THE EFFECTS OF SNOW AVALANCHES ON THE HYDROLOGIC REGIME
OF THE KUNHAR RIVER, WESTERN HIMALAYAN, PAKISTAN:
ANALYSIS AND APPLICATION TO RIVER FLOW FORECASTING.
By
MOHAMMAD INAMULLAH KHAN
B.E. (Civil Engineering),
N.E.D. University of Engineering and Technology, Karachi, Pakistan, 1984
A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF APPLIED SCIENCE
in
THE FACULTY OF GRADUATE STUDIES
Department of Civil Engineering
We accept this thesis as conforming
to the requirement standard
THE UNIVERSITY OF BRITISH COLUMBIA
September 1995
© Mohammad Inamullah Khan, 1995 In presenting this thesis in partial fulfillment of the requirements of an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission.
Department of Civil Engineering
The University of British Columbia
Vancouver, Canada
Date g»b. \.\°l45 ABSTRACT
This study sets out to investigate the significance of snow avalanches on the hydrology
and runoff generation in the Kunhar basin in Northern Pakistan. The objectives of this
research are, to analyze the snowmelt and snow avalanche effects using the U.B.C.
Watershed Model, and to produce a flow forecasting system which takes account of the
snow avalanche effects.
The Kunhar River is a major tributary of the Jhelum River in the western
Himalayas of Pakistan. The basin area is about 2,340 km2 with an elevation range from
800 to 5,300 m above sea level. The watershed has a seasonal snow cover which
develops from early November onwards, reaching a maximum depth in March or April.
Also, the snowpack increases greatly at upper elevations.
In the Kunhar basin the avalanching is a major source of snow redistribution from
higher to lower elevations. It is estimated that on average over 200 x 106m3 water
equivalent of snow is avalanched annually. The percentage of the total affected area
(runout and starting zones) by avalanches in Kunhar basin is estimated to range from
12% to 21%. The starting zone lies at a mean elevation of about 4,000 m and runout
zones are at mean elevations of 2,450 and 2,800 m above sea level. This means that the
avalanche activities in the lower elevations are dependent on the snow precipitation at elevation 4,000 m. This study shows that about 20% of the snowpack at 4,000 m is, on
average, subject to avalanching. Avalanche contribution is found to be very significant in calibrating the watershed model. On average the overall Nash-Sutcliffe coefficient of efficiency of the model was improved from 77 to 84% after introducing avalanches in the calibration which shows improved time distribution of runoff.
Snowmelt pattern in the avalanche areas is significantly modified by avalanche
activity. Firstly, the snowmelt in the runout zones starts about seven days later and lasts
about 30 days more than in areas not affected by avalanches. The snowmelt volume in runout areas is increased by about 200 to 300% in affected areas. The maximum snowmelt from the avalanche runout areas is about 100% higher than the maximum snowmelt in the un-affected areas. The timing of the maximum snowmelt is delayed by
about 15 days in the runout zones of avalanche affected area, due to high accumulation of snow. These results show that the snow avalanches increase both the volume and the period of the snowmelt in the runout zones and also change the time distribution of the
snowmelt. Since the snowmelt increase in the runout zones is compensated by the
decrease in snow in affected areas of the starting zones, the total snow melt from the basin is unchanged.
The above results of flow simulation by using redistribution of snow were used to produce a forecasting system of avalanche activity. Linear regression analyses were performed and the linear relationships for each band were estimated. Regression analyses show very strong correlation between avalanche volume and snowpack accumulation at the upper elevations, i.e., the coefficient of determination (R2) is found to be in a range of
0.9 - 0.95. The extra snow depth acquired at elevations 2,450 to 2,800 m in the form of
iii avalanche is also strongly correlated with the existing snow depth at 4,000 m, R2 ranged between 0.93 and 0.99.
If the snowpack at 4,000 ra elevation is measured then the maximum snow accumulation, which occurs in late March or early April, can be estimated. From the developed equations the total avalanche volume, the snow avalanche depth, and the affected areas for runout and starting zones can be estimated. These estimates can then be used in the U.B.C. Watershed Model to forecast the flow for the coming season.
Application of this procedure showed that the proposed forecasting system gives an improved and reliable estimation of the seasonal flow volume and the time distribution of runoff for the Kunhar river.
IV Table of Contents
Abstract ii
Table of Contents v
List of Tables ix
List of Figures x
List of Symbols xii
Acknowledgment xv
Dedication xvi
Chapter ONE
INTRODUCTION 1
1.1 PREAMBLE 1
1.2 RESEARCH OBJECTIVES 4
Chapter TWO
THEORETICAL BACKGROUND AND LITERATURE REVIEW 6
2.1 INTRODUCTION 6
2.2 GOVERNING FACTORS AND RELEASE MECHANISMS 10
2.3 HYDROLOGICAL ROLE OF AVALANCHE 14
2.4 RESEARCH ON HYDROLOGICAL ASPECT OF AVALANCHE 16
2.4.1 Summary 22
v Chapter THREE
GEOGRAPHY AND CLIMATE 24
3.1 OVERVIEW 24
3.2 GEOGRAPHY OF THE NORTHERN MOUNTAINS 26
3.2.1 Regional Setting 26
3.2.2 The Upper Indus Basin '. 26
3.3 CLIMATE 29
3.3.1 Broad Climatic Controls 29
3.3.2 Local Climate 32
3.4 STUDY AREA 35
3.4.1 Avalanche Activity in Kunhar Basin 38
Chapter FOUR
METHOD OF ANALYSIS 41
4.1 HYDROLOGICAL MODEL 41
4.2 CALIBRATION PROCEDURE 48
4.3 CLIMATIC DATA 50
4.4 PRECIPITATION ADJUSTMENTS 51
4.5 AVALANCHING 53
4.6 SUMMARY OF CALIBRATION 58
4.7 REGRESSION ANALYSES 60
VI 4.7.1 Avalanche Volume 60
4.7.2 Avalanche Depth 61
Chapter FIVE
RESULTS AND DISCUSSIONS 62
5.1 INTRODUCTION 62
5.2 MODEL CALIBRATION 62
5.2.1 Stage-I Calibration 63
5.2.2 Stage-II Calibration 64
5.2.3 Avalanche Contribution 66
5.3 DISTRIBUTION PATTERN OF AVALANCHE 73
5.3.1 Avalanche Volume 73
5.3.2 Avalanche Area 77
5.4 EFFECTS OF AVALANCHES 82
5.4.1 Snowpack Conditions With and Without Avalanche 82
5.4.2 Avalanche Effects on Snowpack Accumulation 89
5.4.3 Avalanche Effects on Snowmelt 93
5.5 AVALANCHE FORECASTING MODEL 104
5.5.1 Avalanche Volumes 104
5.5.2 Avalanche Depth (POPADJ) and Areal Distribution 109
vii Chapter SIX
CONCLUSIONS AND RECOMMENDATIONS 122
6.1 CONCLUSION 122
6.1.1 Calibration Problems 123
6.1.2 Significance of Avalanche 123
6.1.3 Avalanche Forecasting Model 124
6.2 RECOMMENDATIONS 125
6.2.1 Knowledge of the Snowpack at Higher Elevations 125
6.2.2 Snow Course Surveys 126
6.2.3 Testing of the Strategy 126
6.2.4 Model Modification 127
REFERENCES 128
APPENDIX 137
viii List of Tables
Table 4.1. Description of watershed by elevation bands 55
Table 4.2. A typical working sheet showing procedure to calculate the magnitude
of avalanches 59
Table 5.1. Statistics report of Stage-I calibration for year 1979-88 64
Table 5.2. Statistics report of Stage-II calibration for years 1979-88 65
Table 5.3. Statistics report of Stage-I, Stage-II, and Stage-Ill calibrations for the
Kunhar River Basin 67
Table 5.4. Distribution pattern of avalanche volume in each band..... 75
Table 5.5a. Areal distribution of avalanches in each band 78
Table 5.5b. Regression analysis report (A vs. A) 80
Table 5.6. Illustration of calculating the mean snowpack (mm) in band 3 for the
year 1979-80, with and without avalanches 83
Table 5.7. Snowpack depth in millimetres of water equivalent 85
Table 5.8. Snowmelt (Sm) pattern with and without avalanches 95
Table 5.9. Results of regression analyses (A vs. SP'fi) 107 Table 5.10. Extra snowpack (mm) required in bands 3 & 4 and snow to remove from band 6 '. 110
Table 5.11. Results of regression analyses (Sp vs. SP'„) Ill
Table 5.12. Avalanche forecasting for years 1979-80 and 1980-81 115
Table 5.13. Avalanche forecasting for years 1981-82 and 1982-83 115
Table 5.14. Avalanche forecasting for years 1985-86 and 1986-87 116
Table 5.15. Avalanche forecasting for year 1987-88 116
Table 5.16. Comparison of efficiency and deviation in discharge between base• line, avalanched, and forecasted calibrations 121
ix List of Figures
Figure 2.1. A typical sketch of loose-snow avalanche 9
Figure 2.2. Cross section of a typical snow slab 9
Figure 2.3. Stresses in a snowpack due to weight 11
Figure 3.1. Geographical map of the Karakoram-Himalayan Ranges and the Upper
Indus Basin 27
Figure 3.2. Kunhar River Basin 36
Figure 5.1. Comparison of hydrographs before and after avalanches with the observed flow (a) year 1979-80 (b) 1980-81 69 Figure 5.2. Comparison of hydrographs before and after avalanches with the observed flow (a) year 1981-82 (b) 1982-83 70
Figure 5.3. Comparison of hydrographs before and after avalanches with the observed flow (a) year 1985-86 (b) 1986-87 71
Figure 5.4. Comparison of hydrographs before and after avalanches with the observed flow for year 1987-88 72
Figure 5.5a. Avalanche distribution pattern for active bands as a percentage of snowpack in band 6 76
Figure 5.5b. Avalanche in bands 3 & 4 against avalanche in band 6 as a percentage of snowpack in band 6 76
Figure 5.6a. Avalanche area distribution pattern for active bands as a percentage
of total watershed area 79
Figure 5.6b. Avalanche area against volume in bands 81
Figure 5.7. Snowpack depths in mm water equivalent at different elevation band for years, (a) 1979-80; (b) 1980-81; and (c) 1981-82 86 Figure 5.8. Snowpack depths in mm water equivalent at different elevation band for years, (a) 1982-83; (b) 1983-84; and (c) 1985-86 87
Figure 5.9. Snowpack depths in mm water equivalent at different elevation band
x for years, (a) 1986-87; and (b) 1987-88 88
Figure 5.10 Snowpack accumulation in avalanched and non-avalanched part of a band for year 1979-80 91
Figure 5.11 Snowpack accumulation in avalanched and non-avalanched part of a band for year 1980-81 92
Figure 5.12. Snowmelt patterns in elevation bands without and with avalanches for year 1979-80 97
Figure 5.13. Snowmelt patterns in elevation bands without and with avalanches for year 1980-81 98
Figure 5.14. Snowmelt patterns in elevation bands without and with avalanches for year 1981-82 99
Figure 5.15. Snowmelt patterns in elevation bands without and with avalanches for year 1982-83 100
Figure 5.16. Snowmelt patterns in elevation bands without and with avalanches for year 1985-86 101
Figure 5.17. Snowmelt patterns in elevation bands without and with avalanches for year 1986-87 102
Figure 5.18. Snowmelt patterns in elevation bands without and with avalanches for year 1987-88 103
Figure 5.19. Avalanche volume in bands with respect to snowpack in band 6 108
Figure 5.20. Extra snow w.e. required to increase in runout band and to decrease in starting zone with respect to snowpack w.e. in band 6 (a) band 3; (b) band 4; and (c) band 6 112
Figure 5.21. Comparison of observed, avalanched, and forecasted hydrographs for years (a) 1979-80 and (b) 1980-81 117
Figure 5.22. Comparison of observed, avalanched, and forecasted hydrographs for years (a) 1981-82 and (b) 1982-83 118
Figure 5.23. Comparison of observed, avalanched, and forecasted hydrographs for years (a) 1985-86 and (b) 1986-87 119
Figure 5.24. Comparison of observed, avalanched, and forecasted hydrographs for year 1987-88 120
xi List of Symbols
A = Area of the Catchment in m2 (Eq: [2.4])
A = Avalanche Area (km2)
~ 6 3 A = Avalanche Volume Water Equivalent (xlO m )
Av = Avalanche
C = Constant (427.35)
°C = Degree Centigrade
Dv = Deviation in Volume (%)
E! = Nash-Sutfcliffe Coefficient of Efficiency
Eopt = Coefficient of Efficiency for Optimization Process
/ = Yield Coefficient or Proportion of Snow on the Avalanche Path which
Avalanches (t m ) g = Gravitational Acceleration h = Height
H = Catchment Area of Avalanches in Hectares
K = Concentration Factor (Starting Zone Area/Runout Area)
M = Annual Avalanche Mass a m = Meter mm - Millimeter
Q = Discharge (m3)
xii R = Coefficient of Determination
iSa = Amount of Precipitation (water equivalence of snowfall and rain into
snow)
Sm = Snowmelt (mm water equivalent)
Sp = Extra Snow Depth Water Equivalent (mm) Required to Increase or
Decrease From Active bands (From Eqs. [5.5], [5.6], and [5.7])
SP = Snowpack Water Equivalent (x 106 m3)
Sp' = Extra Snowpack Depth water Equivalent in Part of 'b' of a Described
Band form Band CAL File
SP' = Snowpack Depth Water Equivalent (mm)
V = Mean Volume of Avalanches in 1,000 m3 w.e. = Water Equivalent
ASP = Change in Snowpack depth (mm w.e.)
x = Shear Stress a = Normal Stress p = Average Snow Density at Depth h a = Surface Slope Angle Measured from the Horizontal (Degrees)
xiii Band Numbers
Part 'a' of a Band (i.e., Un-affected Area of a Band)
Part 'b' of a Band (i.e., Affected Area of a Band)
Estimated Discharge
Observed Discharge
Total
xiv Acknowledgment
I wish to express my most sincere appreciation and gratitude to Dr. M. C. Quick, who
has had a profound and positive impact on my academic and professional attitude. I
greatly appreciate his advice, continuous guidance, and invaluable encouragement
throughout the course of this research.
Many thanks to Dr. Sakis Loukas, for valuable suggestions and comments.
I am particularly grateful to the Civil Engineering Department of UBC as a whole
and Dr. Warren Bell of BC Hydro for excellent working facilities and cooperation in all
aspects.
I would like to acknowledge Mr. Edmond Yu for his support and assistance in
computer stuff. The support of Heiki Walk is also gratefully acknowledged.
The research was supported financially by IDRC (Canada) and WAPDA
(Pakistan).
xv To my wife and children whose love and encouragement were
constant source of strength
xvi Chapter 1
INTRODUCTION
1.1 PREAMBLE
Pakistan is primarily dependent upon irrigated agriculture in the Indus Basin. This irrigation system includes inter-river link canals and two major storage reservoirs at
Mangla and Tarbela that regulate as well as supplement the water supplies. The Indus
River is Pakistan's main source of water for irrigation, power generation, and urban and industrial water supply.
The waters of Indus derive largely from high-altitude snowfalls and the glaciers of the northern mountain ranges, the Karakoram, Kohistan, and Himalayas. The hydrology of the Upper Indus Basin (U.I.B.) is mainly determined by the snow and ice conditions in these mountains and supply the main stem of the Indus with most of their water.
Spring snowmelt is a significant runoff component in the tributaries of the Indus
River located on the south side of the Himalayan crest line. Heavier snowfalls occur at higher altitudes where slopes tend to be steeper and the snowpacks more unstable.
Therefore these heavier high altitude snowfalls cause a significant fraction of all the snow that falls in the U.I.B. to avalanche through some hundreds of meters (Hewitt, 1988a).
Consequently, part of the snowmelt component is derived from avalanche transported snow. A joint Canada-Pakistan hydrology project of the Indus River System was established in 1985 to investigate high mountain snow and glacier resources. Initially the project was planned as a three-year venture but was then extended to 1989 to investigate the hydrology of the mountain headwaters of the Indus, and especially the snow and ice
conditions above 2,500 m a.s.l. (Hewitt, 1990a).
The project has now entered its second phase which will be completed in 1996.
This is a collaborative project funded jointly by the Water and Power Development
Authority (WAPDA) and the Canadian International Development Research Center
(IDRC). In Pakistan, the project is coordinated by the Hydrology and Research
Directorate of WAPDA. In Canada, the project coordination is undertaken by B.C. Hydro
International Limited (BCHIL). The project objective is to upgrade the capability of
WAPDA to manage the outflows from the Upper Indus Basin. It has involved field work,
with applications of remote sensing, analyses of existing river discharge and
meteorological observations to provide the basis for improving seasonal streamflow forecasting. This work will lead to improved and more effective water management
through effective use of water. Both Pakistani and Canadian engineers are involved in all
phases of the work. After completion of the project in 1996, the inflow forecasts will be
used by WAPDA of Pakistan to plan reservoir releases for irrigation and power
generation.
Field work has been concentrated in two main areas; the Biafo - Hisper and Barpu
- Bualtar glacier areas of the Central Karakofam, and the Nanga Parbat / Kaghan Valley
area of the Western Himalayan. In both areas basic data are being collected to improve knowledge of the meteorology, glaciology, and snow hydrology. Part of the snow
2 hydrology studies involves the investigation of the role of avalanche in seasonal snowmelt especially in the Kunhar River basin.
The Kunhar River basin experiences intense and high magnitude avalanche activity above 1850 m elevation. Most of the avalanche activity occurs in the 4,000 to
2,000 m elevation range, so that high elevation snow is redistributed downwards, sometimes by as much as 2,000 m (De Scally, 1992). The avalanche activity tends to concentrate the avalanche snow into a deeper and more dense snowpack which retards its melting. This avalanching occurs not only in the Kunhar River Basin but also in the surrounding region especially the Neelum River Basin, and it is therefore important to investigate the hydrological influence of this avalanching.
The avalanche snow, on one hand, may increase snow melting rates because of lower elevation & albedo and higher temperatures etc., but on other hand it delays snowmelt runoff because of the large amount of snow accumulation. We not only need to understand the effects of avalanching in hydrologic processes, but we also need methods to predict the streamflow response that results from avalanching. The basis of the present research is to understand the behaviour of avalanches as a fundamental component in the calibration of hydrological model of the Kunhar River basin, and to establish a snow avalanche and a flow forecasting system.
3 1.2 RESEARCH OBJECTIVES
Estimation of peak flows is necessary for the design of any hydro technical project. The flow estimation can be achieved by using a hydrological model along with meteorological
data from a number of stations. Also, good knowledge of the area and the hydrological
processes is needed to simulate the runoff generation. Flow simulation and forecasting
become very difficult in mountain areas mainly because of the lack of reliable databases
with the necessary spatial resolution. Furthermore, because of the limited accessibility of
high mountain areas very little is known of the runoff processes in high elevation
watersheds. For example, avalanches redistribute the snow accumulation and can result in
major time redistribution of the river flow.
The goals of the present research are to determine the significance of avalanche
contribution in Kunhar basin hydrographs by using a watershed model to study the
avalanche effects. This work will assist in establishing a flow forecasting system which
can take account of the snow avalanche effects. Within these broad objectives, the
following specific objectives were adopted;
1. Calibrate the U.B.C. Watershed Model for Kunhar Basin with no
avalanches and get the best possible results;
2. Introduce and investigate the redistribution of snow to simulate
avalanching from higher to lower elevations;
3. Analyze the significance of avalanching by comparing the results of
calibration with and without avalanches input on the basis of hydrographs.
total flow volumes and efficiencies of the analyses;
4 4. Examine results for each year and see if there is a consistent pattern of
avalanching;
5. Establish an avalanche forecasting system on the basis of snowpack
conditions in higher band/s; and
6. Use this system to improve streamflow forecasting for the Kunhar Basin.
While analyzing and simulating avalanches the following important parameters were also given concentration,
i) most active elevation bands regarding avalanche activities (higher
elevation band as starting zone and lower one as runout);
ii) avalanche areal distribution pattern for each elevation band;
iii) avalanche magnitude distribution pattern for each elevation band;
iv) snowmelt patterns before and after avalanching within the elevation bands;
5 Chapter 2
THEORETICAL BACKGROUND AND LITERATURE REVIEW
2.1 INTRODUCTION
An avalanche is a mass of snow transported at high velocities down a mountain slope.
Considerable amounts of snow are displaced by avalanches from higher to lower elevations and are then deposited & concentrated in reduced areas. Avalanche snow is denser, deeper and in much more compact masses than direct snowfall.
Avalanches form when the snowpack resting on a slope undergoes failure. They occur especially in areas of steep slopes - generally between 25° and 60° - but their momentum may carry them on to flatter slopes, particularly if they are channeled into a gully (Goudie, 1993). New dry snow can cling to 40° slopes, where wet slushy snow may slide even on 15° slopes. The critical angle of repose depends on the temperature and density of the snow, which determine its texture and wetness (Barry, 1992).
An avalanche path (Terrain boundaries of known or suspected avalanches) consists of an upper starting zone, the track zone (part of the path between the starting and runout zones) which is often clearly delimited by a swath of grassy or shrubby vegetation running down below tree line, and a lower runout-deposition zone which may have a more or less well-marked debris fan at the foot of the slope (Fig. 2.1). The failure process
6 begins in the starting zone, and then the developed energy and other dynamic characteristics depend on the relief of the track and the amount of material that can be entrained into the avalanche as it gains momentum (Perla. 1978).
Two generic groups of avalanches may be identified: Direct action and delayed action (Armstrong, 1976). A direct action avalanche occurs during or immediately after a storm and is the result of the increased stress applied to the snowpack in the form of new snow. This type of avalanche is the immediate consequence of rate of snow loading in the starting zone. The snow loading rate is a function of many meteorological parameters, among which crystal habit, snowfall intensity, wind transport and snow deposition and temperature are important (Fraser, 1966 and Tesche, 1988). A delayed action avalanche is the result of gradual changes taking place within the snowcover over a longer period of time due to over burden snow load. The thickness of the snowcover exerts a large influence on the proportion of the genetic types of avalanches (Shcherbakov, 1973) and also determines their volumes and to a large extent the moment of avalanche movement.
On the basis of starting zone appearance and general snow structure, snow avalanches are classified into two categories: point avalanches (also called as loose-snow avalanche) and slab avalanches (Fig. 2.2). The point avalanche occurs when snow crystals which adhere poorly to each other collect in a slope steeper than their angle of repose. Failure initiates within a cohesionless layer located immediately below the surface. As soon as it breaks loose, the unstable lump roll down the slope, bulldozing out a widening pattern. During storms, point avalanches occur frequently when the slope angle is steeper than about 45°.
7 Slab release is characterized by an initial spectacular propagation of cracks
followed by the crumbling of a slab-like region of the slope into numerous blocks with
dimensions on the order of about 1 m on slopes of 20-45°. The stability of a slab depends
on the stress state and fracture toughness of a large, cohesive mass. The slab incorporates
larger amounts of snow.
After slab failure, sharply defined fracture surfaces which outline the slab
boundaries remain at the starting zone. These fracture surfaces are designated as follows:
bed surface: main sliding surface of slab (shear type of failure)
crown surface: upslope fracture surface (tension fracture)
stauchwall: downslope boundary (shear fracture)
flank surfaces: two side boundaries of the slab (combination of tension and shear
fractures)
Slab avalanches may involve dry or wet snow, but both are associated with shear
stresses in the snow exceeding the shear strength in some underlying layer. Dry-snow
avalanches are particularly associated with high snowfall amounts over the preceding four
days and with wind redistribution of the snow 12-24 hours before the event (Barry, 1992).
Wet-snow slides, which mainly occur in spring, are associated with the antecedent air
temperature values.
8 Fig. 2.1. A typical sketch of loose-snow avalanche.
Crown /
/ '
Bed surface y^^yy^''
Stauchwall-^ •••
»>™
Fig. 2.2. Cross section of a typical snow slab. (Mears, 1979)
9 2.2 GOVERNING FACTORS & RELEASE MECHANISMS
Avalanches are generated by structural weaknesses in the snow cover which give rise to structural instability (Voight, 1990). When snow lies on a slope, the force parallel to the slope caused by its weight produces shear stresses while the force perpendicular to the slope produces compressive stresses (Fig. 2.3). Failure occurs when the stress exceeds the strength at some point. The shear stress at any depth h, acting parallel to the bottom of the snowpack, is given by the relationship;
x = p g h sin a [2.1]
where; p = average snow density at depth h,
g - acceleration due to gravity,
a = surface slope angle measured from the horizontal.
The normal stress (to slope) at any depth h in the snowpack is given by;
a = p g h Cos a [2.2]
These equations show a direct relation of shear stress to the surface slope steepness and thickness of snowpack. Shear stress will reach a maximum at the base of the snowcover and decline to zero at the surface.
10 Fig. 2.3. Stresses in a snowpack due to weight. (Schaerer, 1981)
11 The sequence of events preceding avalanche release is apt to vary considerably
depending on meteorological conditions, and also on the immediate trigger, which may be
artificial explosive blast of enormous energy or a subtle internal disturbance. The following are a few of the wide variety of avalanche triggers:
/- Precipitation. There is an observed high probability of avalanche occurrence during or immediately after a severe storm. Amount, intensity and duration of snowfall (or rainfall)
are collectively the prime factors of avalanche formation (Armstrong, 1975 and Williams,
1981). Snow strength due to sintering1 cannot keep pace with the increasing stress in an underlying stratum caused by load of additional snowfall (Fraser, 1966). Under a slow rate of precipitation, the snow can absorb the load by changing its shape with a slow
deformation or compression, acting like a flowing or viscous material. Under a rapid load,
on the other hand, there is a less time for the snow to absorb the weight by changing its shapes, it is much more likely to crack under the strain and acts as a brittle or elastic material (Elmeson and Nastaev, 1973).
Winter avalanches are, in general, directly connected with sustained heavy snow• falls; air temperature and other meteorological factors seem to have only a secondary cause (Poggi and Plas, 1966; Perla, 1978). They are widespread and frequent in all mountains on slopes where the snow cover is sufficiently thick (over 30-40 cm) in winter.
1 After the snow is deposited the particle shapes are modified and dendritic (needle-like) crystals decompose into fragments. Simultaneously with the breakup of the dendritic assemblies of newly- deposited crystals is the formation of bonds at the point of contact between snow crystals (Langham, 1981). This process is known as sintering or age hardening which increases the strength of the snow (Adam, 1981).
12 2- Wet snow instability. Avalanches tend to occur during thaw caused by rain or heating.
The relatively high winter temperatures and heavy snowfall produce a deep snowpack
which is generally well settled and mechanically strong, except during and possibly after
periods of thawing. Wet slabs are triggered by the combined effects of water weight and
bed surface lubrication.
3- Rain and temperatures. Rain and high temperatures are also important triggers of
avalanching, reducing cohesion in the snowpack until failure occurs. The precise
significance of each is not clear since they frequently occur together and, in addition, the
rain often accompanies snowfalls. Rain or high temperatures followed by freezing can
create ice-crusts, which when buried by subsequent snowfalls may provide a significant
source of snowpack instability.
The rise in temperature weakens the bond between the crystals; and if the rise
continues it will eventually surround each crystal with a film of meltwater which
lubricates and reduces the static friction. The shear strength of the snow is then reduced to
zero, and in this state wet snow avalanches commonly occur. Rain brings about a rapid
rise in temperature and also provides free water for lubrication.
Strong temperature changes appear to be more important than absolute
temperature values in creating instability (Williams, 1981). When the snowpack is already
critically stressed, a large temperature change in a few hours can increase stress in the
topmost layers and lead to failure. Thus, temperature change can be viewed as the
ultimate trigger when large stress values already exist from other processes.
13 4- Snow drifting. Snow drifting is one of the most important factors of avalanche
formation in the mountains (Kotlyakov, 1966). Drifting causes snow to be re-distributed
and to be concentrated in certain sections of the slopes. The shearing forces exerted by
airflow against the snow surface erode the snow from regions of high wind stress. Eroded
snow is redeposited in sections of low wind stress that become the main snow-collecting
basins for avalanches.
5- Ski loads. A ski traverse across an unstable slab is often an effective way to trigger
instability. A strong downhill push is applied to the slab by the back of the skis to
reinforce fracturing.
6- Shocks. Examples of natural and artificial shock energy which cause avalanche release
are: earthquakes, cornice falls, artillery bursts, and sonic booms.
2.3 HYDROLOGICAL ROLE OF AVALANCHE
The ablation characteristics of avalanche snow differ significantly from the surrounding
undisturbed snow cover, particularly with respect to the undisturbed snow in the starting
zones (De Scally and Gardner, 1988). The changes of the climatic environment and of the
exposed area influence the ablation conditions in a complex manner (Martinec and
Quervain, 1971). Downslope transfer of snow from permanently frozen areas brings it into warmer climates, with a longer melting season. At a given altitude, it takes longer to
14 melt due to reduced surface area and deeper pack, but the flow from it is more concentrated and consistent (Hewitt, 1990b).
The differences in ablation are produced by two important changes which occur as the snow is avalanched to a lower elevation; the ambient air temperature is increased, and the snow is compacted and concentrated by wind, thermodynamic stress and stress from over burden (Wyman, 1995). The physical properties of avalanche snow is entirely different from the properties of fresh snow. The snow deposited in the runout zone is about two to three times denser than the starting-zone snow, and is much harder.
The albedo of avalanche snow is very low as compared to the fresh and undisturbed snow. Whereas, almost 80 to 90% of incident short-wave radiation is reflected by a clean, dry snow surface (Linsley et al, 1986). Albedo of snow surface keeps changing with the age of snow and also with the variation in free water content of the snowpack. As snow ages, its albedo drops to 50% or less because of changes in crystalline structure, density, and amount of dirt on the surface, which is further enhanced by avalanching. These changes in snow properties are the basic features of the avalanches.
Due to all of above factors the following three important behaviours of snowmelt are observed (Zalikhanov, 1975; Martinec, 1976; Perla and Martinelli, 1979; and Bell et al, 1990):
1- Snowmelt rates are increased due to increased ambient air temperature owing to
transport of the snow from higher to lower elevations.
15 2- Snowmelt is delayed as a result of the frequent confinement of avalanche snow,
increasing the snow density and reducing deposit surface area exposed to radiative
and turbulent energy exchanges.
3- The lower albedo of avalanche-snow, caused by entrained debris, increases the
absorption of solar energy which further enhance the ablation rates.
The decrease in area and increase in snowdepth are the most important factors, so that although the melt rate is higher, the melting continues for a longer time. The avalanche snow therefore represents water temporarily withdrawn from snowmelt runoff, producing a decrease in spring runoff but an augmentation of flows during the summer and autumn (Losssev, 1960; Iveronova, 1966; Sosedove and Seversky, 1966). The precise hydrological importance of avalanche snow is dependent on the changes in snowmelt factors between the starting zone and runout zone in addition to the magnitude of avalanches.
2.4 RESEARCH ON HYDROLOGICAL ASPECT OF AVALANCHES
Avalanches in mountain regions transport millions of cubic metres of snow to valley bottoms, and the extent to which the changes affect snowmelt runoff from a basin depends largely on the proportion of the basin's snow cover which is avalanched. A number of
Soviet researchers have reported a favorable effect of avalanches on the runoff regime.
They all showed a maximum ablation rate of snow avalanches which is followed by a prolonged period of gradual decrease. The delay in melting noted in Soviet literature
16 appears to result from the significant concentration of the snow during avalanching, producing a small runout zone.
Iveronova (1966) carried out regular observations (1956-1959) of the snow avalanches at the Tien-Shan station of the Tersky Alatau Ridge in USSR. She discovered that from 3 to 30% of the snow supply is being removed by avalanches. These figures vary strongly from year to year depending on the meteorological features of the winter and particulary on the amount of snow in winter, as well as the temperature conditions during summer. Iveronova found time delays of 2-3 months in melting of the avalanche deposits after the disappearance of the snow cover. The meltwater from the avalanche deposits amounted to 3-11%, of the annual runoff, whereas during the ablation period it amounted to 10 to 27% of the total runoff. This shows that the avalanches can play an essential role in the runoff of rivers.
Sosedove and Seversky (1966) also found a close relation of avalanching with amount of winter precipitation and air temperature regime in Zailiysky Alatau range of
USSR. They estimated 5 to 10% of the amount of snow in 1961 was transported by avalanches. During the snowy 1964 year this proportion was increased and amounted to
20-28%. The avalanche deposits yielded 3-4% and 10-11% of the annual runoff in 1961 and 1964 respectively. During thawing of the deposits their discharge was up to 10-20% of the total runoff, and 30-35% of the total surface runoff respectively.
Shcherbakov (1973) studied the activity of avalanches in Kirghizia, Tian-Shan. He calculated the modules of avalanche flow (1950 to 1953) for the whole of Kirghizia territory based on observations on 45 river basins. Shcherbakov found large regional differences in the degree of intensity of avalanche activity. On an average, about 88 x 106
17 m of snow traveled annually down the mountain slopes in the form of avalanches. This
corresponded approximately to 0.4% of the annual flow of Naryn River. The duration of
the avalanche period varied from 3 to 4 months in Western Tian-Shan to 6 to 8 months in
the northern and Interior Tian-Shan. February-March is the most active period, the
proportion of avalanches in theses months being 25 and 39% respectively. In his studies
of the role of avalanche in feeding mountain rivers and their flow, Shcherbakov indicated
the possibilities of controlling the activity of avalanches in modifying the hydrological
processes by artificially exploding the avalanches.
Zalikhanov (1975) studied the role of avalanches in the Caucasus, USSR.
According to his results, 30 - 64% of the snow accumulation can be transported by
avalanche to the valley bottoms of the Caucasus where the hydrological role of avalanches
is great. In the Kabardivian-Balkar Republic, for example, approximately 2,500
avalanches occurred from 1120 catchment area during the winter of 1962. These
avalanches brought 12.5 x 106 m3 of snow to the valley bottoms in alpine and sub-alpine
zones which resulted in longer period of snow melting. He obtained a relation between the
mean volume of avalanche snow and the catchment area of avalanches. This relation was:
V= 23(H)* [2.3]
where V is the mean volume in 1,000 m3 of avalanches and H is the catchment area of
avalanches in hectares. Similar relations were obtained for the other regions of the
Caucasus which were valid for certain ranges of the elevations.
18 Martinec and de Quervain (1975) presented a model of daily meltwater production and showed that avalanche activity can accelerate snowmelt if 'vertical fall' or 'runout area' of avalanche are large. They established various equations for calculating, i) water equivalent before and after the avalanche event; ii) total daily ablation; and iii) time of disappearing of snow in avalanched and non-avalanched areas. They found increased runoff in April and May by avalanches of the Dischma Valley, Davos and about 10% of area affected by avalanches, calculated by their model.
Martinec and de Quervain's model is unable to account for the rapid melting of that proportion of the deposit which is spread out and exposed, and the slow melting of the remaining protected portion, since it is assumed that the avalanche snow is uniformly distributed on entire area of runout zone and not concentrated in depressions. The field observations indicated that, in the real situation, the deposits on larger paths in normal years generally stay 2 to 3 months after the undisturbed snow cover has melted from the starting zones (de Scally, 1988).
Schearer (1984, 1988) developed equation for estimating the annual avalanche
masses Ma (in tones) by;
M=/A5a [2.4] where / is the yield coefficient or proportion of snow on the avalanche path which avalanches (t m"3), A is the area of the catchment (starting and track zones of the path, in
2 m ), and 5a is the amount of precipitation (water equivalence of snowfall and rain into snow). He found the area of the catchment (A) and the total avalanche-season
precipitation (Sa) to be the most significant determination of avalanche mass. The yield
19 coefficient if) averaged 0.1124 m" at Rogers Pass for all avalanche paths and years, but varies widely (f = .09091 m~3) from year to year and path to path.
Kunhar River Basin
De Scally (1992) investigated the effect of avalanches on snowmelt runoff in the Kunhar basin. The results of modeling based on field measurements show that, of the two main changes occurring during avalanching which affect the subsequent generation of snowmelt runoff - concentration of the avalanche snow and an ambient temperature increase resulting from the avalanches' fall to a lower elevation. As a result, very high rates of surficial melting on avalanche snow were over weighed by the small surface area of the deposits, decreasing the rate of meltwater production and delaying the disappearance of avalanche snow compared with undisturbed snow. Ablation begins earlier in the runout zones but the rate of meltwater production remains low. He found that in year 1986-87, undisturbed snow in the runout zone (i.e. without avalanching) disappears much sooner than undisturbed snow in the starting zone, as a result of greater winter precipitation and lower temperatures in the starting zone. The results from his equations give a delay of 163 days on average in ablation, if all the snow in the starting zone is assumed to avalanche into the runout zone where it is added to the existing undisturbed snow cover. When the volumes of the actual avalanche deposits are added to the existing undisturbed snow in the runout zone the delay reduced to 57 days on average, which is still high.
20 De Scally studied avalanche activities within only 288 km2 (14% of the Kunhar
Basin) and he extrapolated his results to larger area (1372 km2, i.e. 58% of the total
watershed area) potentially affected by avalanching. His extrapolation of results give an
estimation of total volume of avalanche snow (water equivalent) as 212 x 106 m3 (1986)
and 248 x 106 m3 (1987). These represent on average 7.8% of the April to September
runoff and 6.6% of the annual runoff of the Kunhar in these two years. He estimated the
percentages of potential avalanche slope areas to be 15% in moderate winters and 54% in
snowy winters. Following a severe winter (i.e. all potential avalanche slopes are active)
avalanche snow produced 8% of snowmelt runoff and 5-7% of annual runoff. Following a
normal winter theses proportions were estimated to be of the order of 1-2%.
De Scally and Gardner (1989) estimated annual avalanche masses in the Kunhar
basin. The predicted masses were compared to measured avalanche-deposit masses
produced during two winters (1985-86, 1986-87). The predictive equation was based on
data from western Canada (Schearer, 1988). The calculated and measured deposit masses
at the end of each avalanche season showed a wide range. The estimation of annual
avalanche masses using Schearer's equation was successful on the largest paths but
significant differences from the measured masses occurred on the other paths. De Scally
and Gardner attribute these variations to the inaccuracies in the measurement of the
catchment area A and winter precipitation Sa parameters in the equation [2.2].
The prediction of avalanche mass is, however, difficult and uncertain, particularly
in poorly known mountain region such as the Himalayas (De Scally and Gardner, 1989).
From a hydrological point of view, the equation may be useful because it is the large
21 paths which are most important in terms of the area of snow cover affected and thus the amount of water stored in the avalanche deposits at the beginning of the ablation season.
Ablation of avalanched and undisturbed snow was studied by De Scally and
Gardner (1990). They found a much more rapid rate of ablation than in other mountain regions. The higher ablation rate of avalanche snow resulted from its low elevation and high temperature environment. On the other hand this higher amount of ablation rates is frequently offset by reduced surface avalanche area and greater depths, which delay their total melt as compared with undisturbed snow. Air temperature was found to be strongly correlated with snowmelt but they further suggested that energy balance is required to explain this relationship.
De Scally and Gardner (1994) found strong association between snowfalls and avalanche events, with 64% of the 196 events recorded occurring during or within 24 hours of a snowfall. The percentage of snow cover which is transported by avalanches was estimated, on average, to be about 10%.
2.4.1 Summary
Most of the literature on avalanching describe their role as a hazard to property & life and geographical point of view and a little work is done on the basis of their hydrological role.
Some of the Soviet literature refers to the hydrological role of avalanches and all show a direct and favourable influence of avalanches on the river systems. Millions of cubic meters of high mountain snow is transported and deposited from higher to lower elevation
22 each year in the form of avalanches, where they melt faster but for a longer period of time and in a consistent manner due to high degree of compaction and larger depths.
Avalanches play a significant role not only in the feeding of mountain glaciers and rivers but also in the formation of the relief, denudation of the mountain slopes, and growth of vegetation (Iveronova, 1961; Tushinskii, 1963; and Peev, 1965). In mountain slopes, avalanches are directly correlated with the high altitude snowfall intensity and
i magnitude.
23 Chapter 3
GEOGRAPHY AND CLIMATE
3.1 OVERVIEW
Pakistan is an arid to semi-arid region with surface waters derived mainly from the River
Indus and its tributaries. It has an extensive network of irrigation canals, the largest in the world (Abbas, 1967). This irrigation system in the Indus plains is fed through 16 diversion dams and 580 km of inter-river link canals and three major storage reservoirs at
Mangla, Tarbela and Chashma (Tarar, 1982). The Indus River is Pakistan's main source of water for irrigation, power generation, and water supply for urban and industrial units.
The lives of about 125 million people depend on the waters of this huge river system
(Kick, 1978).
The Indus River, with its 860,000 km2 drainage basin and 2,880 km length, is one of the largest of South Asian region. The river and many of its tributaries originate in the
Himalayan, Karakoram and Hindu Kush regions. Its drainage basin covers almost the whole of Pakistan as well as parts of North India, Eastern China and Afghanistan.
The northern mountains of Pakistan provide the only areas of the country with substantial precipitation and an annual moisture surplus. Due to the high elevation, most of the precipitation occurs in the domain of snow and ice and therefore has resulted in extensive glaciation. Most of the remainder of the Indus Basin has low precipitation and a
24 water deficit in most or all of the months of the year. Historically, surface water supply in
Pakistan depended more upon the easterly Indus streams and rainfall; both mainly derived from the summer monsoon.
With the independence and partitioning of India in 1947, most of the Indus Valley became the territory of Pakistan. The international boundary between India and Pakistan cut the irrigation system of the Bari Doab and Sutlej Valley projects, originally designed as one scheme, into two parts. The Indus Water Treaty, signed in 1960, is the basis of water sharing between the two countries. The treaty gives India full control of the eastern tributary streams, Ravi, Beas, and Sutlej; and allows Pakistan to utilize exclusively the flow of the Indus and most of the waters of the Chenab and Jhelum. India has use of the
Chenab and Jhelum for needs within Jammu and Kashmir.
Pakistan, through this treaty, has become increasingly dependent on water from the snow and ice sources in the northern mountains which fed the Tarbela and Mangla
Reservoirs. Just at the time of greatest water need in summer, the supply of meltwater is most plentiful. In recent decades, mainly under the direction of the WAPDA, huge projects have been undertaken to harness the waters of the Indus.
25 3.2 GEOGRAPHY OF THE NORTHERN MOUNTAINS
3.2.1 Regional Setting
The Himalayan region has been considered to encompass the mountain area from the
Pamir region adjoining the Karakoram-Hindukush-Zaskar ranges in the West-northwest, the Tibetan plateau at the center bordered by the Kunlun Shan in the North and the Heng
Tuan in the East and by the Great Himalayan range in the South. The Karakoram
Mountains are situated in the interior of Central Asia. The Karakoram consists of a series of mountain ranges that extends over 2,500 km from the eastern Ladakh to the Hindu
Kush. They are bordered by the Great Himalayan to the south and southwest, the Aghil
Ranges and Kun Lun to the north and northeast, the Pamirs to the northwest, and the
Hindu Kush to the west (see Fig. 3.1).
3.2.2 The Upper Indus Basin
The Upper Indus Basin (U.I.B.) serves as catchment area drained by the Indus River upstream of Tarbela reservoir, the Jhelum River upstream of Mangla reservoir and the
Kabul River, and comprises an area of approximately 250,000 km2 in a mountainous region of the Western Himalayan, Karakoram and Hindu Kush Ranges. Thirteen percent of it is covered by perennial snow and ice. At the end of the winter season, an area of about 200,000 km2 in the mountainous regions of the U.I.B. is extensively snow-covered.
For the
26
Indus main stem, snow may cover more than 90% of the catchment above Tarbela Dam, and commonly more than 70% (Hewitt, 1985a).
The Indus, extends nearly 3,000 km, rises in the southwestern part of the Tibetan
Plateau and flows to the north of the Vale of Kashmir in arid valleys between the
Himalayan and Karakoram mountain ranges. From its origin in Tibet to its terminus in the
Arabian Sea, the Indus drains a total catchment of 933,632 km2 (Ringenoldus, 1975). The river descends south towards the Arabian Sea with a ten year average volume of 38.7 billion m3 discharge past Tarbela dam. The Upper Indus Basin includes the areas upstream of Mangla Dam on the Jhelum River and Kabul River at its mouth.
In the north-western Himalayan, the ranges on both sides of the Indus River are aligned from west-north-west to east-south-east. The crest of the Ladakh ranges about
5400-5700m a.s.l. (Burbank and Fort, 1985). The parallel ridge crest of the north eastern flank of the Zaskar Range rises to similar heights on the south-western side of the Indus.
The sources of the Indus and its tributaries lie high up in the Himalayan,
Karakoram and Hindu Kush mountains. Most of the moisture comes from the river's
Himalayan headwaters. The easterly tributaries of the Indus are fed mainly by monsoonal rains; the westerly ones by snow and ice meltwaters. Meltwater from Himalayan snow and ice at elevations ranging from 2,000 m to 5,000 m dominates the flow of these western rivers (Quick, 1990), and is the largest fraction of annual yields, supplying 75% of the inflow of the Kabul at Warsak; 80% of Swat River; about 70-80% of the inflow of the main Indus to Tarbela Reservoir (Hewitt, 1988a and Wake, 1987); and 65% of the
Jhelum at Mangla Reservoir (Hewitt, 1985b and 1988b). The snow and ice melt
28 streamflow is important because it occurs in the period April to June, before the monsoon rains when it is needed for irrigation and power generation (Quick, 1990).
3.3 CLIMATE
This discussion follows the summary of broad climatic control and local climate mainly described by Barry (1981), Barry and Chorley (1976), Boucher (1975), Hewitt (1988a and
1988b), Lockwood (1974) and Young (1981).
3.3.1 Broad Climatic Controls
The climate of Pakistan is classified as arid or semi-arid. Dense natural forests are present only in some areas at higher altitudes where rainfall is heavy or at a location in close proximity to rivers.
The monsoon of Asia are caused primarily by the differential response of land and sea to incoming solar radiation. In the winter months, the Asiatic land mass gets much colder than the adjoining seas and north-east Asia, and becomes the center of an intense high-pressure system. This Sub Tropical High Pressure (STHP) over Asia extends from
Siberia to the outer fringes of the Himalayan massif.
There are four distinct climatic seasons;
1. Winter Season (December - March)
2. Summer Season (April - June)
3. Monsoon Season (July - September) and
29 4. Post-monsoon Season (October - November).
Precipitation in Pakistan falls during two distinct periods. The first of these seasons is from July to September. Rain during this period is the major component and related mainly to the monsoon depressions. The second rainfall season, originating from winter cyclonic storms from the west, occurs from December to March. The region is dominated by the influx of westerly air masses.
During the winter season, the westerly jet stream lies over southern Asia, with its core located at about 12 km altitude (according to the WMO, any speed exceeding 30 m/s may be called a jet stream, Lydelph, 1985). It splits into two currents in the region of the
Tibetan Plateau around the high mountains, the stronger branch flowing east- southeastward down the Ganges Plain, and the other curving northward and eastward through north China and Mangolia. The two branches reunite again to the east of the plateau and form an immense upper convergence zone over China. These two branches have been attributed to the disruptive effect of the topographic barrier on the airflow, but the northern jet may be located far from the Tibetan Plateau.
This subtropical westerly jet stream steers depressions towards the Karakoram and
Northern India. These lows, which are not usually frontal, appear to penetrate across the
Middle East from the Mediterranean. On an average six to seven western disturbances move across the Himalayan region every month in winter and are an important source of precipitation for the Karakoram and northern India. Over most of the northern mountains, snowfall produces extensive winter snowpacks. These occur above 2,000 m a.s.L, and the great bulk of the moisture is found above 3,000 m a.s.l.
30 During the summer season, the belt of STHP over Central Asia begins to weaken
and temperatures begin to rise rapidly. The land mass of Asia becomes much hotter than the sea areas to the east and south. In March and April western disturbances still occur with a frequency of three or four per month and they are less severe. A strong low- pressure area forms over Pakistan, Afghanistan, and Iran; while high pressure areas build
up over the north Pacific and south Indian oceans. Pakistan lies outside of the tropics and is dominated by shallow incursions of moist monsoon air in summer. The moisture-laden
winds from the oceanic areas move towards the heart of Asia from the south.
In May and June the subtropical jet stream over northern India slowly weakens
and disintegrates, causing the main westerly flow to move north into Central Asia.
Pakistan experiences its highest temperature during these months and rainfall amounts are
generally small. From early July or so there may be a significant influx of monsoonal
rains, especially in the more southerly and westerly basins.
The monsoon season is characterized by the occurrence of general south-west
monsoon current. The Himalayan mountains provide the necessary conditions for the
deflection of the eastern branches of the monsoons in a north-westerly direction along the
Ganges Valley. This can result in the incursion of monsoonal air masses into the
Karakoram, resulting in heavy precipitation at higher altitudes.
Floods are most severe during the monsoon period, when depressions move inland from the Bay of Bengal and occasionally from the Arabian Sea. These depressions move in a northwesterly direction until they reach Rajasthan where some of them recurve
northwards and cause heavy rainfall in the catchment areas of the Indus Basin.
31 In October and May the track of winds lies over the north of Afghanistan and the western Himalayan and Karakoram, but from December to April it shifts further south.
The period from mid-September to mid-November is a transitory one during which monsoon conditions gradually change over to conditions characteristic of the winter season. The northerly winds, which have crossed the great deserts of Central Asia, are free of moisture. October and November are the driest months of the year over the entire
Indus Plains. To the south the high ranges act as a rain curtain, so that little rain reaches the Main Karakoram Chain.
3.3.2 Local Climate
Since the Kunhar Valley is located in the north-western part of the Himalayan, the following discussion is mostly limited to the climate of that part.
The Himalayas have a climatic regime which is characterized by extremes, whether of altitude, aspect, localized relief, or variability in climate, particularly precipitation (Chalise, 1993). The large variation in precipitation at high elevations emphasizes the hydrological importance of these high mountain regions which play a significant role in snowmelt and glacier melt runoff (Singh and Quick, 1993). Intense solar radiation, strong winds, and a great range of temperature are characteristic climatic features of the region. With clearer skies and the apparent northward movement of the sun, the mountain slopes receive greater amounts of solar radiation. The climate of the
Himalayan may be said to consist of four broad and contrasting regions (Mani, 1981).
- the rain forest of the East, ranging in altitude to 2,000 m;
32 - the wet alpine zone above the tree-line rising to 6,000 m or more;
- transitional semi-wet region in the central portion of the mountains; and
- an arid region in the Hindu Kush far to the West.
The Eastern Himalayas has a prolonged monsoon season from June to October, with very little rain from Western disturbances in winter. On the other hand, the Western
Himalayan has a short monsoon from July to August and fairly long wet season from
November to April. The premonsoon season, March to May, is quite dry over the
Western Himalayan except for very occasional thunderstorms. The arrival of monsoon in the West is quite sudden, with an abrupt change in cloudiness, temperature, humidity, wind and rainfall (Bahadur, 1993). The passage of western or extra tropical disturbances moving from west to east during November to April bring relatively more winter precipitation to the Western Himalayas (Thapa, 1993).
During winter the snow accumulates around the Himalayan high peaks and the snowline comes down to about 1,500 m in the Western Himalayas, whereas it is at 3,000 m or above in the Eastern Himalayas. The general direction of the wind over the
Himalayas in winter is from Northeast to Northwest. At higher altitudes, the winds are westerly, averaging about 120 km per hour reaching 160 km per hour or more. At low altitudes, the wind speeds are about 50 to 60 km per hour.
In general, there is very heavy and prolonged precipitation in the Eastern region, rapidly decreasing in both intensity and duration as the monsoon advances westward.
Over the western and northern parts of the Himalayas, the skies are generally clear but the visibility is poor due to dust haze which may extend up to 6 km or more.
33 The seasonal snow-line descends to an altitude of 1,500 m in the Western part of the Himalayas by February in the winter. As the snowmelt commences by the month of
March, the snowline starts receding upwards and by the middle of the June it moves up to
an altitude of 4,500 m (Anand and Prasad, 1993).
Altitude and topography play a significant role in controlling the local climate of the Himalayas extending East to West (Ueno et al., 1993). Both of these factors are major controls on the amount of precipitation and insolation received. Altitude affects both the intensity and the mechanisms of melting of available snow and ice, and the rate of snow
accumulation. Snow accumulation generally increases with altitude
The effect of the orientation of surfaces is due to the differences in the solar
radiation received by slopes. The north facing and south facing slopes commonly have quite different radiation regimes, producing marked differences in local hydrology
(Young and Hewitt, 1988). Altitudinal differences in precipitation are much more marked in the northern than in the southern slopes: There are also marked differences in the
aridity on southerly as against northerly slopes. The transient snowlines are often 700 -
1,000 m higher, and glacial coverage less, on south facing slopes of the same range.
On southerly slopes the surfaces are almost perpendicular to the sun for much of year, while direct sunshine is excluded from many northerly slopes except in mid
summer, therefore air temperature or sensible heat is decisive in melting on northerly
slopes. The sublimation and evaporation losses are much larger in southerly-oriented basins.
34 3.4 STUDY AREA
The River Jhelum originates in Indian Kashmir and flows into Pakistan (Fig. 3.2). The two main headwater tributaries in Pakistan are the Kunhar River and Neelum River
(Kishanganga), both on the south slope of the Northwestern Himalayan and supplying on average 11% and 39% of the total annual discharge of the Jhelum respectively. Only the
Kunhar basin (2,340 km2) is entirely in Pakistan and therefore is vitally important from the point of view of hydrological monitoring by the Water and Power Development
Authority.
The Valley transacts the front ranges of the Himalayan to link the foothills of
North-West Frontier Province with the mountainous Northern Areas in Pakistan. It is located on the south slope of the Himalayan where it is exposed to both winter westerlies and the summer monsoon. The elevation of the Kunhar basin ranges from 800 to 5,300 m, with maximum valley-slope relief on the order of 2,500 m. In the vicinity of Naran village, where avalanche activity is most intense, the vegetation consists primarily of coniferous and deciduous forests up to treeline at approximately 3,600 m elevation (De
Scally and Gardner, 1994). Above this, most areas have a cover of stunted juniper, alpine scrub, meadow and bare rock.
The climate of Kunhar Valley is relatively humid, in contrast to the extremely arid valleys to the north and northeast. A winter snow cover develops above 1,800 m elevation from early November onwards, reaching a maximum depth in march or April. Snow water storage is greatly dependent on elevation, with the annual maximum in the Naran area showing an average increase from 390 mm at 2460 m to 1,010 mm at 3,220 m
35 Fig. 3.2. Kuhar River basin.
36 (WAPDA, 1969) and is produced by disturbances associated with a high level westerly air-stream. Snowfall and snow water storage also increase significantly in a northeasterly direction toward Babusar Pass (De Scally, 1989). At elevation above 1,500 m most of the winter precipitation falls as snow, although heavy rain often accompanies the snowfalls in the valley bottoms. Between storms the skies are generally clear and very large diurnal variations in air temperature are common. January is the coldest month with a mean daily temperature of -2.1 C at Battakundi, with December and February also having below- freezing mean temperatures (De Scally, 1989).
A winter snowcover develops in Kunhar Valley above approximately 1,500 m elevation. Snowcover water equivalents reach a maximum in February or March (April at higher elevations) and range from 990 to 2,650 mm in the Naran area. The large range reflects a sharp increase in snow depth with elevation and distance up-valley.
Snowmelt begins in early April and lasts into July or even later at high elevations where it overlaps with the Indian monsoon period. Temperatures are generally high during the snowmelt period, averaging 13.9° C between May and October at Battakundi
(De Scally, 1992). The amount of monsoon precipitation (mostly rain) decreases markedly to the northeast, with the average May to October totals decreasing from 870 mm at Balakot to 200 mm at Battakundi (De Scally, 1989). The annual flow at Garhi
Habibullah near the mouth of the basin is on average 3.33 x 109 m3.
37 3.4.1 Avalanche Activity in Kunhar Basin
The Kunhar basin experiences intense avalanche activity above 1,850 m elevation which results in a major area-altitude redistribution of winter snow cover. About 70% of the Kunhar Basin lies between elevations 2,500 and 4,500 m. A winter snowpack begins
to accumulate above 1,500 - 2,000 m in early November and reaches a maximum depth in
March or April (De Scally, 1994). Snow accumulation increases markedly with elevation
and towards the northeast direction of Babusar Pass. Snowmelt begins in early April and lasts into July or even later at high elevations. Temperatures are generally high during the
snowmelt period, averaging 14°C between May and October at Battakundi.
The combination of an extremely rugged topography, heavy winter snowfall, and
deforested slopes results in intense avalanche activity up-valley of Kaghan village.
Individual avalanche deposits can represent in excess of 106 m3 of water storage (De
Scally and Gardner, 1989) and frequently persist into late summer and autumn on the larger avalanche paths. Large-scale avalanche activity in Kunhar Valley begins
approximately 50 km up-valley of Balakot and continues to Babusar Pass. The avalanche
season begins in early December and generally ends by mid-May except at higher
elevations, where melt-triggered avalanches continue into June. Following a winter of
maximum avalanche activity, the avalanche snow represents about 10 and 7 % of the
snowmelt runoff and annual runoff, respectively. In normal years these proportions are of
the order of 2 - 3% (De Scally, 1992).
38 The size of avalanche paths in the valley is highly variable with vertical falls
2 ranging from 100 to as much as 2000 meters and starting zone areas in excess of 5 km
(De Scally and Gardner, 1986b). The largest paths in the trunk valley, from Kaghan
village to a point 5 km up-valley, are located on east-facing slopes. Up-valley from this
point they are situated almost exclusively on west- and northwest-facing slopes.
Generally, slopes in the upper valley have greater snow accumulations in the avalanche
starting zones and hence the potential for larger avalanches and greater avalanche travel
distances.
Large avalanches paths in Kaghan Valley are associated with a variety of other
types of geomorphological activity, including streamflow, mudflows, landslides and
debris torrents. The starting zones are generally well above treeline and consist of
numerous sub-basins for each path. The tracks are confined by deep gullies, which are
often broken by cliffs. The runouts generally are on debris fans which have been formed
by either floods or debris torrents. Because the Kaghan valley is very narrow in the
middle reach, many of the large paths extend across its floor and up the opposite slope.
Large avalanches frequently dam the river for up to two days.
As a result of the very large size of many avalanche paths in Kaghan Valley, the
magnitude of individual releases can be very high, exceeding the upper limit (size 5) of
the Canadian snow avalanche size classification (De Scally and Gardner, 1986a). Size 5
releases in this system are described as the "largest snow avalanches known; could destroy
a village or a forest of 40 hectares", and have typical masses of 100,000 tones and impact
pressures of 1000 kpa (McClung and Schaerer, 1981).
39 According to De Scally (1992) 63% of the total mapped area in the main valley is affected by active or potential avalanching whereas in the Saiful Maluk this figure is 73%, with an overall average of 65%. This represents a maximum percentage, since all the avalanche paths are assumed to be active in severe winter. The normal avalanche affected area is around 15% on average. Percentages of 12 - 15% in moderate winters and 42 -
43% in snowy winters are reported from the northern Tien Shan Mountains (Sosedov and
Seversky, 1966).
Small avalanche paths in Kaghan Valley are generally situated on unconfined slopes or in shallow basins. The starting zones are generally below treeline and covered with a mixture of bare rock, brush and immature conifers. The track zones are rarely deeply gullied or markedly confined. Runouts usually extend only a short distance onto the valley floor or terminate in the Kunhar River.
The total winter (November-April) precipitation in Naran (2,400 m) averages
1,080 mm and is mostly in the form of snow (Wapda, 1969). Most avalanches are produced by loading from intense snowfalls and by rain which often accompanies the winter storms (de Scally and Garner, 1986). The north-west aspect represents the most active orientation with respect to avalanche activity and track gradients range from 22° to
34°.
40 Chapter 4
METHOD OF ANALYSIS
4.1 HYDROLOGICAL MODEL
The UBC watershed model has been used in this study. The model was developed by
Quick & Pipes (1977) primarily for mountainous watersheds and calculates the total contribution from both snowmelt and rainfall runoff. The structure of the model is based
on hydrological behaviour as a function of elevation in the watershed. In this study the model was adopted to simulate the redistribution of snowpack caused by avalanching. The maximum and minimum temperatures, and precipitation data are input to the model.
Daily watershed streamflow is the main output of the model along with information on
snowpack water equivalent, snow covered area, current soil moisture status, and
groundwater storage of the watershed. The watershed is divided into area-elevation bands, therefore the above mentioned information can also be obtained for each band separately.
Each elevation band has a separate variable description of the watershed, such as the forested fraction, impermeable fraction, glacierized areas and aspect (see Appendix). The complete structure of the model is discussed by Quick and Pipes (1977), however, some important factors which will be used extensively in the present study are described as follows.
41 Temperature Lapse Rate:
Two lapse rates are specified in the model, one for the maximum temperature and one for the minimum temperature. The lapse rate is calculated for each day using the daily temperature range as an index. In the model, the maximum temperature lapse rate
(TXLAPS) and minimum temperature lapse rate (TNLAPS) are calculated as follows;
TXLAPS = TZLAPS + ( TLXM - TZLAPS ) * TD / AOTERM [4.1]
TNLAPS = TZLAPS - ( TZLAPS - TLNM ) * TD / AOTERM [4.2]
where, TD = daily temperature range (TX - TN), and
TZLAPS = TZ-(PP/PPM) *(TZ-TZP) [4.3] for the above,
PP = daily precipitation
TLXM = 10° C/km
TLMN = 0.5° C /km
TZ = 6.4° C/km (a reference lapse rate for rain-free conditions)
TZP = 3.2° C/km (a reference lapse rate when PP > PPM)
PPM = 5 mm/day
and AOTERM equals the maximum temperature range under open sky conditions
(selected from the data set TX-TN).
42 Form of Precipitation:
The model builds up the snowpack in each band of the watershed using the precipitation
gradients that are determined during the calibration of the model. In the model,
temperatures are used to decide whether the precipitation is in the form of snow or rain;
T = T" / AOFORM [4.4]
and,
T* = Temp + POTASR [4.5]
where, AOFORM is temperature above which all precipitation is rain, model uses the
default value given as 2°C, and POTASR is a increment to the recorded Temp, before
determining the form of precipitation. Now there are three conditions;
i) if T < 0°C
the model will consider all precipitation as snow and POSREP will be activated,
ii) if T > AOFORM
then all precipitation will be rain by using PORREP only,
iii) if A0FORM>T>0°C
then a linear interpolation is used for the form of precipitation between 0 and 1°C, defined
byTl;
Tl = T * PORREP + ( 1 - T ) * POSREP [4.6]
43 POSREP and PORREP are the precipitation adjustment factors at stations for snow and
rain respectively, since precipitation measured at stations may not always be
representative of the areal distribution of precipitation.
Precipitation Adjustment Factor:
Precipitation adjustment factor POPADJ is used in avalanching by increasing or
decreasing the amount of precipitation in a particular band
The modified precipitation PPT" is given by;
PPT" = PPT* * ( 1 + POPADJ ) [4.7]
where,
PPT*=PPB*(1+T1) [4.8]
where, PPB is the precipitation at a particular band obtained from AES station.
Precipitation Elevation Gradients:
The precipitation in any elevation band is calculated from the precipitation in the band
immediately above or below using the equation
T% *-». / + \Aelev/100 w-A
P P 1+C UL+I = UL*( O [4.9]
where, PIJL+1 = precipitation from meteorological station I for day J and elevation band L,
44 a = precipitation gradient (%), and
Aelev = difference in elevations between the AES stations and the band.
The 1 + a multiplier produce a logarithmic increase in precipitation with elevation. The enhancement factor a is separately defined above and below a certain elevation which may be specified by EOLMID. Three precipitation gradients are used;
POGRADL, for bands below EOLMID; POGRADM, for bands between EOLMID and
EOLHI; and POGRADU, for bands above EOLHI.
Snowmelt
The Watershed Model accumulates precipitation falling as snow and then depletes these
snowpacks according to the calculated melt rate. This snow accumulation and depletion is carried out separately in each area-elevation band. The final depletion of snow is a
gradual recession of the snowline from the bottom to the top of a band. The model redistributes the snowpack in such a way that there is more snow in the higher part of the band and less in the lower part of the band. The snowpack is then melted in such a way that less and less band area is covered by snow.
The UBC Watershed Model uses an energy budget approach to calculate
snowmelt. The energy exchange at the surface of a snowpack is made up of four major components:
1. The short-wave radiation exchange, which consists of incoming solar radiation,
and the reflected outgoing short-wave radiation. This short-wave component
depends on the time of year, the site exposure, cloud cover and snow albedo.
45 2. The longwave radiation exchange depends on black body radiation from the
overlying air mass.
3. Convective heat transfer is produced by turbulent heat exchange between the air
mass immediately above the snowpack. This heat transfer is dependent on wind
and air temperature above the snowpack. This component is self limiting and
becomes quite small at higher air temperatures, unless there is a very strong wind.
4. Advective heat transfer or condensation melt, is caused by the transport of
moisture to and from the snowpack. This depends on the relative vapor pressures
of the air and snow surface.
A simple set of equations which expresses the various snowmelt components in terms of millimeters of snowpack per day is given and discussed by Quick (1995).
Model Efficiencies
The model calibration is measured by the Nash-Sutcliffe efficiency, E!, and Coefficient of
2 determination, R . The E! relates how well the estimated hydrographs compare in shape and volume to the observed hydrographs, and is calculated as follows:
HiQobsi-Qesti)2 E\ = l-^ [4.10] 1L(QobSi-Qobs)2
46 where,
1 n Qobs=-J^Qobs [4.11] ni=l n = the number of days for daily runs or hours for hourly runs
Qobsi = the observed flow on day (hour) i
Qest{ = the estimated (calculated) flow on day (hour) i
The coefficient of determination, R, is a factor that relates how well the shape of the estimated hydrographs correspond to the shapes of the observed hydrographs, and it is independent of volume. However, timing does affect the results of this statistic. The coefficient of determination is calculated as follows:
2 *Z{Qobsi-(b*Qesti + a)}
R2 = 1_iEL [4.12]
1L(Qobsi-Qobs)2
where,
1 a = -\ ^Qobs} - b^Qesti [4.13] n U=l i=l
47 H(QobSi)(Qesti) - - ^Qest^Qobsi
n i=l i=l i=l [4.14]
2 T,{Qesti) --\YjQesti\ 1=1 n U=l
In optimization process, the best parameter values are selected on the best efficiency defined as Eopt. This efficiency is slightly less than E!, but gives lesser
deviation in observed and estimated flows. The Eopt is calculated as follows;
Qest Eopt = -\l- + E\ [4.15] Qobs
4.2 CALIBRATION PROCEDURE
The UBC Watershed Model was calibrated using nine years (1979-1988) of flow and meteorological records. The flow was measured at Garhi Habibullah, and the meteorological data was received from two stations, i.e., Battakundi and Astore. The calibration was carried out in three stages. In a first stage, the model was calibrated by using the semi-automatic optimization method of the model without taking into account the influence of avalanching. The important parameters calibrated in this stage were; precipitation adjustment factors (POSREP and PORREP), precipitation gradient factors
(POGRADL, POGRADM and POGRADU), ground water percolation (POPERC), deep
48 zone share (PODZSH), fast and interflow runoff time constants for rain and snow & glacier melt (POFRTK, POFSTK, POGLTK, POFRTK, and POFSTK), time constants for upper and deep ground waters (POUGTK and PODZTK), and impermeable area modification factor (POAGEN), etc.
The results of this calibration showed that the simulated flow deviated
significantly from the observed hydrograph. For this original calibration the total volume
deviated by about -4% of the observed volume and the Nash-Sutcliffe coefficient of
efficiency was about 63%.
The above deviation from the observed data was mainly because of the non-
representativeness of the precipitation measured at the Battakundi station. For this reason,
in second stage, the precipitation representative factor POSREP was adjusted for
individual years from 1979 to 1988 and a significant improvement of the results was
achieved. For this second stage the volume deviation was reduced to -1.9% and the
coefficient of efficiency increased to about 78% for the years 1979-1988. This was the
'base line calibration' (BLC) of stage one of the calibration, so that the results achieved
by producing avalanches in the watershed can be compared with the results of this basic
calibration.
Even though the results of the second stage were improved, the main problem was
the time distribution of the flow. This was done in third and final stage of calibration, the
role of avalanches was introduced in the Kunhar river basin (see section 4.5 for details).
The U.B.C. watershed model was calibrated again using POPADJ, for each avalanche
affected area of each band.
49 4.3 CLIMATIC DATA
The climatic data used in this calibration is from two meteorological stations, i.e. at
Battakundi (elevation 2,660 m above sea level) and at Astore (elevation 2,630 m), and the
data was for a nine year period, 1979-88 (the station locations are shown in Figs. 3.1 &
3.2). In the calibration process, initially the Battakundi data (both precipitation and temperatures) was used for entire watershed but later the Astore temperature data was found to give better results.
Battakundi temperature data gave an under estimation in flow in the period from
February to early April. Whereas using Astore temperature data, this under estimation was controlled and furthermore, the peak flow response was found better as compared to the results of calibration with Battakundi temperature data only.
Detailed research on Mangla Watershed calibration was done by Quick and Pipes
(1989). They found that Battakundi temperature data is not always representative of the temperature conditions in the total Kunhar Watershed, especially because Battakundi is located in a deep valley which is subject to considerable cloud cover. It was found that
Astore temperature data is a much better indicator of Kunhar temperature conditions, except when rainfall is occurring. Quick and Pipes (1979) and Quick (1990) suggested to use Astore temperature data for elevations above 2,500 m.
Therefore, precipitation data was used from Battakundi for the entire watershed, whereas the Astore temperature data was used to calculate snowmelt and glacier melt.
Astore temperature data gave better results for whole watershed except at elevation 4,000 m (band 6), which has the largest band area (i.e. 600 km2, about 26% of whole
50 watershed). When Astore temperature data was used for elevation 4,000 m, an over
estimation was found in late winter, i.e., February- March and early melting season, i.e.,
April-May for some years. At present the database is very limited and these problems
with representing the precipitation & temperature indicates a need for more data stations
in the Kunhar especially at high elevations.
Astore is located to the north of the Kunhar watershed, and on the north slope of
the Himalayan range. Consequently, although Astore temperature appeared reasonably
reliable and representative (Quick and Pipes, 1989), Astore precipitation is much less than
Battakundi, because Astore lies in a dry subsidence zone. For this reason, Battakundi data
is used for all precipitation, both rain and snow. By using Astore temperatures, the
estimated cloud cover is greatly reduced, so that the snowmelt is governed by open sky
(full short wave radiation) whereas, Battakundi temperatures would indicate cloud cover
in the Kunhar valley at times when the river response was more consistent with clear sky
melt. In the model, above 3,500 m, the snowmelt is calculated assuming an exposed open
area, which is a reasonable approximation for these upper valley regions from which
some snow avalanches to the valley floors. Below 3,500 m, the snowpacks are assumed to
be more shaded by the steep valley slopes and by high mountain barriers.
4.4 PRECIPITATION ADJUSTMENTS
Since the highest elevation of the watershed is above 4,600 m, winter precipitation especially the snowfall recorded at Battakundi is not fully representative for higher elevations. There are some years in which Battakundi precipitation is quite representative
51 for the whole watershed; in this study about 4 years out of 9. But there are years when
Battakundi is not fully representative and therefore for these years the snow representative factor, POSREP, had to be adjusted accordingly.
The station recorded less snow fall and because of station elevation much of the precipitation is in the form of rain, therefore, a significant underestimation of snowmelt was found. At higher elevations, the average temperature during winter is below 0°°C and a large portion of precipitation is in the form of snowfall. Therefore, the Battakundi precipitation data were adjusted during the first-stage calibration by adjusting POSREP
(adjustment to precipitation when average temperature < 0, i.e. snowfall) and PORREP
(adjustment to precipitation when average temperature > A0FORM, i.e. rain), where
A0FORM is temperature above which all precipitation is rain. In the first stage calibration the PORREP was kept as -0.8 during optimization and for POSREP the range was given 0 to 0.5.
The hydrographs of this calibration show an underestimation of discharge which is variable from year to year. Hence, an attempt was made to balance the observed and calculated flows for individual years in the second stage of calibration by adjusting
POSREP only, which improved hydrographs as well as efficiency of the calibration. This calibration was named as the 'base-line calibration' (BLC) so that the results achieved after introducing the avalanches can be compared with this calibration.
In this calibration, although the observed and calculated flow volumes were nearly balanced, redistribution of snow from higher to lower elevations was required. This redistribution of snowpack was done in the form of avalanches in third and final stage of calibration.
52 4.5 AVALANCHING
Once the Stage-II calibration was completed no parameters were adjusted other than
POPADJ (precipitation adjustment factor to increase or decrease the amount of
precipitation in a band) i.e. avalanches, so that the results from stage-II and -III can be
compared on the basis of avalanche contribution only.
Originally the whole watershed was divided into 7 bands having mean elevations
as 1150, 2000, 2450, 2800, 3350, 4000 and 4600 meters, and areas as 100, 100, 130, 190,
155, 255 and 70 km2, respectively. These areas are the fraction of 1000 km2, therefore
actual area of each band can be achieved by multiplying with 2.34 (2,340 km2 being total
watershed area).
To simulate the avalanche activity, each band from 2 to 6 was split into two sub-
bands, the unaffected band 'a' and avalanche affected band 'b', having the same elevation
and other watershed descriptions except area. No snowpack was found in band 1,
therefore, it was kept as single band whereas band 7 at 4,600 m was also kept as one band,
because in existing situation the programming interfaces of the model allow a maximum
number of 12 elevation bands. Furthermore, in subsequent stages of calibration, band 7
was found not significant in avalanche activities. Otherwise the programming interfaces
could be modified to accept more than 12 elevation bands.
The advantages of splitting each band into two bands are that the avalanche area
can be controlled. We can create avalanche on a fraction of the area of a particular band
by increasing or decreasing the precipitation multiplier in the runout or starting zone as
the case may be. Also, some other conditions like 'tree cover' (C0TREE) and 'density of
53 the forest canopy' (COCANY) can be controlled accordingly for that part of a particular band elevation which is under Avalanche. The area at 4,600 m elevation, which is
obviously a starting zone, has a single band; it means that if the snow is decreased or increased from this band by putting - or + value for POPADJ, the whole area will be
activated.
Table 4.1 shows an example of the arrangements of elevation bands and some
other watershed descriptions. The shaded portion of the bands is not under avalanche
activity and the sum of the areas in band 'a' and 'b' will be equal to the total area of that
particular band. In Stage-Ill, the avalanched and non-avalanched areas were determined for each year.
54 99 The U.B.C. model was calibrated again for each year by using POPADJ for a portion of area of each band i.e. avalanche affected area. In avalanching, two parameters were dealt with, firstly the area of the avalanche band and secondly the precipitation in each band, controlled by the precipitation multiplier (POPADJ). Both, the area and
POPADJ were increased or decreased gradually, in lower bands, for example, the avalanche was produced in smaller area and for smaller value of POPADJ and then started increasing gradually after each run of the model so that the efficiency and hydrographs could be analyzed. Similarly, the avalanche was initially produced in only one band then other bands were included in avalanching. It means the calibration required a number of run so that the best combination of the avalanche area, POPADJ, and their distribution in bands can be achieved.
Generally, the snow was decreased from upper bands (bands 5b, 6b and 7) and increased at lower bands (bands 2b, 3b and 4b) according to the hydrographs achieved in stage II of calibration. First the snow was removed from only one higher band by giving negative value e.g. -0.5 means that 50% of snow is removed from that particular band. At the same time snowpack was increased first in only single lower band by giving positive value of POPADJ, e.g., 3 means that snowpack is increased by 300% of the existing snow in a band under consideration. In the next step the snowpack was decreased from other higher bands. Similarly the snow was being added to other lower band/s.
The same practice was applied for the area of avalanche, i.e. first the POPADJ was given to a small area then the area was increased accordingly, depending on the shape of the hydrographs. The volume of snow subtracted by avalanches from the upper bands was made equal the volume of snow avalanched to the low and mid elevation bands. After
56 each change in avalanche affected area and POP AD J factor, the model was run for that
particular year and both the hydrograph and statistic report was analyzed before giving
new values of these two parameters.
Once the best efficiency and hydrograph shape (keeping balance between + and -
values of avalanches) were achieved and it was found that no further improvements were
possible, the band CAL files for all bands were created for that particular year. Then from
the plot menu, the values of snowpacks water equivalent, w.e., for each band were
recorded. The actual amounts of snow increased or decreased were calculated by
subtracting 'snowpack without avalanches' from 'snowpack with avalanches'. This actual
amount of snow water equivalent increased or decreased was then multiplied with the
active area, i.e. avalanche area, of the band to achieve actual amount of avalanches. All
6 3
the values of snowpacks and avalanches volume are in '10 m water equivalent' unless
otherwise specified. Table 4.2 explains the method to calculate the avalanche snow
volumes and net increased depth of snowpack water equivalent in a band due to
avalanche.
The runout zone is assumed to include the track zone above it since the track frequently acts as a deposit area for avalanche snow. Further, it was assumed that the
avalanche deposits are uniformly distributed over the entire area of avalanche band and not concentrated in any depressions. The time delay may be low due to this assumption,
since depressions within the watershed provide storage of snowpack for longer time as compared to horizontal surfaces.
57 4.6 SUMMARY OF CALIBRATIONS
The calibration process of the watershed model consists of the following important steps;
Stage-I Calibration: Calibration was attempted using a single value of POSREP which was found good for some years but not for all. The results were not consistent from year to year.
Stage-II Calibration: POSREP was adjusted for each year according to hydrographs and efficiency of the model. This process gave better results but indicated that there is a need to measure precipitation high up in the watershed.
Stage-Ill Calibration: In this final stage of calibration the redistribution of snow from higher to lower bands was carried out for each individual years. Avalanches were created in different bands step by step. Once the best results were achieved in Stage-II, no parameters were altered other than POPADJ and avalanche area.
58 6£
W o 3 3 £ ml eo . o CD w e 3 > P CD Io. £i—if I—»• 3^1 o * * CD CD 5 CD S3 & CD P to < OSS: o © N3 S3 o © P
Bic h o
CD S3 © S3 o 5 o o.
S3 t>3 p + Ul —i o p ON 4^ 4* 1 o *3 S3 00 4^ U3 U3 ui o cr
S3 v3 Ul oc •M S3 Ul © + o p ON U3 OO S3 03 00 4^ S3 S3 Ul o o o ar
U3 ma Ul l*J Ull VO Ul Ul o p
U3 t>3 Ul Ul or o
S3 ui © © ON 4^ Ut S3 © P I—> 1 Ul ON o 1*3 1—" 4^ oVO i—» o ON 4^ Ul o S3 © ON ON © a" <1 ON ©
4*. vo -4 ON ,oo © © ui © 4.7 REGRESSION ANALYSES
4.7.1 Avalanche volume
For the use in an avalanche forecasting system, simple regression analyses were performed for avalanche volumes for each active band. In these analyses, the avalanches
were regarded as the dependent variable, and snowpack in band 6 (SP6) as independent
variable, since SP6 was found to be the single most affected parameter in avalanche production. In the year 1983-84 only, the snow was avalanched from band 7 which was found not typical, therefor this year was excluded from further analyses.
The purpose of these analyses was to determine if a correlation existed between the snowpack in band 6 and the avalanche magnitudes, so that it would be possible to establish equations for avalanche forecasting on the basis of snowpack in band 6.
To investigate whether a linear relationship exists between the avalanche volume and snowpack in band 6 a regression analysis was carried out. It was found that the avalanches could be represented by simple regression equations in terms of the snowpack in band 6. The equations are presented in the general form;
A = a + bSP6±e [4.10]
where; A is the avalanche volume, the intercept a is the avalanche at external variable SP6
(snowpack water equivalent in band 6) of 0 value, the slope b is the variable response for
60 avalanche, and e is the standard error of the regression equation. All values of avalanches
and SP6 are in million cubic meters.
4.7.2 Avalanche Depth
Similarly to the above mentioned procedures the simple regression analyses were performed to investigate the extra snowpack depth, Sp, from avalanching required in runout zones and snow depth to be removed from starting zones, both as a function of existing snowpack depth in band 6. All values of snowpack depth were in millimeters in these analyses. Again, for each affected band three equations were developed which give the extra snow depth in bands 3 and 4 related to the actual snowpack depth in band 6.
These equations will be helpful to find out the value of POPADJ, since the extra snow depth in millimeters is a function of precipitation increasing factor.
By using these values of avalanche volumes and extra snowpack depth, the affected areas can be calculated using equation 5.9 given in section 5.5.2. Therefore with the help of this model not only the volumes of avalanches, but also the precipitation adjustment factor, POPADJ, and avalanche affected area of a band can be calculated for both starting and runout zones.
61 Chapter 5
RESULTS AND DISCUSSIONS
5.1 INTRODUCTION
This chapter mainly consists of four sections. The first section contains results of Stage-I,
Stage-II & Stage III calibrations of the UBC watershed model for Kunhar basin. These results are compared on the bases of model efficiency and hydrographs before and after avalanching. Some important parameters of avalanches (such as areal coverage, volumes and elevation distribution patterns etc.) are discussed in section 5.3. In section 5.4, the effects of avalanches on snowpack, snowmelt, and hydrographs of Kunhar river are shown. Finally, the avalanche forecasting model is presented in section 5.5.
5.2 MODEL CALIBRATION
As described earlier the model calibration for Kunhar basin was done in three stages i.e.
Stage-I, with no allowance for avalanching and with a constant precipitation factor
(POSREP). This basic calibration was carried out using nine years of data (i.e., 1979-88);
Stage-II adjusting POSREP for individual years (Base-line Calibration). It was found in
Stage-I that Battakundi precipitation was not always a good indicator of basin wide
62 conditions. Therefore in this second stage the precipitation' was adjusted for each individual year; and finally
Stage-Ill, investigated the introduction of avalanching in different elevation bands for each individual year.
5.2.1 Stage-I Calibration
The statistic report of Stage-I calibrations is given in Table 5.1. In this calibration the value of PORREP was kept constant as -0.80 during the optimization procedure because this value gave the best results for the rain events. The model optimization indicated that the snow precipitation should be increased by 5% for the entire period of calibration.
Therefore, POSREP was found to be 0.05. This optimization procedure was the Eopt measure as defined in section 4.1. The resulting calibration therefore has a good
coefficient of efficiency and also the seasonal volume agrees reasonably with the measured values. Therefore, good agreement is obtained for both the hydrograph shape
and volume of runoff.
In stage-I, the standard semi-automatic calibration procedure, as described in the
Watershed Model manual, was carried out. This set of best fit parameters was then kept
constant for all subsequent stages except for POSREP in Stage-II and the avalanche factors (affected area and POPADJ) in Stage-Ill.
Results of Stage-I show a significant deviation in simulated flow, i.e. under• estimation, from the observed hydrographs; specially in the months of June-July. From
63 this original calibration the total volume deviated on average by about -6% of the
observed flow volumes with a maximum deviation of -35% for the year 1982-83.
Table 5.1. Statistics report of Stage-I calibration for years 1979-88.
Mean Mean Total Total Dev. in Coeff. Coeff. Qobs
Year Qes, Q of Eff. of Det. Oobs Qobs Qes, Qest
2 3 3 3 3 3 (%) E! R (m /d) (m /d) (m ) (m ) (m )
79-80 43.5 36.3 15,926 13,277 2,648 -16.6 .70 .74
80-81 44.0 40.6 16,049 14,832 1,218 -7.6 .85 .86
81-82 37.1 33.1 13,530 12,081 1,449 -10.7 .69 .71
82-83 44.0 28.4 16,075 10,381 5,694 -35 .59 .82
83-84 38.0 29.0 13,890 10,603 3,288 -23.7 .72 .82
84-85 26.9 27.4 9,820 10,001 -181 1.8 .81 .87
85-86 42.1 44.3 15,363 16,156 -793 - 5.2 .80 .90
86-87 46.6 50.3 17,008 18,345 -1336 7.9 .64 .83
87-88 47.1 59.2 17,254 21,678 -4,424 25.6 -.24 .72
79-88 41.0 38.7 134,916 127,353 7,562 -6.0 .62 .81
5.2.2 Stage-H Calibration
The precipitation representation factor for snow, POSREP, was adjusted by optimization for each individual year in Stage-II. The volume deviation was reduced to -
1.9% on average and Nash-Sutcliffe coefficient of efficiency, denoted by E!, was increased from 62% to about 77% on average for 1979-88 (see Table 5.2). The
64 coefficient of determination, denoted by R2 remained constant at .81% for both stage-I &
II calibration. Stage-I & II results suggest that the precipitation at the Battakundi station is not representative for the whole watershed, specially for snowfall at higher elevation, therefore precipitation had to be adjusted accordingly. This was done first by increasing the POGRADM & POGRADU (i.e., precipitation gradient factors in %tage for elevations below & above EOLHI respectively) in Stage-I calibration for the whole period and then by adjusting POSREP for each year in Stage-II. The calibration after stage II was given name as Base-Line Calibration (BLC) so that the results after avalanching can be compared on the basis of avalanche input only.
Table 5.2. Statistics report of Stage-II calibration for years 1979-88.
Mean Mean Total Total Dev. in Coeff. Coeff. Oobs-
Year Q ofEff. of Det. Qo„s Qobs Qes,
2 3 3 3 3 3 (%) E! R (m /d) (m /d) (m ) (m ) (m )
79-80 43.5 43.0 15,926 15,756 170 -1.1 .70 .73
80-81 44.0 44.1 16,049 16,095 -45 0.3 .82 .84
81-82 37.1 37.1 13,530 13,545 -14.8 0.1 .70 .73
82-83 44.0 42.3 16,075 15,424 651 -4.0 .82 .83
83-84 38.0 37.9 13,890 13,861 30 -0.2 .78 .79
84-85 26.9 27.2 9,820 9,911 -90 0.9 .84 .86 85-86 42.1 41.7 15,363 15,213 151 -1.0 .85 .91
86-87 46.6 44.1 17,008 16,093 915 -5.4 .78 .85
87-88 47.1 44.2 17,254 16,183 1071 -6.2 .67 .79
79-88 41.0 40.2 134,916 132,081 2,835 -2.0 .77 .81
65 5.2.3. Avalanche Contribution
After adjusting the volume to almost negligible differences in simulated and observed flows in the Stages-I and II, the redistribution of snow was done from higher to lower elevations. By applying the procedure for avalanche introduction in the model calibration described in chapter 4, the volume deviation was further reduced to +1.5% on average, while efficiency was found to be 84%. This time, R2 was increased to 85%. The results of
Stages I, II and III are summarized in Table 5.3 for comparison. Due to avalanches the overall efficiency was increased by about 10%. In individual years the improvement ranged from 0 to above 25%. The maximum improvement was found for the year 1987-
88 (i.e., 25.4% increased in efficiency due to avalanches) which shows that avalanche contribution has a significant impact on the runoff for the Kunhar River basin. For years
1985-86; 1986-87; and 1987-88, for example, the model efficiencies increased from 85 to 94%, 78 to 88%, and 67 to 84% respectively (see table 5.3). In addition, the timings and volumes of peak flows were also much better after avalanches.
66 L9
oo 00 00 oo 00 00 oo oo -J vo 1—» ON Ul 4*. ui to o vo CD 00 oo 00 oo oo 00 oo oo oo 00 00 oo -J ON Ul 4^ Ui to o 3 o to ON to Ui o ON © y vo to 00 Ui Ul OS ON o 9*' 3 S 0 i B ON to ON oo 00 Ul ON oo -0 n to 4^ 4^ o to Ul o • 1-1 O ps Ui E?. o oo oo VO 00 oo oo 00 3 to ui o -0 to to ON 4*. w CD »
© CD © o on 1-1 VO o 4^ PS Ui ON Ui o o VO oo 59" 09 4^ Ul s CD 13 CD
o 1 ON Ul 4^ 3 CD to p p p VO 3 II to 4^ to © Ui < r—• o 13 c 1/3 P , ON —J oo oo oo —] oo —] 5. -0 -0 oo Ul 4*. oo to o to o
o *0 00 00 VO 00 -o oo oo © VO Ul ON Ui Ui 4^ Ui vo C/3
+ + + + + + + p Ui ^B- 3. to to Ui Ui Ui p I—> VO © Ul ON to Ul ON 1—" -'Op. Ul I o cro CD 3 I
oo oo oo VO oo 00 oo oo 4^ 4^ 00 4*. Ul to Ui oo •o 8 p o 00 oo 00 VO 00 oo oo 00 Ul ON VO 4*. ON Ui 4^ VO oo -—' CD o e. o 3 to CD VO 4^ B' 8 Vi DC Ul Ul Ui H-> to Ui Ui w Ct> CD o. 00 The adjusted hydrographs before and after avalanches are compared in figures 5.1 to 5.4.
Overall, there was found a lack of early flow response in the base-line calibration and an overestimation of peak flow in the months of May, June and July. Whereas avalanching gave very good early response because of extra snow at lower elevation bands. Also, avalanching gave less mid-season response because of removal of snow from upper band and consolidation into deeper, more limited lower level snow.
It will be noticed that avalanching did not effect the flows in months of August and September and the flow response was found same before and after avalanching during these months. Whereas, avalanching affected largely the peak flows during the months of June and July because of removal of large amount of snow from elevation
4,000 m.
The early response in flows, after avalanching, was mainly due to large amount of snow accumulation in deposition zones and higher melting rates due to much higher temperatures in these lower elevations as compared to the temperatures in upper elevations. So the flow was controlled by larger accumulation of snow and higher temperature rates in lower bands.
68 250
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep
Fig. 5.1. Comparison of hydrographs before and after avalanches with the observed flow, (a) year 1979-80 ; (b) year 1980-81.
69 250 Observed (a) Base-line 200 Avalanched
150 r a> CO o J2 100
50
l j
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep
250 Observed (b) Base-line 200 Avalanched m 150
TO 100 Q 50 I
J I \ L Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep
Fig. 5.2. Comparison of hydrographs before and after avalanches with observed flow, (a) year 1981-82 ; (b) year 1982-83.
70 250
Observed Base-line 200 Avalanched to
" 150 a> CO CO
-j2 100 Q
50
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep
250
Observed (b) Base-line 200 Avalanched to
150 CD CO I—
o J2 100 Q
50
* r Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep
Fig. 5.3. Comparison of hydrographs before and after avalanches with observed flow, (a) year 1985-86; (b) year 1986-87.
71 250
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep
Fig. 5.4. Comparison of hydrographs before and after avalanches with observed flow for year 1987-88.
72 5.3 DISTRIBUTION PATTERN OF AVALANCHE
The avalanche magnitude and areal coverage at each band is calculated. There is found a very good correlation between avalanche volume and high altitude snowpack. While areal distribution is correlated with the avalanche volume in the same band.
5.3.1 Avalanche Volume
The results of analyses showed that the avalanche starting zone is located at band 6 which has a mean elevation of about 4,000 m. The snow then cascades downslope and the track and runout zones are located at bands 3 and 4 with a mean elevation of about
2,450 and 2,800 m, respectively. The most active starting zone was band 6 for all years, except year 1983-84 for which the starting zone was found at 4,600 m in band 7 which was not typical and hence, removed from further analyses. The year 1984-85 was found non avalanche year which received least amount of snowfall in upper bands.
The amount of snow avalanched from band 6 was variable from year to year and ranged between 12 to 27%, depending on the amount of snowfall at higher elevations.
The study shows that on average about 20% of the snowpack at band 6 is subject to avalanching. The deposition of this snow occurs at bands 3 and 4. There is found to be a consistency in percentage of snowpack removed from band 6 and is directly correlated to the amount of snowfall in band 6 (i.e., R2 = .93). The snow avalanched with a higher percentage for higher snowfall years, i.e. 1985, 86, and 87. This means there is a double increase in avalanche amount, i.e., higher percentage of an already high amount of snow.
73 The annual distribution pattern of avalanche magnitude is shown in Table 5.4.
Since, the starting zone lied in band 6 the snowpack volume in the band, denoted by SP6, was the most critical one in producing avalanches. This was found to be the single most important band responsible for avalanching. Therefore, it would be appropriate if the
avalanche deposition magnitudes in band 3 and 4 be correlated with the SP6. It will be noticed in Table 5.4 that the percentage distribution in bands 3 and 4 are more or less equal, i.e. the snow avalanched from band 6 is distributed equally to each runout zone.
The annual distribution pattern of avalanche magnitude expressed as a percentage
of SP6 is shown in figure 5.5a. Although, avalanche magnitude in band 6 is plotted as positive to visualize its magnitude as compared to avalanches in bands 3 and 4, it will be treated as negative, since the snow was removed from this band. Figure 5.5b shows the avalanche distribution of runout zones (i.e., bands 3 & 4) against avalanche of starting zone for each year expressed as a percentage of snowpack in band 6.
74 Table 5.4. Distribution Pattern of Avalanche Volume in each Band.
Avalanches, expressed as
6 3 Avalanche (xl0 m ) %tage of SP6
Year (xloV) Band 3 Band 4 Total Band 6 Band 3 Band 4 Band 6
79-80 927.87 70.39 63.72 134.11 134.16 7.59 6.87 14.46
80-81 1054.37 109.32 93.64 202.96 202.98 10.37 8.88 19.25
81-82 850.30 62.60 38.48 101.08 101.04 7.36 4.53 11.88
82-83 1046.02 76.82 65.69 142.51 142.51 7.34 6.28 13.62
84-85 511.97 0.00 0.00 0.00 0.00 — — —
85-86 1226.22 101.47 145.86 247.33 247.33 8.28 11.90 20.17
86-87 1389.71 181.29 194.43 375:72 375.72 13.05 14.00 27.04
87-88 1213.1 154.81 151.91 306.72 306.73 12.76 12.50 25.28
Average 1101.1 108.1 107.7 215.8 215.8 9.8 9.8 19.60
75 30
25
20
"§ 15 TO
< 10
0 79-80 80-81 81-82 82-83 85-86 86-87 87-88 Year
H Band 3 • Band 4 II Band 6 Fig. 5.5a. Avalanche distribution pattern for active bands as apercentage of snowpack in band 6.
15 20 25 30 Avalanche %tage (Band 6) Fig. 5.5b. Avalanche in bands 3 & 4 against avalanche in band 6 as a percentage of snowpack in band 6 5.3.2 Avalanche Area
The percentage of total area, including starting and runout zones, affected by avalanches in the Kunhar basin is estimated to range from 12% to 21%. These results are in
agreement with the results of a previous study (De Scally, 1992). De Scally studied only
a small area (288 km2) at low and moderate elevation around 2,300 m, at the upper valley
of Kunhar basin, and then extrapolated the results to the high elevations. In the present
study on average about 16% of the total watershed area was found to be influenced by
avalanche activity, of which 9% was in runout zones. The avalanche snow in these
runout zones was distributed as 58% to band 3 and 42% to band 4 (see Table 5.5a). The
areal distribution pattern for each band is shown in figure 5.6a.
It will be noticed here that the 'concentration factor' K, i.e. area of starting zone
divide by the area of runout zone, is very small, which shows that snow was avalanched
and redistributed from a smaller area to a larger area. But, the concentration of snow in
deposition zones is still very high, since the snow was increased greatly above EOLHI
(3,800 m) i.e., POGRADU = 20, where, P0GRADL and P0GRADM were 4 and 5
respectively in original calibration of Stage-I. Furthermore, a high percentage of snow,
from 60% to 80%, was removed from band 6; see last column in table 5.5a for POPADJ
in band 6. This means that the huge amount of snow removed from affected area of band
6 requires larger area to be deposited in runout zones, even though the concentration in
lower zones is much higher. It was found necessary to have the following redistribution
of snow. Therefore, the snow removed from smaller area (but in larger amount) from
77 upper band was sufficient enough to create large amount of avalanches in lower bands, which increased snow about 200 to 600% in elevation range from 2,200 to 3,000 m.
Table 5.5a. Areal distribution of avalanches in each band.
Areal Distribution of Avalanches
(%tage of total watershed, i.e., 2,340 km2) POPADJ
Total Starting Net
Year Band Band Runout Zone, Affected K Band Band Band 3 4 Zone Band 6 Area 3 4 6
79-80 4.70 3.50 8.20 5.26 13.46 .64 3.0 2.0 -.70
80-81 5.49 4.08 9.57 6.10 15.67 .64 4.0 2.6 -.80
81-82 5.00 2.70 7.70 5.05 12.75 .66 2.5 2.0 -.60
82-83 3.96 2.67 6.63 5.80 12.43 .87 2.5 1.8 -.60
84-85 0.00 0.00 0.00 0.00 0.00 — — — —
85-86 3.83 4.56 8.39 8.20 16.60 .98 4.5 3.0 -.60
86-87 6.80 5.25 12.05 8.50 20.55 .71 3.5 6.0 -.80
87-88 8.00 4.60 12.60 8.05 20.65 .64 5.0 4.4 -.80
Average 5.40 3.90 9.30 6.7 16.00 .72 3.6 3.1 -.70
78 79-80 80-81 81-82 82-83 85-86 86-87 87-88 Year
M Band 3 • Band 4 • Band 6
Fig. 5.6a. Avalanche area distribution pattern for active bands as a percentage of total watershed area.
79 The percentage of areal coverage of avalanche has a good correlation with the avalanche volume at the same band. Regression analysis was carried out for the entire period (1979-88) to find out the relationship between area and volume of the avalanche in each affected band. The statistics report for this analysis is given in table 5.5b, whereas figure 5.6b shows the regression of these two parameters. For band 3 the coefficient of
determination (R2) is about 80%, whereas for band 4 and 6 it is found above 83%.
Avalanche area is denoted by A, whereas avalanche volume is denoted by A.
Table 5.5b. Results of regression analyses (A vs. A)
Constant Coefficient Standard Standard Coefficient of (a) (b) Error of Error of A Determination,
Coefficient Estimate (e) R2
Band 3 1.18 0.04 0.008 1.165 0.80
Band 4 1.23 0.02 0.004 0.731 0.83
Band 6 1.93 0.02 0.004 1.208 0.83
80 10 Band 3 rrr- 8 • 2 CO R = .80 • < CD .C o c - ro • co 3
50 100 150 200 6 3 Avalanche Volume (x10 m)
10 Band 4 q> 8 R2 = .83 CD < 6
Ico 4 CO • < 2
-i 0 50 100 150 200 Avalanche Volume (x106 m3)
10 Band 6 8
2 CO R =.83 < 6 CD JZ o c 4 - ra co 3 2
0 i , i 0 100 200 300 400 Avalanche Volume (x10 m3)
Fig. 5.6b. Avalanche area against volume in bands.
81 5.4 EFFECTS OF AVALANCHES
The hydrological conditions of the basin especially the snowpack and snowmelt patterns
were significantly altered by avalanche magnitude and areal distribution in runout and
starting zones. That in turn, affected the Kunhar river hydrographs in positive way, i.e.,
the shape of simulated flow after avalanches agreed more closely with the shape of
observed flow.
The results after avalanching are compared with the hydrological conditions of
the basin before avalanching. Also, a similar comparison is made for avalanche affected
area and unaffected area of the same elevation band, since snow was avalanched in a part
of the band area.
5.4.1 Snowpack Conditions With and Without Avalanche
The snowpack depths in millimeters of water equivalent (w.e.) in avalanche affected and
unaffected area of each band is calculated. Since, the snow was avalanched on a part of
the area of a band, the mean snowpack depth of affected and unaffected area of the band
are also calculated. The mean snowpack depth in active bands after avalanching is given
by equation [5.1] in which the areal coverage of avalanches is taken into account.
82 (SP'a * Aa ) + (SP'h * Ah ) SP' in affected band = -—i -—6 ^ [5.1] At
where, SP' = snowpack depth in mm of water equivalent,
SP'a = snowpack depth (mm) in part 'a' of the band, i.e., in unaffected area
SP'b = snowpack depth (mm) in part 'b' of the band, i.e., in affected area
2 Aa = area under part 'a' of the band, i.e., unaffected area of the band in km
2 Ab = area under part 'b' of the band, i.e., un-affected area of the band in km
2 At = total area of the band in km
The procedure to calculate mean snowpack depth in a band after avalanching is
illustrated in table 5.6.
Table 5.6. Illustration of calculating the mean snowpack (mm) in band 3 for the year 1979-80, with and without avalanches.
Hydrological Year: 1979-80 Band No. 3
Total Area (km2) 304.2
Un-affected Avalanche affected
Area (km2) 194.22 109.98
POPADJ 0 3
Snowpack (mm) 171 811
Mean snowpack depth (811 x 109.98)+ (171 x 194.22) = 402 (mm) in the band 304.2
83 The snowpack conditions in avalanched and non-avalanched areas for each elevation band are shown in table 5.7 which can be visualized in figures 5.7 to 5.9. From these figures it can be observed how much the snow depth is increased or decreased in avalanche runout and starting zones. Note that the dark continuous line shows snow depth in the avalanche area only, which is a portion of that band; while the dotted line is mean snow depth in which the areal coverage of avalanched and non-avalanched zones are considered and calculated by equation [5.1]. These figures show the snowpack conditions at the time of maximum snowpack accumulation, i.e. mid-March for runout zones and mid-April for starting zone. In the year 1983-84 the snow was removed from only band 7 (Fig. 5.8b) showing that the starting zone for this exceptional year lay in band 7. Also note that these figures are only the snow depths in different elevation bands due to avalanches and not the volume of avalanches.
Comparing the snowpack conditions, for example, in band 4 for years 1986-87 &
1987-88 (Fig. 5.9 a and b), although the snowpack is increased in band 4 by 1187 mm and 1607 mm in year 86-87 and 87-88 respectively, the avalanche volume is larger in year 86-87 for band 4. The reason is that the snow avalanche is spread over a larger area
(5.25%) in 86-87 as compared to the areal coverage of snow avalanche in year 87-88
(4.6% area of whole watershed). The volumes avalanched in year 86-87 and 87-88 were
6 3 6 3 194.43 x 10 m and 151.91 x 10 m respectively.
84 £8
oo oo 00 OO OO oo oo OC -o Yea r ON Ul LO to 1—» O oo OO oo oo OO oo oo OO 0v0p oo -o. ON Ul 4^ LO to 1—» o cr g - do tO Ul Ul ON to to '-O ON VO LO LO © Ul to ON Ul 4* s to
t— vC to to to to 1 '-O >3 00 © LO sC ox^h** -J 5T & E mmmUl -J LO P o iiill t—i e
13 H-* 1 cr -J -4 —1 oo vo to LO o > H to 00 LO I—» to 4^ o >— s LO Ul 4^ i—> 4^ % to -J P & 4*. p o- & LO ST Ul 3 jo ON •~4 Ul o to Ul 4*. ON 4^ o 4*. LO 00 4^ i—» oo O ^ ^ I" -J O OO LO ON ON 00 to 3 p o
to to aiWiAiii to Ul LO LO LO to t X NC ON Ul !< & CD LO vC Ul UI LO 4^ OC to P o mmm c »—»•
190 0 3 w 3 i—> to O 1—» i—» h—» h-> 00 vo h—» 4^. —1 »—» 1—1 3 i—' o i—> LO 1—" % LO Ul ON LO © LO LO P & 4^ Ul P o 3 CD 3 ON ON —] O ON -O LO Ul Ul o 4*. ON ON vo p—> VO -J 1—1 © ON -4 4^- *—* 1—« ON Ul P & CD to i-t ON ON -J to >C Ul o» CD to LO ON £ LO $ vC g & § oc to X S Ul LO oc Ul Ul Q. vo Ul 1. < 203 3 to to >3 CD LO Ul to -J 4* -J Ul 3 to Ul JO Ul to ON Ul p o Ul oc o LO Ul —1 Ul s
3 to -o o o -J Ul LO 4^ o it ON o -J 4* ON ON 53 LO Ul Ul 5^ Ul P o P o p ON
3 151 6 3 1—» O i—* 1—> O 1—1 ON ON Ul to 4^ LO VO 4*. 51 1—1 Ul to LO VO O % P 4^ ON ~J O P '* 6 without avalanche 5 (a) 1979-80 ' with avalanche 4 mean
3
2
1
2000 2450 2800 3350 4000 4600 Elevation (m)
without avalanche I 5 o with avalanche o o 4 mean
2000 2450 2800 3350 4000 4600 Elevation (m)
E without avalanche E (c) 1981-82 o with avalanche o o mean
CO
O CO o. o e CO
2000 2450 2800 3350 4000 4600 Elevation (m)
Fig. 5.7. Snowpack depths in mm w.e. at different elevation bands for years (a) 79-80, (c) 80-81, and (c) 81-82.
86 6 without avalanche 5 (a) 1982-83 with avalanche 4 mean
3
2!
1
2000 2450 2800 3350 4000 4600 Elevation (m)
6 without avalanche 5 with avalanche (b) 1983-84 4 mean
3
2
1
2000 2450 2800 3350 4000 4600 Elevation (m)
Elevation (m)
Fig. 5.8. Snowpack depthsin mm water equivalent at different elevation bands for years, (a) 82-83; (b) 83-84; and (c) 85-86.
87 Elevation (m)
Fig. 5.9. Snowpack depths in mm w.e. at different elevation bands for years (a) 86-87, and (b)87-88.
88 5.4.2 Avalanche Effects on Snowpack Accumulation
The snowpack accumulations in affected bands were also greatly altered by avalanches.
It should be noted that avalanching has quite different effects in the starting zone and the
runout zones.
In runout zones the snowpack reached its maximum height in mid-March, while
in starting zone the maximum snowpack was found in mid-April to the end of May.
Figures 5.10 and 5.11 show the avalanche effect on snowpack accumulation for years
1979-80 and 1980-81. For all years, i.e., 1979-88, the snowpack started to develop in
early December and late December for starting and runout zones respectively. Snow
diminished about one month earlier in non-avalanche portion of the runout zones, i.e. at
elevations 2,450 and 2,800 m.
Whereas in starting zone, i.e. at elevation 4,000 m, the snow stayed about one
month longer in non-avalanche areas. For some years, for example year 1986-87, the
snow stayed up to mid-July in non-avalanche area of starting zone. In contrast, the snow
diminished 40 days earlier in this same year in the avalanche areas of the starting zone.
The early diminishing of snowpack in affected areas of starting zone is
compensated by the longer duration of the snowpack in runout zones. But, the higher
melting rates of snow in runout zones, due to lower albedo and higher snow density and
temperatures, gave early runoff in the season. Also, the melting of snow lasts for a longer
period of time in avalanche affected areas due to the high accumulation of snow.
Therefore, the simulated hydrographs are better fitted with the observed hydrographs in
the first half of the melting season, i.e., April through June. On the other hand, in the
89 starting zone, the over estimation of flows in late season is controlled by the removal of snow from part of the higher elevation band.
90 18 16 non avalanched (a) Band 3 14 • avalanched mm ) 2,450 m o o 12 10 3: 8 o CD 6 o c 4 CO 2 0 Dec Jan Feb Mar Apr May 18r non avalanched 16 - (b) Band 4 14 - avalanched mm ) 2,800 m O O 12 -
>< 10 - 91 18r non avalanched 16 - (a) Band 3 14 - • avalanched mm ) 2,450 m O O 12 10 - ai 8 - O - CO a. 6 - o=: c 4 - CO 2 - 0 - Dec Jan Feb Mar 18 non avalanched 16 (b) Band 4 avalanched E 14 E ^^^^ 2,800 m o o 12 10 cu 3: 8 o CO a. 6 o c 4 CO j ^ 2 f " \ Dec Jan Feb Mar Apr May 18r non avalanched 16 - • avalanched 14 - mm ) O (c) Band 6 O 12 IK, 10 - 4,000 m ai 8 - - CoO CL 6 - oS= c 4 - CO 2 - j 0 - Dec Jan Feb Mar Apr May Fig. 5.11. Snowpack accumulation in avalanched and non-avalanched part of an elevation band for year 1980-81. 92 5.4.3 Avalanche Effects on Snowmelt Seasonal snow cover accumulates during weeks or months from consecutive snowfalls and then gradually disappears during the snowmelt season in a yearly cycle. Its typical feature is the gradual decrease of the snow covered area during the snowmelt season which may extend to several months (Martinec, 1987). This phenomenon results from the variable deposition of snow and variable melt rates at different altitudes. Snow accumulation is altered by avalanching which, in turn, affects the snowmelt patterns. The snowmelt patterns were examined for each active band with and without avalanches. These avalanches have a direct effect on the snowmelt patterns of Kunhar River basin. The flow response of each band is directly related to the amount of snow avalanched and its areal distribution. The snowmelt distribution regarding magnitude and timing are tabulated in table 5.8 for each runout band 3 & 4 and starting band 6. These snowmelt patterns with and without avalanches are also shown for these active bands in figures 5.12 to 5.18. The snowmelt in runout areas starts about seven days later and lasts 20 to 30 days more than in the areas not affected by avalanches. In band 3 and 4 the time delay of snowmelt is 20 to 30 and 25 to 35 days respectively, on average. The snowmelt peaks are also shifted away by 30 days on average in the avalanche affected areas of the band. Also, the maximum snowmelt from the avalanche runout areas is about 100-120% higher than the maximum snowmelt from the non affected areas. As a result of the larger snow accumulation in the runout areas the snowmelt volume is increased by 370% and 260% 93 on average in band 3 and 4 respectively. However, this does not change the seasonal snowmelt total, but it does change the time distribution. From the snowmelt graphs it can be seen that the snowmelt rates are much higher in the avalanched areas of the same band mainly due to larger areal coverage of avalanches, since the temperature conditions are the same for both affected and non affected areas in the same elevation band. But, compared to the snow melting in the starting zone, the higher temperature is the main factor for such a higher melting rates in runout zones. In the early melt season, i.e. February to March, the melting rates for both affected and non affected areas are almost equal but huge differences in melting rates occur later in the melting season, i.e. April to May because of much higher temperatures. The average melting rates (based on entire melting period) are about 225% higher in avalanche effected area of band 3, whereas this figure is about 138% for band 4. These results show that the snow avalanches increase both the volume and the period of the snowmelt in runout zones. Note that these percentages are calculated from total melting divided by melting time period, hence areal coverage of snowpack is neglected. In real situations melting rates of avalanche portion will be even higher since avalanches covered smaller portion as compared to non avalanched portion of the band. In other words, the amount of snowmelt, shown in table 5.8, from non affected portion of band 3 & 4 is derived from larger area. In year 1979-80, for example, the non-avalanched area of band 3 is 1.8 times larger than avalanche affected area. As far as the starting zone is concerned, again the melting pattern is greatly modified by snow removal in form of avalanches, but in reverse order as compared to the 94 runout zones.. The snowmelt in affected area of band 6 diminishes 20 to 35 days earlier. The peaks of snowmelt also occur earlier by about 30 days in the avalanche zones of band 6. Table 5.8. Snowmelt (Sm) pattern with and without avalanches. Band 3 Band 4 Band 6 Without Av. With Av. Without Av. With Av. Without Av. With Av. Year: 79-80 Max. Sm 19 32 21 46 64 29 Total Sm 210 844 385 1157 1557 466 Diff. in Sm + 634 mm + 772 mm - 1091 mm Time period 14/2 - 9/4 14/2 - 29/4 22/2 - 4/4 22/2 -10/5 15/4 - 2/6 9/4 - 10/5 Time delay + 20 days + 36 days - 23 days Year: 80-81 Max. Sm 26 42 23 59 67 28 Total Sm 219 1098 378 1353 1771 351 Diff. in Sm + 879 mm + 975 mm - 1420 mm Time period 24/1- 11/4 24/1-8/5 14/2 - 26/4 23/2 -17/5 18/4-9/6 8/4-11/5 Time delay +27 days + 21 days - 29 days Year: 81-82 Max. Sm 15 30 24 42 56 46 Total Sm 212 746 304 918 1425 555 Diff. in Sm + 534 mm + 614 mm -870 mm Time period 28/2 - 7/4 15/3 - 27/4 15/3 -17/4 15/3 - 5/5 5/4 - 28/5 5/4 - 6/5 Time delay + 20 days + 18 days - 22 days continued... 95 Table 5.8. (Continued) Band 3 Band 4 Band 6 Without Av. With Av. Without Av. With Av. Without Av. With Av. Year: 82-83 Max. Sm 28 42 33 65 53 36 Total Sm 322 1152 572 1624 1760 705 Diff. in Sm + 830 mm + 1052 - 1055 Time period 17/2 -19/4 17/2 -12/5 16/3 - 3/5 16/3 - 23/5 15/4 - 24/6 8/4 - 29/5 Time delay + 23 days + 20 days - 26 days Year: 85-86 Max. Sm 20 45 36 66 73 34 Total Sm 245 1368 450 1817 2105 841 Diff. in Sm + 1123 mm + 1367 mm - 1264 mm Time period 11/2-13/4 5/3-14/5 10/3 - 26/4 10/3 - 29/5 17/4 - 5/7 12/4 - 14/6 Time delay + 31 days + 33 days - 21 days Year: 86-87 Max. Sm 18 40 21 56 73 22 Total Sm 147 778 287 1934 2362 468 Diff. in Sm + 631 mm + 1647 mm - 1894 mm Time period 7/2 - 26/3 15/2 - 25/4 7/2 - 26/3 15/2 - 25/4 15/5 -18/7 16/4 - 5/6 Time delay +30 days + 30 days - 43 days Year: 87-88 Max. Sm 15 40 22 56 75 29 Total Sm 164 991 306 1758 2037 407 Diff. in Sm + 827 mm + 1452 mm - 1630 mm Time period 1/1 - 3/4 1/1 - 3/5 25/2 - 19/4 1/1 - 22/5 27/4 -1/7 14/4 - 27/5 Time delay + 30 days + 33 days - 35 days 96 60 Band 3 without avalanche with avalanche 50 E E, 40 % E 30 o CdO 20 10 0 A Feb Mar May Jun Jul 60 Band 4 without avalanche 50 with avalanche f, 40 •*—• CD | 30 0 co 20 10 0 Feb Mar Apr May Jun Jul 60 without avalanche A / ; j Band 6 with 50 avalanche 1 40 CD | 30 o co 20 10 0 i i i ^ S 1 Feb Mar Apr Ma\ y 1 1 Jun 1 Jul 1 Fig. 5.12. Snowmelt patterns in elevation bands without and with avalanches for year 1979-80. 97 70 Band 3 without avalanche 60 with avalanche 50 E ±^ 40 CD 30 O c CO 20 10 0 A Mifm Feb Mar May Jun Jul 70 without avalanche 60 Band 4 with avalanche _ 50 E f 40 | CD 30 | Io " 20 10 0 Feb Mar Jun Jul 70 without avalanche Band 6 60 with avalanche _ 50 E f 40| CD | 30 | 5 20I 10 o Feb Mar Jun Jul Fig. 5.13. Snowmelt patterns in elevation bands without and with avalanches for year 1980-81. 98 60 Band 3 without avalanche with avalanche 50 I 40 CD | 30 o co 2fJ 10 0 Feb May Jun Jul 60 without avalanche Band 4 50 with avalanche i 40 ±i CD | 30 o co 20 10 0 .A. Feb Mar Jun Jul 60 without avalanche Band 6 with avalanche 50 E E, 40 CD 30 O| CO 20 10 0 Feb Mar May Jun Jul Fig. 5.14. Snowmelt patterns in elevation bands without and with avalanches for year 1981-82. 99 60 - Band 3 without avalanche with avalanche 50 - E E, 40 - CD E 30 - O CO 20 - 10 - o - A A A/) Feb Mar Jun Jul 60 - without avalanche Band 4 50 - with avalanche E* E, 40 - ±± CO 1 30 - o CO 20 - 10 - o - Feb Mar Jun Jul 60 without avalanche Band 6 with avalanche 50 £ 40 ±i 10 0 Feb Mar Jul Fig. 5.15. Snowmelt patterns in elevation bands without and with avalanches for year 1982-83. too 70 " Band 3 without avalanche 60 with avalanche E 50 E, ±? CD 40 E 1 30 CO 20 10 i i i 0 Feb Mar Apr May Jun Jul 70 without avalanche Band 4 60 with avalanche "E 50 E ±± CD 40 E 1 30 CO 20 10 0 Feb Mar Jun Jul 70 without avalanche Band 6 60 with avalanche E* 50 E, ±; 40 CD E | 30 CO 20 10 0 Feb Mar Apr Jul Fig. 5.16. Snowmelt patterns in elevation bands without and with avalanches for year 1985-86. 101 70 without avalanche Band 3 60 with avalanche E 50 E, ±i 40 CD f— 30 CO 20 10 0 Feb May Jun Jul 70 without avalanche Band 4 60 with avalanche *E 50 E, CD 40 E I 30 CO 20 10 0 Feb Mar Apr May Jun Jul 70 " without avalanche — - Band 6 60 - with avalanche — E~ 50 E, ±i 40 CD 30 CO 20 10 0 i i , A/ i \ i ; i Feb Mar Apr May Jun Jul Fig. 5.17. Snowmelt patterns in elevation bands without and with avalanches for year 1986-87. 102 70 without avalanche Band 3 60 with avalanche E 50 E, CD 40 E § 30 Cf—O 20 10 0 1A. Feb May Jun Jul 70 without avalanche Band 4 60 with avalanche 50 CD 40 O 30 CO 20 10 0 Feb Mar Apr May Jun Jul 70 without avalanche Band 6 60 with avalanche E 50 §40 CD 30 oI " 20 10 0 Feb Mar Jun Jul Fig. 5.18. Snowmelt patterns in elevation bands without and with avalanches for year 1987-88. 103 5.5 AVALANCHE FORECASTING MODEL The results of the flow simulations, obtained by modeling the avalanche redistribution of snow, were used to produce a forecasting system of avalanche activity. In validation this model is then used to forecast the flow from Kunhar basin for the entire period of analyses i.e., 1979-1988. This forecasting model uses regression relationships between the amount and extent of avalanches as a function of the winter seasonal snowpack accumulation in elevation band 6 (4,000 m). 5.5.1 Avalanche Volumes Simple regression analyses were performed for avalanche magnitudes for each active band. As described in section 4.6, in these analyses the avalanches were regarded as the dependent variable, and snowpack in band 6 (SP6) as the independent variable. The equations are presented in a general form as given in Eq. [4.10], i.e.; A = a + b SP6 ± e where; A is the avalanche magnitude, SP6 is the maximum snowpack water equivalent volume for elevation band 6, which usually occurs in mid-April, the intercept a is the avalanche at external variable SP6 of 0 value, the slope b is the variable response for 104 avalanche, and e is the standard error of the regression equation. All values of avalanches and SP6 are in millions of cubic metres water equivalent. Initially, the regression analyses were performed between values of avalanches, volumes and SP6 (see table 5.4) for eight years. In the next step the threshold value of SP6 to generate the avalanches was found by giving different values of SP6 for year 1984- 6 3 85, which had the original value as 511.97 x 10 m , since this was the non- avalanche year. The value which gave the best coefficient of determination was then selected i.e., 6 3 600 x 10 m , keeping this value of SP6 for the year 1984-85; again the regression analyses were repeated for the entire eight years. Analyses of the results showed that total avalanche volume and the snow water equivalent due to avalanching for each affected band (band 3, 4, and 6) is highly correlated to the maximum snow accumulation of snowpack at band 6. The linear equations [5.2] to [5.4] were established. The coefficient of determination of these equations were found to be between 90 - 95 %. This high correlation shows that there is a very strong relationship between the avalanche activity and the snowpack at band 6. Figure 5.19 shows the correlation between avalanche volume, A, and SP6. The regression analyses results are presented in table 5.9 for the following derived equations; 105 A3 = -129.23 + 0.216 x SP6 [5.2] A4 =-171.65 +0.256 xSP6 [5.3] A6 = -300.89 + 0.472 x SP6 [5.4] where, A, and A4 are avalanche volumes in runout bands 3 and 4 respectively; A6 is the avalanche from the starting zone band 6; and SP6 is the maximum snowpack volume water equivalent in band 6. 6 3 The threshold value of SP6 to generate avalanches was found to be 640 x 10 m . From figure 5.19 the following two conditions were observed; If, 640 < SP6 < 660 ; all snow calculated from Eq. [5.4] will be avalanched to band 3, i.e., only band 3 will be activated. If, SP6 > 660 ; band 4 will also be activated and avalanche volume from band 6 (Eq. 5.4) will be distributed in both band 3 and band 4 according to equations [5.2] and [5.3] respectively. 106 Table 5.9. Results of regression analyses (A vs. SP6). Standard Standard Coefficient of Constant Coefficient Error of Error of A Determination, (a) (b) Coefficient estimate (e) R2 Band 3 -129.23 0.216 0.03 19.73 0.90 Band 4 -171.65 0.256 0.03 16.45 0.95 Band 6 -300.89 0.472 0.05 30.22 0.95 107 400 (a) "E 300 BAND 3 (runout zone) R2 = .90 i 200 o c 5 100 I.I. 400 600 800 1,000 1,200 1,400 1,600 SP6(x106m3) 400 (b) co E 300 BAND 4 (runout zone) R2 = .95 % 200 .c o c _co 5 100 0 400 600 800 1,000 1,200 1,400 1,600 SP6(x106m3) 400 (c) co E 300 BAND 6 (starting zone) O R2 = .95 ~ 200 x: o c • TO « 100 m / 00 600 800 1,000 1,200 1,400 1,600 SP6(x106 m3) Fig. 5.19. Avalanche volume in bands with respect to snowpack in band 6. 108 5.5.2 Avalanche Depth (POPADJ) and Areal Distribution Once avalanche volumes are calculated using equations [5.2], [5.3], and [5.4] for each active zones, the next step to find out the depth of avalanche, which is governed by POPADJ, and avalanche affected area in a particular band . That is, i) how much extra snow depth is required in band 3 & 4, to produce avalanches' volume calculated using equation [5.2] and [5.3], ii) how much snow (depth) is to be avalanched downslope from band 6 to generate the avalanche volume calculated using equation [5.4], and iii) how much area is required both in the runout and the starting zones to accommodate the calculated volume of avalanche snow for a calculated depth. For i) & ii), POPADJ values for bands 3, 4, and 6 are required; whereas for iii) the affected band area (denoted by A), under which the snow is either increased or decreased must be estimated. Simple linear regression analyses were performed between the extra snowpack depth, Sp, required in an affected band and the existing snowpack depth (mm) in band 6, SP'6. Table 5.10 shows the extra snowpack depth (mm water equivalent) in band 3 & 4, denoted by Sp'3 and Sp'4 respectively, snow to be removed from band 6 denoted by Sp'6, found during Stage-Ill calibration, and the original snowpack depth mm water equivalent in band 6, denoted by SP'6. 109 Table 5.10. Extra snowpack depth (mm) required in band 3 & 4 and snow to remove from band 6. Year Sp'3 Sp'4 Sp'6 SP'6 1979-80 640 778 1090 1555 1980-81 851 982 1422 1767 1981-82 535 609 855 1425 1982-83 829 1053 1050 1753 1984-85 0 0 0 858 1985-86 1131 1367 1289 2055 1986-87 1140 1582 1889 2329 1987-88 827 1411 1628 2033 Again a strong correlation was found between snowpack depth increase and original snowpack depth in band 6. Almost 100% correlation was found for band 4, while for band 3 and band 6 it was 93% and 92% respectively. Three equations were then derived to find the extra snowpack depth required in band 3 &4 and snow depth to be removed from band 6. The regression analyses results are presented in table 5.11 which corresponds to the following derived equations; 110 Sp3 = -599.28 + 0.78 SP'6 [5.5] Sp4 = -963.28 + 1.12 SP'6 [5.6] Sp6 = -930.01 + 1.21 SP'6 [5.7] where; Sp3 and Sp4 are snowpack depth in millimetres required in bands 3 and 4 respectively, Sp6 is snowpack depth required to be avalanched to the lower bands, and SP'6 is actual snowpack depth in millimetres before avalanching. Figure 5.20 shows the correlation between the avalanche snowpack depths, Sp3, Sp4, Sp6, and SP'6. Table 5.11. Results of regression analyses (Sp vs. SP'6). Standard Standard Coefficient of Constant Coefficient Error of Error of Sp Determination, (a) (b) Coefficient estimate (e) R2 Band 3 -599.28 0.78 0.09 102.18 0.93 Band 4 -963.73 1.12 0.05 55.24 0.99 Band 6 -930.01 1.21 0.14 173.90 0.92 111 2,000 E E (a) 1,500 co ~ BAND 3 (runout zone) ai R2 = .93 3: I 1,000 c - CO xCD: o 500 c m co 3 300 1,000 1,500 2,000 2,500 Snowpack w.e. in band 6, SP'6 , (mm) 2,000 E E (b) 1,500 CO BAND 4 (runout zone) 2 ai R = .99 3i 1,000 cI co ai> xz o 500 c co 3 300 1,000 1,500 2,000 2,500 Snowpack w.e. in band 6, SP'6 , (mm) 2,000 E E, (c) df 1,500 co BAND 6 (starting zone) ai R2 = .92 • 3: 1,000 - o c co d) x: o 500 c JO CO 3 1 00 1,000 1,500 2,000 2,500 Snowpack w.e. in band 6, SPe , (mm) Fig. 5.20. Extra snow w.e. required to increase in runout bands and to decrease in starting zone band with respect to snowpack w. e. in band 6. a) band3; b) band 4; and c) band 6. 112 Once the extra required snow depth is calculated from equations [5.5] to [5.7] for each band, the next step is to find out the corresponding value of the precipitation increase factor, POPADJ, needed to calculate Sp, given by; POPADJ =-^P- [5.8] SPa where; Sp is extra snow required at band under consideration, and SP'a is snow depth in part 'a' of that band i.e., under un-affected area of band. The UBC model was then run using these POPADJ values calculated from equation [5.8] for the affected part of each band. SP'3b, SP'4b, and SP'6b were noted from the band CAL files. Now the new value of extra snow depth was calculated from the following equation; Sp' = SP'b - SP [5.9] This value of Sp' will be slightly different from Sp and will be used to calculate the affected area given by; A=(A/Sp')*C [5.10] where; A = avalanche affected area of a band, A = avalanche volume of that band calculated from equation [5.2] to [5.4], Sp' = extra snow depth calculated from equation [5.9], and 113 C = a constant to give area in km for each 1000 kmz of watershed. Since the avalanche volume A (in million cubic metres) is for the actual area of band, and Sp' is in millimeters, thus, for example, C = 1000/2.34 = 427.35. Since no data was available after 1988 for validation of the derived equations, the period of data, i.e., 1979-88 was used to forecast the avalanche volumes and avalanche affected areas. Calculations of these two parameters and results are given in tables from 5.12 to 5.15. Figures from 5.21 to 5. 24 show the observed, avalanched and forecast hydrographs, and their efficiency and deviation in discharge are compared in table 5.16. 114 Table 5.12. Avalanche forecasting for year 1979-80 and 1980-81. Year: 1979-80 Year: 1980-81 Parameters Band 3 Band 4 Band 6 Band 3 Band 4 Band 6 i) A (x 106m3) 71.19 65.88 137.06 98.5 98.28 196.79 ii) Sp (mm) 614 778 955 779 964 1089 iii) POPADJ 4 2.7 -.62 =Sp /SP'a 3.6 2.1 -.61 iv) SP'b (from band CAL file) 938 1189 605 1047 1378 657 950 851 1020 1110 v) Sp' = SP'b - SP'a 767 817 vi) A=(A/Sp')* C 39.66 34.46 61.66 49.46 41.18 75.76 Table 5.13. Avalanche forecasting for year 1981-82 and 1982-83. Year: 1981-82 Year: 1982-83 Parameters Band 3 Band 4 Band 6 Band 3 Band 4 Band 6 i) A (x 106m3) 54.43 46.03 100.45 96.71 96.13 192.82 ii) Sp (mm) 512 632 794 768 1000 1191 iii) POPADJ 2.5 2.1 2.6 1.8 =Sp /SP'a -.56 -.68 iv) SP'b (from band CAL file) 745 944 627 1155 1616 562 640 798 862 1053 1191 v) Sp' = SP'b - SP'a 535 vi)A = (A/Sp')*C 43.48 30.74 53.79 47.95 39.01 69.19 115 Table 5.14. Avalanche forecasting for year 1985-86 and 1986-87. Year: 1985-86 Year: 1986-87 Parameters Band 3 Band 4 Band 6 Band 3 Band 4 Band 6 i) A (x loV) 135.63 142.26 277.88 170.95 184.11 355.05 ii) Sp (mm) 1004 1338 . 1557 1217 1645 1888 iii) POPADJ 3 8.5 7.2 =Sp /SP'a 5 -.76 -.81 iv) SP'b (from band CAL file) 1459 1807 422 1682 2148 418 1367 1633 1538 1919 1911 v) Sp' = SP'b-SP'a 1256 vi)A=(A/Sp')*C 46.15 44.47 72.72 47.5 41 76.4 Table 5.13. Avalanche forecasting for year 1987-88. Year: 1987-88 Parameters Band 3 Band 4 Band 6 i) A (x 106m3) 132.80 138.90 271.69 ii) Sp (mm) 986 1313 1530 iii) POPADJ =Sp /SF 10.1 4.5 -.75 iv) SP'b (from band CAL file) 1766 1747 507 1444 1526 v) Sp' = SP'b - SP'a 1668 vi)A=(A/Sp')*C 34.02 41.11 76.09 116 Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep 200 Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Fig. 5.21. Comparison of observed, avalanched, and forecasted hydrographs for years (a) 1979-80; (b) 1980-81. 117 200 Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Fig. 5.22. Comparison of observed, avalanched, and forecasted hydrographs for years (a) 1981-82; (b) 1982-83. 118 200 Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep 200 0 ^ Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Fig. 5.23. Comparison of observed, avalanched, and forecasted hydrographs for years (a) 1985-86; (b) 1986-87. 119 200 Observed i Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Fig. 5.24. Comparison of observed, avalanched, and forecasted hydrographs for year 1987-88. 120 Table 5.16. Comparison of efficiency and deviation in discharge between base-line, avalanched and forecasted calibrations. Year Base-line Avalanched Forecasted 1979-80 E! .70 .77 .77 Dv< {%) -1.1 -0.5 -0.29 1980-81 E! .82 .88 .88 Dv {%) 0.3 3.2 2.85 1981-82 E! .70 .73 .72 Dv {%) 0.1 1.7 1.6 1982-83 E! .82 .82 .81 Dv (%) -4 -3.6 -3.5 1985-86 E! .85 .94 .93 Dv (%) -1 2 3 1986-87 E! .78 .88 .86 Dv (%) -5.4 0.96 5.3 1987-88 E! .67 .84 .85 Dv (%) -6.2 2.7 2.2 Average E! .76 .84 .83 Dv {%) -2.5 0.79 1.3 121 Chapter 6 CONCLUSIONS AND RECOMMENDATIONS 6.1 CONCLUSIONS The Kunhar River basin experiences intense snow avalanche activities above 2,000 m elevation. These avalanches significantly modify the hydrology of the basin, hence, become very important phenomena in controlling the volume and especially the timing of the peak flow from the basin. In this thesis the following main objectives have been achieved; 1- The significance of avalanches in calibrating the U.B.C. watershed model has been shown. 2- The effects of avalanches on the hydrologic regimes in the Kunhar River basin have been investigated. 3- Some important parameters of avalanches, such as; areal coverage, depth, active elevation bands, etc. have been identified. 4- An avalanche forecasting model has been developed. 122 6.1.1 Calibration Problems During the calibration processes, two main problems are identified. The first one is the non-representativeness of the snow precipitation recorded at Battakundi station, especially for the high elevations of the watershed where the temperatures are very cold during winter and most of the precipitation is in the form of snow. Hence, first the precipitation gradient is increased for elevations above 3,350 m, then it was found necessary to increase the precipitation adjustment factor for individual years. It is also verified that the temperature data from the Astore station gives a better representation of the snowmelt conditions of the Kunhar basin. The second problem is the redistribution of the snow from higher to lower elevation. This is done by producing avalanches in lower bands and subtracting the equal snow volumes from a higher band. 6.1.2 Significance of Avalanche In model calibration, the hydrographs were modified by avalanche contribution and on average the model efficiencies were improved from 77% to 84%, i.e. on average the avalanches increase the model efficiency by about 10%. The results of this analysis show that the avalanche starting zone is located at band 6 (elevation 4,000 m). The snow then cascades downslope and the runout and track zones are located at bands 3 and 4 with a mean elevation of about 2,450 and 2,800 m, respectively. About 12% to 21% of Kunhar basin is estimated to be affected by avalanche activity. 123 On average, over 200x10 m of snow is avalanched from higher to lower bands annually, which has a direct influence on the snowmelt pattern of the region. As a result of large avalanche activity the snowmelt volume is increased by about 200 to 300% in runout zones, and this delays the snowmelt period by about 30 days in the avalanche areas. The snowpack at band 6 is found to be the single most responsible parameter producing avalanching. On average 16% of the snowpack at this band is estimated to be subject to avalanching, the deposition of which is distributed about equally in bands 3 and 4. The avalanche volumes in these runout zones are strongly correlated with the snowpack accumulation at band 6. The linear regression analyses show very high coefficient of determination, that is, about 90 to 95% of the variation in avalanching is explained by the snowpack volume in band 6. Moreover, the extra snow depth due to avalanche in the runout zones is also heavily dependent on the existing snow depth in band 6 (R2 = 0.93-0.99). The threshold value of snowpack at band 6 to generate 6 3 avalanches is found to be about 600x10 m water equivalent. 6.1.3 Avalanche Forecasting Model On the basis of the above analyses an avalanche forecasting model is developed. The model facilitates estimation of the volume, concentration factor, and areal distribution of snow avalanches. Therefore, if the snowpack volume in band 6 (4,000 m) is known the avalanches volumes alongwith their areal distribution, both in the starting zone and the 124 runout zones can be estimated. The application of the developed avalanche forecasting system for a period of nine years (1979-88) shows a reliable estimation of flow volume and distribution. 6.2 RECOMMENDATIONS It is mentioned above that the avalanche activity in Kunhar basin is related to the high altitude snow conditions. If the knowledge of the snowpack extent at elevations above 3,000 m can be obtained and used in a hydrologic model alongwith the climatic data, then estimates of flows can be made more accurately by applying the avalanche forecasting strategy. 6.2.1 Knowledge of the Snowpack at Higher Elevations As the key factor for avalanche forecasting is high altitude snow conditions, a permanent station above 3,500 m in addition to one low altitude station is necessary. It is commonly experienced that only light valley bottom precipitation may occur when significant precipitation may be occurring at high elevation. Therefore, a high altitude station is required to avoid extrapolation of lower elevation precipitation to higher elevation. Also, the temperature conditions vary significantly in mountain regions and as already mentioned, the Battakundi temperature data is not fully representative to the conditions at higher elevations, more stations are required at different elevations. 125 This important factor is already been taken care off during second phase of the Snow and Ice Hydrology Project. Before commencement of the project, almost all the Hydrometeorological data in Pakistan was for valley bottoms sites between elevations 1,000 m and 2,000 m (BCHIL, 1991). Now the Upper Indus Basin has extensive network of remote stations at least one at low altitude and one at high altitude above 3,500 m. Four climate stations have been established in the Kunhar basin during the second phase of the Project: at Battakundi (El. 2,260 m), Burawai (El. 2,900 m) 20 km northeast of Naran, at Shogran (El. 3,000 m) 24 km northeast of Balakot near the divide between the Kunhar and Neelum basins, and at Babusar Pass (El. 4173). 6.2.2 Snow Course Surveys Snow course measurement especially at higher elevations above 3,500 m is another important factor. If the knowledge of the snow depth, density, and areal extent is combined with the existing snow conditions, more reliable forecasting of the potential avalanche activity can be obtained. 6.2.3 Testing of the Strategy The same strategy for development of avalanche forecasting system should be applied to other similar sub-basin within the region, such as Neelum River basin. This river is the largest tributary of the Jhelum within Pakistan and Azad Kashmir and also fed by avalanches. It has more or less same meteorological and hydrological features as the 126 Kunhar basin. Therefore, the testing of the present forecasting system can be done by calibrating the watershed model for the Neelum River Basin using avalanche contribution. 6.2.4 Model Modification The avalanche estimation system developed in this thesis can be made automatic, because all factors have been found to be linearly related to snowpack water equivalent at the 4,000 m level. On the basis of this developed system the U.B.C. Watershed Model can be easily modified to carry out the avalanche analyses automatically. After modification the model will take account of the important avalanche parameters on the basis of high altitude snowpack. These parameters are areal coverage of avalanche, active avalanche zones, and volume displaced by avalanches. ********* 127 REFERENCES Abbas, B.M. 1967. Planning Water Development in Pakistan. Indus Vol. 10 No. 9 October 1967. Wapda Printing Press Pakistan. Adam, K.M. 1981. Travel over snow. In: Handbook of Snow - Principles, Processes, Management & Use. Ed: Gray, D.M., and Male, D.H. Division of Hydrology, University of Saskatchewan, Saskatoon, Canada. Pergamon Press, pp. 521-561. Akkouratov, V.N. 1966. Meteorological Conditions of Avalanche Formation in the Khibiny. In: International Symposium on Scientific Aspects of Snow and Ice Avalanches. 5-10 April, 1965 DAVOS, Switzerland. Pub. No. 69 IASH. Gentbrugge (Belgique). 35-42. Anand, Verdhen & T. Prasad. 1993. Snowmelt Runoff Simulation Models and their Suitability in Himalayan Conditions. Snow and Glacier Hydrology (Proceedings of the Kathmandu Symposium), Nov. 1992. IAHS Publ. No. 218, 1993. pp. 239- 248. Armstrong, Betry and Williams, Knox. 1975. The Avalanche Book. Fulcrum, Incorporated. Golden, Colorado, p. 231. Armstrong, Richard L. 1973. Avalanche studies in the San Juan Mountains of Southwestern Colorado. Proceedings of the Western Snow Conference, 41st. Annual Meeting. April 17-19, 1973. Colorado State University, Colorado pp. 44- 51. 128 Armstrong, Richard L. 1976. The application of isotopic profiling snow gauge data to avalanche research. Proceedings of the Western Snow Conference, 44th. Annual Meeting., April 20-22, 1976. Colorado State University, Colorado pp. 12-23. Bahadur, Jagdish. 1993. The Himalayas: A Third Polar Region. Snow and Glacier Hydrology (Proceedings of the Kathmandu Symposium), Nov. 1992. IAHS Publ. No. 218, 1993. pp. 181-190. Barry, Roger G. 1981. Mountain Weather and Climate. Methuen London and New York. 313 p. Barry, Roger G. 1992. Mountain Weather & Climate. Routledge, Chapman and Hall, Inc. New York. 402p. Barry, Roger G. and Chorley R. J. 1976. Atmosphere, Weather and Climate. Methuen & Company Limited. 432 p. Batura Glacier Investigation Group. 1977. The Batura Glacier in the Karakoram Mountains and its Variations. Scientia Sinica. Vol. XXII, No. 8, April 4, 1977, pp. 958-978. BCHIL. 1991. Pakistan Snow and Ice Hydrology Project. Annual Report 1991. B.C. Hydro International - WAPDA. Report No. BCHIL.007 October 1991. Bell, Ian; Gardner, James; and De Scally, Fes. 1990. An estimate of snow avalanche debris transport Kaghan Valley, Himalayas, Pakistan. Arctic and Alpine Research. 22(3), 1990. Burbank, Douglas W. and Fort, Monique B. 1985. Bedrock control on glacier limits: examples from the Ladakh and Zanskar Range, North-Western Himalayan. Journal ofGlaciology. Vol. 31, No. 108. 143-149. 129 Chalise, S.R. regional Cooperation in Hydrological Research and Training in the Hindu Kush - Himalayas. Snow and Glacier Hydrology (Proceedings of the Kathmandu Symposium), Nov. 1992. IAHS Publ. No. 218,1993. pp. 37-47. De Scally, F. A. 1992. Influence of avalanche snow transport on snowmelt runoff. Journal of Hydrology, 137 (1992). Elsevier Science Publishers B. V., Amsterdam. 73-97. De Scally, F. A. 1994. Relative importance of snow accumulation and monsoon rainfall data for estimating annual runoff, Jhelum basin, Pakistan. Hydrological Sciences- Journal, Vol. 39. No. 3. pp. 199-216. De Scally, F. A. and Gardner, J.S. 1986a. Avalanche hazard in Kaghan Valley, Himalayan Range, Pakistan. Proceedings of International Snow Science Workshop, 22-25 October 1986, Lake Tahoe, California. ISSW Workshop Committee, Homewood, CA/University of California, Santa Barbara, CA. pp. 21- 28. De Scally, F. A. and Gardner, J.S. 1986b. Avalanche hazard in Kaghan Valley, Pakistan. Working Paper No. 2, Snow and Ice Hydrology Project, Wilfrid Laurier University. Waterloo, Ontario. De Scally, F.A. 1989. The role of avalanche snow transport in seasonal snowmelt, Himalayan Mountains, Pakistan. Ph. D. Thesis, University of Waterloo. De Scally, F.A. and Gardner, J.S. 1988. The Hydrological Importance of Avalanche Snow, Kaghan Valley, Himalayan Mountain, Pakistan. In: Proceedings of the International Snow Science Workshop. Canadian Avalanche Association, October 1988. Whistler, B.C. Canada, pp. 277-283. De Scally, F.A. and Gardner, J.S. 1989. Evaluation of Avalanche-mass determination Approaches: An Example from the Himalayan, Pakistan. Journal of Glaciology. Vol. 35, No. 120,1989. pp. 248-252. 130 De Scally, F.A. and Gardner, J.S. 1990. Ablation of avalanched and undisturbed snow, Himalayan Mountains, Pakistan. In: Snow and Ice Hydrology Project, Upper Indus Basin - Final Report Volume II. S.I.H.P. Published Papers, Canadian Centre. De Scally, F.A. and Gardner, J.S. 1994. Characteristics and Mitigation of the Snow Avalanche Hazard in Kaghan Valley, Pakistan Himalayas. Natural Hazards No. 9. Kluwer Academic Publishers. Printed in Netherlands, pp. 197-213. Elmesov, A.M. and Nastaev, M.Kh. 1973. Snow cover and the possibilities of forecasting its collapse from the slopes. Physics of snow, avalanches and glaciers. Ed: G.K. Sulakvelidze and M.A. Dolov. Indian National Scientific Documentation Centre, New Delhi, pp. 166-189. Fraser, Colin. 1966. The Avalanche Enigma. John Murray. London. William Colowes and Sons, Limited, p.301. Goudie, Andrew. 1993. The Nature of the Environment. Blackwell. Oxford UK & Cambridge USA. 397p. Hewitt, Kenneth. 1985a. The Snow and Ice Hydrology Project: An Overview. Snow and Ice Hydrology Project Annual Report. 1985. Hewitt, Kenneth. 1985b. Snow and Ice Hydrology in remote, high mountain basins: the Himalayan sources of the River Indus. Snow and Ice Hydrology Project, Working Paper #i, Wilfrid Laurier University, Waterloo, Canada, 29p. Hewitt, Kenneth. 1988a. Snow and Ice Hydrology in Remote High Mountain Regions: The Himalayan Source of the River Indus. Snow and Ice Hydrology Project Working Paper No. 1. 29p. Hewitt, Kenneth. 1988b. The Snow and Ice Hydrology Project: Research and Training for Water Resource Development in the Upper Indus Basin. Journal of Canada - Pakistan Cooperation. Volume-2, No.l. pp. 63-72. 131 Hewitt, Kenneth. 1990a. Himalayan Glaciers and Snowfields: Research into a major water resource. Snow and Ice Hydrology Project: Final Report II. SIHP Published Papers, Canadian Centre. Hewitt, Kenneth. 1990b. Snow and Ice Hydrology Project - Upper Indus Basin. Overall Report. WAPDA - IDRC - WLU, Final Report 1 1990. Iveronova, M.I. 1961. The role of snow avalanches in the formation of relief in Tersk Alatau mountain ranges (quantitative characteristics). IZV. AN SSSR, Seriya Geogr., No. 3, 1961. Iveronova, M.I. 1966. The hydrological role of avalanches. In: International Symposium. Scientific Aspects of Snow and Ice Avalanches. International Association of Scientific Hydrology, Gentbrugge, Publ. 69. pp. 73-77. Kick, W. 1978. Material for a glacier inventory of the Indus drainage basin - The Nanga Parbat Massif. World Glacier Inventory. IAHS-AISH Publ. No. 126 1980, pp. 105-109. KotTyakov, V.M. and M.Ya. Plam. 1966. The Influence of Drifting on Snow Distribution in the Mountains and its Role in the Formation of Avalanches. In: International Symposium on Scientific Aspects of Snow and Ice Avalanches. 5-10 April, 1965 DAVOS, Switzerland. Pub. No. 69IASH. Gentbrugge (Belgique). 53-60. Lachapelle, E. 1966. Avalanche Forecasting - A modern synthesis. In: International Symposium on Scientific Aspects of Snow and Ice Avalanches. 5-10 April, 1965 DAVOS, Switzerland. Pub. No. 69 IASH. Gentbrugge (Belgique). 350-356. Langham, E.J. 1981. Physics and Properties of Snowcover. In: Handbook of Snow - Principles, Processes, Management & Use. Ed: Gray, D.M., and Male, D.H. Division of Hydrology, University of Saskatchewan, Saskatoon, Canada. Pergamon Press, pp. 275-337. 132 Linsley, Ray K.; Kohler, Max A.; and Paulhus, Joseph L.H. 1986. Hydrology for Engineers. Mc Graw-hill Book Company. McGraw-Hill Series in Water Resources and Environment Engineering, p. 508. Lossev, K.S. 1960. Avalanches as the hydrological factor. Meteorology and Hydrology. No. 5. Lydolph, Paul E. 1985. The Climate of the Earth. Rowman & Allanheld Publishers. U.S.A. 386 p. Mani, A. 1984. The climate of the Himalayas. The Himalayas - Aspects of Change by J.S. Lai. India International Centre, new Delhi, pp. 3-15. Martinec, J. 1976. Snow and Ice. Facets of Hydrology, Ed: J.C. Rodda. Wiley, Bristol, pp. 85-118. Martinec, J. and M.R. de Quervain. 1975. The effect of snow displacement by avalanche on snow melt and runoff. In: Snow and Ice Symposium. Proceedings of the Moscow Symposium, August 1971, IAHS. Pub. No. 104,1975. pp. 364-377. McClung, David M. and Schaerer, Peter A. 1981. Snow Avalanche Size Classification. In Proceedings of Avalanche Workshop, 3-5 November 1980, Vancouver B.C. Technical Memorandum No. 133, Associate Committee on Geotechnical Research, National Research Council of Canada, pp. 12-30. Mears, Arthur I. 1979. Colorado Snow-Avalanche Area Studies. Colorado Geological Survey. Dept. of Natural Resources Denver, Colorado, 1979. p. 124. Peev, CD. 1965. Geomorphic activity of snow avalanches, Symposium on Scientific Aspects of Snow Avalanche, Davos, Switzerland, April, 1965. Perla, R.I. 1978. Failure of snow slopes. In; Rockslides and Avalanches - Natural Phenomena, by Barry Voight. Elsevier Scientific Publishing Company New York. pp. 732 - 752. 133 Pipes, A and Quick, M.C. 1987. Modelling large scale effects of snow cover. Large Scale Effects of Seasonal Snow Cover: Proceedings of the Vancouver Symposium, August 1987. IAHS Publ. no. 166. pp. 151-160. Pipes, A and Quick, M.C. 1989. Review of the Mangla Catchment Calibration. UBC Vancouver, Canada. 39p. Poggi, A. and J. Plas. 1966. Conditions Meteorologiques Critiques pour Le Declencheraent Des Avalanches. In: International Symposium on Scientific Aspects of Snow and Ice Avalanches. 5-10 April, 1965 DAVOS, Switzerland. Pub. No. 69 IASH. Gentbrugge (Belgique). 25-34. Quick, M.C. 1990. Modelling and Forecasting. Runoff Hydrology of Glacier and Snow- fed Streams. Handbook of Snow & Ice Hydrology IV, Runoff Hydrology. Snow and Ice Hydrology Project. Upper Indus Basin, Pakistan. Cold Regions Research Centre (WLU). pp. 25-43. Quick, M.C. 1995. UBC Watershed Model Manual. Version 4, 1995. University of British Columbia, Department of Civil Engineering. Vancouver, B.C. Quick, M.C. and Pipes, A. 1977. UBC Watershed Model. Hydrological Science Bull. 22(1). pp. 153-161. Ringenoldus, James C. 1975. Flood Forecasting and Flood Warning Systems. Appraisal of Flood Management Systems in Pakistan. Vol. 1. Harza Engineering International, Pakistan. Schaerer, P.A. 1981. Avalanches. Handbook of Snow: Principles, Processes, Management & Use. Ed: D.M. Gray and D.H. Male. Division of Hydrology, University of Saskatchewan, Saskatoon, Canada. Pergamon Press. N.Y. pp. 475- 518. Schaerer, P.A. 1988. The yield of avalanche snow at Roger Pass, British Columbia, Canada. Journal of Glaciology. Vol. 34, No. 117, pp. 188-193. 134 Singh, Pratap and Quick, M.C. 1993. Streamflow Simulation of Satluj River in the Western Himalayas. Snow and Glacier Hydrology: Proceedings of the Kathmandu Symposium, November 1992. IAHS Publ. no. 218, 1993. pp. 261-271. Sosedov, IS. and Seversky, I.V. 1966. On hydrological role of snow avalanches in the northern slope of the Zailiyusky Alatau, In: International Symposium, Scientific Aspects of Snow and Ice Avalanches. International Association of Scientific Hydrology, Gentbrugge. Publ. 69. pp. 78-85. Tarar, Riaz Nazir. 1982. Water resources investigation in Pakistan with the help of Landsat Imagery - Snow surveys 1975-1978. Hydrological Aspects of Alpine and High Mountain Areas. Proceedings of the Exeter Symposium. July 1982. IAHS Publ. No. 138. pp. 177-190. Tesche, T.W. 1988. Numerical simulation of snow transport, deposition, and redistribution. Proceedings of the Western Snow Conference, 56th. Annual Meeting. April 19-21, 1988. Kalispell, Montana. Colorado State University, Colorado pp. 93-104. Thapa, K.B. 1993. Estimation of Snowmelt Runoff in Himalayan catchments Incorporating Remote Sensing Data.. Snow and Glacier Hydrology (Proceedings of the Kathmandu Symposium), Nov. 1992. IAHS Publ. No. 218, 1993. pp. 69- 74. Tushinskii G.K. 1960. Avalanches, snow banks, and glaciers. Geografgiz, Moscow, 1963. Voight, B. 1990. Snow avalanche hazards and mitigation in the United States. Panel on Snow Avalanches. National Research Council. National Academy Press, Washington, D.C. p. 84. Wapda. 1969. Snow Surveys of West Pakistan. 1961-1968. Lahore Surface Watyer Hydrology Project, WAPDA. 135 Williams, Knox. 1981. Weather and Avalanche Forecasting. In: Proceedings of Avalanche Workshop, November 1980. Vancouver, B.C. Ed: Canadian Avalanche Committee. National Research Council of Canada. Technical Memorandum No. 133 Ottawa, pp. 31-41. Williams, Mark W. and Clow, David W. 1990. Hydrologic and biologic consequences of an avalanche striking and ice-covered lake. Western Snow Conference, 58th. Annual Meeting. April 17-19, 1990, California. Wyman, Robin R. 1995. Modeling Snowpack Accumulation and Depletion. Mountain Hydrology: Peaks and Valleys in Research and Application. Proceedings of a conference, Canadian Society for Hydrological Sciences. Publ. Canadian Water Resources Association, Cambridge Ont. Canada. Young, GJ. and Kenneth Hewitt. 1988. Hydrology Research in the Upper Indus Basin, Karakoram Himalayan, Pakistan. A working paper of geography department, Wilfrid Laurier University, Waterloo, Ontario. Zalikhanov, M.Ch. 1975. Hydrological role of avalanches in the Caucasus. In: Snow and Ice Symposium, Moscow, 1971. International Association of Scientific Hydrology, Gentbrugge. Publ. 104, pp. 390-394. 136 APPENDIX o o co 00 1 II <» y ^ 3 55 U ^ P. S r3*8 137 138 s o a § •aa o s I T3 s •a § o s 6 I II g o 1 a o 1—1 eS 2 th e <*) s 4> •s 'S. 4> a o .a . «> I s canop ; o f th e 1 are a i n =Nort h o. © i G o u 3 la 1 | .S c tio n efo r & nde x •a a (4-1 rt § 8 •s § •a o a o o c a s are a t y o f 11 a tatio r o o •a « «i t •a •a Mea n Dens i Orie n Fore s 3 3 52 £3$ o o o o o >o r- o o o o O IO io © r-- >o ©' CN ©' II CN © ©' -H ~ O - /—\ •3 O IO >o io c/3 H < <• on © H H © © © © IX £ 2 u u U © 139 o o if § •a a PH 111 sC/2 .5 « U 5 cl 5 o -j •s 0 a 01 o 1 I s o o It § •a •8 •a a •s OH a a u O I 0 6 CO T3 o a sag? 1 0 £ * £ 15 § •s I I •a H w S J D 1 o o 5/3 I .... x z S 5 < 3 H H . i 33 3333££ ££ § ££ £ I 140 © © if o § .•a3 §» rt 3 © © •a 60 " §1 rt © c rt « a O 23B © o S3 xi o s if *2 . •« rt N •3 ! 50 o S 2 1 a a ° £ g o rt -.4 -»M S3 •a I C9 M 141 o o p o 1 o §°: g « o rt o o CA a o cu a o 11 13 8 a ON 5 o tS if 3 a CA rt 0) _ a a CO ^ bCA _; CA C •S s rt 3 rt II* cu 60 60 60 X rt - a a 8 ^ >^ rt cu rt g xi x) i> Q o rt a to X! a g 60 CU CU v O rt « « OH fa ^1 X) XI X] 1o ta ta OJ u o o cu 3 § 2 6 XI XI rt 3 CA § •5 £ «2 CA 3 3 I* >, >• rt rt o O rt $ 1 rt ^ s rt 60 CU o a g OH CM 8 CU •a •15 a 3 s a O O 3 o X) «3 <3 73 la CA CA o 1 i i N 1-1 Hi rt 2 T3 "<3 cu I 0-30 to O O ^ XS XI 3 § U rj u ^4 60 60 Ia a rt 2 <*> on o o s & § —1 3 u H H H •S ^ o U U U CM i £££ 00 00 33 142 i © CN if © ^ if CN if CN II* fe II* O. O § O «= S* g to e c a 9 - ^3 « •§ o 3 ?o a w60 s - .8 © s •S Si is lis I § 8 K « u 3 P W u i h 6 „ g § O 5 i II .9 -9 S? 3» a: y £ ^ H « W on on < H h n H 5 is sss :3 £££££££ 3? 143 S cu , « 13 CM O § f•a CO o o © © © © vo © © © t © 33© © © ©' © ©' ©' © ©'©'©' © © ©' ©' ©' © ©' © © © © © © © © cn © © © v-> © •* © © ©• © © © © ©' odd ©* © © ©• ©'©'©' ©' cn cn © © © © © © © © © © © >/n © un un © © ©' ©' ©' © ©' © © © © ©' © ©' ©' ©'©'©' ©' © © © © © © © © © © © vo © m io o © 3 © © © © © ©* ©'©'©' © © ©' ©' ©'©'©' ©' © © cn © © © © © © © vo © © © VO © VO VO © © © © © © ©' © ©'©'©' © © ©' © © © © © © © © © © © © o\ © © © r-~ © S S o © © © © © © © ©'©'©' oo © ©'(NO © © ©' © O un un © © © © un © © © © © © © © VO vo © © © © © ©' ©' © © © © © © © © © a oo ©* 00 144 c a o o R 3 '3 ~ '3 c 2rt rt| 2 ^ I 3 145 a "» 2 +-* rt a> U 2? P 13 5 XI rt III © 0 & 1 o § rt s Q A u 4-1/5> w