Rebecca Harrison (2003)

8/12/2019 Rebecca Harrison (2003)

1/72

Meltwater Flow in a Snowpack,

Niwot Ridge and Berthoud Pass, Colorado

Rebecca Harrison

This thesis is submitted in part fulfilment of the requirements for the B.Sc degree in

Environmental Science at the University of Lancaster. January 2003


2/72

1

Abstract

Niwot Ridge (NR) and Berthoud Pass (BP) are both situated in the Eastern Front

Range of the Rocky Mountains, Colorado, USA. Distinguishing the presence of

meltwater pathways is an important research issue in these areas due to local towns andcities relying on snowmelt for a large fraction of their water supply throughout the

summer months. A long established technique in soil, dye tracing was applied to snow in

order to investigate whether or not meltwater flowpaths were present. Experiments were

carried out twice in the two different locations, NR and BP showing how different

snowpack properties affect the meltwater flow. Red food colouring was used as a dye,

added to natural snowmelt water and applied to the snow surface and then left to allow

infiltration to occur. After a short period of time snow was removed in thin one and twocentimetre sections and the resulting snow profile photographed. Stratigraphic and

temperature profiles of the snow were also identified with the use of snowpits. With the

help of the computer program MATLAB the centre of mass of dye in each section,

spread of dye around this point and the roughness of each section in terms of the

presence of dyed and non-dyed areas of snow were established. Comparisons were made

between images, statistical results and stratigraphic snow profiles to help understand the

processes involved. Preferential flowpaths were seen to exist mainly in a snowpack not

isothermal at 0 oC. When the snow became warmer and had a more uniform temperature

structure dye tended to move as one large plume through the snow spreading horizontally

with depth. High amounts of lateral flow were also observed when there was a change in

snowpack properties in the vertical direction. There did not appear to be a lot of

connectivity between flowpaths, vertical movement appeared to occur mainly when a

weak spot was found between layers of different properties, for example ice layers.


3/72

2

Contents

Abstract 1Contents 2List of Figures 4

List of Tables 5

1. Introduction 6- 1.1 Background 6- 1.2 Flow Processes and Hydrology of a Snowpack 7- 1.3 Dye Tracing 9

- 1.3.1 Dye Tracing in Soil 10- 1.3.2 Dye Tracing on Glaciers 11- 1.3.3 Dye Tracing in Snow 11- 1.3.4 Dyes 12

- 1.4 Research Aims 12- 1.5 Report Structure 13

2. Experimental methods 14 - 2.1 Research Team 14

- 2.2 Field Sites 14- 2.3 Dyes 16- 2.4 The Guillotine 17- 2.5 The Experiment 20- 2.6 Problems 21- 2.7 Statistical Analysis 22

3. Results 25- 3.1 Image Problems 25- 3.2 Fieldwork at Niwot Ridge 25- 3.3 Fieldwork at Berthoud Pass 29

- 3.3.1 BP1 29- 3.3.2 BP2 34

4. Analysis 39- 4.1 Centre of Mass 39

- 4.1.1 Centre of Mass for NR 40- 4.1.2 Centre of Mass for BP 1 42- 4.1.3 Centre of Mass for BP 2 44- 4.1.4 Comparison between Two Snowpacks Centre of Mass 47

- 4.2 Variance around the Centre of Mass 47- 4.2.1 Variance for NR 48- 4.2.2 Variance for BP 1 49- 4.2.3 Variance for BP 2 51- 4.2.4 Comparison between Two Snowpacks Isothermal at 0 oC 52

- 4.3 Roughness 53- 4.3.1 Roughness in all Experiments 53- 4.3.2 Roughness for NR 55- 4.3.3 Roughness for BP 1 56- 4.3.4 Roughness for BP 2 57


4/72

3

- 4.4 Summary 58

5. Conclusion 60- 5.1 Conclusions 60- 5.2 Suggestions for Further Work 60

References 62Acknowledgements 65Appendices 66

- Appendix A Code used in MATLAB for statistical analysis 66- Appendix B Raw data after statistical analysis 68


5/72

4

List of Figures

Figure 1.1 Movement of meltwater through snowpack 6Figure 1.2 Representation of channelled and water-film flow in snow 7Figure 1.3 Development of flow fingers 9Figure 1.4 Dye tracing example image 10

Figure 2.1 Location map 15Figure 2.2 Dye sprayed onto snow surface 16Figure 2.3 Mechanism of the guillotine 18Figure 2.4 Image of guillotine 19Figure 2.5 Using the guillotine 20Figure 2.6 Sequential removal of sections in snowpack 20Figure 2.7 Example of image from dye tracing experiments 23Figure 3.1 Image 2 from Niwot Ridge 26Figure 3.2 Image 12 from Niwot Ridge 27Figure 3.3 Image 28 from Niwot Ridge 27Figure 3.4 Image 44 from Niwot Ridge 28Figure 3.5 Image 48 from Niwot Ridge 28Figure 3.6 Image 94 from Niwot Ridge 29Figure 3.7 Image 1 from Berthoud Pass 1 31Figure 3.8 Image 11 from Berthoud Pass 1 31Figure 3.9 Image 20 from Berthoud Pass 1 32Figure 3.10 Image 23 from Berthoud Pass 1 32Figure 3.11 Image 27 from Berthoud Pass 1 33Figure 3.12 Image 40 from Berthoud Pass 1 33Figure 3.13 Image 1 from Berthoud Pass 2 35Figure 3.14 Image 21 from Berthoud Pass 2 35Figure 3.15 Image 25 from Berthoud Pass 2 36Figure 3.16 Image 38 from Berthoud Pass 2 36Figure 3.17 Image 60 from Berthoud Pass 2 37Figure 3.18 Image 78 from Berthoud Pass 2 37Figure 4.1 Centre of mass graph for three experiments 39Figure 4.2 Centre of mass in X an Y co-ordinate, May 9th 40Figure 4.3 Centre of mass as a function of distance through snowpack, May 9th 40Figure 4.4 Centre of mass in X an Y co-ordinate, May 18th 42Figure 4.5 Centre of mass as a function of distance through snowpack, May 18th 43Figure 4.6 Centre of mass in X an Y co-ordinate, June 1st 45Figure 4.7 Centre of mass as a function of distance through snowpack, June 1st 46

Figure 4.8 Variance in centre of mass, May 9th 48Figure 4.9 Variance in centre of mass, May 18th 50Figure 4.10 Variance in centre of mass, June 1st 51Figure 4.11 Schematic diagram of ice layers in the snowpack 52Figure 4.12 Roughness as a function of distance through snowpack 54Figure 4.13 Roughness in X and Y co-ordinates, May 9th 56Figure 4.14 Roughness in X and Y co-ordinates, May 18th 57Figure 4.15 Roughness in X and Y co-ordinates, June 1st 58Figure 4.16 Summary diagram of meltwater flow processes 59


6/72

5

List Of Tables

Table 1 Snowpit data for Niwot Ridge 26Table 2 Snowpit data for Berthoud Pass 1 30Table 3 Snowpit data for Berthoud Pass 2 34

Table 4 Centre of mass, variance and roughness data for NR 68Table 5 Centre of mass, variance and roughness data for BP1 69Table 6 Centre of mass, variance and roughness data for BP2 70


7/72

6

1. Introduction

1.1 Background

A reliable water source becomes increasingly important as the population rises inthe western United States. Alpine snowpacks provide a major source of annual

streamflow especially in spring contributing a large fraction of water to the hydrological

system (Michaels, 1985). Knowledge of meltwater routing through snow is important in

the determination of seasonal runoff from snowpacks and glaciers (Pfeffer and

Humphrey, 1996)

Initially, preferential water movement through snow is important for energy

transfer in the liquid phase, solute transport in wet snow and snowpack stability(Schneebeli, 1995). As the snow begins to melt physical movement of water becomes

more important, particularly release of meltwater at the base of a snowpack as in

Colorado the main source of water is from snowmelt (figure 1.1). The rate of release of

meltwater affects the river hydrograph and is especially important for the amount and

timing of the release of water from dams.

Figure 1.1 Movement of meltwater through all stages in the snowpack

The onset of snowmelt can vary by up to two months annually and it is therefore particularly important to understand the processes involved (Kattelmann and Dozier,


8/72

7

1999). Spatial distribution, solute transport, chemical movement, avalanche forecasting

and hydropower potential are also all important consequences of meltwater flow

(Schneebeli 1995; Williams et al., 1999; Boggild, 2000; Petersen et al., 2001). Despite

research there is still no clear understanding of preferential flow and we are unable to

simulate it (Schneebeli, 1995).

1.2 Flow Processes and Hydrology of a Snowpack

Melt is produced at the surface of the snow initially draining vertically and

developing into heterogeneous flow as vertical movement occurs. This is due to variable

hydraulic conductivity caused by different density and grain structures, further modified

by refreezing and changes in liquid water content (Pfeffer and Humphrey, 1996). Liquidwater movement is thought to occur in distinct flow paths through the snow as opposed

to uniform flow through a homogeneous porous medium (op. cit.). The distribution and

size of these preferential flowpaths depends upon the structure of the snowpack and local

weather conditions (Schneebeli, 1995). Percolating meltwater moves in two ways (figure

1.2): (i). 'Channelled water flow', where meltwater flows as a single body of water and

(ii). 'Water-film flow', where meltwater flows slowly surrounding snow grains in a thin

film of liquid water (Wakahama, 1974). Small-scale spatial variation and macroscopiclayering exist in snowpacks due to variations in the bulk density and grain structure

(Tseng et al., 1994).

Figure 1.2 Schematic representation of (i) channelled water flow and (ii) water- film flow in snow. Black dots and areas represent water moving through the snow.

Adapted from Wakahama, 1974.


9/72

8

Research has also been carried out into the weakening of the mechanical strength

and densification of snow relating to levels of meltwater within it (Wakahama, 1974). It

was found densification rate was double during the day compared to overnight and

suggested this might be due to a decrease in pressure from run-off of meltwater or

mechanical weakening of the snow during the warmer daytime temperatures. Over thecourse of the above research it was found that snow grains would continue to grow even

during the melt season, perhaps having an implication on meltwater flow. Schneebeli

(1995) also noted after an infiltration event snowpack properties may have changed

altering the location of flowpaths in the snowpack.

Tseng et al. (1994) have shown natural deposits of snow rarely remain isothermal

throughout their lifetime due to the wide range of physical processes occurring. Melting

and infiltration cause thinning of the snowpack throughout the melt season, varyinglevels of saturation and refreezing of meltwater both at the surface and within the snow

change the snows hydraulic and thermal properties.

The wetting front is the point at which snow changes from the wet top layer to a

dry bottom layer before the snow becomes isothermal. Macropores (open channels within

the snowpack) and flow fingers, lead to preferential flow of water through the snow

before the entire snowpack has become wet (Marsh and Woo, 1984; Schneebeli, 1995).

After the wetting front has reached the base of the snowpack water can infiltrate

throughout the whole mass of snow.

Kattelmann and Dozier (1999) noted that as the snowmelt season begins liquid

water enters the snowpack and it becomes 'ripe' at a rate and spatial variability influenced

by feedback mechanisms. During this ripening procedure water is retained by capillary

pressure in pore spaces at 0 o C. When pressure becomes equal between the upper and

lower layers of snow water is released and flows with gravity through the snow.

Observations of the snowpack show natural processes to include grain growth and

rounding, ice layer formation, warming of the snow to melting temperature,

densification, capillary retention of liquid water and creation of a water flow network.

After the onset of melt water is either retained by the snowpack or lost as runoff

depending on the temperature and physical structure of the snow (Pfeffer and Humphrey,

1996).


10/72

9

Figure 1.3 Movement of water through a snowpack showing the development of

flow fingers. Adapted from Marsh and Woo (1984).

Marsh and Woo (1984) showed that ponding occurs at stratigraphic boundaries

within a snowpack due to position of the wetting front and abrupt changes in snow

texture. Ponded water then begins to flow in a downhill direction and spread laterally

forming a wet layer, as water continues to accumulate vertical flow fingers begin to

develop (figure 1.3). Water may also spread sideways forming narrow strips in the snow

according to minor differences in snow properties.

Pfeffer and Humphrey (1998) noted the formation of ice layers at stratigraphic

boundaries in a snowpack when the rate of freezing is greater than the force of water

accumulating and pushing across the boundary. Following the formation of this ice layer,

water drains onto the impermeable boundary and is diverted down slope or laterally

within the snowpack until an area is found where water is able to flow through the

boundary. As a body of ice is formed between the wet and dry layers of snow, latent heat

is released internally, warming the surrounding snow (Marsh and Woo, 1984). In the wet

portion of the snowpack ice decays as this latent heat is released and further melting canoccur. Unusually high rates of water input, infiltration and refreezing causing the

formation of ice layers indicates unusually cold initial conditions or early onset of melt

(Pfeffer and Humphrey, 1998).

1.3 Dye Tracing

Quantitative determination and prediction of preferential flowpaths is difficultdue to spatial and temporal variability involved in different flow processes (Petersen et


11/72

10

al., 1997). In the past dye tracing experiments have been undertaken to observe primary

flow paths of water and solutes at a high spatial resolution (figure 1.4; Petersen et al.,

2001). Use of this dye tracing technique has enabled researchers to carry out detailed

studies of preferential flow through soil and glaciers. Existence of preferential meltwater

flowpaths in snow are generally recognised, however only limited quantitativeinvestigations have been carried out (Schneebeli, 1995). Understanding flow in different

mediums is important for water velocity and storage within the medium, flow and

transport of water and solutes, for example pesticides, through the medium and release of

water and solutes to the surrounding environment. Within snow, formation of preferential

flowpaths and meltwater movement, including macropore flow and water retention

becomes important for developing models of snowmelt run-off.

Figure 1.4 Example of a dye tracing experiment image taken in soil (Baveye et al., 1998)

1.3.1 Dye tracing in Soil

Dye-tracing work has been undertaken in soil to help improve our understanding

of soil drainage systems for irrigation, chemical leaching, water and nutrient availabilityfor plants and flood potential. The majority of research has concentrated on the use of


12/72

11

coloured and fluorescent dyes showing preferential flow pathways and transport of water

and solutes through cracks, fissures, fingers, root systems and earthworm burrows in the

soil profile (Ghodrati and Drury, 1990; Petersen et al., 1997). Ghodrati and Drury (1990)

used a dye tracing experiment to characterise water and solute transport in the matrix of a

soil system in three dimensions. The study showed the presence of preferential flowchannels with water moving both laterally and vertically under different soil conditions.

Other soil related studies have investigated water infiltration into different types

of soil for example clay, sand and agricultural soils (Ritsema et al., 1993; Heppell et al.,

2000; Petersen et al., 2001). Soil has also been studied under different environmental

conditions, through dye tracing, including forested regions, frozen ground and hillslopes

(Luxmoore et al., 1990; Stadler et al., 2000).

1.3.2 Dye Tracing on Glaciers

Recently dye-tracing experiments have been used for research into glacial

meltwater. Niewnow et al., (1996; 1998) studied the Haut glacier d'Arolla, Switzerland

by carrying out 415 dye-tracer experiments during 1990 and 1991. The aim of the study

was to determine spatial patterns of meltwater flow and evolution of the drainage system

throughout the summer months. Analysis of the results showed specific characteristics of

the englacial and subglacial drainage systems including mean flow velocity, dispersion

coefficient and cross sectional area of flow. Channelled or distributed flow regime and

the response of seasonal run-off changes through these 2 regimes were also investigated

using dye-tracing experiments.

Subglacial drainage characteristics have also been studied at Dokriani glacier,

India through dye tracing. Dispersion, meltwater velocity, passage geometry and

channelled or distributed flow regimes were all investigated concluding flow is

dependant on the level of meltwater present (Hasnain et al., 2001).

1.3.3 Dye Tracing in Snow

The study of snow is a well-established science, however there appears to have

been few experiments carried out using the dye tracing technique. Comparable with soil

different snowpacks have different characteristics, which can be investigated through dye

tracing. Evidence from experiments so far suggests there is no distinct pattern of flow


13/72

12

paths and preferential flow depends mainly upon the boundary conditions (Schneebeli,

1995). Schneebeli (1995) carried out dye tracing experiments using four different

coloured dyes to investigate the development and stability of flowpaths. Observations

were made of the existing flowpaths concluding that they do not remain stable in time or

space. Boggild (1999) has also investigated meltwater flow and water retention insnowpacks concentrating research on snow in West Greenland. Ink and water were

mixed to create a dye and sprinkled on the snow surface, ten centimetre sections were

then removed and photographed to create a three dimensional profile of the flow paths.

1.3.4 Dyes

The dye used in these experiments varies depending on the characteristics of themedium on which the dye tracing is performed. Within soil a variety of dyes have been

used including Acid-Red 1, Dispersed-Orange 3, Rhodamine B and Brilliant Blue FCF

(Ghodrati and Drury, 1990; Luxmoore et al., 1990; Petersen et al., 1997; Stadler et al.,

2000; Petersen et al., 2001;). Specific studies have been carried out in order to test

Rhodamine wt, Lissamine FF, Amino G acid and Brilliant Blue FCF as dye tracers for

their toxicity, mobility, absorption and fluorescence in the soil environment (Trudgill,

1987; Flury and Fluhler, 1995). Other tracers are also used in soil such as KBr (Ritsema

et al., 1993).

During the summers of 1989, 1990 and 1991 Rhodamine -B and Flourescine dyes

were used on Haut glacier d'Arolla, Switzerland and in 2000 Rhodamine - wt was used

on Dokriani glacier, India during a series of glacial dye tracing experiments (Niewnow et

al., 1996; Niewnow et al., 1998; Hasnain et al., 2001). There doesnt appear to be any

specific research published on dye tracers in snow, however, some tracers used in the

past are Brilliant Blue FCF, azofloxine, uranine and Lissamine yellow (Schneebeli,

1995).

1.4 Research Aims

This series of experiments attempts to help understand melt water movement

through snow using the technique of dye tracing. Spatial representation of snowmelt

processes is a research problem yet to be solved and few studies so far have dealt with

the problem of spatially distributed snowmelt models (Horne and Kavvas, 1997). A


14/72

13

series of dye tracing experiments were designed to try and understand water movement

through snow. The main aim of the study is to better physically understand this water

movement enabling improvement of snowmelt models in the future.

Dye tracing involves mixing a selected dye with natural snowmelt water from the

snowpack and allowing it to infiltrate through the snow over a period of approximatelytwo hours. After this time a cutting device was placed above the dyed area of snow and

used to cut away thin sections of snow to expose the infiltrated dye. A sequential series

of digital images were then taken and used for studying general characteristics of flow in

order to establish relevant patterns. Using the computer program MATLAB the centre of

mass was identified to determine specific flow areas and the variance around the centre

of mass in a particular section compared to other sections within the same snowpack.

Comparisons were also made between different snowpacks.This project has benefited from the involvement in an active research program in

the USA. Members of INSTAAR (Institute of Arctic and Alpine Research) are currently

undertaking several research projects at the University of Colorado into snow and

hydrology. Daily and weekly measurements are taken at the INSTAAR Mountain

Research Station including snowmelt runoff level, snow properties (for example depth,

density and temperature) and meteorological data. Annual data has also been collected

and compiled by this research team since 1994. The specific experiments were carried

out with Tyler Erickson, a PhD student at the University of Colorado and part of the

INSTAAR snow hydrology research group.

1.5 Report Structure

The report initially covers background to the project, why it was carried out and

what previous work has been done in soil, glaciers and snow to initiate this particular line

of research. Chapter two then covers the experimental methods describing field sites,

equipment, the experiments and problems encountered during the experiment and their

solutions. Chapter three shows all the results obtained during the experiment leading to

an analysis and discussion of the results in chapter 4. Finally the work is summarised and

concluded ending with suggestions for further research work.


15/72

14

2. Experimental M ethods

2.1 Research Team

The overall project aims to gain a better understanding of spatial continuity and

connectivity on varying scales in a snowpack using radar tomography, dye tracing and

lysimeters. Research support came from a grant by NSF (National Science Foundation)

the research being organised and run by Mark Williams, a snow hydrologist at the

University of Colorado. Other members of the research team include Tyler Erickson, a

PhD student with M. Williams leading fieldwork at the research sites and carrying out

geostatistical analyses. Tissa Illangasekare is a groundwater hydrologist from the

Colorado School of Mines working on model development and parameter estimation.Tad Pffefer, a glaciologist at the University of Colorado will work on radar tomography

instrumentation and analysis in the future on this project. Similar work is also being

undertaken at the Mammoth Mountain Ski Area (MMSA) site in California directed and

run by Rick Kattelmann for comparative data collection and analysis. It is also expected

that other graduate and undergraduate students in the future will work in the research

areas mentioned in collaboration with this project.

2.2 Field Sites

Two field sites were chosen for this experiment to enable the study of two

snowpacks with differing characteristics (figure2.1).


16/72

15

~30km

Figure 2.1 Small area of Colorado map to show research site location

Initially experiments were carried out at the University of Colorado MountainResearch Station, Niwot Ridge (NR), Eastern slope of the Colorado Front Range of the

Rocky Mountains, 5km east of Continental Divide (Williams et al., 1999). The 'Soddie'

site used in this experiment is located at an elevation of 3150 m, below treeline and

surrounded by a forest preventing lateral inflow of meltwater. Snowmelt occurs relatively

fast at NR, by the time the first experiment was carried out the snowpack was isothermal

at 0 o C.

Experiments were also carried out at Berthoud Pass (BP) ski area at an elevation

of 3482m and a slope angle of 18 o, this site is also surrounded by a forested area below

treeline. Snowdrifts at BP were relatively deep at approximately 1.4m depth compared to

those at NR at approximately 90cm depth. During the first experiment (BP1) on 18th

May the snow was not isothermal at 0 o C, by June 1st (BP2) the snowpack had become

isothermal at 0 oC.

These two field areas were chosen due to easy access and undisturbed sites

consisting of several deep snowdrifts within easy travelling distance of Boulder and each

other. Both areas contain snowpacks with different characteristics mainly due to the


17/72

16

climate and elevation. However, the setting of the specific site within the trees helps to

control other factors such as sunlight, wind and inflow of meltwater. The two field sites

do not retain snow all year round as in other US states and countries, across the Colorado

Rockies complete melting of snow occurs during the spring and summer months.

2.3 Dyes

A dye tracing technique was used during this series of experiments as it was

relatively inexpensive, quick and easy to set up at different field sites without the need

for large amounts of equipment. Food colouring mixed with natural snowmelt water was

used as a tracer during this experiment. The choice of dye is very important for any study

of this kind, as several factors could influence the results. The selected dye must havelow toxicity while maintaining high solubility and visibility. The dye used was red food

colouring as it is non-toxic, inexpensive and easy to obtain with high visibility and

solubility in snow. The ingredients in the dye were water, propylene glycol and red 40.

Other dyes were not tested, however features represented in the snow may have proved to

be different due to different absorption properties.

Figure 2.2 Dye sprayed onto snowpack across a two by two metre area


18/72

17

30 ml of red food dye were mixed with one (US) gallon (3.78 l) of natural snow

meltwater and spread over an undisturbed two by two metre surface using a weed sprayer

and allowed to infiltrate into the snow (figure 2.2).

2.4 The Guillotine

A cutting device was used during the experiment to produce accurately spaced

and clean sections through the snow pack while studying the flowpaths. The 'guillotine'

(figure 2.3) used was designed and built by Tyler specifically to run the dye tracing

experiments accurately, avoiding the potential hazard of smearing dye on the snow

surface and in order to gain clean cuts through the snow. No one was allowed to disturb

the area to be sprayed with dye before the dye was applied, after application the dyedarea was left completely undisturbed for approximately two hours. After this time, the

guillotine was set up across the dyed area and secured firmly to ensure there was no

movement, which could affect the snow surface, dye infiltration or inaccurate

measurements when cutting the slices.

The guillotine consists of a tubular plastic frame lying on the snow surface

supporting the rest of the structure. Two hollow metal tubes are placed around this plastic

frame and allow forward and backward movement of the guillotine across the snow.

Above these metal tubes is the framework holding a large rectangular metal frame with a

1-cm wide blade at the base in order to cut through the snow. A small snow pit was dug

directly in front of the dyed area and the guillotine used initially to gain a clean section

until the dyed area of snow was reached. The snow was cut away using the guillotine in

one or two cm sections to reveal the flowpaths over an area of approximately 1m 2 (figure

2.4). Initially 2 cm sections of snow were removed however ice layers proved to be a

problem and it was found that 1 cm slices were easier, more accurate and provided a

smoother section to be photographed.


19/72

18

Figure 2.3 Mechanism of the guillotine from front and side views.


20/72

19

Figure 2.4 Image showing guillotine device at NR.

Three people are required to operate the guillotine successfully, two people to

pushed the guillotine from the top (figure 2.5) and moved it through the snow as new

sections are required. The third person stood in the snow pit to verify that straight

sections were cut and to operate the digital camera. The camera was fixed relative to the

position of the guillotine frame and the timer used to ensure identical image size and

resolution.


21/72

20

Figure 2.5 Using the guillotine to cut a section from the snowpack at NR.

2.5 The Experiment

The experiment was carried out on four separate occasions on May 9th and May

15th at NR and May 18th and June 1st at BP (figure 2.1). Sections were removed every 1

or 2 cm by pushing the guillotine back after each cut, an image taken using a digital

camera and the images sequentially analysed (figure 2.6). At BP the slope was

considerably steeper than at NR so the guillotine had to be firmly anchored before it

could be used. Dye was sprayed slightly uphill of the area to be sampled since the dye

was expected to move down hill as well as percolating vertically through the snow this proved to be the case.

Figure 2.6 Inferred snowpack showing how sequential sections are removed

throughout experiment


22/72


23/72

22

2.7 Statistical analysis

All the images taken in the field were transferred to a computer for analysis.

Firstly the blurred and completely out of focus images were discarded and the rest

corrected using the sharpen tool with a graphics package. The images were then all cut

out so the same area appeared on every image, this was done by using the guillotine

frame as a reference point and cutting out the snow area seen on every image. Resizing of

all the images then took place to ensure identical pixel sizes the images were 666 by 654

pixels, a pixel being 0.11 by 0.16cm. An example of an image cut out ready to be

analysed is shown in figure 2.7.

Figure 2.7 Example of an image ready for statistical analysis

Spatial moment analysis was then carried out on each of the images for NR, BP1

and BP2 using the computer program MATLAB (programming details can be seen in

appendix A). Each pixel has a value in the red, green and blue spectrum between 0 and

255, where 0 is white and 255 is pure red for example, these are known as digital


24/72

23

numbers and were the values used during analysis. Figure 2.7 shows the point of the

origin used for each image during the analysis. Two comparative pixels are also

highlighted to show how dye concentration varies between adjacent pixels. The two

pixels highlighted are at co-ordinates (i,j) 624,320 and 625,320 with digital number

values in the red of 216 and 187, the higher value indicating a higher concentration of dye in that particular pixel.

The first moment involves looking at the centre of mass in each image to

determine where the main mass of flow is in the snowpack and whether this varies

through the snowpack using:

X = (X ij P ij) (1) P ij

Where X is mean centre of mass in x co-ordinate, X ij is the distance from the origin in the

x (i) and y (j) co-ordinates and P ij is the pixel value at the X ij co-ordinate in the pixel

value array. The first moment is calculated in the same way for the y co-ordinate:

Y = (Y ij P ij) (2) P ij

Where Y is the mean centre of mass in y co-ordinate, Y ij is the distance from the origin inthe x (i) and y (j) co-ordinates.

The second moment is based upon variance around the centre of mass using the

equations:

x = (Xij - X) 2 P ij (3) P ij

y = (Yij - Y)2

P ij (4) P ij

The variance attempts to illustrate the spread of dye around the centre of mass, a high

value for the variance indicating a wide spread around the centre of mass.

After analysis using the first and second moments an assessment of the roughness

of each section was carried out in order to investigate the existence of specific flow areas.

The roughness is defined as:


25/72

24

R x = (P ij - P i+1j)2 (5) n

R y = (P ij - P i+1j)2 (6) n

where R x is the roughness value in x, R y is the roughness value in y, P ij is the pixel value

to the left of pixel P i+1j and n is the number of pixels. All the results were then plotted in

various different ways and the results compared and analysed.


26/72

25

3. Results

3.1 Image Problems

Certain problems with the images were identified and solutions found. The main

problem found during the fieldwork was that the sun caused a low contrast ratio on someimages reducing the visibility of the flowpaths. A solution to this might have been to

carry out this tracing experiment at night (Schneebeli, 1995) however under the climatic

conditions of Colorado the temperature dropped at this elevation so therefore melt would

not occur as the snowpack refreezes. Cloud also proved to be a problem as scattered

clouds on the research days forced some photographs to be taken during sunny periods

and others during a cloudy spell again altering the contrast ratio within the image.

Computer manipulation of the images helped to remove some of the contrast ratio problems and improved visibility of the flowpaths for statistical analysis.

After the fieldwork was complete some of the images appeared to be blurred as

the automatic focus mechanism on the camera did not appear to work correctly. To

correct for this the camera had to be checked after every photo and the blurred images

sharpened using a computer. Some of the images were not correctable and another

experiment would need to be carried out under the same conditions in order to gain a full

data set for analysis.

3.2 Fieldwork on Niwot Ridge (NR)

The dye tracing experiment was carried out at NR on a cold but sunny day with

an air temperature of 3 oC and snow surface temperature of 0 oC with a few clouds and

slow wind speeds. Unfortunately snowpit data for this experiment is unavailable however

the snow had similar properties on May 15th and therefore the data for this day is shown

in table 1.

48 images were taken and used for statistical analysis at NR, a small sample of

these are seen in figures 3.1 to 3.6. Each image has a number associated with it indicating

the distance from the start of the cutting in centimetres. Each section was cut 2 cm

behind the last so for example image 12 (figure 3.2) was 12 cm behind the first cut and

the 6th image taken in the sequence.


27/72

26

Table 1 Snowpit data for NR.

Date 15/05/02Location Niwot

Ridge

Temp Net Weight Height above Height above grain(C) (g) ground (cm) ground (cm) shape

183 182- wet snow, slush

180 1770 513 IC discontinuous ice

170 1740 489 ET

160 1680 533 IC distant layer of ice

150 1670 493 ET

140 1510 506 IC ice layers

130 1500 550 ET

120 1180 553 IC very thin ice layer

110 1170 542 ET

100 700 530

900 540

80

Figure 3.1 Image 2 from Niwot Ridge on May 9th 2002. Note the slightly dipping layersin the snowpack forcing the dye to move laterally rather than vertically.


28/72

27

Figure 3.2 Image 12 from Niwot Ridge on May 9th 2002. The stripe down the centre of the image corresponded to a small piece of ice falling out of the snowpack.

Figure 3.3 Image 28 from Niwot Ridge on May 9th 2002. Image is slightly blurred however this should not affect results as statistics concentrates on contrast between dyed

snow and non-dyed snow.


29/72

28

Figure 3.4 Image 44 from Niwot Ridge on May 9th 2002.

Figure 3.5 Image 48 from Niwot Ridge on May 9th 2002.


30/72

29

Figure 3.6 Image 94 (final image) from Niwot Ridge on May 9th 2002. Note theincreased concentration of dye lower in the snowpack comparative to figure 3.1.

3.3 Fieldwork at Berthoud Pass

3.3.1 BP1

Fieldwork was carried out for two days at BP. A snowpit record was also created

and particular properties identified including temperature, net weight, stratifying layers

and grain shape (table 2). Abbreviations used in the snowpit sampling of grains include

ET (equi-temperature), IC (ice), TG (depth hoar) and CR (crust). At an elevation of

3482m, the air temperature was 8 oC with an identical snow surface temperature of 0 oC.

Wind speed was again low and low cloud cover observed although the experiment was

carried out in a tree covered area so changes in sunlight would not greatly affect the

experiment. The guillotine was pushed down to a depth of one metre, 60cm above the

snowpack base creating sections 1cm apart.


31/72

30

Table 2 Snowpit data for BP1.Date: 18/05/02 12:50

Location: Berthoud Pass1

Local Slope: 18 DegreesAspect: 343 degrees from

north

GPS N: 4405646 MetersGPS E: 433684 Meters

Temp Net Weight Height above Height above Grain(C) (g) ground (cm) Ground (cm) Shape

0 160 1610 454 ET well rounded, wet

150 153-1 484 IC Bonded grains

140 150-1 426 ET well bonded, rounded, wet

130 1450 506 IC

120 143-1 464 ET well bonded

110 136-1 438 ET well rounded, bounded

100 1320 422 ET Slushy

90 1220 391 ET well banded clusters

80 1180 442 ET well banded, rounded

70 1070 455 ET well rounded

60 420 464 ET Rounded facets

50 240 418 CR

40 170 419 ET Rounded angular facets

30 120 420 ET Rounded angular facets

20 00 376

100 376

00

During the BP1 experiment 37 images were taken, however due to focusing

problems with the camera images between 11 and 20cm were not analysed. Figures 3.7

to 3.12 show six of the images used in statistical analysis.


32/72

31

Figure 3.7 First Image from Berthoud Pass on May 18th 2002. Note the well-defined areas of dye towards the top of the section and the lateral flow across layers within the

snowpack.

Figure 3.8 Image 11 from Berthoud Pass on May 18th 2002. Specific plumes of

meltwater movement are observed on the left of the image and an anomalous patch of dye seen towards the right of the section.


33/72

32

Figure 3.9 Image 20 from Berthoud Pass on May 18th 2002.

Figure 3.10 Image 23 from Berthoud Pass on May 18th 2002. Higher concentrations of

dye are observed compared to figure 3.8 also more layering is seen towards the top of the section.


34/72

33

Figure 3.11 Image 27 from Berthoud Pass on May 18th 2002.

Figure 3.12 Image 40 (final image) from Berthoud Pass on May 18th 2002.


35/72

34

3.3.2 BP2

BP2 experimental images consisted of 78 sections split 1cm apart in the

snowpack. The weather was cooler during this experiment relative to BP1 although the

temperature was the same as at NR at 3o

C although under cloudy skies on this occasion.During the time between the two experiments snow properties changed from a varied

temperature stratification to a snowpack isothermal at 0 oC as seen from snowpit data for

BP2 in table 3. Snow depth at BP2 was 14cm lower than at BP1, the guillotine was used

to 1m depth in the snow, 46cm above the snow base.

Table 3 Snowpit data from BP2. Note the isothermal at 0 oC temperature structure of the snowpack

Date: 01/06/02Location: Berthoud Pass

2Local Slope: 18 Degrees

Aspect: 343 degrees from northGPS N: 4405646 MetersGPS E: 433684 Meters

Temp Net Weight snowpack height Layer height grain shape(C) (cm) above ground (cm) Above ground (cm)

146 146ET

140 141

0 429 IC130 140

0 479 ET120 138

0 441 IC110 135

0 434 ET100 132

0 411 ET90 117

0 403 IC80 114

0 471 ET70 0

0 40960

0 46250

0 41340

0 46230

0 42420

0 445 100


36/72

35

A selection of the images taken at BP2 are observed in figures 3.13 to 3.18. Note

the increase in concentration and spread of dye in the snowpack in later images.

Figure 3.13 First Image from Berthoud Pass (i.e. 0cm from start) on June 1st 2002. Notetwo layers dominating in the centre of the sections and the high dye concentration

between them.

Figure 3.14 Image 21 from Berthoud Pass on June 1st 2002.


37/72

36




38/72

37


Figure 3.18 Image 78 (final image) from Berthoud Pass on June 1st 2002. Much wider spread of dye was observed in this image compared to figures 3.13 to 3.17.


39/72

38

After all the data was collected the images were cut out and adjusted to the same

resolution as seen in figures 3.1 to 3.18. All the images then underwent statistical

analysis as described in chapter two. Results of the statistical analysis are seen and

analysed in chapter 4 and plotted to enable comparisons through individual snowpacks,

between the two different snowpacks and between the same snowpack at different times.


40/72

39

4. Analysis

4.1 Centre of Mass

Figure 4.1 shows the first moment, centre of mass in each section through the

snowpack during the three experiments: NR, BP1 and BP2. Each data point on the graphindicates one sections centre of mass (Raw data for these analyses is seen in appendix B).

0.000

10.000

20.000

30.000

40.000

50.000

60.000

70.000

80.000

90.000

100.000

0.000 10.000 20.000 30.000 40.000 50.000 60.000 70.000 80.000 90.000

Centre of Mass in X coordinate (cm)

C e n

t r e o

f M a s s

i n Y c o o r d

i n a

t e ( c m

)

BP2

BP1

NR

Figure 4.1 Centre of mass in x and y co-ordinates for all three experiments on May 9th at Niwot Ridge (NR), May 18th (BP1) and June 1st (BP2) at Berthoud Pass. Axis indicate

size of each section seen in cm.


41/72

40

4.1.1 Centre of Mass for Niwot Ridge

30.0035.00

40.00

45.00

50.00

55.00

60.00

65.00

70.00

30.00 32.00 34.00 36.00 38.00 40.00 42.00

Centre of Mass in X (cm)

C e n

t r e o

f M a s s

i n Y ( c m

)

Figure 4.2 Co-ordinates for the centre of mass in x and y according to the dyeconcentration, variation in centre of mass was shown by concentrating on specific area

of graph (figure 4.1). Niwot Ridge, May 9th 2002

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

0 20 40 60 80 100

Sections through snowpack (cm)

C e n

t r e o

f M a s s

( c m

XY

Figure 4.3 Centre of mass in x and y according to the dye concentration. Axis showcentre of mass varying as a function of distance from the start i.e. number of sections cut

(from 0 cm). Niwot Ridge, May 9th 2002

Figures 4.2 and 4.3 were compared with the images taken at NR in order to

establish how the centre of mass varies through the snowpack statistically and visually.The images initially showed a concentration of dye towards the top of the section with


42/72

41

two main layers slightly tilted towards the right (figure 3.1). A definite vertical

movement of the dye was seen in the snowpack, in which there appeared to be a fairly

uniform spread in the horizontal axis. As the centre of mass decreased in x there was a

slight increase in y indicating a vertical rise associated with the shift in the centre of mass

to the left. In sections about 12cm from the start of the sample area the movement of dyeappeared to trend slightly towards the left of the section seen in figure 4.3 as centre of

mass increased in the horizontal direction. 28 centimetres from the beginning of the cut

area a slight drop was seen in the centre of mass in the x and y directions. This was

difficult to distinguish on the images however the concentration of dye appeared to be

slightly higher in image 28 and therefore may slightly affect the results (figure 3.3).

Another rise in both x and y co-ordinates was seen at roughly 48cm, therefore shifting

the centre of mass vertically upwards and horizontally to the right (figure 3.5). Nosignificant movement was seen on the images however dark images causing a low

contrast ratio might have proved to be the problem (e.g. figure 3.4).

Figure 4.3 shows a general upwards trend in the y co-ordinate indicating the

centre of mass moving vertically from roughly 40 cm above the ground to 60 cm above

the ground. This was not obvious in the images as it was thought there was a gradual

drop in the main area of dye throughout the snowpack (figures 3.1 to 3.6). This drop may

be due to the extra time required to cut earlier sections of snow away and therefore

allowing further infiltration of the dye through the snow. However higher concentration

of dye retained higher in the snowpack may have caused this rise in the centre of mass

not easily observed. As reflected in figure 4.3 there was a fairly uniform distribution of

dye in the x co-ordinate.

Later on in the sample area there was a plume of dye concentrated about halfway

down the section in the centre of the horizontal axis reflecting the average point for the

centre of mass seen in figures 4.1 and 4.2. Plumes of dye were apparent on the images

but seen towards the left and right of the sections horizontally resulting in a centre of

mass towards the middle although observed to be slightly to the left of the centre. There

appeared to be a strong negative correlation between the centre of mass in the x and y co-

ordinates as seen in figure 4.2, as the concentration of dye in x moves horizontally to the

right, vertically the centre of mass falls. Observations of the images show this was the

case, as dye concentration appeared to move vertically downwards with time and shift

towards the right of each section.


43/72

42

4.1.2 Centre of Mass for Berthoud Pass 1

15.0020.00

25.00

30.00

35.00

40.00

45.00

50.00

36.00 37.00 38.00 39.00 40.00


C e n

t r e o

f M a s s

i n Y c o o r d

i n a

t

( c m

)


of graph (figure 4.1). Berthoud Pass, May 18th 2002

Figure 4.1 appeared to show a wide spread in the centre of mass for BP1 that was

also seen in the experiments carried out on NR (figure 4.1). When compared to the

images early in the snowpack a concentration of dye was seen towards the top of the

sections where there were three main layers observed and highlighted by dye (figure 3.7).

Towards the top of the snowpack movement appeared to be concentrated towards the

right of the section, as you move vertically downwards the centre of mass remained

concentrated to the right. This is shown in figure 4.4 where there was a slight negative

correlation between the x and y co-ordinates as the centre of mass in y moves upwards

the centre of mass in x shifts to the left. The pattern is confirmed for the horizontal in

figure 4.5 as in the x co-ordinate there was a constant level for the centre of mass. This

was not reflected in the y co-ordinate however and there does not appear to be any

pattern in the centre of mass through the snowpack.


44/72

43

0.00

5.00

10.00

15.00

20.00

25.0030.00

35.00

40.00

45.00

50.00

0 5 10 15 20 25 30 35 40


C e n

t r e o

f M a s s

( c m

XY


(from 0 cm). Berthoud Pass, May 18th 2002

Unfortunately fewer sections were made during BP1 experiment compared to the

other two experimental days reflected in figures 4.4 and 4.5 due to problems with the

camera. Useful results were still obtained, only across a shorter sample area of snow.

Observed on the images small patches of red dye appear lower than the main area of flow, which may shift the centre of mass vertically but not horizontally due to the

occurrence of these patches on both the left and right of the section. The main patch

appeared to be on the right side of the images and became more concentrated as more

sections were removed (e.g. figures 3.8 and 3.11). There was no evidence of a flow path

for this movement seen during any part of the experiment. Explanations for this anomaly

may be an area of flow around the side of the section seen due to an ice layer for example

or due to fast vertical movement in which there was no adhesion to the snow above the

point at which the dye was seen. Concentration of dye at this anomalous point may be

due to a change in snow properties reducing or stopping the flow of water through the

snow.

As more sections of snow were removed flow fingers were noted forming in the

snow. This may have been due to some areas of snow where water was retained while in

other areas flow fingers may have developed due to differences in snowpack properties.

As seen in figure 4.5 there were a few results missing between 12 and 20 cm, this

was due to out of focus images being taken. Between the missing data points a change in


45/72

44

snow properties was noted with more layering seen in the snowpack after 20cm and the

dye concentrated towards the top of these sections relative to the lower concentration

observed in the first set of sections (figure 3.8 compared to figure 3.9). Two low centres

of mass were noted after 20cm at 22 and 23cm from the start as seen in 4.5. However this

was not easily seen when comparing images and was evidently due to a difference in theway the dye was concentrated with respect to the snow (figure 3.10).

Towards the end of the sampled snow area the dye appeared to be concentrated at

the top of the snow sections with a few lower patches of dye seen but no obvious

pathways. In this area of snow the centre of mass appeared to rise in the vertical direction

unusually fast, when compared to the images dye was seen in high concentrations

towards the top of each section reflecting the retention of meltwater held in layers

towards the top of the snowpack (figure 3.11). Snowpit data showed that towards the topof the snowpack at BP1 the snow was at a temperature of -1 oC at a depth of 7 to 15cm

and 17 to 28cm (table 2) explaining the retained meltwater.

4.1.3 Centre of Mass for Berthoud Pass 2

The centre of mass at BP2 (figure 4.6) was seen trending towards the right of the

section between 30 and 40cm in the horizontal scale compared to BP1 where the centre

of mass was trending towards the left of this area (figure 4.4). Variation in the centre of

mass appeared to be fairly small and there was a definite concentration towards one

specific area of the snow. The pattern seen in the images was two main layers roughly

halfway down each section concentrating the dye and evidently retaining meltwater at

this point. Small plumes of dye were observed below these layers moving meltwater

vertically through the snowpack (figure 3.13). Moving through the snow with the

removal of more sections meltwater concentration appeared to be between the two layers

mentioned above however, there was development of plumes below these layers not

altering the centre of mass as they were spread fairly evenly across the section.


46/72

45

48.000

49.000

50.00051.000

52.000

53.000

54.000

55.000

56.000

35.000 36.000 37.000 38.000 39.000 40.000


C e n

t r e o

f M a s s

i n Y c o o r d

i n a

t e ( c m


of graph (figure 4.1). Berthoud Pass, June 1st 2002

Experiments carried out at BP2 showed the smoothest pattern in terms of

movement of the centre of mass through different sections in the snowpack (figure 4.7).

In both the x and y co-ordinate there was nearly a straight line with a slight rise towards

the end of the sample area showing the increased height of the centre of mass. Vertical

differences in the concentration of the dye was seen in the images with dye highlighting

the presence of layers in the upper parts of the section between 21 and 29cm (figures

3.14 and 3.15). This was not reflected in figure 4.7 as there were vertical plumes also

seen towards both the left and right of the section towards the snowpack base therefore

changes in the horizontal centre of mass were small with a slight change in the vertical

centre of mass.


47/72

46

0.000

10.000

20.000

30.000

40.000

50.000

60.000

0 20 40 60 80


C e n

t r e o

f m a s s ( c m

Max mass in XMax mass in Y


(from 0 cm). Berthoud Pass, June 1st 2002

Further on in the snowpack a more uniform spread of dye was seen across each

section increasing towards the surface. This was surprising as it was expected dye would

continue to infiltrate through the snow and a downward trend seen in the centre of mass

as more vertical infiltration occurs. Figure 4.1 showed BP2 having the smallest change inthe centre of mass, over the course of the experiment this centre of mass appeared to

remain at the same point skewed slightly left horizontally but remaining in the centre

vertically. In comparison to the images this vertical distribution was due to the two layers

in the centre of the snowpack retaining the meltwater for a relatively long period of time

(figure 3.13). Early in the experiment from observations of the images it seemed the

centre of mass would be towards the right of the sections, however moving through the

snowpack leads to a bias towards the left of the sections especially with the formation of

the meltwater plumes towards the bottom left of the section.

Towards the end of the sample area, centre of mass in the vertical moved upwards

and in the horizontal moved to the right of the section, however more snow was seen in

the sections compared to earlier relative to the concentration of dye. The movement of

the centre of mass noted in figure 4.7 from 60cm through the section to the end at 78cm

was also seen in the images (figures 3.17 and 3.18). There was continued development of

plumes in the right and centre of the section although movement was also noted towards

the left of the sections.


48/72

47

4.1.4 Comparison between the Centre of Mass for Berthoud Pass on May

18th and June 1st

At BP1 the snowpack had two distinct layers at -1o

C with the rest at 0o

C. Thecolder layers measured at BP1 were towards the top of the snowpack, 7 to 10 cm and 15

to 17cm below the top of the snowpack. Figure 4.5 shows the centre of mass in the

horizontal to have been at a depth between 60 and 65cm from the top of the snowpack

more than 40cm below the point at which there were colder layers present. There are also

a couple of layers highlighted by a concentration of dye towards the top of the images,

which may correspond to the lower temperature layers mentioned above (figure 3.7).

When studying the snow grains in each layer it was discovered that one particular layer in the snow was described as 'slushy' at a depth of 60cm the same depth as the main

centre of mass. Well-banded clusters of snow grains underlay this layer of slushy snow,

evidently a meltwater retaining layer in the snowpack.

By the time BP2 experiment was carried out the snow had become isothermal at

0oC and 14 cm of snow had melted and removed from the snow depth during the short

period between the experiments. The experiment was still carried out to 1m depth

however fewer distinct layers were observed when the snowpack was examined and

below 32 cm from the surface no changes in snowpack properties were noted (table 3).

Layers observed in the snowpack were however seen lower than this and the centre of

mass was clustered around 50 to 55cm depth. The snowpit was examined after the

experiment had taken place and by the end of the experiment the main meltwater layers

had migrated towards the snowpack surface and therefore were not seen during the

snowpack analysis. To correct for this the snow should have been observed at both ends

of the pit left and compared to remove these sources of error.

4.2 Variance Around the Centre of Mass

After the centre of mass for each section was complete, the variance of this centre

of mass was also calculated to show the spread in centre of mass through the snowpack

in the x and y co-ordinates. Variance has been calculated for every section and plotted for

each experiment in figures 4.8, 4.9 and 4.10.


49/72


50/72

49

in variance around the centre of mass between 0 and approximately 36cm was noted and

in the images there appeared to be a wider spread of dye as you move through the

snowpack as opposed to clustering around one central point in the snow, for example

figure 3.1 compared to figure 3.3. This area may have been at a slightly different

temperature or the snow had slightly different properties to the surrounding snowpack for example grain shape, size or density where the experiment was also carried out due to

this higher spread of meltwater through the snow. Later in this sample area there was a

plume of dye towards the bottom right of the sections, however slightly less dye was

observed to the left of the sections which may explain the slightly lower variance seen in

the x and y directions towards the end of the experiment (figure 3.5). After 78cm there

appeared to be relatively high infiltration compared to the surrounding sections and

slightly less dye was observed in the upper part of the image, this was seen where therewas a slightly lower centre of mass in figure 4.3. This lower centre of mass and reduction

in surrounding dye concentration has reduced the variance seen in figure 4.8 and the last

section appeared to be an anomaly as through the rest of the sample area dye was spread

across the whole of each section.

4.2.2 Variance for Berthoud Pass 1

Variance in the x direction at BP1 was observed to be steady and fairly uniform

throughout the sample area indicating while the centre of mass and the variance around

this point varies in the vertical there was very little or no change in the horizontal (figure

4.9). Due to the properties of the snow during this experiment the dye did not move in

one large mass through the snowpack but instead through specific channels and flow

fingers for example. Variance in the centre of mass was therefore harder to distinguish

than on NR where obvious areas of dye existed or not depending on the section although

there was a uniform horizontal layering system seen in the images which was evidently

areas where there was a change in snowpack properties.


51/72

50

15.00

17.00

19.00

21.00

23.00

25.00

27.00

29.00

31.00

33.00

35.00

0 5 10 15 20 25 30 35 40 45

Distance from start (cm)

V a r

i a n c e

( c m

2 )

Variance in XVariance in Y

Figure 4.9 Variance in centre of mass as a function of distance from the start i.e. number of sections cut (from 0 cm). Berthoud Pass, May 18th 2002

In the y co-ordinate there was quite a lot of difference between the variances for

different sections reflected in the centres of mass seen in figure 4.5. Further through the

sample area, dye appeared to become more concentrated within the snowpack varying in

the centre of mass and therefore altering the variance around this central point. The pattern observed after 20cm in the vertical axis was unusual although there appeared to

be a reasonably constant variance with the dye concentrated towards the top of each

section. Figure 4.9 shows a very low variance in the vertical at 23cm, surprising as

sections cut both before and after show very similar dye distributions (figures 3.9, 3.10

and 3.11).

After sequential analysis of all the images associated with this experiment there

was an obvious difference in the dye concentration spread through the snowpack betweenwidely spaced sections, for example 11cm and 27cm (figures 3.8 and 3.11). There

appeared to be a lower centre of mass vertically and very little variation at 11cm and a

higher dye concentration in the vertical direction at 27cm. Interestingly these two

sections had a very similar centre of mass (figure 4.5) but completely different variances

in the vertical axis indicating a much higher spread around the centre of mass at 27cm

(figure 4.9). Images from sections directly adjacent to one another appeared to have a

fairly uniform variance compared to the sections split apart with different variances even

though they were only a few centimetres apart in the snowpack.


52/72

51

4.2.3 Variance for Berthoud Pass 2

Of the three experiments there was least change in variance through the snowpack

at BP2 compared to the two other experimental days as seen in figure 4.10. An almostconstant spread was seen in each section in the horizontal direction possibly due to layers

observed in the snowpack retaining meltwater at one specific snow depth. Through the

horizontal section area it was not obvious from the images how the variance around the

centre of mass remains reasonably constant as the concentration of dye increases through

the snowpack. As seen in figure 4.7 centre of mass also stayed fairly constant moving

slightly to the right and up in the vertical direction with distance through the snowpack,

consistent with the slight increase in variance around this centre of mass.In the vertical axis there were also fairly uniform variances however the variance

slightly increased further through the sample area. The increase in variance around the

centre of mass corresponded with the slight vertical rise in the centre of mass seen in

figure 4.7. This was observed also in the images as the dye became more dispersed

throughout the sample area, for example figure 3.13 compared to figure 3.18. Causes of

this might be due to the snowpack being isothermal at 0 oC and plumes of meltwater

moving as one mass through the snow becoming deeper towards the end of the sample

area as more time was allowed for infiltration to occur.

1 5 . 0 0 0

1 7 . 0 0 0

1 9 . 0 0 0

2 1 . 0 0 0

2 3 . 0 0 0

2 5 . 0 0 0

2 7 . 0 0 0

2 9 . 0 0 0

3 1 . 0 0 0

3 3 . 0 0 0

3 5 . 0 0 0

0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0

D i s t a n c e f r o m s t a r t ( c m )

V a r

i a n c e

( c m

2 )

Va r i a n c e i n X

Va r i a n c e i n Y

Figure 4.10 Variance in centre of mass as a function of distance from the start i.e.number of sections cut (from 0 cm). Berthoud Pass, June 1st 2002


53/72

52

The spread of meltwater through the snow highlighted the large area of the

snowpack being responsible for meltwater flow. Variance seen at BP2 was relatively

high indicating an even spread of meltwater throughout the whole snowpack.

4.2.4 Comparison between Two Different Snowpacks Isothermal at 0 oC

On NR and at BP2 the snowpacks had achieved equilibrium with the whole

snowpack remaining at a temperature of 0 oC. As observed in the images the dye appeared

to move as one plume through the snow as it had uniform temperature and properties.

Snow grains were equi-temperature throughout the snowpack on both days with a couple

of ice layers also observed. At BP2 these ice layers were fairly shallow at depths of 5cmand 8cm beneath the snow surface however these were not highlighted by the dye. Lower

in the vertical direction two layers were observed, one of these might have corresponded

to the ice layer observed in the snow pit survey at 30 cm depth. Meltwater movement

through the snow was seen to be similar in both snowpacks as layers existed in both,

although slightly higher in the vertical direction at NR compared to BP2, as seen in

figures 3.6 and 3.18. Below these layers very similar development of meltwater plumes

was also observed most likely due to weak areas in these layers allowing the downwards

movement of meltwater (figure 4.11).

Figure 4.11 Schematic diagram of ice layers in the snowpack showing weak areas and meltwater flow. Adapted from Kattelmann and Dozier, 1999.


54/72

53

The two snowpacks evidently had differing properties despite having a common

temperature, location was also important with a different elevation affecting particular

melt processes. Dye concentration in the snow was higher on BP2 compared to NR

however this may be due to photographic exposure, clouds or infiltration time for the dye

and is unlikely to affect the overall demonstration of meltwater movement.

4.3 Roughness

4.3.1 Roughness in all Experiments

Contrast between dyed areas and non-dyed areas within the snow were shown

with an estimate of the roughness. Firstly roughness is seen in a direct comparison across

all experiments moving through the snowpack. In later sections roughness is plotted for

each individual experiment.

A comparison between roughness in the different snowpacks in the x and y co-

ordinates are shown in figure 4.12. Low roughness values indicate a smooth, uniform

flow of dye through the snowpack and high roughness indicates a high contrast ratio

between the snow and the dye indicating the existence of specific meltwater channels.

The highest roughness was expected to be in a snowpack non-isothermal at 0 oC due tothe presence of meltwater flow channels i.e. During BP1 experiment according to the

snowpit data collected and observed in the sections for example figure 3.8 shows the dye

following distinct vertical pathways.


55/72

54

Roughness Comparison

0.000

0.500

1.000

1.500

2.000

2.500

3.000

0 10 20 30 40 50 60 70 80 90 100Sections through snowpack (cm)

R o u g

h n e s s

BP2 - X BP2 - Y NR - X

NR - Y BP1 - X BP1 - Y

Figure 4.12 Roughness in dye passing through the snowpack. Comparison of roughness

between the three experiments in x and y co-ordinates as a function of distance from the

start.

Figure 4.12 shows on NR there was firstly a negative trend in the roughness

values, leading to a slight positive correlation where roughness increased. Indication of an increasing value of roughness was seen in both the horizontal and vertical axis

although the sections appeared to be rougher in the horizontal direction as opposed to the

vertical. Changes in roughness were difficult to observe when the images were seen as a

sequential set, however the roughness value differences were fairly small in comparison

with each other and will account for changes in intensity of the dye not seen by the

human eye.

Roughness at BP1 was highly scattered between adjacent centres, a similar

pattern to that observed in the centre of mass and the variance in the y co-ordinate

(figures 4.5 and 4.9). As opposed to NR, roughness in the horizontal was lower than the

roughness in the vertical direction indicating a slightly more defined pattern in the

vertical between clean and dyed snow areas. Later in the snowpack as you go past 20cm

there was a definite vertical shift upwards of dye in the snow, which appeared to be a

more concentrated area of dye relative to the earlier sections observed. This vertical shift

was seen when comparing figures 3.7 and 3.9. Roughness values decreased towards the

end of this experiment indicating a smoother spread of dye with a vertical rise in the


56/72

55

centre of mass. This was confirmed by the higher variance observed in the vertical

direction in figure 4.9.

As observed in the centre of mass and variance data from BP2 showed a smooth

pattern throughout the snowpack in terms of roughness. Roughness in the vertical

direction was seen to rise initially until around 25cm then become more uniform asopposed to vertical roughness, which rose until around 38cm before becoming steady.

Patterns seen in the roughness were reflected in observations of the images as moving

further through the snowpack specific patches of dye and plumes of meltwater movement

were seen. Figures 3.13, 3.14 and 3.17 show this gradual progression of dye developing

meltwater plumes within the snowpack. Experiments carried out on NR showed higher

roughness in the horizontal direction than the vertical as was the case at BP2, this may be

an indication of higher horizontal spread through the snowpack due to the isothermaltemperature (0 oC) properties of the snow.

4.3.2 Roughness for Niwot Ridge

When comparing horizontal and vertical roughness values a similar pattern was

seen. Indications from figure 3.4 compared to figure 3.5 appeared to show that with an

increase in dye in the vertical direction an increase was also observed in the horizontal.

Towards the end of the sample area plumes of dye were observed low in the sections,

these appeared to move as one mass through the snow shown in figure 3.6 and

corresponding well with figure 4.13 where there was an increase in the x direction there

is also an increase in the y direction. A strong positive correlation was noted between

roughness in the x and y directions on all three days but particularly on NR and BP2

where an increase was observed in the roughness in the horizontal, an increase was also

seen in the vertical direction. The direct relationship seen between the vertical and

horizontal direction indicates an isotropic structure to the snowpack with equal

distributions in both the horizontal and the vertical direction.


57/72

56

0.5

0.7

0.9

1.1

1.3

1.5

0.5 0.7 0.9 1.1 1.3 1.5

Roughness in X

R o u g

h n e s s

i n Y

Figure 4.13 Roughness at Niwot Ridge, May 9th 2002 with roughness in x plotted against roughness in y.

4.3.3 Roughness for Berthoud Pass 1

At BP1 the snow was not isothermal at 0 oC and therefore it was expected this

experiment would show development of flow paths in the snowpack. This appeared to be

the case when studying the images as dye was concentrated in particular areas throughout

the upper part of the snowpack an example of which is seen in figure 3.8. Concentration

of the dye was mainly in the horizontal plane however specific channels appeared to be

present vertically also.

According to the calculations for roughness the existence of flow paths was not

obvious, it was expected that the roughness values would be highest in this data set due

to the high contrast between dyed and non-dyed snow. The reasons for this may have

been due to the fact that a lot of snow was exposed towards the bottom of the section

with no dye present indicating a smooth surface at this level. The top of the snowpack is

likely to have been rough due to the high contrast between snow and dye. A short test

was performed for this by cutting away the lower part of a few images and doing

roughness calculations once again. The results showed the roughness to be on the order

of 1.5 to 2.7 as opposed to the roughness values seen in figure 4.14 on the order of 0.5 to

1.6.


58/72

57

Roughness at BP1 was slightly anisotropic opposed to NR as figure 4.14 shows

there is a higher roughness in the vertical direction compared to the horizontal indicating

the dominance of a horizontal layering structure from the roughness values observed.

1.0

1.2

1.4

1.6

1.8

2.0

2.2

2.4

1.0 1.2 1.4 1.6 1.8 2.0

Roughness in X

R o u g

h n e s s

i n

Figure 4.14 Roughness at Berthoud Pass, May 18th 2002 with roughness in x plotted against roughness in y.

4.3.4 Roughness for Berthoud Pass 2

In contrast with NR and BP1, during experiments at BP2 (figure 4.15) there

appeared to be higher roughness values in the horizontal than the vertical and a much

more constant vertical roughness. Roughness values were also higher for BP2 on the

order of 2 to 2.5 compared to 0.5 to 2 seen during the previous experiments. Higher

roughness values seen here may be due to the wider spread of dye observed in the

snowpack with variable patches of dye and snow as seen in figure 3.16 for example. Theother two experiments may have had lower roughness values due to the more uniform

spread of dye on NR and very small areas of dye at BP1 causing a high roughness in

these areas but relatively smooth elsewhere in the section.

Images taken of these sections showed an increase in dye concentration through

the snowpack with a higher spread of dye observed towards the end of the sample area as

seen in figure 3.18 compared to figure 3.13. Also seen was the fact that initial vertical

movement of dye was not obvious but later in the snowpack more specific dye plumes

became apparent. Increases in vertical movements were accompanied by some spread in


59/72

58

the horizontal direction. As reflected in figure 4.15 horizontal spread did not increase

greatly as initially observed on the images as there was a high concentration of dye

roughly halfway down snowpack spread completely across each section between two

layers of differing snow properties (figure 3.13). As in the other isothermal snowpack at

NR a direct relationship is seen between roughness in the horizontal and the verticaldirections indicating a steady structure of flow through the snow.

.

1.5

1.6

1.7

1.8

1.9

2.0

2.1

2.2

1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8

Roughness in X

R o u g

h n e s s

i n Y

Figure 4.15 Roughness at Berthoud Pass, June 1st 2002 with roughness in x plotted against roughness in y.

4.4 Summary

Overall experiments carried out NR, BP1 and BP2 showed some similar and some

different characteristics of snow under varying conditions. Figure 4.16 attempts to

summarise the general characteristics observed in these snowpacks.


60/72

59

Figure 4.16 Summary diagram to show meltwater flow processes in a snowpack.


61/72

60

5. Conclusion

5.1 Conclusions

Dye tracing in snow was carried out in this experiment in order to establish the

presence of preferential flow and better physically understand spatial movement of water.

Non-toxic food colouring mixed with natural snowmelt water was applied to a two by

two metre area of the snowpack surface and allowed to infiltrate. After this time one or

two centimetre sections of snow were cut and a sequential set of images taken as more

sections were removed. The experiment was repeated twice at two different locations in

order to compare different snowpacks. After completion of the fieldwork images were

transferred to a computer for analysis using the computer program MATLAB.It appears flowpaths depend on nature of the snow and boundary conditions.

Statistical analysis showed meltwater moves through a snowpack isothermal at 0 oC as

one general plume unless stratigraphic layers are encountered where lateral spread of

meltwater occurred. In comparison to this snowpacks with layers of differing temperature

appeared to show specific flow channels. Small-scale changes were common and

different rates of flow observed within very small areas of the snowpack. Spatial scale

observations need to be carried out studying the snowpack in detail as opposed to over large areas of melt in order to understand both microscale and mesoscale melt processes.

Overall flowpaths exist in a snowpack moving water through preferential

channels towards the snowpack base. Flow rates increase as the snowpack becomes

warmer and then meltwater movement is continuous through the entire snowpack

moving in large plumes spreading horizontally as movement progresses. The level of

connectivity of flowpaths appears to be higher when the snow has uniform properties,

connectivity between flowpaths in a non-isothermal snowpack is through lateral spread

of meltwater across layers of different snow properties.

5.2 Suggestions for Further Work

As a result of the information found here it is suggested that further dye tracing

experiments are carried out at the same locations under similar snow conditions to see if results are related between different years of snowfall. It would also be useful to carry


62/72

61

out further experiments earlier in the melt season especially at NR to gain further

understanding of melt processes prior to the snowpack becoming isothermal at 0 oC.

For a more thorough investigation a technique known as radar tomography should

be used in the future as it is less destructive to the environment and therefore repeatable

in the exact location, however this technique is quite expensive. The radar techniqueinvolves high frequency radar sensing melt water features in the snowpack at a distance

up to one metre as tested by Albert et al. (1999). A snow trench is dug along the line of

investigation and portable tracks installed along which the radar can be pulled at a

constant rate. Old refrozen melt fingers as well as new wet fingers are observed in the

study of meltwater flowpaths. Investigation at Niwot Ridge and Berthoud Pass would be

possible using this technique along with the dye tracing experiments in order to compare

different results using the two different techniques.After the data has been collected more geostatistics should be applied in order to

attempt to more fully understand the snow. An example of further statistical analysis

might be using variograms to characterise heterogeneity of each section. This process

was used successfully by Rea and Knight (1998) in correlating data from radar and

hydrogeology of an area using dielectric and hydraulic properties in the subsurface of a

sedimentary bed. This could be applied in the same manner to snowpack data in order to

study spatial variation in meltwater flow in the snow. Due to the low roughness values

seen during experiment BP1 due to the low level of infiltration further statistical work

should be carried out in the higher levels of the snowpack where the dye is present. This

would ensure that the areas of infiltrated dye are properly represented. The other

alternative would be to wait longer for dye to infiltrate during periods when the snow is

at a colder temperature.

A lot of research work has been carried out into the need for specific dyes in soil

dye tracing experiments according to the specific study. Research into the use of different

dyes in snow would be useful to establish if any particular dyes absorb more fully to the

snowpack and therefore indicate flowpaths better than a different dye for example.

In conclusion future research work into snow is vital for water supply, flood

management, recreation and tourism and modelling problems in relation to water supply

and long and short-term climate change and stability.


63/72

62

References

Albert M., Koh G., Perron F., 1999 Radar investigations of melt pathways in a natural

snowpack. Hydrological Processes, p 2991 -3000

Baveye P., Boast C.W., Ogawa S., Parlange J., Steenhuis T., 1998 Influence of image

resolution and thresholding on the apparent mass fractal characteristics of preferential

flow patterns in field soils. Water Resources Research , p 2783 - 2796

Boggild C. E., 2000 Preferential flow and meltwater retention in cold snow packs in

West-Greenland. Nordic Hydrology , p 287 - 300

Flury M., Fluhler H., 1995 Tracer characteristics of Brilliant Blue FCF. Soil Science

Society of America , p 22-27

Ghodrati M., Jury W.A., 1990 A field study using dyes to characterize preferential flow

of water. Soil Science Society of America , p 1558 - 1563

Hasnain S.I., Jose P.G., Ahmed S., Negi D.C., 2001 Character of the subglacial drainage

system in the ablation area of Dokriani glacier, India as revealed by dye-tracer studies.

Journal of Hydrology , p 216 - 223

Heppell C.M., Burt T.P., Williams R.J., 2000 Variations in the hydrology of an

underdrained clay hillslope. Journal of Hydrology , p 236 - 256

Horne F.E., Kavvas M.L., 1997 Physics of the spatially averaged snowmelt process.

Journal of Hydrology , p 179 - 207

Kattelmann R., Dozier J., 1999 Observations of snowpack ripening in the Sierra Nevada,

California, U.S.A. Journal of Glaciology , p 409 -416

Luxmoore R.J., Jardine P.M., Wilson G.V., Jones J.R., Zelzany L.W.

Documents

Rebecca Harrison (2003)