indications of completely cold glacier
tongues. However, the rest of the glacier, is warm based and the warm ice is even up to 200 m
thick. The study further showed that in the accumulation zone of Potpeschniggbreen (which is a
name for a small glacier contributing to Von Postbreen, indicated as red star in Figure 9) the cool
top layer just stops at an altitude of around 725 m above sea level (Sevestre et al., 2015). Von
Postbreen corresponds to the polythermal glacier of type E meaning it is mainly temperate ice
with a cold surface layer around the ablation zones (Pettersson et al., 2003). The maximal glacial
extent of Von Postbreen was observed before its surge in 1870 (Hagen, 1993).
4.4.4. Drønbreen
Drønbreen is located in Isfjorden and represents one of those glaciers that are known to have
surged in the last century (around 1900). The glacier stretches from 300 m up to 1000 m in
elevation (Hagen, 1993). It is a cold based glacier despite an ice depth of 186 m at the deepest
part (UNIS Glaciology Course, 2019). A wind drift study was obtained in 2005 in the area of
Drønbreen. It showed that accumulation is highly influenced by local topography, leading to
additional accumulation loading on the west-facing slopes of Drønbreen (Jaedicke and Gauer,
2005). Figure 10 shows the location and extent of the glacier and on the Satellite image in C the
ELA is visible.
Figure 9: Von Postbreen Location Map, A shows the glacier with the topographic map and B indicates the
loaction. C shows the glacier from a sattelite image. The red star indicates Potpeschnigbreen, where most
of the fieldwork was carried out for this thesis (Norwegian Polar Institute, 2018).
21
5. Methods
Several different surveying methods were used to answer the research questions. Trying to
understand the dynamics of one accumulation season and the state of the accumulation zone is
a complex task, which required a combination of several different methods, applied during
different times of the year. The fieldwork was carried out in spring and summer 2019 and included
snow profiles, snow-depth surveys, GPR radargrams (GPR is the Abbreviation for Ground
penetrating radar), firn cores, iButtons and a thermistor string station.
Figure 11 indicates an overview of the methods used, the data gained from the methods and the
way each method contributed to the results. The MALÅ Radar with a 500 MHz shielded antenna
was used to collect radar transects to gain knowledge about the snow and firn depth and about
the possibility of liquid water pockets, or warm firn.
In total 15 snowprofiles were dug and analysed to compare the snow properties over elevational
changes, including density, stratigraphy and temperature. Furthermore, the snow stratigraphy
allows to see spatial distributed differences in layering and local climate patterns.
Two different temperature loggers have been used in this work. First an iButton probe was built
and installed at Nordenskiöldbreen to obtain more information about subsurface melt, snow
temperature evolution and to simply test the durability of iButtons as measurement instruments
in harsh environments.
Figure 10: Drønbreen Location Map, A shows the glacier with the topographic map and B indicates the
loaction. C shows the glacier from a sattelite image (Norwegian Polar Institute, 2018).
22
Second, a station with two highly sensitive thermistor string cables was installed at Foxfonna in
order to obtain detailed data about the subsurface temperature development over the season and
also pick up latent heat release and small-scale temperature variations within the snow and firn.
Using the thermistor string and iButton datasets, the subsurface temperature evolution was
plotted. Further with the continuous temperature measurements and the data from two weather
stations, the density evolution was modelled. Finally, the refreezing rate was modelled using the
calculated density and the temperature evolution.
The fieldwork on Drønbreen was carried out in February and the data was collected together with
the students from the AG-825 Glaciology course at UNIS. This includes the snow profile data, the
radar data and the core data from Drønbreen. All the other data was collected by the author of
this thesis with the help of fieldwork assistants and the supervisor.
5.1. GPR (Ground penetrating radar)
A GPR survey was obtained using a MALÅ Radar with a 500 MHz shielded antenna. The aim
was to gain knowledge about the firn and snow depth and distribution. Further the radargrams
were used to identify possible warm firn zones or liquid water. With the ability to see subsurface
structures through the GPR, the distribution and state of the accumulation zone can be
understood.
Figure 11: Fieldwork methods and data collection overview
23
5.1.1. Characteristics
To be able to interpret collected radar images some basic principles need to be understood.
Ground penetrating radar is using electromagnetic waves to penetrate the Earth’s surface. The
radar consists of two antennas, one which is transmitting a wave along the wire and generates
an electromagnetic field in which an electromagnetic pulse is released and travelling through the
underlying medium. Depending on the ground properties, the pulse is reflected and is detected
by the receiving antenna. The signal recorded by the receiving antenna can be used to convert
the two-way travel time into time and depth of the scanned medium (Bogorodsky et al., 1985).
Depending on the electromagnetic properties of the scanned area, the velocity and attenuation
of the waves is varying with the medium. Fortunately, in snow and ice the waves loose little signal
allowing deep wave propagations, ranging from several kilometres with low frequency systems
and several metres for high frequency systems. Changing reflection signals in snow and ice are
due to variations in the dielectric constant, the permittivity and these are mostly because of density
changes. In pure ice, the electric permittivity is around 3.14 to 3.18 (high frequency limit), whereas
the permittivity of polycrystalline ice with a density of 917 kg m-3 at 20 °C is 3.15 (Kovacs et al.,
1995).
5.1.2. Survey design
The GPR data were collected using a MALÅ RAMAC radar system (MALÅ Geoscience Ground
Penetrating Radar instruments, Malå, Sweden), which is operating as a combination of a shielded
antennae with a frequency of 500 MHz together with a MALÅ ProEx control unit and a Monitor
seen in Figure 12 and in Figure 13.
To increase the signal to noise ratio during the data collection, 8 traces were stacked at each
shot-point location. The 8-fold stack is proposing a decent compromise between measuring
velocity and data-quality. The acquisition mode of time triggering was set to 0.2 s for travelling on
snowmobile at a speed of around 15 m s-1. These settings together with the 500 MHz antenna,
yield a maximum resolution of 0.11 m vertically and 0.21 m horizontally. GPS measurements were
taken separately with a Garmin Montana Handheld GPS system (Montana® 680t, Garmin
International, Inc., Olathe, KS, USA) to track the radargram path.
For the sites Drønbreen, Potpeschniggbreen and Nordenskiöldbreen the snow scooter was used.
The GPR was transported in a pulka and attached with a rope to the snowmobile as seen in
Figure 13. At Foxfonna glacier the survey was conducted with skis, because of restricted
snowmobile allowance in the area. The GPS coordinates were collected separate to the GPR,
except for the Drønbreen site. For the Data collection at Foxfonna the radar was also in a plastic
pulka and attached with ropes to the person carrying it, seen in Figure 12.
GPR, on Potpeschnigbreen was performed due to the interesting radar scattering observed in the
study from Sevestre et al. (2015) which showed that the firn was getting temperate up to the
surface at an elevation of around 800 m a.s.l..
24
5.1.3. Processing
Editing and post processing of the radar data was done using Reflex W (Version 9.1, Sandmeier
geophysical research, Karlsruhe, GER), program. First processing step was to remove files which
were recorded while standing. The GPS dataset was adjusted to the measurements of the GPR
to know the exact location of the measurement and to match a GPS position to every single trace
recorded. This was done by creating a new Cor file in an Excel sheet matching the collected
traces with the GPS positions.
First the radargrams were imported into Reflex W. The first processing step was a 1D Dewow
filter (subtract mean) application. The Dewow filter is used when the data shows a low- frequency
content or DC-shift (Sandmeier, 2019). The plot scale was adjusted to see the radargram content
and further the start time was corrected, using the correction of max phase wrap. Additionally,
each radargram was corrected to the time of the first correct signal arrival which also removes
drifts in the first arrival times of individual traces. Further the data was improved using the
equidistance traces function. For some radargrams a gain function (energy decay) was applied
to enable the visibility of signals from deeper areas. Finally, the radargrams were corrected for
the topography to see if the observed details match with topographic features.
5.2. Coring
Firn cores were drilled with the aim of (i) measuring snow temperature and density, and to (ii)
compare the firn stratigraphy with the collected GPR radargrams. The cores were collected on
Von Postbreen (only on the part of the glacier named Potpeschniggbreen) on Nordenskiöldbreen
and on Drønbreen.
The cores were excavated using a Kovacs Mark V coring system (Kovacs Enterprise LLC,
Roseburg, OR, USA), which is able to retrieve a 0.14 m diameter core up to 1 m long as seen in
Figure 14C. The temperatures were measured right after retrieving the core with Lollipop
Thermometers (TraceableÒ LollipopÔ waterproof/shockproof thermometer, TraceableÒ Products,
Webster, TX, USA). The temperature was measured not in equal distance or at each stratigraphic
Figure 12: RAMAC radar system setup in pulka
attached with ropes to skier. Radar is in the
pulka, monitor in the hands attached to the
backpack with the control unit. Photogr aph A, B
and C © Kristin Fredheim
A B
C
Figure 13: RAMAC radar system with the radar
setup in the pulka, the monitor and the control
unit on the snow-scooter.
25
layer because some of the layers were pure ice or too hard to get the thermometer in. In order to
get the best resolution, several thermometers were used in short distances (0.02 – 0.05 m) to
derive the temperatures as seen in Figure 14A. Some of the thermometers completely went
through the ice core and therefore were not correct. Another difficulty was that the temperatures
of the core and the outside air temperatures were not the same. This led to quick adaption of the
core temperatures to the outside temperatures and the measurements needed to happen
extremely quickly in order to measure undisturbed data.
5.3. Snow profiles and snow depth survey
Several snow profiles have been dug and analysed on all of the studied glaciers. The aim of
collecting the snow profiles was to understand the accumulation in detail on different locations,
ranging from low to high elevation and showing the effect of glaciers in with different aspects.
Additionally, the density measurements allowed insights in the winter accumulation and the snow
stratigraphy showed the snow metamorphisms.
Further the snow depth was measured using a simple probe and the aim was to get an overview
of the current spatial accumulation patterns of snow with elevation and to further validate the GPR
surveys.
5.3.1. Experiment Setup
First, a good site was selected, representative for the respective elevation, and in a second step
the depth of the snow profile was measured using a snow probe. When the location seemed
Figure 14: The coring work is shown in these photographs. A shows the measuring
of the firn cores directly after retrieval, B shows the location on Nordenskiöldbreen
with the whole core in front. C shows Molly Peek drilling the core and standing on
the base of the snow profiles. Finally, D shows a detail of the extracted core on
Nordenskiöldbreen with melt layers.
A B
CD
26
representative, a hole was dug, using a snow shovel, until firn or glacier ice was reached. The
first measurement was the temperature in 0.1 m distances, starting at the top and ending at the
bottom. Afterwards, the snow stratigraphy was observed, including crystal shape, size and
sometimes hardness seen in Figure 15C. Finally, the density was measured also every 0.1 m
using a Snow Rip Cutter 1000 with lid and then the snow was weighted with a digital weight scale
(A and D Company HT-5000scale), all from the Snowmetric SnowProKit (ProSnowKit,
Snowmetrics, Fort Collins, CO, USA). All snow samples were extracted from a pit wall which is
not directly facing the sun to have no changing effects from the solar radiation. The temperature
measurements were done, similar to the firn cores, with the Traceable® Lollipop thermometers.
The Lollipop thermometers can be used indoors or outdoors, and the measurement range lies
between minus 50 °C and 300 °C. Temperature resolution is 0.1 °C from - 20 °C to 200 °C,
otherwise 1 °C. Accuracy is ± 0.4 °C or ± 1 °C between 0 °C and 100 °C.
To perform a snow depth survey, an eTrex 10 Garmin GPS was used to yield the coordinates and
a simple avalanche probe was used to get the snow depth. The survey was conducted with
snowmobile or walking on skis depending on the location allowances.
5.4. Temperature loggers
5.4.1. Measuring temperature with iButtons
Five iButtons were mounted on an aluminium pole to measure continuous subsurface
temperature data within the snowpack on Nordenskiöldbreen. The iButton temperature data was
used to (i) plot the subsurface temperature evolution and (ii) to model density evolution and further
use it to model refreezing of meltwater.
Figure 15: Snow profile sampling. A shows the snow profile wall with the typical
holes after taking snow out with the Rip Cutterfor density measurments.
Photograph B shows snow profile examination on Drønbreen (SPD1) and
photograph C shows the crystal stratigraphy collection. (Photograph B © Enaut
Izagirre; Photograph C © Lukas Hillisch)
A
B C
27
Characteristics
The Thermochron iButton, (Thermochron® iButton® device, DS1921G, Maxim Integrated, San
Jose, CA, USA), can measure temperatures from minus 30 °C to plus 70 °C and records the
result in a protected memory section. The record timing can be adjusted, and 2048 temperature
values taken at equidistant intervals ranging from 1 to 255 min can be stored. Data is transferred
through the 1-Wire® protocol, which requires a single data lead and a ground return. Each one
has a unique registration number. The cover is made of stainless-steel package, which is resistant
to moisture and dirt, but not waterproof. The DS1921G measures temperatures with a resolution
of 0.5 °C and automatically wakes up to record temperature in the programmed interval. The
inspiration to use this iButtons came from a study where the iButtons recorded nice temperature
gradients within the snowpack over a winter season which were used with in situ observations to
detect small scale snow metamorphisms (Gagne, 2018).
Experiment Setup
First the iButtons were tested if they measure correct values and compared with the Lollipop
Thermometer. After the validation of the accuracy, they were put on a normal avalanche probe,
with a length of 2.2 m and in equidistance intervals of 0.25 m.
To prevent heat transfer from the metallic probe, the probe was insulated with a silver fabric. The
iButtons were programmed to record temperature every 80 minutes, to yield the maximum data
for the given field period and storage. To waterproof the iButtons, simple Balloons were used
which was shown to work successful in other works with water contact (Tully, 2007).
Unfortunately, one balloon broke already before setting up of the device, so further silver tape
was used to ensure that the iButtons stay waterproof as seen in Figure 16C Figure 16D.
The setup was installed on the 11th of May 2019 on Nordenskiöldbreen at an elevation of around
639 m a.s.l. as seen in Figure 16A. A snow profile was made on the same day and the elevation
was chosen as compromise for a measurement within the area of the ELA and the walking
distance in summer, to collect the loggers again. At the end of the melt season the iButtons were
collected to interpret the data. The iButtons were found half buried seen in Figure 16B, at
N 78°38.4392 E 17°23.5147 and 631 m a.s.l. measured by a Garmin eTrex 10 (eTrex® 10,
Garmin International, Inc., Olathe, KS, USA). The air temperature was - 0,4 °C and the snow
surface temperature was at 0 °C. After the download the data was ready for the interpretation and
the iButton data was used from 12th of May 2019, 10:40 PM until 22nd of August 2019, 02:40 PM
for further interpretations. In total the iButtons were mounted on the glacier from the 11th of May
2019 until the 22nd of August 2019 and also recorded the whole time.
The software used to download the data was One-Wire-Viewer for Mac, which is a Java™
demonstration application still released in BETA version. To download the data from the iButton
and connect to the One-Wire-Viewer, the DS9490 Adapter was required.
28
5.4.2. Measuring temperature with thermistor strings
The purpose of the thermistor string installation was the continuous measurement of
temperatures to detect melt, refreezing and the length of the snow cover. Therefore, a complete
station was built on Foxfonna glacier and two highly sensitive thermistor strings were drilled into
the snow and ice. After downloading the data was further used to model the subsurface
temperature evolution and refreezing within the snow.
Experiment setup
On Foxfonna glacier, a station was setup with two thermistor strings (CS225-L SDI-12
Temperature Sensor String, Campbell Scientific, Inc., Logan, UT, USA) entering the snow, firn
and glacier. The used thermistor consisted of temperature sensors mounted in rugged steel-
reinforced cable, which allowed the burial.
The station was built on the 3rd of May and finally downloaded on the 3rd of August 2019. The
datalogger used was a CR 1000 (CR1000, Rugged Data Logger, Campbell Scientific, Inc., Logan,
UT, USA). Before the thermistor string was brought into field, it was calibrated in a water bath and
wired to the several relay multiplexors. The thermistor resolution is 0.0078 °C, according to the
manufacturer.
Figure 16: Installation of iButtons at Nordenskiöldbreen. Photograph A shows the iButton Probe
right after mounting, in the afternoon of the 11th of May 2019. Photograph B shows the installation
on the 22nd of August 2019. The probe is lying, with two iButtons buried in snow. Photograph C
shows the original waterproofing with balloons, but the balloon got a hole and additional tape
was used to ensure waterproofing as seen in photograph D, the final setup.
A B
CD
29
The setup of the station was done using a Kovacs Corer to drill a hole for the string and further,
reuse the drill chips, snow and water was filled into the hole. Usually such a setup is best to install
before the onset of the snow season, but it was logistically and timewise not possible for this
project. Water was cooked in a stove and collected in a big bucket for filling. The 24 loggers on
each string have a distance of 0.25 metres and the string length was 6 metres. During the drilling,
the power drill gave up and one extension rod for the corer broke, which resulted in the whole
setup being drilled 4.5 metres deep.
The station was put on several single connected aluminium poles and drilled into the glacier.
Unfortunately, two of the poles where not mounted well, so the station started leaning towards
the mountain side quite quickly with the local weather seen at Figure 17B. Figure 18 shows the
glacier on the day when the data was downloaded, the 3rd of August 2019.
Further the glacier surface is well visible on Figure 18A, Figure 18C and Figure 18D from above.
After downloading, the temperature values were calculated using the measured voltage and
resistances with the Steinhart Equation, which is given by the manufacturer. The recorded data
were checked against observations. In the result and methods section the thermistor data were
named Foxfonna thermistor 1 for the first thermistor and Foxfonna thermistor 2 for the second
thermistor string to perform the modelling without mixing datasets. The two cables of the
thermistors are visible after melting out in Figure 17B and Figure 18B.
Figure 17: The thermistor string station. The station, photograph A
on the 28th of May 2019 and B on the 24th of July 2019.
A B
30
5.5. Density evolution modelling
The aim of the density evolution modelling was that the resulting dataset was necessary as input
to calculate refreezing. The density was calculated using the measured starting density, the
measured temperature data and freely available data from weather stations. The density was
modelled for each timestep for the temperature dataset from the thermistors and also from the
iButtons data. Finally, the density evolution is also interesting to compare with the weather data
and snow depth development.
5.5.1. Model choice
Generally, density evolution is hard to measure, because it would need in situ observations nearly
every other day. Furthermore, once the density is measured, the sample location is actually
destroyed, therefore the evolution is also hard to measure in situ. Another possibility is to simply
assume the density development using the density that was measured at the beginning. However,
for this project, the density data was not available at any other point during the observation time.
Therefore, an assumption of densities was not carried out, because an interpolation from the
starting density would lead to extremely imprecise results. Nevertheless, a model approach was
used to achieve the density evolution dataset.
Most scientific work that quantifies refreezing using advanced meteorological models that solve
multiple physical calculations at the same time combined with in situ measurements. Since none
of the advanced models was available, because most of them are not open source, another
approach to solve the density evolution was used. However, a publication was found, which
calculates the density development of a snowpack, based on simple meteorological parameters
available from weather stations (Meløysund et al. 2007). Ultimately, the approach for the density
Figure 18: Revisit of thermistor string station on the 3rd of August 2019. A shows the summer surface of the location,
B the thermistor station after a repair and proper demounting and the thermistor strings, melted out in front. C
shows the thermistor string station in the red circle, showing the steepening of the slope and the Firn layer above.
D is showing the station, one more time from above and very good visible the bare ice, Si and firn zone.
31
development modeling for this thesis was chosen based on the publication by Meløysund et al.
(2007). The main reason to use this publication was, that the weather data is freely available for
download from the weather stations and the initial density needed was measured in situ.
The data used for the modelling was taken from the Unis
weather stations, Breinosa and Adventdalen, the location
is indicated in Figure 19. From Adventdalen the
precipitation and radiation measurements were used. From
the Breinosa weather station, relative air humidity, wind
velocity, the atmospheric pressure and the air temperature
was used as input data for the modelling. The amount of
rain was calculated by precipitation data measured as sum
of precipitation per day in Adventdalen and as snow
events, using the air temperature from Breinosa station as
threshold.
5.5.2. Model description
In the publication from Meløysund et al. (2007) the density is calculated using a multiple
regression analysis based on several meteorological datasets complemented by 608 snow
density measurements in Norway measured between 1967 – 1986. The climatic region, where
the density measurements are from, is quite stable in winter. The analysis resulted in a good
correlation between meteorological data and measured snow density. The statistical model
resulted in density values with an accuracy of 70 % and a standard deviation of 24 kg m-3.
In detail, Meløysund et al. (2007) evaluated meteorological parameters for their significance of
density evolution. The parameters are listed in Table 1.
Table 1: Evaluated parameters checked for a correlation with density by Meløysund et al. (2007).
General
parameters
Parameter details
Velocity
Wind velocity [m s-1]
Wind velocities during snow events [m s-1]
Average wind velocity [m s-1]
Average wind velocities during snow events [m s-1 ]
Precipitation
Precipitation [mm]
Amount of precipitation during snow events [mm]
Amount of precipitation during rain events [mm]
Humidity
Relative air humidity [%]
Relative air humidity during snow events [%]
Energy
Absorbed radiation by the snow cover [W m-2]
Time
Daily sun hours [h]
Temperature
Snow temperature [°C]
Air temperature [°C]
Average air temperature [°C]
Positive temperature [°C]
Pressure
Atmospheric pressure [hPa]
Amount of atmospheric pressure [hPa]
Density
Starting density [kg m-3 ]
Height
Snow depth [cm]
Factors
Snow drift, amount of times wind velocities exceeded
4 m s-1
Albedo, ratio of incoming and outgoing radiation
Figure 19: Location of the Unis weather
stations, Breinosa (close to Foxfonna) and
Adventdalen (Norwegian Polar Institute,
2018).
32
To reveal possible correlations between the parameters listed in Table 1 and the density,
Meløysund et al. (2007) used a multiple regression analysis. The analysis revealed that only
seven parameters had a significant linear correlation in terms of impact to the calculated density
with high significance (P = 0.000). Therefore, the seven significant parameters listed in Table 2
received a defined bias.
During the analysis the correlation of 70 % (R²) and a standard deviation of 24 kg m-³ where
achieved, further the bias
W
was introduced, which weights the different parameters. Using the
results of the multiple regression analysis, following seven parameters were defined as significant
for the density development with the consecutive biases set as seen in Table 2 (Meløysund et al.
2007).
Table 2: Parameters correlating with density, model results by Meløysund et al. (2007).
Parameter
/XY
Bias defined by Meløysund
et al. (2007).
WY
Starting density ZT [kg m-³]
Relative humidity during snow events 86E')? [%]
0.0167
Snow drift, number of times the wind velocities exceeded 4 m s-1 [\D&R+
2.23
Amount of the atmospheric pressure ]+)+ [hPa]
−4.66
Snow depth U [cm]
−1.84
Absorbed solar energy by the snow cover ^E*' [W m-2]
2.40
Amount of precipitation during rain events 8C [mm]
−1.89
Following the defined parameter biases the equation to retrieve the density is shown in ( 13 ).
5.5.3. Model workflow
The following section explains very carefully, the exact workflow used to get the density evolution
in this thesis (Figure 20).
In order to calculate the density for this work, the significant parameters defined by Meløysund et
al. (2007) were used. The program used to run the model was MatLab (R2019b, The MathWorks
Inc., Natick, MA, USA). The parameters and the origin of the data used are explained here.
Figure 20: Density evolution workflow.
33
The starting density,
ZT
[kg m-³] was estimated by the stratigraphy. The density was taken in a
snow profile with an accuracy of 0.1 metres, close to the thermistor strings on Foxfonna and for
the iButtons, the density from the location next to the installation was used.
The Relative humidity during snow events,
86E')?
[%] was calculated with the relative air humidity
RH [%] and the amount of precipitation during snow events snow Day [mm] (Equation 8):
86E')? #
_
`abcZ<d@86
+
&ef ///////
0
=
5
( 8 )
Every time period with precipitation was related with the air temperature and classified. The
needed data was downloaded from UNIS Breinosa weather station and UNIS Adventdalen
weather station, which were collected every five minutes or once a day. The bias was
WghLijM #
;k;lmn
, as defined by Meløysund et al. (2007) see Table 2.
The snow drift is the number of times the wind velocities exceeded 4 m s-1 and is called
[\D&R+
in
the equation. The bias was set as
WRopqrs # tktu
, as defined by Meløysund et al. (2007) (Table 2).
The snow drift data was derived from wind velocity data from Breinosa, gathered with a frequency
of five minutes (( 9 )):
[\D&R+ # [$v?&'\ w x/2/`3f,//////0(5
( 9 )
The snow depth,
U
[cm] was derived from data analysis of the measured subsurface temperature
evolution, which was gathered every ten minutes at Foxfonna glacier. If the snow layer was cooler
than 0 °C, the layer was considered to be snow and if it was warmer no snow was assumed. The
bias for snow depth was set
W\# lkyx
as defined by Meløysund et al. (2007) see Table 2.
The amount of the atmospheric pressure,
]+)+
[hPa] was calculated from sampled data with a
frequency of five minutes, using the atmospheric pressure data
]
[hPa] at Breinosa weather
station (( 10 )). The bias was set as
WCsjs # xkmm@l;3z
as defined by Meløysund et al. (2007)
(Table 2).
]+)+ #_]
+
&ef ///////0J{<5
( 10 )
The solar energy absorbed by the snow cover
^E*'
[W m-2] was calculated by the sampled daily
sun hours s [h] and the albedo a [%]. The data was provided by the weather station in
Adventdalen, measured with a frequency of five minutes (( 11 )):
^E*' #_`@$l(<,
+
&ef ///////01/2345
( 11 )
34
Sun hours
`
were defined as following:
Due to unknown ground circumstances at the weather station, the albedo a was estimated
constant with
< # ;ky|/
0
=
5. During the modelling another threshold was tested but didn’t work
great, because the location specific albedo values where unknown. When the amount of incoming
short wave radiation
`c&'
[W m- 2] exceeded a threshold of 120 W m-2 then sunshine can be
assumed (WMO, 2003). At this point the bias was set at
W+L}i # tkx
as defined by Meløysund et
al. (2007) see Table 2.
The amount of rain precipitation
8C
[mm], was calculated by the measured precipitation data as
sum of precipitation per day in Adventdalen (( 12 )). Precipitation as snow, has been identified
using the air temperature in the same region as threshold. The threshold temperature for whether
the precipitation falls as rain or snow was set to 0
~
. The bias was set as
Wg•# lky€/
as defined
by Meløysund et al. (2007) see Table 2.
8C#_A<•aZ<d
+
&ef ///////0225
( 12 )
To obtain a better daily resolution, all of the data was interpolated to a time step of five minutes.
It is known that interpolation blurs the data and is only, in this case a linear, a rough estimation of
the timesteps between two known data points, but as refreezing occurs in very short time
windows, this step had to be undertaken.
Including all the mentioned parameters above, the density
Z
[kg m-3] could be estimated with
following equation (( 13 )):
Z'#/Z'3f -WghLijM @86E')? -WRopqrs @[\D&R+ - W\@U-WCsjs @]+)+ -Wg•@8C//0H‚/23ƒ5
( 13 )
For Svalbard the values were overestimated around three times, so instead of analysing and
changing the particular biases, the whole model outcome was multiplied by a factor of 0.35.
When estimating the density for Foxfonna, the thermistor strings were buried also in glacier ice,
but the equation doesn’t apply for ice as it based on snow density. Therefore, the glacier ice was
assumed as steady with the same density throughout the model run.
5.6. Calculating the amount of refreezing
The aim of this modelling was to quantify the amount of water which refreezes within the snow or
firn pack on the glacier surface. This is crucial to understand the changes in the surface mass of
a glacier. The estimation of refreezing was based on the density evolution modeling from
Meløysund et al. (2007) and a thermodynamic equation proposed by Ward Van Pelt in personal
communication for the measured dataset. The workflow diagram in Figure 21, shows the single
steps to model refreezing. Each step is explained in detail in the following section.
35
5.6.1. Model background
The refreezing rate can be estimated from the temperature evolution by drawing attention at the
rapid increases. The time evolution of subsurface temperatures is described by the following
thermodynamic equation describing local heating through heat diffusion and refreezing
(Equation 14; van Pelt et al., 2016):
Z@B] @U9
U^ #U
UP„H@U9
UP…-†@.[//////
( 14 )
where
‡
is the density [kg m-3],
ˆ‰
is the specific heat capacity of snow [J kg-1 K-1],
\Š
\+
is the
temperature change in time [K s-1],
H
is the effective conductivity [J s-1 K-1 m-3],
P
is depth [m],
†
determines the refreezing rate [kg s-1 m-3] and
.R
is the latent heat of fusion (~ 3.3 x105) [J kg-1].
In this thesis case it was assumed that conduction is negligible, since refreezing is a very fast
process, while conduction takes much more time. This assumption is true for the sharp spikes of
temperature changes within the measured temperature dataset. The equation could be rewritten
as following proposed by Van Pelt in personal communication (( 15 )):
† # l
.R@Z@B] @U9
U^///////0H‚/`3f/23ƒ5
( 15 )
Therefore, the equation results in the refreezing rate
†
[kg s-1 m-3] as a function of snow density
‡
and temperature change
\Š
\+k/
5.6.2. Model approach
Using the modelled density dataset and the measured subsurface temperature evolution, the
refreezing rate F [kg m-2 s-1] could be modelled too. To complete this, the following parameters
Figure 21: Refreezing work flow diagram.
36
were needed. First, the latent heat of fusion
.R
[J kg-1] which is in general defined as constant
.R# uku@ l;‹k
The density input dataset
Z
[kg m-3] used in this equation was modelled using the equations from
Meløysund et al. (2007). The specific heat capacity of snow
BC
[J kg-1 K-1] was defined as constant
BC#t;€;/
J kg-1 K-1. The workflow used in this thesis is shown in a detailed workflow chart in the
Appendix Figure 62.
The temperature change over time
\Š
\+
[K s-1] was calculated with data measured every ten minutes
at Foxfonna (( 16 )):
U9
U^ # 9U9
U^///////0Œ/`3f5
( 16 )
With all the known parameters, the actual refreezing rate F could be calculated using the
thermodynamic equation shown in ( 15 ).
Finally, to calculate the total amount of refreezing in
0225
the equation was rewritten as following
(( 17 )):
† # l
.R@>E
>?@JE@B]@U9
U^///////0225
( 17 )
The final results of using this model outcome needed to be adjusted. The model calculates
refreezing based on temperature changes over time. However, within the snowpack little
temperature changes happen all the time resulting in a background noise which leads to
overestimated refreezing values. Therefore, a filter was applied which removes all refreezing
events below 0.1mm. This adjustment was applied by decreasing all the data by 0.1mm.
6. Results
In the following chapter, the results from the fieldwork are presented and described. Each sub-
section is representing a result related to the methods.
6.1. GPR
GPR profiles from each of the four studied glaciers and adjacent areas are presented in this
section. The snow depth surveys have been performed according to the GPR track, to validate
the snow depth with the obtained radargrams. Drønbreen was the first surveyed location on 26th
of February 2019. Secondly on Von Postbreen, the survey was conducted on 26th of April 2019
and the exact location was in two outlet glaciers which merge into Von Postbreen, named
Potpeschniggbreen and Phillipbreen. On Nordenskiöldbreen the GPR radargram was collected
on the 7th of May 2019 and the last survey was conducted on Foxfonna on the 28th of May 2019.
37
6.1.1. General results
The original setting of wave propagation velocity of 0.2 m ns-1 corresponded well with the
observed snow depth and core surveys, compared to the basal reflections at the snow/firn and
snow/ice interface. Therefore, the travel velocity of the radar waves was not calculated through a
CMP (common midpoint analysis) because especially the snow depth and the high-density
accumulation found at the bed of each snow profile aligned with the reflections in the collected
radargrams. This dense reflection can be observed in every single radargram collected at each
glacier and can be seen in the following chapters.
The presented profiles are not elevation corrected, which can lead to a blurred vision of details
throughout the long datasets. However, all data have been checked with elevation corrected and
it appears that a “bowl” shape of firn and Si, which can be seen in almost all radargrams, lower
than the current ELA, always forms in topographic depressions where it smoothed out the
topography. Therefore, the radargrams can be found elevation corrected in the Appendix (Figure
78, Figure 79, Figure 80, Figure 81, Figure 82 and Figure 83). The following Table shows a
summary of the collected results. The firn depth and snow depth in metres should give an
overview and is estimated from the radargrams. The result of the ELA height seen in Table 3 is
estimated because of the GPR results. However, it was a winter survey and therefore, to yield
more accurate ELA locations, the summer glacier surface needs to be examined carefully.
Table 3: Overview of the GPR results.
Nordenskiöld
breen
Phillipbreen
(Von Postbreen)
Potpeschniggbreen
(Von Postbreen)
Drønbreen
Foxfonna
Firn
depth
[m]
max 18 m
highly variable
general 2 m
max 8 m in
1800-metre-long
surface
depression
10 m at the upper
3500 m of the GPR
track
General
around 10 m
max 20 m
No firn only SI
or old firn with
a thickness of
few
centimetres
Snow depth
range [m]
1– 2.1
2 – 2.44
2 – 2.75
1.2 – 2.4
1.50 – 2.0
ELA
[m a.s.l.]
600
600
600
600 – 700
650 – 700
6.1.2. Wet firn
The radargrams have been compared with publications showing wet firn and water tables
(Christianson et al., 2015). To detect a water table, a strong near horizontal reflector crossing
other reflectors needs to be apparent. As snow and firn is layered, the crossing reflectors should
be easy to spot, and therefore are absent in the collected radargrams. Nevertheless, the large
white zones observed within each of the “bowl” structures can be interpreted as refrozen
meltwater storage. No indications of a water table lead to an assumption of no present firn
aquifers, which was expected, because firn aquifers need high accumulation rates and a lot of
latent heat exhausts from refreezing meltwater which percolated deep down in the firn layer, to
exist. Nevertheless, meltwater is indeed draining into these distinctive “bowls” and refreezing due
to shallow depths leads to clear seasonal layering with huge differences in density.
38
6.1.3. Glacier hydrology
Larger channels are seen as hyperbolas in unmigrated data and can be found in no dataset
except for Drønbreen.
On Foxfonna, indications of a supraglacial drainage network are visible. On Von Postbreen and
Nordenskiöldbreen, none of these features were observed. Further to note is, that while collecting
the iButtons on Nordenskiöldbreen in summer, only very little supraglacial streams were visible,
except for the glacier near the margin. Further, a large moulin was observed on the northeast end
at the part of Nordenskiöldbreen where it continues towards the crevassed area.
6.1.4. Foxfonna
At Foxfonna, the snow depth picked up by the radar, is aligned with the in-situ measurements
done by probing. What can be seen is that the dense layers found on the bottom of every snow
profile on Foxfonna can be observed through scattering in the radargram see Figure 22. Some
subglacial features were picked up by the radar, like a drainage channel typically seen as
hyperbola, but no crevasses. At the end of the GPR radargram, which was just next to the
thermistor station, a little noise could be observed below the snow depth. It shows typical
reflections of a Si layer or very old firn which is existent on the side of the glacier where the radar
was measuring, and also visible in summer. Nevertheless, the scattering reaches down not more
than 2 m from the glacier surface and is general very weak, therefore quite dense and probably
Figure 22 shows the location and the last radargram obtained at Foxfonna, 0628. At the end of the radargram a little scattering
can be observed, which is reaching down a bit deeper than the rest. At the radargram, the x-axis (left) shows the two-way
travel time of the radar wave (in ns), the second axis the depth at the chosen velocity (in m) (Norwegian Polar Institute, 2018).
39
meltwater infused which refroze randomly and left some spaces with less density. The elevation
corrected radargram of Foxfonna can be found in the Appendix at Figure 64.
6.1.5. Nordenskiöldbreen
On Nordenskiöldbreen, a GPR survey was conducted on 7th of May 2019. The radar transects
were collected over an elevational range of 500 – 800 ma.s.l.. Unfortunately, the Ethernet cable
connection to the Monitor was susceptible to failure, therefore only three short sections could be
collected with the radar named 0617, 0618 and 0620. The radar was collected driving uphill and
the first section, 0617 was the longest collected profile, showing a little firn in the beginning and
then a bigger firn patch corresponding to the bedrock topography with a mean depth of 6 metres
as seen in Figure 23. The elevation corrected radargram detail of 0617 and 0618 can be found in
the Appendix in Figure 65 and in Figure 67. In the bedrock depression where the firn layer of
profile 0617 sinks, a big layer of Si could be observed with a layer of nearly one metre deep
refrozen meltwater.
Figure 23:A shows the whole radargram collected on Potpeschniggbreen. B shows the same just in an overview and
with the whole radar (together with Phillipbreen). C shows the whole radargram, starting with a “bowl” and ending
with a deep layer of Si and firn. D shows a detail capture of the “bowl” structure. In C and D, the x-axis (left) shows
the two-way travel time of the radar wave (in ns), the second axis the depth at the chosen velocity (in m) (Norwegian
Polar Institute, 2018).
40
6.1.6. Potpeschniggbreen (Von Postbreen)
The radargram at Potpeschniggbreen starts at 500 m a.s.l. and ends at around 730 m a.s.l., with
a length of 6 km seen in Figure 24.The glacier stretches from east towards southeast. On the
map in Figure 24 D, a detail was pointed out, which corresponds with a surface depression and
has the shape of a bowl. On Potpeschniggbreen, crevasses can be seen two times on the radar
through deep scattering, close to the “bowl” structure in the beginning of the radargram, in Figure
24. At the aerial images from the area, an area with many crevasses can be observed too. In
the GPR it is not visible that the glacier is steepening a bit on the side which probably leads to
crevassing. The radargram of 0610 can be found elevation corrected in the Appendix in Figure
66.
6.1.7. Phillipbreen (Von Postbreen)
The obtained GPR track on Phillipbreen is 7 km long, starts at an elevation of around 720 m a.s.l.
and ends at 500 m a.s.l. as seen in Figure 25A andFigure 25B. The glacier stretches from north
towards south and is quite long and flat, and therefore has only small and little increases in
elevation. The accumulation on Phillipbreen is around 1.6 m deep with spatial variations. The
“bowl” structure observed at the beginning of the track, has a three- metres-deep younger
accumulation layer and further down a dense probably refrozen water layer, on top of a that at a
depth of around 0.7 m beneath the other layer, an older and denser layer is reaching down 9 m
Figure 24: The radar path is shown in the maps on Nordenskiöldbreen, with the three obtained
profiles.0617 and 0620 are examples of the obtained radargrams. The x-axis (left) shows the two-way
travel time of the radar wave (in ns), the second axis the depth at the chosen velocity (in m) (Norwegian
Polar Institute, 2018).
41
from the surface, which can be seen in Figure 25D. The “bowl” structure stretches over a length
of 1600 m.
On Phillipbreen crevasses have been observed at the middle of the radargram, slightly visible in
Figure 25C on the right side of the radar image as long, deep and thin reflections. Further, also
the snow profile was dug into a crevasse and on the aerial images from summer, more horizontal
crevasses can be seen. The whole track length can be found elevation corrected in the Appendix
at Figure 68.
Figure 25: A represents the GPR track with an overview map seen in B. C shows the collected radargram and D
the detail of the big “bowl” structure. In C and D, the x-axis (left) shows the two-way travel time of the radar wave
(in ns), the second axis the depth at the chosen velocity (in m) (Norwegian Polar Institute, 2018).
42
6.1.8. Drønbreen
Several radargrams were recorded on Drønbreen on the 26th of February 2019. In this result
section, only three of the profiles are shown and discussed, because they are the only ones which
show the accumulation zone as seen in Figure 26. The profiles are divided into 0028, 0029, 0030
and the last profile, 0030 is crossing the first, 0028 in the track what can be seen in Figure 26.
Nevertheless, profile 0028, which is stretching over a length of 1400 m, shows very random
scatterings, which can be explained by the moraine dividing both glacier accumulation zones. As
the glaciers meet a lot of sediments and rocks are present in the ice, which can be seen as
random scatterings of different densities. The profile was adapted using a trace gain to make the
deep reflections visible, which led to a very intense looking snow cover, where details are not
visible. Profile 0029 has a recorded length of 1800 m. Starting at 240 m profile length, a six-
metres-deep firn layer can be seen on the radargram, which continues 660 metres and has a
maximum depth of around 18 m. After a short section with no deep firn, only some Si, a firn layer
starting at 1000 m profile length, picks up again and can be observed until the end. The firn layer
at the end section of profile 0029 has a maximum depth of 24 m.
The Profile 0030 has a length of 1800 m, showing a quite continuous firn layer with a mean depth
of ten metres and an approximately three metres deep Si layer dividing the snowpack and the
firn. The firn and Si can be seen very good on this radargram but only without applying an energy
gain filter see Figure 26 0030.
When comparing profile 0028 and 0030 which drive along nearly the same elevation path, it can
be observed, that profile 0030 has a big SI and Firn layer while at profile 0028 it is nearly absent.
This can be explained through the very local accumulation patterns, with a tendency towards the
mountain side and because the glacier slope is steepening a bit towards the margin.
43
Figure 26: This figure shows three of the radargrams collected at Drønbreen glacier. The map indicates
the location where the profiles were collected using a snow scooter. The third profile, 0030 was marked
to make the layering visible. The signal between the blue and red line is showing firn, while the layer
between the red lines is indicating Si. In all radargrams, the x-axis (left) shows the two-way travel time
of the radar wave (in ns), the second axis the depth at the chosen velocity (in m).
44
6.2. Snow profiles
Snow profiles were collected at all field study sites. To get an overview of the data, all the snow
profiles were plotted for further analysis. This overview of all snow profiles can be found in
Table 4. It indicates that the profiles were collected over the whole accumulation season and
therefore more or less accumulation has happened after the profiles were recorded. The mean
snow water equivalent of all observed snow profiles is 669.36 mm with a mean standard
deviation of 244.80 mm. The mean of the average density of each snow profile is 372.93 kg m-3
with a standard deviation of 49.90 kg m-3. A complete table of the snow profiles including the
GPS location can be found in the Appendix in Table 6.
All presented snow profiles were plotted with the Snow profiler from LAWIS (European Avalanche
Warning Services., 2020). The Hardness column and the lemons section can be ignored as it was
not observed in the field and just interpolated by the program. The plotted profiles show nicely the
temperature profiles and melt layer distribution.
Table 4: Overview of collected snowprofiles.
profile
name
date of data
collection
profile location
elevation
[m a.s.l.]
total snow water
equivalent [mm]
average
density [kg m-3]
snow
depth [cm]
SPF1
25.02.2019
Foxfonna
613.00
302.30
233.00
144.00
SPF2
03.04.2019
Foxfonna
550.00
562.40
391.00
145.00
SPF3
03.04.2019
Foxfonna
620.00
620.80
403.00
155.00
SPF4
11.04.2019
Foxfonna
632.00
546.90
378.00
145.00
SPF5
11.04.2019
Foxfonna
660.00
593.40
386.00
145.00
SPF6
02.05.2019
Foxfonna station
638.00
733.80
367.00
200.00
SPN1
07.05.2019
Nordenskiöldbreen
841.00
613.80
400.00
165.00
SPN2
11.05.2019
Nordenskiöldbreen
635.00
810.00
405.00
200.00
SPV1
10.04.2019
Potpeschnigbreen
700.00
1123.00
405.00
275.00
SPV2
26.04.2019
Potpeschnigbreen
605.00
---
---
200.00
SPP3
24.04.2019
Phillipbreen
708.00
1251.00
461.00
244.00
SPD1
28.02.2019
Drønbreen
343.00
419.00
349.00
120.00
SPD2
26.02.2019
Drønbreen
442.00
480.60
325.00
150.00
SPD3
26.02.2019
Drønbreen
588.00
659.40
361.00
180.00
SPD4
27.02.2019
Drønbreen
700.00
654.70
357.00
206.00
mean
618.33
669.36
372.93
178.27
standard
deviation
111.43
244.80
49.90
41.05
45
6.2.1. Foxfonna
In total, six snowprofiles were collected (Figure 27 Figure 28), in elevations ranging from the
lowest 550 m a.s.l. up to 660 m a.s.l.. Four were collected in April 2019, the first one in February
2019 and the last at the beginning of May 2019. Very interesting to see is the red temperature
line in all profiles, which is significantly negative in the upper snow profile part performed in early
season with very cool temperatures. Further it is visible that this seasonal warming or cooling
effect from the surface air is affecting the temperature generally approximately 1 metre deep into
the snowpack starting at the surface. Other features seen in all of the snowprofiles are the melt
or rain layers close to the snow profile bottom. All the snow profiles indicate such layers, which
are probably from autumn rain or snowfall in high temperatures. The deepest high-density ice
layer ca be observed at around 0.5 m snow profile depth, but at the high elevation snow profiles,
it is not visible anymore, indicating a precipitation event as rain at lower elevation and snow at
higher elevation.
SPF1 SPF2
SPF3 SPF4
3.4.2019
Figure 27: Snow profiles collected on Foxfonna glacier SPF1- SPF4.
46
6.2.2. Nordenskiöldbreen
On Nordenskiöldbreen, data from two snowprofiles were collected (Figure 29). The first one,
SPN1 was done on the 2nd of May 2019 and the second one on the 11th of May 2019. The
elevation of the first profile is higher with an altitude of 841 m a.s.l. and the profile temperature is
extremely cool at the top, indicating a cool period just before measuring. On the day the snow
profile was performed, the sun was out and on the surface of the snow layer little ice droplets
were observed. This little ice droplets can be explained by melt due to high sun radiation, which
is indicated as melt-freeze crust in the profile.
On the second profile, the crusts were not observed, which can be explained by the location of
the slope which was rather flat, and the sunlight penetrates the snow surface differently depending
on slope angle. The rest of the profile looks extremely similar, especially the hardness sketch and
the location of melt layers and facets.
Figure 28: Snow profiles collected on Foxfonna glacier SPF5 and SPF6
SPF5 SPF6
47
6.2.3. Von Postbreen
On Von Postbreen, three snowprofiles were collected, two on Potpeschniggbreen and one on
Phillipbreen. Again, extremely good visible in Figure 30, the temperature gradient shown as red
line in the snowpack are influenced by the surface air temperature up to 1 metre deep. The two
profiles from Potpeschniggbreen show similar characteristics and variations one would expect
with elevation. Additionally, the snow hardness sketch is very well comparable, besides from
different layer thicknesses, the hardness looks nearly identical, which is surprising, because the
terrain was indeed different from profile 1 compared to profile 2.
On Phillipbreen, the profile SPP shown in Figure 31, does differ from Potpeschniggbreen, which
is not only due to the different elevation, but because mountains protect the glacier from receiving
vast amounts of snow from the east and probably a local wind pattern is significant in redistribution
of snow from north to south rather than east. The profile showed similar melt-freeze crust as on
Potpeschniggbreen, but in between the crystals almost in all layers built up to facets. Considering
the fact that under this profile, a crevasse opened up, this explains the faceting due to high
temperature gradients from the cool surface air and the warmer air in the crevasse.
Figure 29: Snow profiles plotted from Nordenskiöldbreen named SPN1 and SPN2.
SPN1 SPN2
48
Figure 30: Snow profiles collected on Potpeschniggbreen (Von Postbreen)
named SPP1 and SPP2.
SPP1 SPP2
SPP3
Figure 31: Snow profile from Phillipbr een (Von
Postbreen) named SPP3.
SPP1 SPP2
SPP3
49
6.2.4. Drønbreen
The four snow profiles show a big difference in snow crystal stratigraphy (Figure 32), which can
be explained by the fact that most of the data collection at Drønbreen glacier was part of the AG-
325 Glaciology course at UNIS and the observers were still in training. Additionally, snow
stratigraphy observations are always biased and influenced by the observing person. Through a
period of three days in February, (25th – 27th of February 2019) the snow profile data were
collected. Nevertheless, the temperature recordings are fully reliable and are quite interesting as
Figure 32: Snowprofiles on Drønbreen.
SPD1 SPD2
SPD3 SPD4
50
a large temperature was present through the relative shallow snowprofiles. SP1 show the effect
of this large temperature gradient, a lot of faceting is observed within the snowpack.
6.3. Snow depth surveys
Snow depth probing was performed on three glaciers. The depth surveys were always collected
in the GPR Trek for comparison. This chapter presents the results of the snow depth probing
results in comparison with the elevation of the probing sites.
6.3.1. Foxfonna
The snow depth sampling took place on 28th of May 2019, together with the GPR measurements.
The coordinates were taken with the GPX eTrex 10. All the points measured were collected
exactly on the GPR track at Foxfonna which can be seen in Figure 22. The snow depth is also
increasing on Foxfonna with elevation, but this varies a lot as seen in the plot in Figure 33. In
Figure 34 there is again a correlation between convex terrain and little snow depth. The reason
for the last two measurements which show very deep snow can be explained through the
topography. At the top end of Foxfonna the glacier is rather flat and sheltered by valleys.
Additionally, the two measurements were recorded on the west side of the glacier just before the
steep mountain starts. Therefore, due to the prevailing easterly winds, this can be expected and
can be interpreted as normal.
Figure 33: The plot shows the collected snow depths in
according to the elevation on Foxfonna.
51
6.3.2. Nordenskiöldbreen
On Nordenskiöldbreen the snow depth survey
was performed on 7th of May 2019. The Garmin
Montana GPS was used to collect the GPS
coordinates. Figure 35 shows the location of the
snow depth survey with the collected points.
Figure 37 shows the increase of snow with
elevation and a decent trendline. However, when
comparing the elevation with the snow depth line
as seen in Figure 36, it is visible that compared to
the top, the lower part of the glacier has quite high
snow depths. This could be explained by the
easterly winds, which transport the snow down
the glacier but also by the fact that the end of
survey was on a convex hill and therefore
resulted in very little snow depths.
Figure 35: Location of the snow depth survey on
Nordenskiöldbreen (Norwegian Polar Institute, 2018).
Figure 34: The plot shows the elevation (red) and the snow depth (blue) as single
line to compare on Foxfonna.
52
Figure 36: The plot shows the collected snow depths according to the elevation
on Nordenskiöldbreen.
Figure 37: The plot shows the elevation (red) and the snow depth (blue) as single
line to compare on Nordenskiöldbreen.
53
6.3.3. Drønbreen
The snow depth survey on Drønbreen was performed on the 27th of February 2019, which is less
comparable to the other surveys because February is still the beginning of the accumulation
season. Therefore, the snow height was probably much higher at the end of the accumulation
season. The Garmin Montana GPS was used to collect the coordinates. Figure 38 shows the
elevation and snow depth relation. It can be seen that the snow depth increases linear with the
elevation except for some points especially around 550 to 650 m a.s.l.. The location where this
happens, is having less snow due to a topographical feature as seen in Figure 39. The area is
rather convex, and the wind can transport snow more easily and move it to more concave terrain.
Figure 38: The plot shows the collected snow depths
according to the elevation on Drønbreen.
Figure 39: The plot shows the elevation (red) and the snow depth (blue) as single line to compare
on Drønbreen.
54
6.4. Firn cores
In the following chapter the results from the firn cores are presented. Table 5 shows an overview
of the collected cores. Each location is described, and the core data was plotted using the depth,
stratigraphy and temperature data.
Table 5: Firn core locations overview.
Coring Location
Core depth
Elevation (m a.s.l.)
Drønbreen
N 78°07.388 E 016°48.314
4.70 m
700.00
Potpeschnigbreen (Von Postbreen)
N 78°28.695 E 018°02.388
6.05 m
560
Nordenskiöldbreen
N 78°38.470 E 017°28.196
3.65 m
84
6.4.1. Nordenskiöldbreen
A firn core was drilled into the snow profile SPN1 at an elevation of 841 m a.s.l. on
Nordenskiöldbreen (Figure 41). The air temperature measured on this day was - 12.2 °C. The
snow profiles had a depth of 165 m and the drilled core reached further 3.77 m down. At the end
of the core temperatures of - 3.7 and - 3.8 °C were measured. The corer stopped to get a good
grip in the ice and no more core could be excavated. When looking at the stratigraphy, at the end
of the core, a lot of clear ice layers can be observed, suggesting a high density as seen in Figure
41 Further the core stratigraphy can be seen in comparison to the real photographed core in the
Appendix in Figure 69.
6.4.2. Potpeschniggbreen (Von Postbreen)
On Potpeschniggbreen, one core was drilled at an elevation of 600 m a.s.l. The 5.45 m long core
was drilled starting from the end of the snow profile. The snow profile depth on the location was
2.51 m. The air temperature on the day was - 5.6 °C, therefore cooler than the recorded core
temperatures as seen in Figure 40. The most interesting feature of this core is the temperature
gradient. The highest temperatures recorded were -0.7°C indicating a warm firn zone. The core
was drilled quite late in season, so the
BB
could be removed already, but the warm temperatures
indicate an active accumulation zone. Additionally, the temperature line seen in Figure 40 is
showing a trend towards 0 °C.
55
6.4.3. Drønbreen
The 5 m long core on Drønbreen was also drilled in the base of a snow profile (Figure 43). The
warmest recording at the core was - 2.4 °C, which is also quite warm. The core was drilled with
the Glaciology course at Unis within several days of fieldwork. The core was drilled in February
and therefore the firn was in the middle of the cooling period where the latent heat released from
freezing meltwater is slowly cooling again, even in great depths. At a depth of 3.90 m the core
starts to get extremely dense with more ice than firn with temperatures between - 2.4 and - 5.8 °C
(Figure 42).
Figure 40: The firn core on
Potpeschniggbreen starts in the
snow profile base with a total length
of 5.45 m.
Figure 41: The firn core from
Nordenskiöldbreen with the total
length of 3.77 m.
56
Figure 43: Core Stratigraphy on Drønbreen. UNIS Glaciology Course (2019)
Figure 42: Core Temperature in comparison with the depth measured at Drønbreen.
-25
-20
-15
-10
-5
0
-500 -400 -300 -200 -100 0100 200 300
T (°C)
Height (cm)
T (°C) vs. Height (cm)
T (°C)
57
6.5. Thermistor string and iButton temperature measurements
This chapter presents the results of the subsurface temperature development from the thermistor
string installation on Foxfonna and the iButtons installation on Nordenskiöldbreen. The
measurement periods of the thermistor strings and the iButton varied a little bit, therefore to
compare the data a plot was made to see the precipitation in each sampling period in detail. This
precipitation plot can be found in Figure 74 in the Appendix section.
6.5.1. Subsurface temperature evolution from the thermistor strings
The thermistors were installed on the 3rd of May and downloaded on the 3rd of August 2019. All
plots in this chapter show this time period. The Installation had two thermistor strings, named
Foxfonna thermistor 1 and Foxfonna thermistor 2.
The raw thermistor string data can be found in the Appendix in Figure 70 for Foxfonna thermistor
1 and in Figure 71 for Foxfonna thermistor 2.
The temperature threshold to determine snow or not snow was set to 0°C for Figure 45 Figure
46. However, in Figure 72 Figure 73 in the Appendix, the result of the subsurface temperature
evolution for the thermistors is shown without removing any data. However, this needs to be
reviewed critically as the thermistors melted out.
Nevertheless, all the white data in Figure 45 Figure 46 is refereed to no data meaning
temperatures above 0°C were measured. To complete the subsurface temperature modelling a
plot was made using 1.4°C as snow threshold. The changed subsurface temperature evolution
plot for thermistor 1 can be found in the Appendix in Figure 76.
The general pattern of subsurface temperature evolution seen in Figure 45 Figure 46 agrees with
the results of measurements from the Breinosa weather station, seen in Figure 44. In May and
June 2019, a gradual warming occurs due to the increase of air temperature. Nevertheless, a
sharp increase in snowmelt can be observed caused by infiltrating water in the beginning of July.
In the temperature plot at Figure 44 an incredibly warm period starting on July 5th and ending on
July 12th with temperatures above 10°°C measured on the Breinosa mountain plateau can be
seen.
Before the beginning of June temperatures are mainly below zero, but afterwards a continuous
rise in temperatures can be observed with the mentioned high temperatures in July.
Figure 47 shows the precipitation development in the observed period. Rain events start in mid-
June with the highest amount of rain between July the 2nd and July the 17th, which is in line with
the extreme warm temperatures over 10 °C measured on the mountain plateau Breinosa. Another
peak of rain can be observed on July 25th – 28th, where also the temperature plot hits a high again.
58
When comparing the temperature and precipitation plots to the subsurface temperature evolution,
it can be observed that at Foxfonna thermistor 1, the first melt starts at around mid-June and then
gradually continues towards August. In July a big rain event can be observed in Foxfonna
thermistor 1, which led to a big decrease in the snow pack, seen in Figure 45. The first melt which
happens mid-June is correlated to a rain event with 0.005 mm measured rainfall, but with
temperatures of 8 °C on the mountain, which brings in a lot of heat.
Figure 45: Subsurface Temperature Evolution. Foxfonna thermistor 1.
Figure 44: Temperature data derived from Breinosa weather station.
.
59
Thermistor 2 shows similar patterns, shown in Figure 46. The first melt hits the snowpack in mid-
June and then it gradually decreases. A rather big melt event can be clearly seen here in the end
of July which corresponds to the precipitation data and is considered as rain. On thermistor 2
piping of water down the thermistor might have happened, as the thermistor gradually melted out.
Figure 47: Precipitation sum on Foxfonna. Data derived from Adventdalen weather station
and combined with temperature data from Breinosa weather station.
Figure 46: Subsurface Temperature Evolution. Foxfonna thermistor 2.
60
6.5.2. Subsurface temperature evolution from the iButtons
In the subsurface temperature data derived from the iButtons, the raw data is displayed in Figure
48 and shows the timing when each logger melted out. Figure 49 shows the stepwise decrease
of the snowpack. Because the installation was only in the upper metres of the snowpack, it cannot
be interpreted if all the accumulation melted away on the sample location.
Nevertheless, the first melt onset can be observed mid-June, similar to the thermistor strings
(Figure 45 Figure 46). Melt lowers the snowpack about 0.25 m and takes away the
BB
quite
quickly which leads to a steeper curve compared to both thermistor strings. End of July a period
with completely melted out loggers is seen and then in the beginning of August, the accumulation
starts again (Figure 49).
Figure 49: Subsurface Temperature Evolution raw data on Nordenskiöldbreen, iButtons.
Figure 48: Subsurface Temperature Evolution on Nordenskiöldbreen using the iButtons.
61
6.6. Density evolution modelling
The measured density in snow profiles, the measured temperature values and the weather station
data made it possible to model the density evolution for each time step of the measurement period
by using the methodology from Meløysund et al. (2007). The following chapter explains the results
of the different datasets.
6.6.1. Foxfonna
The density evolution on Foxfonna largely corresponds with the weather data seen at Figure 44
Figure 47. Further the density evolution plot is showing similar surface lowering as the subsurface
temperature evolution. It is good visible in Figure 52 Figure 53, that the superimposed refrozen
high-density layer at the bottom of the snow profile survives until mid-July. In this model it can be
observed that a big portion of the snow profile is vanished already in mid-June, actually more than
half of the snow. The density modelling didn’t apply for the glacier ice which can be seen blue
coloured in Figure 50, Figure 51, Figure 52, andFigure 53 show the density evolution for each of
the in situ measured layer compared to the time. Further it can be observed that layers with a
higher density, still increase in density and melt out slower. The results of the density evolution
were evaluated through in situ observations and through comparison with the measured
subsurface temperature evolution. The result is in good agreement with the measured and
observed data.
Figure 50: Density Evolution Foxfonna thermistor 1 raw data.
62
Figure 52: Density Evolution Foxfonna thermistor 1.
Figure 51: Density Evolution Foxfonna thermistor 2 raw data.
63
6.6.2. Nordenskiöldbreen
The density evolution on Nordenskiöldbreen shows a plot with not very high resolution. Due to
the low measurement frequency compared to the thermistors, the data looks different.
Nevertheless, it can be observed that the warm event starting in the beginning of June was
enough to melt the snowpack over more than 1 m and that the layers are increasing the density.
However, this result could not be evaluated through in situ observations through the remote
location but a validation by comparing the results with the measured subsurface temperature
evolution shows that the results agree with the measured data. The plot showing the density
development over time is shown in the Appendix in Figure 75.
Figure 54: Density evolution on Nordenskiöldbreen.
Figure 53: Density Evolution Foxfonna thermistor 2.
64
6.7. Refreezing
This chapter present the results of the refreezing modelling. The refreezing is based on a
thermodynamic equation using the density evolution modelling results and the measured
temperature data from the thermistors and iButtons. The results can be compared with the overall
precipitation in the same period, which is 48.8 mm. From the 48.8 mm, precipitation in form of
rain contributed with 38.1 mm and snow with 10.7 mm.
6.7.1. Foxfonna
The refreezing value for Foxfonna thermistor 1 is 593 mm which can be seen in Figure 55 and
Figure 56 . For Foxfonna thermistor 2 seen in Figure 57 and 58, refreezing results in 221 mm.
Both model outcomes were adjusted using a threshold of decreasing all refreezing with the value
0.1 mm to remove the background scattering of daily temperature fluctuations. (The original
refreezing, without using a threshold, resulted in 1671 mm for the first thermistor on Foxfonna and
1099 mm for the second thermistor string). The refreezing on Foxfonna was derived from a total
measurement period of 90 days and 12 hours.
Both thermistor strings are very close to each other, showing a big difference in measured
refreezing values.
The station was close to a steeper mountainside, which could lead to increased meltwater
drainage from the steep side. Further the high values of refreezing in thermistor 1 could be
explained by meltwater draining down the thermistor on the warm and rainy period mid-June.
What can also be seen is that the big refreezing events relate with the rain and warming events,
whereas all the other refreezing is probably due to surface melt.
Figure 55: Refreezing rate at Foxfonna thermistor 1 using a threshold to remove the
background scattering.
65
Figure 56: Refreezing total in mm at Foxfonna thermistor 1. The results are plotted with using
a threshold to remove the background scattering by decreasing all results with 0.1mm.
Figure 57: Refreezing rate on Foxfonna thermistor 2 using a threshold to remove the
background scattering by decreasing all results with 0.1mm.
66
6.7.2. Nordenskiöldbreen
The following plot in Figure 60 shows the refreezing results using the iButtons data collected on
Nordenskiöldbreen. Figure 59 shows the amount of refreezing in mm. The refreezing value for
Nordenskiöldbreen is 389 mm using a threshold of decreasing all refreezing with the value
0.1 mm to remove the background scattering of daily temperature fluctuations. (The original
model output, without the equation resulted in 1137 mm refreezing). The measurement duration
of the experiment was 101 days and 16 hours.
However, refreezing is hard to detect in this case, because the snow is melting out quite fast.
Thus, the installation didn’t reach the snow/firn interface, the quantification has its limitations.
Nevertheless, it can be observed that the refreezing happening on the 10th of June is just after
the first rain precipitation event happening in the beginning of June seen in Figure 60 Figure 61.
Figure 61 shows the total measured precipitation from the Adventdalen station. When comparing
Figure 59 Figure 61, the relationship between refreezing and precipitation is getting obvious.
Figure 58: Refreezing in mm on Foxfonna thermistor 2 using a threshold to remove the
background scattering.
67
Figure 59: Amount of refreezing from the iButtons on Nordenskiöldbreen in mm using a
threshold to remove the background scattering by decreasing all results with 0.1mm.
Figure 60: iButtons refreezing plot using a threshold to remove the background scattering by
decreasing all results with 0.1mm.
68
7. Discussion
In this chapter the results are discussed and critically analysed. And some sub-chapters are
presenting combined interpretations of all the results, for example the glacial thermal regime
switch.
7.1. GPR surveys
The GPR survey on Foxfonna (Figure 22) revealed hardly any or just a tiny little spot left with old
firn and Si. Drønbreen (Figure 26) which has a higher accumulation zone compared to Foxfonna,
has more firn left. However, similar firn developments in terms of where the firn is left can be
observed on both glaciers. This can be explained through the location of both glaciers. Both
glaciers are located in a valley and ranging from south the upper part to north the lower part of
the glaciers. Therefore, easterly winds lead to snow transport and the snow is accumulated in
front of the east facing slopes which leads to more firn in these areas. Nevertheless, Foxfonna
glacier (excluding the ice cap on top) has a higher ELA than its top is reaching to. Drønbreen has
still some firn left, to draw conclusions on its future development, more years of monitoring of the
firn zone is needed. However, the firn core on Drønbreen (Figure 43) with 5 metres depth showed
the warmest temperatures at - 2.4 °C and it showed dense layers at the bottom. Further, the GPR
radargrams indicate a firn depth of around 10 metres depth (Figure 26) which could indicate a
densification and thinning of firn due to increased melt processes. In contrast to Foxfonna,
Drønbreen was not visited during summer and therefore, the location of the ELA represents a
rough estimation in this study.
The GPR surveys on Nordenskiöldbreen(Figure 24) and on Von Postbreen (Figure 23 Figure 25)
can also be compared for the reason that the glaciers are both located in a valley rising from west
towards east. Therefore, the accumulation patterns are similar. Further, the collected GPR data
on Nordenskiöldbreen corresponds well with the description of Pälli et al.(2002), who also
Figure 61: Precipitation from Adventdalen station, classified by temperature data by Breinosa
and time period of measured data from the iButton installation.
69
obtained radar data on Nordenskiöldbreen and found the “bowl” structures. In addition, all the
“bowl” like firn structures on Nordenskiöldbreen and Von Postbreen can be explained by the
bedrock topography beneath the deep firn patch. The firn is smoothing out the natural topography
which is in a depression. This effect can also be observed in the snow depth surveys, which show
that the snow is smoothing out the topography.
Another interesting observation is that in this “bowl” like firn structures also big layers of ice can
be seen in the radargrams indicating a quite big refreezing of meltwater. This refreezing of
meltwater in surface depressions even beneath the ELA could contribute significantly to glacier
mass accumulation especially because the topographic areas where this occurs are stretching
over several kilometres.
Furthermore, the radargrams have been checked for water or water tables. However, on both
glaciers no liquid water or water tables could be observed. Especially on Potpeschniggbreen (part
of Von Postbreen; Figure 23), no big scattering, which could indicate water pockets, was
observed, therefore the cold winter wave could deeply penetrate into the firn and take out the
heat from the previous year. The radar was driven there because of the previous study from
Sevestre et al. (2015) which showed that the glacier has a big temperate firn layer even reaching
up to the surface in the area around 800 m a.s.l.. In this work, this warm firn layer reaching up to
the surface couldn’t be. Several theories could explain this. First the radar survey stopped maybe
just before the warm firn, second the date of the survey let the firn cool off. Third, the firn in order
to be warm towards the surface needs to witness a huge melt or rain event just before
measurements or buried deep in the snow from previous season, which just didn’t happen for this
survey. The last theory is that the radar which Sevestre et al. (2015) used was a different antennas
and therefore the results need to be interpreted carefully and cannot be compared directly to the
radargrams collected in this study.
Finally, the drilled core at an elevation of 600 m a.s.l. on Potpeschniggbreen indicated warm
temperatures of - 0.7 °C (Figure 41). Together with the core results, and the GPR survey results,
which indicate a rather thin firn cover, a warming and thinning of the firn cover could be
interpreted, which would correspond with the modelling results performed on Nordenskiöldbreen
by Van Pelt et al, (2017). However, to be able to draw a concrete conclusion, the firn cover and
temperatures need further monitoring.
Interpretation of radar imagery is extremely subjective and may introduce errors. For example,
the accuracy of the depth measurement is always restricted by the ability to determine the first
signal return from the real surface and not the air during the data processing.
7.2. Snow profiles and firn cores
The snow profiles and snow depth distribution on Drønbreen (Figure 32) and on Foxfonna (Figure
27 Figure 28) show a prevailing snow deposition in the southeast facing mountain slopes similar
to the snowdrift study performed by Jaedicke and Gauer, (2005). Generally, the observed snow
density agrees with other studies performed on Svalbard. For example, Winther et al. (2003)
reported average snowpack density values of 374 kg m-3 in Svalbard, which is comparable the
results obtained in this study seen in Table 4.
Similar to other glaciers in Svalbard, wind and topography play an important role of snow
accumulation and redistribution on all observed glaciers.
70
The stratigraphy corresponds well with the crystal types shown in the study performed by
Eckerstorfer and Christiansen, (2011). However, the stratigraphy is surprisingly similar on
Foxfonna and Drønbreen. The same similarities can be observed for the stratigraphy on Von
Postbreen (Figure 30 Figure 31) and Nordenskiöldbreen (Figure 29) indicating that the location
of the glacier and its aspect is playing a huge role in the snow metamorphisms and local weather
forcing.
On Drønbreen snow depth is increasing a lot with elevation (Figure 39) whereas on Foxfonna
(Figure 34) snow depth is increasing but not as much as on Drønbreen. This can be explained by
local topography which is favourable for snow transport on Foxfonna due to an opening towards
the east. In comparison to Foxfonna, Drønbreen is surrounded by steep mountain slopes on the
top, which prevent the snowdrift and even help snow accumulation. This little difference could
lead to the fact that Drønbreen still has a firn depth of 10 metres and Foxfonna has lost nearly all
of its firn. Further, this shows that accumulation is highly depend on local topography and this can
determine whether a glacier gains, stagnates or loses mass.
Finally, the firn core from Nordenskiöldbreen with the total length of 3.77 m (Figure 40) and high
density in the end of the core also suggests meltwater refreezing happening. The temperature
curve however, is not directing completely towards 0°C suggesting that the firn cooled through
the wintertime. Which could be easily possible, because the data was collected in the beginning
of May and therefore after the coldest period.
7.3. Refreezing
7.3.1. Density modelling
The goal of the density modelling was primarily to have an input dataset for the refreezing
calculation. Nevertheless, the results are not bad given the method of using a model calibrated
for mainland Norway in Svalbard. The adjustment was that the whole data was multiplied by the
factor 0.35 which yielded good results because the values were overestimated around three times
for the climate in Svalbard.
However, to be more accurate, each of the single biases
W•
would need correction to be valid in
the “arctic tundra maritime climate setting” that Svalbard is in. Norway has more precipitation and
probably a less dense snowpack due to warmer temperatures and less wind. Further the albedo
and precipitation values would need evaluation, because of errors in the measurements.
Therefore, the main limitations for this computational model are that it was designed for Norway,
and not for an Arctic climate setting. However, to adjust the biases, a multiple regression analysis
would need to be performed for Svalbard which was not done for this project due to lack of density
data.
Finally, the density evolution modelling indicates that a little rainfall can lead to a fast decrease in
the snowpack (Figure 52 Figure 53) and therefore is having a huge impact on the glacier
accumulation. Further the rain might also shorten the snow cover time of the glacier surface
substantially. Nevertheless, rain can also contribute to snow covered glacier surface, especially
when it rains in autumn on the first snow on the glacier and refreezes on the surface. This creates
a very dense layer of Si which can prolong the snow cover season on the glacier surface in the
following summer and therefore might have a slight melt reducing effect. This dense Si layer could
be observed on all surveyed glaciers in 2019. However, this melt reducing effect was also present
71
in the density evolution seen in Figure 52 for Foxfonna thermistor 1. It shows that the refreezing
happening in autumn which creates an ice layer from rain events, melt out late in the summer
season, and therefore might have a protective effect on the glacier surface due to changing of the
albedo.
7.3.2. Refreezing modelling
By using the thermodynamic equation, refreezing could be quantified through the fast temperature
changes within the snow column. The refreezing plots (Figure 55, Figure 57 and Figure 60) show
well and distinctive when the refreezing happened.
Besides the fact that refreezing could be quantified, a relation can be observed in the model
results. The big refreezing events seem to always relate to a rain event or a period with extreme
warm temperatures. Therefore, it can be assumed that refreezing in the arctic needs to have
temperatures above zero for meltwater or rain to get available. However, the results also show
that refreezing is spatially highly variable, because Foxfonna thermistor 1 and 2 have different
refreezing plots, despite the setup was very close.
For the refreezing, it needs to be discussed that the threshold determining if snow was present or
not was set to 0°C. However, snow can be present above 0 °C with the right atmospheric
boundaries (Figure 77). Therefore, the result of the modelling might be a little bit lower than it
actually is, because of less snow assumed due to the threshold.
Further, the result of the modelling needed to be adapted because the equation mainly tracks the
temperature changes. However, temperature changes happen all the time within the snowpack
and creating a background noise scattering. This scattering in the background leads to an
overestimation of refreezing. Therefore, a filter was applied on the results. This filter removed all
temperature changes (minus 0.1 mm) which resulted in the refreezing values presented in this
thesis. However, this adjustment needs to be reviewed critically. On one side, it removes the
background noise successfully, but on the other side it can lead to an underestimation of
refreezing through deleting the 0.1 mm refreezing values when the actual refreezing happened
too. The underestimation gets visible when comparing to the original refreezing results, without
threshold which resulted in 1671 mm for the first thermistor on Foxfonna, 1099 mm for the second
thermistor string on Foxfonna and 1137 mm refreezing on Nordenskiöldbreen. Therefore, the
refreezing plots without the threshold are attached in the Appendix (Figure 78, Figure 79, Figure
80, Figure 81, Figure 82 and Figure 82), to compare and the actual refreezing, which is probably
a little bit higher than stated in the results.
The refreezing calculation obtained from the iButtons data, shows very blurred results, which can
be explained by having only five iButtons available for the installation on Nordenskiöldbreen.
However, the iButtons proved to be a very useful, cheap and easy tool to measure temperature
data even in a glacial environment.
72
Summarizing, the most important learnings derived from this modelling are:
- Water infiltration happens in an extremely heterogenous way, especially when rain events
occur. Surface melt and slow water drainage is happening far more homogenous.
- Rain and temperatures far above the freezing point are the most important factor for
lowering the snowpack, introduce heat, meltwater and further lead to refreezing.
- The refreezing of rain on snow events in early autumn, could protect the glacier from early
melt out in the season and prolongs the snow cover from bare ice exposure due to
changed albedo.
- The loggers melted out on both installations, indicating that the refreezing happening
within the snowpack in summer and spring, is hardly significant for accumulation, in
locations lower than the current ELA, but it still prolongs the snow cover season for at least
a little bit.
- Intense melt or rainfall remove a huge amount of the snowpack and lead to more decrease
than increase of mass due to refreezing. Therefore, for refreezing within the snow to be
positive for mass accumulation, it has to happen with only a little bit of rain or surface melt.
If too much rain or melt occurs, all the snow cover is gone. This shows, that the line
between accumulation and ablation is highly sensitive and dependent on the seasonal
rainfall events. Meaning that an increase of rain on the island could have a huge potential
of decreasing mass and rising the ELA to extremely high on the glaciers due to the fast
removal of the full snow cover.
7.3.3. Glacier thermal switch from temperate to cold glacier in Svalbard
Temperate glaciers strongly depend on a deep firn cover where water can percolate and release
heat in small portions over several years. In this way, mass can increase through this idealized
balanced system. However, when the firn is decreasing, and melt increasing, the meltwater fills
up pores (densification) and also decreases the overall firn extent, which leads to temperature
rising due to latent heat release to the surroundings. This effect was also observed in this thesis
through the dense ice cores, and the GPR radargrams. This could imply that a little change, like
the increase of rain could possibly mean a huge change in this sensitive mass accumulating
system.
Further this would lead to less heat transported deep into the firn, because of the firn thinning,
which could lead to a complete switch of the glacier’s thermal regime.
When the glacier changes its thermal regime, in this case it would turn from warm to cold, the
whole glacier dynamics are changing too. With glacier dynamics, the glacial hydrology, ice
velocity, microbial community behaviour and a lot of other processes will change.
Finally, temperate ice and firn, which is dependent on water percolation within the firn column,
could turn cold because of the densification and thinning of the firn due to increased air
temperatures and rain precipitation in Svalbard. The thinning and densification of the firn could
further imply glacier mass loss. However, the thermal switch of the glacier leads to dynamic
changes which can hardly be understood without further investigations.
73
7.3.4. Influence of refreezing on the mass balance
Unfortunately, the observed refreezing cannot be compared to the overall mass balance
measurements, as only the spring and summer refreezing within the snow was observed using
the thermistor strings and iButtons.
However, if the dense layers on the glacier surface account for autumn refreezing, it could be
added to the summer values.
Nevertheless, Foxfonna might be not a representative study location to account for the effects of
refreezing on the mass balance, as it was observed that all accumulation at the chosen station
location melted at the end of the summer season. Refreezing can only be significant for glacier
mass balance when the summer melt is not leading to immense runoff throughout the season,
which is only present in higher accumulation zones.
Further to witness a mass buffering effect, refreezing needs to occur in a cool environment,
preferably in a deep firn column or in a high elevated snowpack, where the latent heat release
through refreezing is not imply an increase of temperatures towards the freezing point.
Nevertheless, the observations in this study conclude that refreezing is a process happening all
the time, but the surroundings where it occurs make the difference and determine whether it leads
to mass gain or loss.
8. Conclusion and outlook
8.1. Conclusions
Little research which includes detailed snow and firn observations on glacier accumulation zones
has been conducted before, whereas considerable gaps in the understanding of the accumulation
drivers on arctic glaciers are existent. Therefore, this thesis aimed to establish a deeper
understanding of the accumulation processes by data collection and analysis at four different
accumulation zones of Svalbard glaciers. The main objectives of this thesis, to increase the
understanding of the snow and firn processes of glaciers in Svalbard and to quantify the refreezing
of melt or rainwater could be answered thoroughly through the modelling and processing of the
collected data.
Q1: How much Snow and Firn is present in the accumulation zone and what are their properties?
The collected data to answer this research question showed that the glaciers Foxfonna and
Drønbreen have similar accumulation trends through their location. However, Drønbreen has
more firn left than Foxfonna through its unique surface topography. Both glaciers have a relatively
high ELA compared to the glaciers Von Postbreen and Nordenskiöldbreen which also show
similarities in accumulation. Nevertheless, density and stratigraphy data in the observation period,
showed high similarities on all four glaciers with respect to local variability.
A general trend of densification and thinning of firn could be observed around the ELA at all
observed glaciers. Further the annual snow accumulation depends highly on the glacier aspect
and its bedrock topography. Accumulation in concave surface topography structures could lead
to mass accumulation even lower than the current ELA and might be significant to study in the
future in terms of mass balance.
74
Q2: Can the amount of melt or rainwater, which refreezes within the snow and firn, be quantified?
The modelling of the refreezing of melt and rainwater within the snow was carried out successfully.
The study results show the quantification of refreezing modelling using two different methods on
two different locations. The iButtons and thermistor strings proved to be a good instrument for
subsurface temperature measurements. Further, the density modelling using the equation of
Meløysund et al. (2007). proved to work good by using an introduced factor of multiplying the
result with 0.35 due to the differences in local climate of Svalbard and mainland Norway. With the
help of the modelled density dataset, the amount of refreezing could be quantified. However, the
refreezing results always need to be interpreted carefully. The results show the refreezing within
the snow column which yielded values of 593 mm for Foxfonna thermistor 1, 221 mm for Foxfonna
thermistor 2 and 389 mm for Nordenskiöldbreen. Thus, the refreezing modelling outcome suggest
a high relation with previous periods of high temperature or rainfall.
Nevertheless, the measured refreezing on Foxfonna happened below the ELA and on
Nordenskiöldbreen, the installation was only in the top of the snowpack, which in both cases
resulted in a complete melt out of both installations during the season. The experiment setup of
using two thermistor strings close to each other, further proved the high spatial variability in
meltwater percolation.
Finally, the data showed that the impact of refreezing depends largely on the local weather and
precipitation regime and therefore the monitoring is crucial to understand changes in glacier
accumulation patterns.
8.2. Outlook
To be able to fully understand the condition and dynamics of Svalbard glaciers in the future, the
effects of climate change upon the extent and thermal conditions need to be examined carefully.
The collected data outcome from the accumulation zones, GPR, firn cores and snow profiles
suggest a densification and thinning of firn especially around the ELA. However, to fully
understand the amount of firn densification and firn mass decrease further monitoring of the firn
is necessary. Monitoring of firn below, around and above the ELA on different glaciers is
needed. The monitoring using automated temperature measurements together with GPR and
snow observations over several years would allow to draw a distinct conclusion if the trend of
decreasing and densification of firn is significant for glacier mass loss and changes in glacier
dynamics.
Overall, the state of the accumulation zones strongly depend on the seasonal precipitation and
temperatures. One late season rainfall can melt all seasonal accumulation, proposing a high
sensitivity of glaciers towards future changing climate patterns.
Concerning the quantification of refreezing, the effect of rain and high temperatures on the
timing and amount of refreezing needs to be investigated further. Additionally, the role of
refreezing within the snowpack needs to be monitored in different elevations together with the
refreezing happening in the firn column. This extensive refreezing monitoring would help to
understand the effect of refreezing within the snow but also the amount of refreezing in the firn
which contributes to the internal accumulation. Given the increased glacier mass loss and rise
of temperatures and precipitation in previous years it becomes apparent that understanding the
processes of accumulation through refreezing need careful monitoring rather sooner than later.
75
This is crucial to understand the effects of climate change on glacier mass balance and glacier
dynamics. Generally, this study and also others (Marchenko et al., 2017; van Pelt et al., 2016)
showed a high variability in glacier surface mass accumulation due to variability in Svalbard’s
weather patterns. Therefore, a long-term monitoring project of data collection is absolutely
necessary.
Based on this study, suggestions for future investigations are outlined below:
- Further in situ measurements of the location and properties of cold and temperate firn
zones need to be observed because snow and firn are the number one drivers in terms
of glacier survival.
- The usage of different methods in terms of temperature data collection (iButtons,
thermistors) should be carried out repeated and with the same amount of loggers to be
able to compare the methods thoroughly.
- Modelling is necessary to understand the complex interactions happening and to
simulate the refreezing and especially the rate of change the glaciers undergo at the
moment. Additionally, to understand future changes, models coupled with climate
models is crucial.
- In terms of modelling, in situ data of water percolation is highly demanded, because
most of the models at the moment assume that the vertical water percolation is laterally
uniform, however field observations have shown that water flow has huge spatial
variations.
- Integrative field sampling (connect of experts within the snow hydrology, climatology and
glacier hydrology field) should be carried out, which would be highly beneficial to get a
big picture of the underlying processes leading to changes in glacier mass.
- More in situ observations are necessary to detect changes in firn cover and
accumulation patterns and to study the glacier response to a changing climate.
Additionally, a combination using aerial and satellite images could be obtained together
with the in-situ measurements to detect and understand changes in the recent climate
period and in the future.
76
9. References
Armstrong, R.L., Brun, E., 2008. Snow and Climate: Physical Processes, Surface Energy
Exchange and Modeling.
Bamber, J., Payne, A., 2003. Mass Balance of the Cryosphere: Observations and Modelling of
Contemporary and Future Changes. Cambridge University Press.
Benn, D.I., Evans, D.J.A., 2010. Glaciers & Glaciation, 2. ed. ed. Hodder Education, London.
Bintanja, R., Andry, O., 2017. Towards a rain-dominated Arctic. Nature Clim Change 7, 263–267.
https://doi.org/10.1038/nclimate3240
Bogorodsky, V.V., Bentley, C.R., Gudmandsen, P.E., 1985. Radioglaciology. D.Reidel Publishing
Company.
Charalampidis, C., Van As, D., Colgan, W.T., Fausto, R.S., Macferrin, M., Machguth, H., 2016.
Thermal tracing of retained meltwater in the lower accumulation area of the Southwestern
Greenland ice sheet. Ann. Glaciol. 57, 1–10. https://doi.org/10.1017/aog.2016.2
Christiansen, H.H., French, H.M., Humlum, O., 2005. Permafrost in the Gruve-7 mine,
Adventdalen, Svalbard. Norsk Geografisk Tidsskrift - Norwegian Journal of Geography
59, 109–115. https://doi.org/10.1080/00291950510020592
Christianson, K., Kohler, J., Alley, R.B., Nuth, C., Pelt, W.J.J., 2015. Dynamic perennial firn
aquifer on an Arctic glacier. Geophys. Res. Lett. 42, 1418–1426.
https://doi.org/10.1002/2014GL062806
Cogley, J.G., Hock, R., Rasmussen, L.A., Arendt, A.A., Bauder, A., Braithwaite, R.J., Jansson,
P., Kaser, G., Möller, M., Nicholson, L., Zemp, M., 2011. Glossary of glacier mass balance
and related terms.
Colbeck, S.C., Akitaya, E., Armstrong, R.L., Gubler, H., Lafeuille, J., Lied, K., McClung, D.M.,
Morris, E.M., 1990. The International Classification for Seasonal Snow on the Ground.
International Commission on Snow and Ice.
Cuffey, K., Paterson, W.S.B., 2010. The physics of glaciers, 4th ed. ed. Butterworth-
Heinemann/Elsevier, Burlington, MA.
Dang, H., Genthon, C., Martin, E., 2017. Numerical modelling of snow cover over polar ice sheets.
Ann. Glaciol ., 25 , 170–176. Annals of Glaciology 25, 170–176.
https://doi.org/10.3189/S0260305500013987
DeGeer, G., 1912. A geochronology of the last 12000 years, in: Congr. Géol. Int. Stockholm 1910.
pp. 241–253.
Dowdeswell, J.A., 1984. , in: Airborne Radio Echo Sounding of Sub-Polar Glaciers in Spitsbergen,
182 from Skrifter (Norsk Polarinstitutt). Norsk Polarinstitutt, p. 41.
Eckerstorfer, M., Christiansen, H.H., 2011. The “High Arctic Maritime Snow Climate” in Central
Svalbard. Arctic, Antarctic, and Alpine Research 43, 11–21. https://doi.org/10.1657/1938-
4246-43.1.11
European Avalanche Warning Services., 2020. Lawis Version: 2.0.826 [WWW Document]. URL
https://lawis.at/profile/edit.php?table=true&zoom=6&intro=false&last_sid=238¢er=47
.264320080254805%2C11.392822265625&zeit=132&last_pid=11364 (accessed
1.25.20).
Fierz, C., Armstrong, R.L., Durand, Y., Etchevers, P., Greene, E., McClung, D.M., Nishimura, K.,
Satyawali, P.K., Sokratov, S.A., 2008. The International classification of seasonal snow
on the ground.
Førland, E.J., Benestad, R., Hanssen-Bauer, I., Haugen, J.E., Skaugen, T.E., 2011. Temperature
and Precipitation Development at Svalbard 1900–2100. Advances in Meteorology 2011,
1–14. https://doi.org/10.1155/2011/893790
Førland, E.J., Hanssen-Bauer, I., 2003. Climate variations and implications for precipitation types
in the Norwegian Arctic. Norwegian Meteorological Institute report No. 24/02 klima, 22.
Forwick, M., Vorren, T.O., Hald, M., Korsun, S., Roh, Y., Vogt, C., Yoo, K.-C., 2010. Spatial and
temporal influence of glaciers and rivers on the sedimentary environment in Sassenfjorden
77
and Tempelfjorden, Spitsbergen. Geological Society, London, Special Publications 344,
163–193. https://doi.org/10.1144/SP344.13
Gagne, G., 2018. IButton Thermochrons: An Affordable and Effective Technique for Measuring
Temperature Gradients. Presented at the ISSW, Utah Avalanche Center.
Goosse, H., Kay, J.E., Armour, K.C., Bodas-Salcedo, A., Chepfer, H., Docquier, D., Jonko, A.,
Kushner, P.J., Lecomte, O., Massonnet, F., Park, H.-S., Pithan, F., Svensson, G.,
Vancoppenolle, M., 2018. Quantifying climate feedbacks in polar regions. Nat Commun 9,
13. https://doi.org/10.1038/s41467-018-04173-0
Hagen, J.O. (Ed.), 1993. Glacier atlas of Svalbard and Jan Mayen, Meddelelser / Norsk
Polarinstitutt. Nork Polarinstitutt, Oslo.
Hagen, J.O., Kohler, J., Melvold, K., Winther, J.-G., 2003. Glaciers in Svalbard: mass balance,
runoff and freshwater flux. Polar Research 22, 145–159.
https://doi.org/10.3402/polar.v22i2.6452
Hagen, J.O., Liestøl, O., 1990. Long-Term Glacier Mass-Balance Investigations in Svalbard,
1950–88. Annals of Glaciology, 14, 102-106.
https://doi.org/10.3189/S0260305500008351
Hock, R., 2005. Glacier melt: a review of processes and their modelling. Progress in Physical
Geography: Earth and Environment 29, 362–391.
https://doi.org/10.1191/0309133305pp453ra
Jaedicke, C., Gauer, P., 2005. The influence of drifting snow on the location of glaciers on western
Spitsbergen, Svalbard. Ann. Glaciol. 42, 237–242.
https://doi.org/10.3189/172756405781812628
Kinar, N.J., Pomeroy, J.W., 2015. Measurement of the physical properties of the snowpack. Rev.
Geophys. 53, 481–544. https://doi.org/10.1002/2015RG000481
König, M., Nuth, C., Kohler, J., Moholdt, G., Pettersen, R., 2014. A digital glacier database for
Svalbard., in: Global Land Ice Measurements from Space. Springer-Verlag Berlin
Heidelberg.
Kovacs, A., Gow, A.J., Morey, R.M., 1995. The in-situ dielectric constant of polar firn revisited.
Cold Regions Science and Technology 23, 245–256. https://doi.org/10.1016/0165-
232X(94)00016-Q
Koziol, K.A., Moggridge, H.L., Cook, J.M., Hodson, A.J., 2019. Organic carbon fluxes of a glacier
surface: A case study of Foxfonna, a small Arctic glacier. Earth Surf. Process. Landforms
44, 405–416. https://doi.org/10.1002/esp.4501
Liestøl, O., 1974. Glaciological work in 1972. Norsk Polarinstitutt Arbok 1974 125–135.
Marchenko, S., van Pelt, W.J.J., Claremar, B., Pohjola, V., Pettersson, R., Machguth, H., Reijmer,
C., 2017. Parameterizing Deep Water Percolation Improves Subsurface Temperature
Simulations by a Multilayer Firn Model. Front. Earth Sci. 5.
https://doi.org/10.3389/feart.2017.00016
Mätzler, C., 1996. Notes on microwave radiation from snow samples and emission of layered
snowpacks. Institute of Applied Physics Dept. of Microwave Physics.
Meløysund, V., Leira, B., Høiseth, K.V., Lisø, K.R., 2007. Predicting snow density using
meteorological data. Met. Apps 14, 413–423. https://doi.org/10.1002/met.40
meteoblue, 2020. Svalbard Airport, Longyear [WWW Document]. meteoblue. URL
https://www.meteoblue.com/de/wetter/woche/svalbard-airport%2c-longyear_svalbard-
und-jan-mayen_6296766 (accessed 1.26.20).
Norwegian Polar Institute, 2018. Map Data and Services [WWW Document]. URL
https://geodata.npolar.no/#terms-of-use (accessed 3.1.20).
Nuth, C., Moholdt, G., Kohler, J., Hagen, J.O., Kääb, A., 2010. Svalbard glacier elevation changes
and contribution to sea level rise. J. Geophys. Res. 115, F01008.
https://doi.org/10.1029/2008JF001223
Oerlemans, J., 2001. Glaciers and Climate Change, 1st Edition. ed. CRC Press.
Overland, J.E., Wang, M., 2018. Resolving Future Arctic/Midlatitude Weather Connections.
Earth’s Future 6, 1146–1152. https://doi.org/10.1029/2018EF000901
78
Pälli, A., Kohler, J.C., Isaksson, E., Moore, J.C., Pinglot, J.F., Pohjola, V.A., Samuelsson, H.,
2002. Spatial and temporal variability of snow accumulation using ground-penetrating
radar and ice cores on a Svalbard glacier. J. Glaciol. 48, 417–424.
https://doi.org/10.3189/172756502781831205
Paterson, W.S.B., 1994. The physics of glaciers. Oxford: Pergamon Press.
Pettersson, R., Jansson, P., Holmlund, P., 2003. Cold surface layer thinning on Storglaciären,
Sweden, observed by repeated ground penetrating radar surveys. Journal of Geophysical
Research: Earth Surface 108. https://doi.org/10.1029/2003JF000024
Pfeffer, W.T., Humphrey, N.F., 1996. Determination of timing and location of water movement
and ice-layer formtation by temperature measurements in sub-freezing snow. Journal of
Glaciology 42, 292–304. https://doi.org/10.3189/S0022143000004159
Pfeffer, W.T., Meier, M.F., Illangasekare, T.H., 1991. Retention of Greenland runoff by refreezing:
Implications for projected future sea level change. J. Geophys. Res. 96, 22117.
https://doi.org/10.1029/91JC02502
Pörtner, H.-O., D.C. Roberts, V. Masson-Delmotte, P. Zhai, M. Tignor, E. Poloczanska, K.
Mintenbeck, M. Nicolai, A. Okem, J. Petzold, B. Rama, N. Weyer, 2019. IPCC, SROCC
Special Report on the Ocean and Cryosphere in a Changing Climate.
Rachlewicz, G., Szczuciński, W., Ewertowski, M., 2007. Post-“Little Ice Age” retreat rates of
glaciers around Billefjorden in central Spitsbergen, Svalbard. Polish Polar Research 28,
159–186.
Reijmer, C.H., van den Broeke, M.R., Fettweis, X., Ettema, J., Stap, L.B., 2012. Refreezing on
the Greenland ice sheet: a comparison of parameterizations. The Cryosphere 6, 743–762.
https://doi.org/10.5194/tc-6-743-2012
Rutter, N., Hodson, A., Irvine-Fynn, T., Solås, M.K., 2011. Hydrology and hydrochemistry of a
deglaciating high-Arctic catchment, Svalbard. Journal of Hydrology 410, 39–50.
https://doi.org/10.1016/j.jhydrol.2011.09.001
Sandmeier, K.J., 2019. Introduction to the processing of GPR-data within REFLEXW.
Serreze, M.C., Barry, R.G., 2011. Processes and impacts of Arctic amplification: A research
synthesis. Global and Planetary Change 77, 85–96.
https://doi.org/10.1016/j.gloplacha.2011.03.004
Sevestre, H., Benn, D.I., Hulton, N.R.J., Bælum, K., 2015. Thermal structure of Svalbard glaciers
and implications for thermal switch models of glacier surging. J. Geophys. Res. Earth Surf.
120, 2220–2236. https://doi.org/10.1002/2015JF003517
Sommerfeld, R.A., LaChapelle, E., 1970. The classification of snow metamorphism. Journal of
Glaciology 9, 3–18. https://doi.org/10.3189/S0022143000026757
Stuecker, M.F., Bitz, C.M., Armour, K.C., Proistosescu, C., Kang, S.M., Xie, S.-P., Kim, D.,
McGregor, S., Zhang, W., Zhao, S., Cai, W., Dong, Y., Jin, F.-F., 2018. Polar amplification
dominated by local forcing and feedbacks. Nature Clim Change 8, 1076–1081.
https://doi.org/10.1038/s41558-018-0339-y
Tully, R., 2007. The use of low cost “iButton” Temperature Logger Arrays to Generate High Spatial
Resolution Tidal Inundation Regime Data. (master thesis). Marine Resource Management
Oregon State University.
UNIS Glaciology Course, 2019. Fieldreport 2019.
van As, D., Box, J.E., Fausto, R.S., 2016. Challenges of Quantifying Meltwater Retention in Snow
and Firn: An Expert Elicitation. Front. Earth Sci. 4.
https://doi.org/10.3389/feart.2016.00101
van Pelt, W.J.J., Oerlemans, J., Reijmer, C.H., Pohjola, V.A., Pettersson, R., van Angelen, J.H.,
2012. Simulating melt, runoff and refreezing on Nordenskiöldbreen, Svalbard, using a
coupled snow and energy balance model. The Cryosphere 6, 641–659.
https://doi.org/10.5194/tc-6-641-2012
van Pelt, W.J.J., Pohjola, V.A., Reijmer, C.H., 2016. The Changing Impact of Snow Conditions
and Refreezing on the Mass Balance of an Idealized Svalbard Glacier. Front. Earth Sci.
4. https://doi.org/10.3389/feart.2016.00102
79
Warren, S.G., 1982. Optical properties of snow. Rev. Geophys. 20, 67.
https://doi.org/10.1029/RG020i001p00067
Wendl, I.A., 2014. High resolution records of black carbon and other aerosol constituents from
the Lomonosovfonna 2009 ice core. Bern.
Winther, J.-G., 1993. Short- and Long-Term Variability of Snow Albedo. Hydrology Research 24,
199–212. https://doi.org/10.2166/nh.1993.0022
Winther, J.-G., Bruland, O., Sand, K., Killingtveit, Å., Marechal, D., 1998. Snow accumulation
distribution on Spitsbergen, Svalbard, in 1997. Polar Research 17, 155–164.
https://doi.org/10.3402/polar.v17i2.6616
WMO, 2003. Manual on the Global Observing System. Presented at the WMO-No. 554, Geneva.
Wright, A.P., Wadham, J.L., Siegert, M.J., Luckman, A., Kohler, J., Nuttall, A.M., 2007. Modeling
the refreezing of meltwater as superimposed ice on a high Arctic glacier: A comparison of
approaches. J. Geophys. Res. 112. https://doi.org/10.1029/2007JF000818
80
10. Appendix
Figure 63: Nordenskiöldbreen seen from the south. (Norsk Polar
Institute, 1936)
Figure 62: Workflow chart showing all data and processing steps used to get the refreezing calculation results.
81
Figure 65: Elevation corrected GPR radargram number 0618 detail on Nordenskiöldbreen. The profile shows the
detailed view of the snow profile and the firn. The big white line can be interpreted as the high-density rain layer
from autumn which could be observed in the snow profiles on the glacier surface.
Figure 64: Elevation corrected GPR radargram on Foxfonna. The profile collection is from left to right and the
purple scattering shows the snow accumulation. In the end of the radargram a little scattering can be seen which
can be interpreted as Si or old firn.
82
Figure 67: The radargram 0617 was collected on Nordenskiöldbreen in the beginning of May 2019. The
radargram is elevation corrected and the effect of the surface topography is good visible. A huge firn “bowl”
structure built up in the surface depression whereas on the surrounding are only little firn is visible.
Figure 66: This radargram shows the elevation corrected radargram 0610 collected on Potpeschniggbreen (Von
Postbreen). The radar survey collection happened from left to right. The radargram shows nicely the firn
distribution on Potpeschniggbreen. However, no big scattering is visible which would indicate liquid water.
83
Table 6: The complete overview of all snow profiles including the GPS location of the obtained profiles.
profile
name
date
location
elevation (m
a.s.l.)
snow water
equivalent (mm)
average density
(kg/m³)
snow depth
(cm)
GPS
SPF1
25.2.2019
Foxfonna
613
302,3
233
144
N 78°08.281 E 016° 09.745
SPF2
3.4.2019
Foxfonna
550
562,4
391
145
N 78°08.637 E 016°08.451
SPF3
3.4.2019
Foxfonna
620
620,8
403
155
N 78°08.446 E 016°09.292
SPF4
11.4.2019
Foxfonna
632
546,9
378
145
N 78°08.171 E 016°10.260
SPF5
11.4.2019
Foxfonna
660
593,4
386
145
N 78.08.068 E 01610689
SPF6
2.5.2019
Foxfonna station
638
733,8
367
200
N78 08.105 E 016°10.332
SPN1
7.5.19
Nordenskiöldbree
n
841
613,8
400
165
N 78 38 470 E017 28 196
SPN2
11.5.19
Nordenskiöldbree
n
635
810
405
200
N 78° 38.434 E 017°23.542
SPV1
10.4.19
Potpeschnigbreen
700
1123,3
405
275
N78 28 768 E18 03 404
SPV2
26.4.19
Potpeschnigbreen
605
no data
no data
200
N 78°28.695 E 018°02.388
SPP3
24.4.19
Phillipbreen
708
1250,7
461
244
N 78° 30.702 E 017°50.665
SPD1
28.2.19
Drønebreen
343
419
349
120
N 78°08.131 E 016°49.807
SPD2
25.2.19
Drønebreen
442
480,6
325
150
N 78°07.389 E 016° 48.332
SPD3
26.2.19
Drønebreen
588
659,4
361
180
SPD4
27.2.19
Drønebreen
700
654,7
357
206
mean
618,33
669,36
372,93
178,27
std
111,43
244,80
49,90
41,05
84
Figure 68: The elevation corrected radargram represents the GPR collected on Phillipbreen (Von Postbreen).
The radargram shows that the accumulation of firn is dependent on surface depression at this altitude. The
radargram was collected between 500 m a.s.l. (right) and 720 m a.s.l (left).
Figure 69: This Figure shows the actual firncore in comparison with the
stratigraphy from Nordenskiöldbreen. The core couldn’t be transported back into
the cold room for further analysis or good photographs, therefore it is shown here.
85
Figure 70: Raw temperature recordings of the thermistor string number 1 on Foxfonna. Each color
shows one logger. It can be seen that starting early June, a lot of the loggers melted out and start
to record air temperatures with daily fluctuations.
Figure 71: Raw temperature recordings of the thermistor string number 2 on Foxfonna. The plot
shows each single logger with a different color. It can be observed that at the beginning of June
the loggers start to melt out
.
86
Figure 72: Subsurface temperature evolution on Foxfonna thermistor 2 over the whole data
collection period. At this figure no values were set as “no data” therefore, the positive data values
can be observed. However, it is important to remember that the loggers which melted out were
lying on the glacier surface and therefore, the depth of the logger location is not accurate on the
right side of the plot.
Figure 73: Subsurface temperature evolution on Foxfonna thermistor 1 using all of the collected
data. All of the data is visible, and no data was set to be invisible. The depth of the installation is
correct for the left part of the plot but on the right part, the thermistors which melted out were lying
on the glacier surface. Therefore, the interpretation needs to be carefully.
87
Figure 74: The precipitation over the whole observation period. Snow shows the blue column and
the orange one is rain. The blue line left indicates the starting time of the iButton recordings and
the beginning of the scale indicates the start of the thermistor recordings. The red line right
indicates the download date of the thermistor string data and the end of the graph indicates the
end of the iButton data collection.
Figure 75: Density evolution on Nordenskiöldbreen measured with the iButtons. The graph
shows the density trend of each single starting density layer. The loggers melted out quickly,
therefore the lines end rather abruptly.
88
Figure 76:Subsurface Temperature evolution of Foxfonna thermistor 1. This is the result when all
the temperature values above 1.4°C are set invisible. It shows the snowpack is lowering more
gradually and not as sharp as in the model results when all values above 0°C are set invisible.
Figure 77: The temperature data recorded by the iButtons including positive temperatures.
Interestingly the iButtons which were still on the glacier surface even though they melted out,
recorded temperatures of over 20°C. The graph also shows that the light blue section which is
between zero and 2°C could still be interpreted as snow. However, this is discussable and for the
modelling in this thesis only data below 0°C was interpreted as snow.
89
Figure 78: Refreezing in mm on Foxfonna thermistor 1.
.
Figure 79: Refreezing in mm on Foxfonna thermistor 1.
.1
90
Figure 80: Refreezing in mm on Foxfonna thermistor 2.
.
Figure 81: Amount of refreezing from the iButtons on Nordenskiöldbreen in mm.
91
Figure 83: Refreezing in mm on Foxfonna thermistor 2.
.
Figure 82: iButtons refreezing plot.
.
92
11. Erklärung / Affirmation
Hiermit bestätige ich, dass:
- ich die vorliegende Masterarbeit selbstständig verfasst habe,
- ich keine anderen Quellen und Hilfsmittel benutzt habe als die angegebenen und mich
keiner unerlaubten Hilfe bedient habe,
- ich dieses Masterarbeitsthema bisher weder im In- noch im Ausland in irgendeiner Form
als Prüfungsarbeit vorgelegt habe,
- diese Arbeit mit der vom Begutachter beurteilten Arbeit vollständig übereinstimmt.
Hereby I certify, that:
- the master thesis was written by me, not using sources and tools other than quoted and
without use of any other illegitimate support.
- I clearly marked and separately listed all of the literature and all of the other sources
which I employed when producing this academic work, either literally or in content
- I have not submitted this master thesis either nationally or internationally in any form
- the version of this thesis is the same as the work judged by the supervisor.
Wien / Vienna 9.4.2020
