Investigation on the character of the subglacial drainage systemin the lower part of the ablation area of Storglaciären, northernSweden

Investigation on the character of the subglacial drainage system in the lower part of the ablation area of Storglaciären, northern

Sweden

Fanny Ekblom Johansson

2013

Abstract

The study in this thesis concerns Storglaciären, a very well known and studied glacier in northern Sweden. The glacier has been an object for research since the end of the 19 ^th century. During the meltseason of 2012 25 dyetracing experiments were executed. These experiments were conducted to investigate the internal drainage system of Storglaciären in the lower ablation area. Similar studies were done in 1989 by Regine Hock and Roger Leb Hooke (1993). The outcome of the study in 2012 has been compared with their results to see if any changes in the drainage system have occurred. The results have also been compared to the results of Seaberg et al. (1988) from their experiments in 1984 and 1985. Studies of glacier behaviour are important since they have a large impact on the local and global environment.

Moreover it has been observed that smaller glaciers (such as Storglaciären), that are easier to reach and to work on, have similar behaviour as bigger glaciers, making them good objects for research (Jansson, 1996).

The experiments were conducted between the 6 ^th and 24 ^th of august and ex- ecuted by rst injecting dye into moulins on the glacier and then measuring the concentration of dye in the proglacial streams merging out from the front of Stor- glaciären. Rhodamine WT was used as dye. Storglaciären has three main pro-glacial streams named Nordjåkk, Centerjåkk and Sydjåkk. Nordjåkk merges from the north side and the other two from the south side of the glacier front. Measurements were in the beginning taken in all of the streams but since no concentration was visible in Nordjåkk the focus was at the end of the eldperiod only at Centerjåkk and Syd- jåkk, which both had detectable dye concentrations. Both manual and automatical measurements were done.

Breakthrough curves (concentration vs. time) were plotted for each experiment and for both Centerjåkk and Sydjåkk. From these curves calculations were done following the methods in Willis et al. (2011). The main parameters calculated were: transit velocity, dispersivity and dye recovery. Breakthrough curves were also modelled for each experiment using the method in Willis et al. (1990).

Overall the drainage system in the lower part of the ablation area of Storglaciären has not changed signicantly during the past 20 years. But the drainage system seems to be divided into dierent parts using both a straight channel system and a distributed system. The distributed system of 2012 seems to be more homogeneous than in 1989 but whether the system is braided or consists of a linked cavity system is hard to tell.

Dierences seen this year compared to previous investigations are that the tran- sition from an early to a late season drainage system occurred later in the meltsea- son. The dominating subglacial stream in 2012 was Centerjåkk and not Sydjåkk as in previous investigations (Nordjåkk dominated north as before). The meltseason lasted only a few weeks in 2012 because of the cold conditions and low precipitation.

This may have had a big inuence on the behaviour of the glacier.

Sammandrag

Studien i detta masterarbete berör Storglaciären som är en välkänd och sedan länge studerad glaciär i norra Sverige. Glaciären uppmärksammades för första gån- gen i början av 1900-talet och sedan dess har glaciären ingått i ett ertal studier. Un- der dess smältsäsong 2012 utsattes Storglaciären för 25 st spårämnesförsök. Dessa experiment utfördes med syftet att undersöka Storglaciärens interna dräneringssys- tem i dess nedre ablationsområde. Liknande studier utfördes år 1989 av Regine Hock och Roger Leb Hooke (1993). Resultaten från studien 2012 jämförs med deras resul- tat för att se om några förändringar i dräneringssystemet har skett. Resultaten har också jämförts med resultaten från Seaberg et al. (1988) från deras spårämnesförsök utförda år 1984 och 1985. Studier av hur glaciärer beter sig är viktiga eftersom de har en stor inverkan både på sin lokala och den globala miljön. Dessutom har det observerats att mindre glaciärer har liknande beteende som större glaciärer, vilket gör dem till passande studieobjekt (Jansson, 1996).

Spårämnesförsöken utfördes mellan den 6 och 24 augusti genom att hälla ner spårämne i glaciärbrunnar lokaliserade på glaciären, för att sedan mäta efter dess koncentrationer i de proglaciära jåkkar som strömmar ut från Storglaciärens front.

Storglaciären har tre jåkkar: Nordjåkk, Centerjåkk och Sydjåkk. Nordjåkk benner sig på glaciärens norra sida och de andra två på dess södra sida. Mätningar gjordes från början i alla tre jåkkar men eftersom inga koncentrationer uppmättes i Nord- jåkk yttades fokus till enbart Sydjåkk och Centerjåkk som båda gav resultat.

Både manuella och automatiska mätningar utfördes. Rhodamine WT användes som spårämne. Så kallade breakthrough curves plottades (koncentration mot tid) för varje experiment och jåkk. Med utgångspunkt i dessa gjordes era beräkningar bl.a. transit velocity, dispersivity och dye recovery efter metoderna i Willis et al. (2011). Breakthrough curves modellerades också för varje experiment enligt Willis et al. (1990).

På det hela taget har dräneringssystemet i den nedre delen av Storglaciärens ablationsområde inte förändrats mycket de senaste 20 åren. Systemet verkar dock vara uppdelat och glaciären använder sig både av ett kanalsystem och ett ätat system för dränering. Det ätade systemet 2012 visar tecken på att det är mer homogent än 1989, men om systemet består av länkade kaviteter eller tätt ätade kanaler går inte att fastställa.

Övergången från säsongens tidiga till säsongens sena dräneringssystem skedde

senare under smältsäsongen än tidigare observerats. Den dominerande subglaciala

jåkken var Centerjåkk och inte Sydjåkk som tidigare. Smältsäsongen 2012 varade

endast i några veckor på grund av de kyliga förhållandena och lite nederbörd. Detta

kan starkt ha påverkat glaciärens beteende detta år.

Contents

1 Introduction 1

2 Glacial Hydrology 3

2.1 The glacial drainage system . . . . 3

2.2 Proglacial runo . . . . 7

3 The dye tracing method 7 3.1 Rhodamine WT and Uranine . . . . 8

4 Storglaciären 10 4.1 Hydrology of Storglaciären . . . . 10

4.2 Recent dyetracing experiments on Storglaciären . . . . 13

5 Method 17 5.1 Field work . . . . 17

5.2 Discharge measurements . . . . 18

5.3 Available data - temperature, precipitation and discharge . . . . 18

5.3.1 Temperature and precipitation during previous studies . . . . 19

5.4 Calculations on each experiment . . . . 20

5.5 Modelled breakthrough curves . . . . 21

6 Result 22 6.1 Field observations . . . . 22

6.2 The breakthrough curves . . . . 22

6.3 Summary tables . . . . 24

6.3.1 Centerjåkk . . . . 25

6.3.2 Sydjåkk . . . . 26

6.4 Breakthrough curve shapes . . . . 27

6.5 Moulin groups . . . . 27

6.5.1 Dispersivity and transit velocity . . . . 27

6.5.2 M1 . . . . 29

6.5.3 M2 . . . . 29

6.5.4 Moulin 8 . . . . 29

6.5.5 Moulin 5 . . . . 30

6.5.6 Moulin 12 . . . . 31

6.5.7 Neglected experiments . . . . 31

6.6 Sources of error . . . . 31

7 Discussion 33 7.1 Stream vs. clean calibration . . . . 33

7.2 Manual vs. automatic sampling . . . . 34

7.3 Dye recovery . . . . 36

7.4 Breakthrough curve shapes . . . . 36

7.5 Comparison to previous experiments . . . . 37

7.5.1 Transit velocity . . . . 38

7.5.2 Dispersivity . . . . 40

7.5.3 The proglacial streams . . . . 40

7.6 Distribution of the system . . . . 41

7.6.1 Comparing to Hock and Hooke (1993) and Seaberg et al. (1988) . 42 8 Concluding remarks 43 9 Thank you 44 10 References 44 10.1 Data . . . . 44

10.2 Figures . . . . 44

10.3 Internet . . . . 44

10.4 Manuals . . . . 45

10.5 Printed sources . . . . 45

11 Appendix A 47

12 Appendix B 56

List of abbreviations

• Exp. = experiment, each experiment were given a number in the order of execution.

• Date , the date when the experiment occurred, all in august 2012.

• P eak spec. = peak specication and how the breakthrough curve appears: 1 p = the curve has only one distinct peak, s = the curve consists of more than one peak (they are separated from each other), m = a multipeaked curve and er = the cuve has a erratic behaviour.

• Sampling method , tells if the samples were picked and measured manually (manual) or automatically (auto).

• Injection time gives the exact time when the dye were injected (hour:min).

• w 0 = Amount of dye injected in ml.

• M , stands for Moulin and says which moulin was used as the injection point for each experiment respectively.

• x = Distance between injection and sample point in meters.

• Q = Total daily mean discharge of both Centerjåkk and Sydjåkk, in m ³ s ⁻¹ .

• Q _c /Q s = Daily mean discharge of each stream respectively, in m ³ s ⁻¹ .

• Q _p = Discharge measured when the peak of the rTW-concentration occurred, in m ³ s ⁻¹ . This value is specied for each stream respectively.

• P eak , is the time when the concentration peak occurred during the experiment.

• u = Transit speed through the glacier in ms ⁻¹ . u is corrected for pressure dierence when the water emerges from the glacier front.

• D = The dispersion coecient in m ² s ⁻¹ .

• d = The dispersivity in meters.

• w = Weight of the dye passing in kilos, calculated using Q.

• p = Percent of dye recovered after travelled through the glacier, calculated using Q.

• ¹ Discharge not recorded during this experiment, discharge from 8 ^th of august were

used during calculations.

1 Introduction

Glaciers have long been a subject to the human interest especially in areas depending on the seasonal melting of glaciers to ll rivers with fresh water or the areas using water for hydro electrical power. They also aect glaciated catchments in shorter time scales by even out precipitations peaks etc. Glaciers are studied to understand these processes but also to prevent or to be prepared for catastrophic events caused by glaciers behaviour, such as outburst oods from lakes. They are also studied to understand how they may contribute to climate change or how they might respond to the changing climate and what will happen if they disappear (Benn and Evans, 2010). Glaciers represent a huge water storage and their disappearing or survival has a large inuence on the global sea level, which would increase by almost 70 meters if all glaciers present on Earth were to disappear (Holmlund and Jansson, 2003). It is known that studies on smaller glaciers such as Storglaciären can be compared with other bigger ones since they have similar properties (Jansson, 1996). This is also very practical since large glaciers may be dicult to work on and some glaciers are just hard to reach. The glacier environment is inuenced by the climate which controls the melting of snow and ice and the glacial runo. The presence of water in the glacier system is very important, particularly at the glacier bed (in the subglacial drainage system) which has a large impact on the motion of the ice. Also due to transport of water from the glacial surface through the glacier, climate processes occurring at the surface indirectly inuence processes occurring at the glacial bed (Benn and Evans, 2010).

Storglaciären, a well known glacier in northern Sweden, has been the object of research since the end of the 19 ^th century when the rst photo of the glacier was taken. Surface mass balance measurements of the glacier were initiated in 1946 and are still carried out on an annual basis, making the mass balance record of Storglaciären the longest continuous record in the world (Koblet et. al., 2011). Several experiments and investigations have been conducted on the glacier including studies of the glacier dynamics, ice velocity, bedrock topography and glacial hydrology (Jansson, 1996). One of these studies was done in 1989 when ten successful dye tracing experiments were performed. The purpose of this study was to investigate the seasonal, diurnal and spatial variations in the internal drainage system of the lower ablation area of Storglaciären. This was done by Regine Hock and Roger LeB. Hooke who published an article on their results named Evolution of the internal drainage system in the lower part of the ablation area of Storglaciären, Sweden (1993). The presented Master's thesis is based on the named article in such that it aimed to conduct similar dye tracing experiments on Storglaciären and to compare the

ndings with the conclusions drawn by Hock and Hooke (1993).

The goal of this thesis is to estimate whether any changes occurred in the lower abla-

tion area of Storglaciären during the last 20 years. Seaberg et al. (1988) similar to Hock

and Hooke (1993) also conducted dye tracing experiments on Storglaciären during the

years of 1984 and 1985. Their results are also considered in the comparison to the results

gained from this study. The eldwork for this thesis was conducted between the 6 ^th and

24 ^th of August in 2012. Dye tracing experiments were preformed using Rhodamine WT

dye as tracer. The measurements of dye concentrations after injection on the glacier were

conducted in all three pro-glacial streams emerging from Storglaciären during each ex-

periment. Dye concentrations were measured both manually and automatically using two

dierent uorometers. Several moulins (>25) were located on the glacier and 11 of them

were used during 25 dye tracing experiments. Twenty of the experiments showed clear

time-concentration curves and were used in further calculations based on the methods

proposed by Willis et al. (2011) and Willis et al. (1990). This study focuses on the lower

ablation area of Storglaciären since all moulins and crevasses used in previous studies

were located there. The summer of 2012 was relatively cold and only a small amount of

precipitation occurred when the experiments were conducted.

2 Glacial Hydrology

Melting of snow and ice is one of the main sources of water for glaciers, it concerns melting of the surface but also within the glacier and at its bed. The latter two are caused by ice deformation or sliding of the glacier, which causes friction heat and gives little contribution to seasonal water production. Melting at the surface is the result of solar radiation and turbulent heat ux. Water can also be added from external sources like precipitation, groundwater or surface runo. Since glaciers exist in several dierent environments the contribution of these factors vary a lot from region to region (Benn and Evans, 2010). For temperate glaciers, such as Storglaciären, the most important source of water is the melting at the glacier surface (Shreve, 1972). The water ow is determined by elevation and pressure dierences: water ows from high elevation towards lower elevation and from higher pressure to lower pressure. Thus, water ows along gradients of hydraulic potential (also known as hydraulic head). For streams at the glacier surface the hydraulic potential depends on the elevation and the mass of the water whereas for the water ow through a glacier (in conduits) the dierence in water pressure gets involved.

The hydraulic potential is given by equation 1, with φ as hydraulic potential, ρ as water density, g as gravity, z as elevation and P w as the water pressure (Benn and Evans, 2010).

φ = ρ _w gz + P _w (1)

The two terms on the right side of equation 1 are called elevation head and pressure head respectively. The pressure head for a still standing water body is given by equation 2 where h w equals the elevation of the water surface (Benn and Evans, 2010).

P _w = ρ _w g(h _w − z) (2)

If the sum of the pressure and elevation head is constant the hydraulic potential is equal over the whole water body. This occurs in a still standing water body since the water only moves along a change in hydraulic potential. The friction caused by the water

owing through conduits results in a drop in the pressure head, ∆φ f (see equation 3) while the gradient is determined by the resistance to ow (Benn and Evans, 2010). Inside the glacier several locations can have the same hydraulic potential. These areas make equipotential surfaces where the hydraulic gradient is equal. This results in that the water is moving perpendicular to these equipotential surfaces as the hydraulic potential is largest in this direction. The direction of the ow is towards lower potential, resulting in a water movement downwards through the glacier in the direction of the ice (Holmlund and Jansson, 2003).

P _w = ρ _w g(h _w − z) − ∆φ _f (3)

2.1 The glacial drainage system

The glacial drainage system is divided into four parts: supraglacial, englacial, subglacial

drainage and storage (see a simple illustration in gure 1). The supraglacial drainage

system includes the water on the glacier surface and is comparable with a owing river

system. Undisturbed glacier ice is considered impermeable and channels develop in net-

works of small streams with the ow controlled by gravity (Shreve, 1972). The network

pattern may change or develop following ice structures on the glacier such as crevasses or foliation (Benn and Evans, 2010).

Figure 1: The gure is a simple illustration of the glacial hydrological system.

The englacial drainage system of a glacier consists of a network of conduits and channels of dierent sizes. Opening of conduits is a result of the frictional heat created by owing water. This heat makes conduits bigger at higher discharges since faster

owing water produces more friction heat. In addition, the conduits are kept open due to the pressure of the owing water. If the ice pressure exceeds the water pressure, the conduits will close by ice creep. Water movement in the englacial system is slow and many conduits are relatively small and not always straight channels (Holmlund and Jansson, 2003). Since the water has no contact with the bed it is mostly free of sediments and appear as clear water.

The subglacial drainage system occurs along the interface of the glacier with the

underlying bed resulting in water enriched with sediment. The suspended load in the

subglacial runo and the contact with the bed make this the most interesting of the three

systems (Shreve, 1972). The subglacial drainage system is generally divided into two

main categories: channelized systems and distributed systems. The channelized system

transports water eectively with fast ows through an eciently developed network of

channels. The distributed system on the other hand has smaller conduits and a winding

network for the water to ow through making it a slow transporting system. To the

channelized system belong channels that are incised into the ice, so called R- or H-

channels, into rock or sediment (so called N-channels) and tunnel valleys which typically

have the same shape as N-channels but are larger (Benn and Evans, 2010). The channels

can have dierent shapes ranging from straight and single branched channels to channels

with several side branches to accommodate the water surplus during high ows. These

channels can also develop an arborescent network similar to stream networks (Fountain

and Walder, 1998). The tunnels are maintained by that higher water pressure develops in the smaller channels and lower pressure in the larger once, forcing the larger channels to attract water from its surroundings (Benn and Evans, 2010). Subglacial tunnels are believed to be wide and to have a low dome shape, not semicircular as may be seen at the front of a glacier (Hock and Hooke, 1993). A semicircular shape resists closing better and based on hydraulic properties a subglacial tunnel is able to close rapidly at low water pressures (Jansson, 1996). Since the hydraulic potential is determined by the water pressure and elevation, subglacial channels do not need to follow the slope of the glacial bed. This gives the channels the freedom of owing both uphill or across the slope (Benn and Evans, 2010). The distributed system consists of linked cavity networks between the ice and the bed, braided channel networks between ice and sediments, lm

ow and groundwater ow. Linked cavity system represent a way for the glacier to be able to hold large amounts of water below the glacier under high pressures which increases the glacier's motion while maintaining low water ow velocities at the same time. This would be impossible in a system consisting only of tunnels where the water pressure decreases with higher discharges.

The velocity of the water is controlled by the size of the cavities and it is possible that cavities can exist at the same time as channels. It is suggested that the subglacial drainage system at the end of the winter season has the structure of linked cavities and as soon as the melt period starts the system develops tunnels (Holmgren and Jansson, 2003). Cavities develops at the interface between the glacier and its bed where the local ice pressure is lower than the basal water pressure, for example, on the downglacier side of irregularities in the glacier bed. Groundwater ow is usually neglected since hard bedrock has often a very low permeability even though exceptions exists, for example limestone (Benn and Evans, 2010). Between the glacier ice and the bed a thin (µm) lm of water is owing at very low velocity. This is known as lm ow and inuences the motion of ice, through a process named regelation (melting and refreezing of ice due to contact pressure dierences between the ice and the bed caused by its irregularities). The water transported at the glacier bed is believed to be produced by local basal melting, which occurs because of high pressures and energy derived from friction and geothermal heat. The water emerging from subglacial sources is turbid and enriched with sediment since the water has contact with the bed (Holmlund and Jansson, 2003).

Storage occurs in both liquid and solid form. Water can get trapped in cavities or form lakes. Firn aquifers may develop when water at the surface percolates through the snow and refreezes when reaching a lower temperature and become superimposed ice. Latent heat is released in the refreezing process and the snow above melts from underneath. The water accumulates in the remaining snow, forming a rn aquifer (Benn and Evans, 2010).

Water travels between these systems and may also exit the glacier by proglacial streams emerging at the front of the glacier or by lateral streams at the glaciers side (Stenborg, 1973).

One way to route water from the supraglacial system into the englacial system of

a glacier is by water inow into moulins and crevasses. This is possible during the

melt period when the englacial and subglacial system are adapting to increases in the

production of surface melt water (Shreve, 1972). Crevasses develop from pre-existing

cracks in the glacier surface. For a crack to start expanding it has to be lled with water,

so that the pressure on the walls inside the crevasse overcome the ice strength and the

overburden pressure. Crevasses and moulins can stretch from the surface down to the glacial bed (Benn and Evans, 2010). Moulins develop preferably in the upglacial part of crevasse elds when supraglacial water drains into some part of a crevasse. New moulins prevent surface water from owing further along the glacier surface and existing moulins placed on the same route terminate when the water supply ceases. Water lled holes (old moulins etc.) do not develop into draining moulins since running water has to be present for this to happen (Stenborg, 1972).

Englacial conduits are ecient in transporting surface water down to the bed making the englacial drainage system an important connection between the glacial surface and the bed of the glacier. In temperate glaciers when surface meltwater reaches the bed and increases the water storage an accelerated basal motion is initiated. This demonstrates that processes acting at the bed are inuenced by processes acting at the surface, which are directly inuenced by changes in climate and the local environment (Benn and Evans, 2010). Most glaciers experience a annual melting period and the change through the melt season aects among other things the ice dynamics and the sliding speed of the glacier.

This is because the ice velocity is strongly connected to the subglacial water pressures (Jansson, 1996). High water pressures in the subglacial system result in large movements of the glacier (Seaberg, 1988). However, at the end of the winter season the conduits and channels are shrinking due to the ice deformation and absences of owing water.

During the winter only a small amount of water transport through the drainage system occurs. When the melt period starts in spring, water is pushed into the shrunken and narrow conduits and they opens. The conduits open by friction heat due to the owing melt water (Hock and Hooke, 1993). Since there is not enough space the water pressure increases, which also increases the sliding speed and the surface velocity of the glacier.

During the melt season the conduits grow and the system becomes less arborescent and is able to accommodate more water (Seaberg, 1988). As the conduits get larger it is harder to maintain a high pressure. As the pressure decreases so does the velocity of the water, especially when a colder period starts with decreasing water supply towards the end of the melt season. The conduits start to shrink to smaller size and the system nally collapses by ice creep. When the system is smaller it is again easier to build up higher water pressures to increase glacier velocity. This response to changes in the conduit system makes the velocity of glaciers very dynamic (Holmlund and Jansson, 2003). During the summer the water pressures experience large dierences even during a 24-hour period.

The maximum amplitude is near or above ice overburden pressure and the minimum value can go to near the atmospheric pressure. During the winter, as the conduits close, the water pressure gradually increases (Jansson, 1996).

Depending on climate, thermal regime, glacier geometry and such, the glacial hydrol-

ogy system changes a lot in both time and space. The changes aect the glacier itself and

reect the environmental changes both locally and globally. By understanding glacial

responses we might be able to understand and get a hint about the future climate.

2.2 Proglacial runo

The proglacial runo is inuenced by many factors. The sources for baseow are sub- glacial melt waters, water stored within the glacier and groundwater, which normally vary little during the day but more over longer time scales. The diurnal temperature cycle drives the daily discharge from glaciers into streams and adds discharge to the basal ow. This water often derives from eciently transported meltwater from quickly bared ice surfaces. The peak in the daily discharge uctuations is determined by the time of maximum melting at the glacier surface but typically lagged since the water has to travel through the glacier terminus from its melting source. The lag time is long for poorly developed glacier drainage systems and shorter if the conduit system is well devel- oped and the surface consists of ice rather then snow. This also explains the occurrence of shorter lag times at the end of the melt season when the drainage system is more ecient. Rainfall can also contribute to melting of snow and ice and this water can di- rectly contribute to glacier runo. On glaciers where the seasonal snow cover is removed, high magnitude rainfall events for example can result in large rainwater contributions to glacier runo through direct surface runo (Dahlke et al., 2013). In climate regions with maximum precipitation occuring during the winter and maximum temperature oc- curing during the summer, the maximum discharge is seperated from the precipitation peak. This is because the precipitation stored during the winter and released during the warmer summers produces higher runo. As the storage decreases towards the end of the summer, the runo decreases as well. In areas where the precipitation maximum occurs during the summer along with the maximum temperature, the discharge follows the seasonal pattern of the precipitation and the glacial inuence on the seasonal runo

cycle is not distinguished (Benn and Evans, 2010).

3 The dye tracing method

To investigate the englacial and subglacial drainage system dye tracing test may be per-

formed. A tracer is injected into a moulin or crevasse and at the front of the glacier

attempts are made to detect and monitor the concentration of the injected tracer as it

passes through the system. The measured tracer concentration is then plotted versus

time, and the shape of the graph (also called breakthrough curve) gives a hint of how the

drainage system may be structured. Dye tracing method have been used in hydrological

experiments for many years and go by several dierent names, such as tracer dilution

method, sudden injection method or the dye tracing method. The dye tracing method

has a wide range of applications including ow and discharge measurements, to mea-

sure pump performance, dispersion studies, leak detection, to measure mixing of water

of dierent origin, characterization of fractures in bedrock and wells, characterizations of

glacial drainage systems and to track pollution (Turner designs, 998-5121, 2012, Inter-

net). Dye tracing experiments can be performed in two dierent ways: the continuous

method where the dye is injected at a constant rate and the sudden injection or slug

injection method when a known amount of the tracer is injected instantaneously. The

latter is further described as it was used in this project. A known mass of the tracer is

quickly injected upstream or into a moulin and continuous measurements of the tracer

concentration in the streamwater are taken downstream at certain time intervals. The

distance between the injection point and measurement location has to be long enough to give the tracer time to properly mix with the water (Schnegg et al., 2011). Measurements of the dye concentration can be done manually or automatically. The result is a so called breakthrough curve showing concentration versus time. From this plot the amount of discharge, dispersion, transit velocity, and dye recovery can be calculated. The method works similarly when used on a glacier. When researchers started to use the tracer di- lution method sodium chloride was widely used as a tracer, but today there are several tracers to choose from. A big advantage of using ourescent dye instead of sodium chlo- ride is the lower amount that has to be used to get a useful breakthrough curve (Schnegg et al., 2011). Both liquid and powder dyes can be used.

Concerning dye tracing experiments on a glacier, it is common that tracers poured early in the glacier melt season typically emerge at the front in diuse curves, charac- terized by wide or multiple concentration peaks. This indicates a slow and inecient transport of water. However, as the meltwater increases during the melt season and the system develops, the breakthrough curves become more even and sharp. The velocity through a glacier (the transit speed) is also lower at the beginning of the melt season and increases as the drainage system develops (Benn and Evans, 2010). Hydraulic parame- ters such as dispersion and dispersivity are often calculated from the experiment and are derived from the width of the breakthrough curve. It is a measure on how distributed the subglacial system might be. The breakthrough curve from one injection may show more than one peak. These so called multipeaked curves indicate the presence of anabranches in the system, which are small or long enough so that the dye following two dierent channels does not arrive at the same time at the terminus. The distance between the peaks decreases when the length or the velocity dierences in the branches decreases, thus, multipeaks suggest a distribution of the glacial drainage system. When homoge- neous distribution of the drainage system occurs multipeaks merge and may result in larger dispersivities. This occurs when the system has a large number of anabranches of dierent sizes and lengths. The curve may also show several peaks, which are more or less separated and they often suggest inhomogeneous braiding. Secondary peaks may indicate storage of glacial water, that could have been stored in cavities etc. in the glacier conduit system and contributed to the glacier runo later than expected (Hock and Hooke, 1993).

3.1 Rhodamine WT and Uranine

Rhodamine WT (water tracable) dye has the color of red/pink and is soluble in water. The substance molecular formula is C 29 H ₂₉ N ₂ O ₅ ClN a ₂ and it has the CAS Number: 37299- 86-8. (Rhodamine WT, abbey color, 2012, Internet). The dye has a grand uorescent capacity which makes it very useful for dye tracing experiments. Uranine is another

uorescent dye that is derived from uorescein (an orange powder with the molecular

formula C 20 H ₁₂ O ₅ ) but dissolved in water. It has an orange/bright yellow color (Uranine,

abbey color, Internet). Compared to rhodamine uranine has two disadvantages, the dye

is very sensitive to sunlight and also to pH (Turner designs, 998-5103, 2012 Internet).

Figur e 2: Map illustr ating Stor glaciär en.

4 Storglaciären

Figure 3: The blue dot is were Storglaciären is located in Sweden ( c Lantmäteriet, i2012/921).

Sweden has about 300 glaciers and several of them are located in the area of the highest mountain in Sweden, mountain Kebnekaise. Storglaciären is lo- cated on the east side of the Kebnekaise massif with the front facing the Tarfala valley (67 ^o 55' N, 18 ^o 35' E, Figure 3) (Jansson, 1996). Storglaciären is a small polythermal valley glacier with a length of 3.2 km. The area of the glacier is approximately 3 km ² and the average thickness is 95 m. The max- imum thickness is found in the upper part of the ablation area with approximately 250 m. The front is at an elevation of 1120 m a.s.l. and the top of the accumulation area lies at 1730 m a.s.l. (Jans- son, 1996). The main part of the glacier is tem- perate but in the ablation area there exists a cold surface layer of 20-60 m thickness. The thickest part of the layer is at the front and it becomes thin- ner towards the equilibrium line (Schneider, 1999).

The bedrock in the area of Storglaciären consists mostly of amphibolite and gneiss (Andréasson and Gee, 1989). Beneath the ablation area, about 1 km West from the glacier front there is a bedrock ridge consisting of rock more resistant to glacial erosion than its surroundings. This ridge is called the riegel and is important concerning the drainage pattern of the glacier and therefore also its dynamics (Jansson, 1996). A map of Storglaciären is presented in g- ure 2, see a picture of the glacier in gure 4.

4.1 Hydrology of Storglaciären

Storglaciären has three main pro-glacial streams emerging at the front of the glacier, which are named Nordjåkk, Centerjåkk and Sydjåkk. In previous studies Centerjåkk is only occasionally mentioned. Centerjåkk was rst mentioned by Hock and Hooke (1993) as a branch that diverted from Sydjåkk. They named this new branch Centrumjåkk and noted that the stream merged with Nordjåkk downstream. Centrumjåkk is called Centerjåkk today. They concluded that the branch originated from Sydjåkk because of its high sediment content. The pro-glacial streams are dierent in appearance, especially Nordjåkk, which has a lower sediment load. The water in Nordjåkk is clear and is believed to carry englacial water from the accumulation area and supraglacial water from the glacier surface, i.e. water that has not been in contact with the glacier bed. Sydjåkk and Centerjåkk carry a visibly larger sediment load and are suggested to be lled with subglacial water from the ablation area (Hock and Hooke, 1993 and Jansson, 1996).

Sydjåkk also receives some supraglacial water owing down the south side of the glacier

Figure 4: The picture illustrates Storglaciären and its proglacial streams owing into the Tarfala valley.

front. The bed of the streams consists of cobbles and boulders and they are relatively stable even though they change continually through the seasons, especially after bigger rain events (Hock and Hooke, 1993). A close-up picture of the pro-glacial streams are seen in gure 5.

Crevasses are common on Storglaciären and one major crevasse eld is normally cre-

ated where the glacier ows over the bedrock riegel (see gure 2). Moulins are often

created in or close to these crevasse elds (Stenborg 1969). To get to the englacial and

subglacial drainage system the water has to ow through these moulins and crevasses

because of the cold layer in the ablation area, which prevents the percolation of water

(Hock and Hooke, 1993). The water can also percolate through the temperate rn area to

enter the internal drainage system (Jansson, 1996). Based on the hydrological properties

of Storglaciären Jansson (1996) divides the glacier into three parts: the rn area above

the equilibrium line, the upper part of the ablation area reaching from the equilibrium

line down to the riegel and the lower part of the ablation area reaching from the riegel

towards the front. The rn area is not well investigated but has water owing through

a rn aquifer and englacially drains into Nordjåkk. The upper part of the ablation area

drains slowly through the englacial system with water draining mainly from the rn area

into Nordjåkk. The lower part of the ablation area has a faster subglacial drainage sys-

Figure 5: The picture shows a close-up of the three streams. From left: Sydjåkk, Center- jåkk, Nordjåkk in two branches. The picture is taken in the beginning of the eld period when a lot of snow was still remaining and the streams were not owing as much.

tem and the water source is the surface water running from the upper ablation area down into moulins and crevasses in the area of the riegel (Jansson, 1996). This water has most denitely contact with the glacier bed and leaves the glacier via Sydjåkk.

The riegel mentioned before is important for the drainage pattern of Storglaciären.

Glacial erosion have created overdeepenings in the bedrock topography on the riegels sides. Storglaciären has in total three overdeepenings which alter the subglacial system.

The overdeepenings makes it is hard for the subglacial water to keep the tunnels open.

This since the water needs to travel upwards and the energy needed for the upward

transport would result in a lowering of the water temperature and eventually a closure of

the tunnels. Thus, it is assumed that this area does not have well developed subglacial

system but rather englacial conduits. Since the water in Storglaciären's englacial system

is transported with low velocity the pressure dierences due to the overdeepenings do

not become as large. The dispersion of the drainage system in the lower ablation area is

assumed to be high at the beginning of the melt season and gradually decreasing when

the melt increases, usually during July (Seaberg, 1988). The reason is that the drainage

system is poorly developed after the winter season. Conduits and tunnels have shrunk or

even closed making the system braided with small conduits. As the season moves past

the summer the conduits slowly grow and straighten making the system less braided.

The active component is the owing water which widens the conduits through friction and energy and results in decreased transfer time of water through the system (Hock and Hooke, 1993).

4.2 Recent dyetracing experiments on Storglaciären

There are three previous studies that are relevant for the experiments done in this study.

These include articles from Hooke et al. (1988), Seaberg et al. (1988), and Hock and Hooke (1993). The main article used for comparison is Hock and Hooke (1993) which discusses ten successful dye tracing experiments on Storglaciären performed during the melt season of 1989. The purpose of the 1989 study was to investigate the seasonal, diurnal and spatial variations in the internal drainage system in the lower ablation area of Storglaciären. Sodium uorescein and rhodamine WT were used as dye and poured into moulins, crevasses and boreholes. Concentrations were measured at four places along the three pro-glacial streams. Hock and Hooke (1993) drew the conclusion that the lower part of the glacier consists of a multi-branched arboresecent network of passages. This conclusion is based on the observed tracer-transit times and the fact that moulins located close to each other gave dierent results. A braided system is suggested because secondary peaks were seen in several breakthrough curves along with diurnal variation. Secondary peaks may also occur due to temporal storage of dye instead of braiding, since the dye may got trapped during high discharge events and released when discharge decreased again. However, the possible reservoir had to be dammed by pressure dierences for the dye not to return to the main ow as soon as the concentration in the main stream dropped below that in the reservoir. Thus, they concluded that the reservoirs were not likely the reason for the occurrence of the secondary peaks. Multi-peaked breakthrough curves also occurred and indicated a braided system. When comparing Sydjåkk and Centerjåkk the dye concentrations were signicantly higher in Sydjåkk but still present in Centerjåkk. This suggests an unstable or irregular subglacial connection between the two streams with dilution of the water reaching Centerjåkk. This happens when the water owing towards Centerjåkk takes another path than the water owing directly into Sydjåkk. The dispersivity (the spreading of dye in the system) values derived from the breakthrough curves ranged between 2 and 55 m (Sydjåkk) with the highest value (55 m) measured in July 1989. Only one measurement was done in July and the values in August 1989 were signicantly lower (the second highest value were 17 m). It is possible that Storglaciären has an inhomogeneous braided drainage system at the end of the winter and at the beginning of the melt season, when low discharges occurs. As the season progresses the system becomes more ecient with higher discharges and the divides between the anabranches get water lled, thus, decreasing the braiding. This also explains why the occurrence of multipeaks decreases towards the end of the season (Hock and Hooke, 1993).

Seaberg et al. (1988) also preformed dye tracing experiments on Storglaciären to

investigate the character of the englacial and subglacial drainage system in the lower

ablation area. The experiments were conducted during the melt seasons of 1984 and

1985. The moulins used were located just above the riegel. Rhodamine WT was used as

dye. The dispersivity values ranged between 1.5 and 5.1 m (Sydjåkk) in 1985. Seaberg et

al. (1988) also concluded that the drainage system was braided in the lower ablation area

and that the anabranches were small at the beginning of the melt season. As the melting increased the velocity increased and the braiding decreased; a big drop in dispersivity can be seen when the conduits reach nearly their maximum size. The transit velocity from both articles are presented in gure 6, the equations for the relations are v = 0.44Q ^0.75 for Hock and Hooke (1993) and v = 0.26Q ^0.98 for Seaberg et al. (1988). Seaberg et al.

(1988) came also to the conclusion that the conduit system inside Storglaciären was more or less fully occupied with water as indicated by the exponent close to 1. Hock and Hooke (1993) drew the conclusion that there is a possibility of an increase in cross-sectional area of the ow in at least part of the system as discharges increases, this since they estimated a smaller exponent.

Figure 6: The graphs shows the velocity in relation to discharge of both Hock and Hooke (1993) and Seaberg et. al. (1988)

Seaberg et al. (1988) write in their conclusion that The drainage system leading from the moulins in the middle of the ablation area of Storglaciären to sampling site S-1 seems to consist of a single homogeneously braided stream under most conditions. S-1 is a measuring point on the south side of the glacier margin. They concluded that one quarter of the water in Sydjåkk emerges at the front which follows a separate drainage system but is connected to the one owing passed S-1. Hock and Hooke (1993) state that

the drainage system beneath the lower part of Storglaciären consists of a multi-branched

arborescent network of passages. The passages are believed to be braided, low and wide

in there appearance to be able to close rapidly with low ows. Figure 7 shows the map of

Storglaciären from the experiments of Hock and Hooke (1993) and gure 8 shows the map

of Seaberg et.al. (1988). Both maps includes injection points and sampling locations.

Figur e 7: Map of Stor glaciär en of Ho ck and Ho oke (1993).The inje ction points ar e th e moulind name d M1, M2 and M3. The sampling points use d was S1, C1 and N1.

Figur e 8: Map of Stor glaciär en of Se ab er g et.al. (1988). The in je ction points ar e name d M-1 thr ough M-3, JM-4, JM-5 and ST-1. The main sampling lo- cation wer e S-1 but also S-2 and N-1 was use d for dyetr acing exp eriments.

Figur e 9: The map is a close up on the Stor glaciär en ar ea wher e the exp eriments we re exe cute d.

5 Method

The dye tracing method was used with rhodamine WT as a tracer. 11 moulins were used in 25 dye tracing experiments, 20 of these experiments were analysed closer. The eld work is described below followed by a brief description of how the discharge measurements were done in the three pro-glacial streams. Available meteorological data from Tarfala Research Station is also presented and the last section provides a summary of how the data analysis has been done.

5.1 Field work

Figure 10: Rhodamine wt poured into moulin 7.

Active moulins were located and mapped on the glacier using a hand-held GPS. An active moulin was dened as to have a considerable amount of wa- ter owing into it, no pool of water visible or audible and preferably deep. A good sampling location in front of the glacier was located and marked at each stream. Since all proglacial streams divert into sev- eral anabranches after emerging at the glacier termi- nus conuence points were chosen as preferred sam- pling locations. Each moulin and sampling location can be seen in gure 9. The moulins are numbered 1-12 and the sampling spots named S for Sydjåkk, C for Centerjåkk and N for Nordjåkk. Dye was prepared in the lab together with calibration solu- tions. Four calibration solutions were made, one us- ing tap water, denoted as clean calibration solution with low turbidity and one using water from each stream, (further denoted as stream calibration) to be able to compare and get an understanding of the turbidity. Dye was poured in the water owing into a chosen moulin in order to avoid dye being spread by the wind or getting trapped on the moulin wall.

Injection was done by a person not taking part in the sampling. The exact time of in- jection was noted and the sampling started soon after the dye was injected (gure 10).

The sampling was performed in all three streams using dierent sampling intervals that were adjusted after analyzing the results of the rst experiments. Nordjåkk did not show any sign of dye and therefore was given a lower sampling frequency (10 or 15 minutes) throughout the whole period. The main focus was on Centerjåkk and Sydjåkk. The sam- pling interval used for Centerjåkk and Sydjåkk ranged between 10 sec and 30 minutes.

The more frequent measurements were taken during peak concentrations. Manually taken

samples were brought back to the lab to let sediment settle down and to have the wa-

ter temperature adjust to the room temperature before analysis with the Turner Design

Aquauor handheld uorometer. The manual samples were measured both with clean

and stream calibration solutions. When using the automatic ourometer, the instrument

was calibrated in the eld using streamwater and a known amount of rhodamine WT dye

before each start of the experiments. The automatic measurements were conducted by two dierent automatic urometers, GUNN-FL30 (Albilia Co., Switzerland) and a sea- point rhodamine uorometer connected to a campbell CR10x datalogger (Willis et.al., 2012), which were deployed in Sydjåkk and Centerjåkk respectively. Data about the moulins, distance to each sampling location and their coordinates are in Appedix B.

5.2 Discharge measurements

Discharge was also measured in the proglacial streams using the dye tracing method, uranine dye and a handheld uorometer (AquaFluor). The aim was to measure the discharge at least twice a day, before and after a dye tracing experiment to determine whether the stream discharge was increasing or decreasing during the experiment. A few days after the measuring period started pressure transducers were installed in Nordjåkk and Centerjåkk, which together with the breakthrough curves from the discharge dye tracing experiments could be used to estimate the discharge continuously by establishing a so called stage-discharge curve.

5.3 Available data - temperature, precipitation and discharge

The data presented in gure 11 and table 1 was received from Tarfala Research Station (TRS), 2012. Graph 11A shows the total precipitation, graph 11B the temperature and graph 11C shows the discharge of the three streams respectively. Markers in gure11C show discharge volumes estimated manually using the Uranine dye injections. Over the course of the study period precipitation was very low and occurred only during the second half.

Table 1: Average temperature and precipitation of relevant years.

Temperature [ ^o C] Precipitation[mm]

Year June July August Jun-Aug Jun-Aug

1984 4.64 6.73 4.38 5.06 343

1985 6.18 7.64 3.76 5.86 472

1988 4.42 9.39 5.76 6.78 280

1989 2.81 5.34 6.04 4.75 407

2011 6.98 9.86 7.42 8.10 574

2012 2.71 5.58 5.77 4.38 437

The temperature varied between 0 and 11 ^o C. The discharge in Nordjåkk varied the

most and also showed the greatest peak discharge with 1.2 m ³ s ^-1 . Sydjåkk had the lowest

discharge throughout the measuring period and varied between 0 and 0.3 m ³ s ^-1 . All of

the three streams decreased in discharge towards the end of the eld period, especially

Centerjåkk, which had a discharge of approximately 0.6 m ³ s ^-1 before the 19 ^th of august

but only 0.08 m ³ s ^-1 at the end of august. Since the precipitation was low most of the water

from the glacier was caused by glacier melting. The temperature shown in gure 11 was

measured with a temporary weather station on Storglaciären at an approximate elevation of 1600 m a .s.l.

Figure 11: The gure displays the precipitation, temperature and the discharge of the three pro-streams respectively during the eld period.

5.3.1 Temperature and precipitation during previous studies

The average temperature (June-August) for the years when dyetracing experiments were

conducted on Storglaciären are similar (table 1). Ranging between 4.4-5.9 ^o C with the

coldest year being 2012. The development of the englacial and subglacial drainage system

is not only dependent on the current year's summer temperatures but also on the previous

year's temperatures. The total precipitation (June-Aug) ranges between 343-472 mm with

the lowest 1984 and highest 1985 but these are not extreme values compared to the years when no experiments were done. Most precipitation fell during July and August in the years of the previous experiments.

5.4 Calculations on each experiment

For each experiment the concentration of dye measured in the streams was plotted against the time where time = 0 at the injection of the dye. Successful experiments showed a clear breakthrough curve of the Rhodamine WT dye in either or both Sydjåkk and Centerjåkk. These breakthrough curve were used as a base for several calculations. The following calculations were done for each experiment using the same methods as proposed by Willis et al. (2011):

1. Transit distance, x in meters, is the distance between the moulin used (the injec- tion point) and the sampling location. The distance is measured as the horizontal straight line in ArcGIS using the GPS coordinates that were collected in the eld.

2. Residence time, t m , given in seconds as the time between dye injection and peak concentration.

3. Transit speed, u ms ^-1 gives an approximate value on the mean velocity of the ow through the glacier. Given by

u = x

t _m . (4)

For Centerjåkk and Sydjåkk u has to be corrected for the velocity dierence when the water exits the glacier and is able to ow in an open channel. Thus, u estimated with the above equation was corrected for the higher ow velocity in the stream outside the glacier using the following equation:

u _corrected = x − x stream

t _m − t _stream . (5)

Where x is the transit distance, x stream is the distance in meters from the terminus of the glacier to the sampling location in the stream. t m (time in seconds to peak concentration) is adjusted for each experiment. t stream is the time for the water to travel in the stream. t stream is given by

t _stream = x _stream

v _stream , (6)

where v stream is, for Sydjåkk, v syd = 0.843 ∗ ln(Qpeaks + 0.4968) , were Qpeaks is the peak discharge. For Centerjåkk, v center = 0.55 , an average from previous discharge estimates. (The new u should be smaller than the non-corrected u.)

4. The dispersion coecient, D, is a measure of the width of the breakthrough curve

in m ² s ^-1 and gives an idea about how much the dye spread out during the transport

from the moulin to the pro-glacial stream. D was calculated as equation 7 where t i

is the time in seconds to reach half the peak concentration on the rising and falling limb respectively on the breakthrough curve used. These are named t 1 (value on the rising limb) and t 2 (value on the falling limb).

D = x ² (t _m − t _i ) ² 4t ² _m t _i ln(2 ^t _t

i

)

, (7)

Equation 7 is rst solved using t i = t 1 to estimate D1 and then solved for D2 using t _i = t 2 . Thus, two equations are solved simultaneously until D1 = D2 by changing the value of t m .

5. Dispersivity, d in meters, is the measure of the spreading of dye in relation to the transit speed and is as equation 8. (D and u m are determined for each experiment using the previous equations.)

d = D

u _m , (8)

6. Dye recovery, w in kg, is the weight of dye passing the sampling location. Given by

w =

n

X

t=1

cQdt, (9)

where c equals the measured concentration of dye (minus the background concen- tration) in ppb = µl ⁻¹ and Q (ms ⁻¹ ) is the stream discharge at the time. For the calculation of the dye recovery in Sydjåkk and Centerjåkk the total discharge of both of the streams was used. This is because they seem to have the same path through most parts of the glacier and split up at the terminus. dt is the time in- terval in seconds between each measurement. The result is in µg. To get the dye recovery in kg equation 9 is multiplied with 10 ⁹ . w is expressed in percent, p, by equation 10 where w 0 is the weight of the dye injected in the moulin.

d = 100 w

w ₀ . (10)

5.5 Modelled breakthrough curves

The modelled breakthrough curves can be calculated by the equation Willis et. al. (1990) used, see equation 11.

c(t) = u Q _peak

v 0

(4πDt)

exp − (x − ut) ²

dDt (11)

where u (here not corrected), x and D is determined according to previous calcula-

tions. Qpeak is the discharge at the time of the peak concentration. v 0 is the volume (m ³ )

of the dye injected. t equals the time in seconds. For t a vector of increasing time with a

time interval of 10 second was used. If the observed concentration values were cut o in

time, t was articially lengthen. The amplitude of the modelled curves were modied by

adjusting the Qpeak, this to t the modelled breakthrough curve to the measured one.

6 Result

6.1 Field observations

The glacier was still snow covered in several areas at the beginning of the eld period at the end of July 2012. Also not many active moulins were found during the rst days.

After about one week the moulins opened up properly and the snow was almost gone.

Snow was also present at the front of the glacier covering the streams at the terminus, especially Nordjåkk which became more visible as the season progressed. Centerjåkk had two bigger snow patches covering the stream, one at the glacier front and one a bit further downstream, which disappeared towards the end of the season. Sydjåkk was snow covered downstream, where it merged with the other pro-glacial streams and remained snow cov- ered for the most part of the eld period. Sydjåkk stream had a notable sediment load at the beginning that increased during the measuring period. The ow also increased visibly during the rst week. However, Sydjåkk also showed contributions of clear supraglacial water that was contributed from the southern side of the glacier. Centerjåkk showed very little braiding, mostly only at the beginning of the eld period but when the snow melted away the stream carried noticeably more water and was less braided. However, the most visible change was the increasing amount of sediment load in Sydjåkk observed over the eld period. At the beginning of the season Centerjåkk showed more discharge than Sydjåkk. Nordjåkk originates from the northern part of the glacier surface and emerges in several anabranches at the northern side of the glacier terminus. This year the stream was very braided in the front, though two bigger branches were visible and merged downstream just before their conuence with Centerjåkk. Approximately 200 m south of the conuence of Nordjåkk and Centerjåkk is Sydjåkk joining the two streams (see gure 5 in section 4.1).

6.2 The breakthrough curves

The experiments resulted in almost 40 breakthrough curves including all experiments and both streams, in this chapter curves from two experiments are presented. See all of the breakthrough curves in appendix A (organized according to the grouping done in section 6.5). Figure 12 through 17 represents the breakthrough curves from two exper- iments, number 3 executed on the 9 ^th of August and number 11 executed on the 19 ^th of August. Experiment 3 were manually sampled and then analysed with the handheld

uorometer back in the lab while number 11 were sampled and analysed automatically.

Comparing Centerjåkk and Sydjåkk the dye concentrations were signicantly higher in

Centerjåkk than Sydjåkk for most of the experiments. Sydjåkk showed higher concentra-

tions in experiments 9, 10, 15, 21 and 25. During experiment 9, number 1 on the 18 ^th of

August, the automatic measurements from Centerjåkk is lower than observed in Sydjåkk

but the manual samples show higher concentration than Sydjåkk (gure 18 and 19). This

suggests that the uorometer measuring in Centerjåkk did not measure the correct con-

centration values. The same situation occurred during experiment 10, however, manual

samples were not taken during experiment 15, 21 and 25. Since Centerjåkk showed higher

concentrations than Sydjåkk during all other experiments it is possible that the values

were not correctly measured during these experiments due to some calibration error.

Figure 12: Centerjåkk experiment 3. Figure 13: Centerjåkk experiment 11.

Figure 14: Sydjåkk experiment 3. Figure 15: Sydjåkk experiment 11.

Figure 16: Both streams experiment 3. Figure 17: Both streams experiment 11.

Figure 18: Experiment 9. Figure 19: Experiment 10.

6.3 Summary tables

Information about each experiment and the result from the calculations are presented in two tables. Table 2 and table 3 summarize results for Centerjåkk and Sydjåkk respec- tively. All data presented are based on stream calibration unless marked with a star (*) indicating that the clean calibration was used. The transit speed in Centerjåkk decreased during the last experiments and ranged between 0.038 and 0.25 ms ⁻¹ . Sydjåkk had lower speeds throughout the whole period with a maximum value of 0.18 ms ⁻¹ . The dispersion coecient ranged between a minimum value of 0.011 m ² s ⁻¹ and a maximum value of 5.44 m ² s ⁻¹ for Centerjåkk and was generally smaller for Sydjåkk (0.21 - 7.11 m ² s ⁻¹ ).

Except for the extreme values, Sydjåkk showed generally lower or similar values to those

observed for Centerjåkk. This also applies for the dispersivity values. Sydjåkk showed

higher extreme values but had otherwise similar or lower values than Centerjåkk. The

range of Centerjåkk's and Sydjåkk's dispersivity values is 0.044 to 21.26 m and 1.1 to

27.36 m respectively. Centerjåkk had also higher dye recovery rates than Sydjåkk and

even three experiments with a dye recovery above 100 %. Centerjåkk had an average

recovery rate of 80.44 % during the eld period while Sydjåkk's average recovery was

54.6 %.

6.3.1 Cen te rjåkk Table 2: The table shows the result of Centerjåkk. E xp. D ate P eak spec.

S ampl ing method I nj ection time w 0 ml

M x m Q m 3 s − 1 Q c m 3 s − 1 Q p m 3 s − 1 P eak u m/s D m 2 s − 1 d m w k g P % 1 06 Aug m ulti man ual 11:40 50 2 1303 0 .1733 1 0 .1044 1 See Qc 14:01 0.12* 2.21* 14.55* 0.036* 63.65* 3 09 Aug 1 p, er man ual 12:30 75 2 1303 0.1149 0.1149 0.2200 15:30 0.097 0.55 4.56 0.12 139.39 8 17 Aug 1 p man ual 11:04 100 7 1341 0.8234 0.6044 0.6023 12:30 0.22 1.19 4.57 0.23 211.62 9 18 Aug 1 p man ual 11:22 150 4 1355 0.7516 0.5449 0.5716 12:55 0.21 0.63 2.59 0.23 136.93 9 18 Aug 1 p auto 11:22 150 4 1355 0.7516 0.5449 0.5716 12:52 0.22 0.011 0.044 0.089 53.06 10 18 Aug m man ual 14:56 80 4 1355 0.7516 0.5449 0.5211 16:30 0.22 1.71 6.59 0.11 124.51 10 18 Aug 1 p auto 14:56 80 4 1355 0.7516 0.5449 0.5211 16:16 0.24 0.32 1.13 0.039 43.42 11 19 Aug 1 p man ual 12:36 80 8 1289 0.5725 0.4012 0.4169 14:10 0.19 0.31 1.36 0.14 157 11 19 Aug 1 p auto 12:36 80 8 1289 0.5725 0.4012 0.4169 14:13 0.19 0.3 1.35 0.061 67.55 13 20 Aug s_1, er Auto 10:29 70 5 992 0.4611 0.3140 0.3356 11:48 0.16 1.56 7.45 0.0084 10.72 13 20 Aug s_2, er auto 10:29 70 5 992 0.4611 0.3140 0.3493 11:48 0.09 0.6 4.75 0.059 75.17 14 20 Aug 1 p auto 14:20 70 4 1355 0.4611 0.3140 0.3588 15:50 0.22 0.38 1.51 0.068 86.89 15 20 Aug 1 p, er auto 15:58 70 9 1269 0.4611 0.3140 0.3441 17:09 0.26 5.44 18.26 0.066 83.44 16 21 Aug 1 p, er auto 11:20 70 2 1303 0.4000 0.2669 0.2924 14:16 0.099 1.94 15.72 0.026 33.04 17 21 Aug 1 p auto 15:43 70 4 1355 0.4000 0.2669 0.2841 17:09 0.23 0.48 1.83 0.049 62.89 18 22 Aug 1 p auto 10:34 70 10 1242 0.3314 0.2150 0.2217 12:24 0.15 0.31 1.65 0.06 76.95 19 22 Aug 1 p auto 13:13 70 5 992 0.3314 0.2150 0.2165 14:57 0.11 1.14 7.17 0.039 50.33 20 22 Aug 1 p auto 16:30 70 12 842 0.3314 0.2150 0.1950 18:22 0.085 1.66 13.24 0.03 38.21 21 23 Aug 1 p, er auto 11:10 70 8 1289 0.2198 0.1332 0.1284 13:40 0.069 1.88 21.26 0.03 50.41 22 23 Aug 1 p, er auto 13:53 70 8 1289 0.2198 0.1332 0.1477 19:50 0.046 0.84 14.03 0.065 83.2 23 23 Aug 1 p, er auto 20:40 68 8 1289 0.2198 0.1332 0.1336 03:54 0.038 0.7 14.14 0.072 93.75 25 24 Aug 1 p, er auto 10:54 70 12 842 0.1961 0.1164 0.1077 16:47 0.06 1.709 18.5 0.022 27.51

6.3.2 Sydjåkk Table 3: The table shows the result of Sydjåkk. E xp. D ate P eak spec.

S ampl ing method I nj ection time w 0 ml

M x m Q m 3 s − 1 Q c m 3 s − 1 Q p m 3 s − 1 P eak u m/s D m 2 s − 1 d m w k g P % 1 06 Aug 1 p man ual 11:40 50 2 1263 0 .1733 1 0 .0689 1 See Q 14:20 0.099* 1.02* 7.75* 0.012* 22.16* 3 09 Aug 1 p, er man ual 12:30 75 2 1263 0.5735 0.0459 See Q 10:50 0.054 0.32 3.04 0.032 37.79 8 17 Aug 1 p man ual 11:04 100 7 1312 0.8234 0.2203 0.2202 12:30 0.17 0.87 3.42 0.08 75.59 9 18 Aug m auto 11:22 150 4 1323 0.7516 0.2067 0.2132 12:55 0.17 0.34 1.43 0.16 93.58 10 18 Aug 1 p auto 14:56 80 4 1323 0.7516 0.2067 0.2014 16:23 0.18 0.33 1.3 0.16 93.58 11 19 Aug 1 p man ual 12:36 80 8 1242 0.5725 0.1213 0.1754 14:15 0.15 0.23 1.1 0.068 40.47 11 19 Aug 1 p auto 12:36 80 8 1242 0.5725 0.1395 0.1754 14:16 0.15 0.28 1.35 0.056 33.54 12 19 Aug 1 p, er auto 14:37 70 3 1257 0.5725 0.1213 0.1750 15:33 0.089 0.25 2.11 0.027 33.77 13 20 Aug m,er,s? auto 10:29 70 5 924 0.4611 0.1471 0.1573 12:49 0.074 0.21 1.91 0.017 21.34 14 20 Aug 1 p auto 14:20 70 4 1323 0.4611 0.1471 0.1590 15:58 0.16 0.3 1.33 0.041 52.21 15 20 Aug m auto 15:58 70 2 1230 0.4611 0.1471 0.1538 17:19 0.18 6.75 26.67 0.064 81.11 16 21 Aug 1 p, er auto 11:20 70 2 1263 0.4000 0.1331 0.1409 14:20 0.087 1.59 13.67 0.013 17.11 17 21 Aug 1 p auto 15:43 70 4 1323 0.4000 0.1331 0.1384 17:15 0.16 0.49 2.07 0.054 69.3 18 22 Aug 1 p auto 10:34 70 10 1312 0.3314 0.1164 0.1188 12:32 0.13 0.38 2.05 0.056 70.89 19 22 Aug 1 p auto 13:13 70 5 924 0.3314 0.1164 0.1170 14:54 0.09 1.76 12.34 0.032 40.91 20 22 Aug m, er auto 16:30 70 12 773 0.3314 0.1164 0.1108 18:27 0.065 0.78 7.32 0.04 51.15 21 23 Aug 1 p, er auto 11:10 70 8 1242 0.2198 0.0866 0.2013 11:14 0.039 1.18 23.8 0.031 39.3 22 23 Aug 1 p, er auto 13:53 70 8 1242 0.2198 0.0866 0.0929 20:04 0.03 0.91 23.43 0.034 43.1 23 23 Aug 1 p, er auto 20:40 68 8 1242 0.2198 0.0866 0.0900 23:04 0.056 1.083 14.86 0.061 77.73 24 24 Aug 1 p, er auto 10:54 70 7 1312 0.1961 0.0797 0.0784 13:07 0.12 1.2 7.3 0.078 99.82 25 24 Aug 1 p, er auto 14:15 70 12 773 0.1961 0.0797 0.0755 16:54 0.051 0.34 4.44 0.013 15.93

6.4 Breakthrough curve shapes

Since the appearance of a breakthrough curve may indicate dierent types of drainage systems, common features are used to describe the curves. These features can exist within the same breakthrough curve. Also the width of the curve matters. The features are:

• Plain, when the curve has a clear well-dened concentration peak.

• Multipeaks, when the curve has more than one distinct peak.

• Secondary peaks, when the curve has more than one peak separated from each other. It might even be as two curves.

• Erratic behaviour. A noisy pattern on the whole or parts of the curve, with many random small peaks on short time intervals. This pattern can also be seen as minor secondary peaks superimposed on the main peak.

• Tail, if the curve has a tail it means that some dye came through later than the main pulse.

6.5 Moulin groups

The moulins have been divided into two groups, M1 (moulin 2, 3 and 9) and M2 (moulin 4, 7 and 10) to make analysing easier. The moulins in each group are very close to each other and is probably connected to the subglacial drainage system in a similar way.

Moulin 8, 5 and 12 are not grouped with other moulins. The two groups and moulin 8 are located in the middle of the glacier's north and south margins while 5 and 12 are further south and east (see gure 9). Group M1 is located above the riegel, between 1340-1350 m elevation. Group M2 is located north of M1 less then 100 meters away. Moulin 8 is located less than 100 meter south from of M1. Moulin 5 and 12 are located below the riegel at 1290 and 1280 m elevation respectively. Number 1 and 11 were not used in any successful experiment and are excluded from this grouping.

In gure 20 the experiments are plotted together with the temperature and pre- cipitation. This gure is used in the following sections along with the features of the breakthrough curves, dispersivity and transit velocity for each moulin group.

6.5.1 Dispersivity and transit velocity

The dispersivity values between the groups of moulins is dierent in range but generally

the dispersivities increased after experiment number 11 (19 ^th of August). This is referred

to as the generalisation further on. The increase seems to be followed by a decrease occur-

ring at dierent times depending on the dierent groups. A summary of the dispersivity

results according to the moulin grouping can be seen in table 4, for details see the result

tables for each stream, in table 2 and 3 respectively. Concerning the transit velocity it

generally decreases between experiment 10 and 15 e.g. the 18 ^th - 20 ^th of August.

Figure 20: Temperature and precipitation data together with the experiments plotted on the air temperature when the dye was injected.

Table 4: Dispersivity and Transit velocity range for each stream.

Group M1 M2 8 5 12

Centerjåkk

d range (m) 4.56-18.26 0.044-4.57 1.35-21.26 6.1-7.17 13.24-18.5 u range (ms ⁻¹ ) 0.097-0.26 0.22-0.24 0.11-0.16 0.038-0.19 0.06-0.085 Sydjåkk

d range (m) 2.11-26.67 1.3-7.3 1.35-23.8 1.91-12.34 7.32-4.44

u range (ms ⁻¹ ) 0.097-0.18 0.18-0.22 0.074-0.09 0.03-0.15 0.051-0.65