REGIONAL FLOW ESTIMATION USING A HYDROLOGIC MODEL

By

ZORAN MICOVTC

B.Sc.(Eng.), The University ofNovi Sad, Yugoslavia, 1994

A THESIS SUBMITED 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 required standard

THE UNIVERSITY OF

July 1998

©ZoranMicovic, 1998 In presenting this thesis in partial fulfilment of the requirements for 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

The University of British Columbia Vancouver, Canada

DE-6 (2/88) ABSTRACT

The twelve watersheds analyzed in this study are heterogeneous in terms of drainage area, climate, topography, soil type, vegetation, geology and hydrologic regime, which indicates that any attempt at a statistical regionalization of streamflow characteristics for these watersheds would be unreliable unless based on a very large number of watersheds. Therefore, the hydrological behavior of these watersheds was analyzed using the UBC Watershed Model.

The watersheds were calibrated until a maximized efficiency was achieved. A sensitivity analysis showed that the model was most sensitive to precipitation parameters and thus, precipitation was the most important factor. Given good precipitation data, the next most important parameter was found to be the fraction of impermeable area in the watershed. Therefore, several methods for estimating this parameter were examined and surficial geology maps gave the best results.

Analysis of all the parameters for each watershed revealed that there was quite a consistent set of parameters for everything except the precipitation gradients and the fraction of impermeable area.

Lack of variability of the parameters affecting the time distribution of runoff with watershed size supports the idea that the land phase controls the runoff process and the channel phase is secondary and appears almost negligible even for the watersheds larger than 1000 km2 in size.

Thus, the watersheds were re-run using the fixed set of parameters and inputting precipitation and fraction of impermeable area for each watershed. The results obtained by this simplified method were then compared with the results of the original calibration for each watershed. The comparison showed that this fixed parameter set provided reliable flow estimates, because the reduction in overall model efficiency ranged from 0 to 10%, and in most cases stayed within 5%.

In addition, this set of parameters considerably simplifies model calibration, and is an excellent first step in obtaining a full calibration. Therefore, this method is very useful for estimating runoff from an ungauged watershed, provided meteorological input is available. ii TABLE OF CONTENTS

ABSTRACT ii

LIST OF TABLES vi

LIST OF FIGURES vii

LIST OF SYMBOLS ix

ACKNOWLEDGEMENTS xi

1. INTRODUCTION 1

1.1 General Introduction 1

1.2 Regional Flow Estimates 2

1.3 Objectives 3

1.4 Study Outline 5

2. LITERATURE REVIEW 7

2.1 Introduction 7

2.2 Land Phase Routing 8

2.3 Channel Phase Routing 11

2.4 Previous Studies 13

2.5 Summary 20

iii 3. STUDY WATERSHEDS 21

3.1 General Information 21

3.2 Physical Description of the Watersheds 21

3.2.1 Barlow Creek 21

3.2.2 Bone Creek 23

3.2.3 Bridge River 24

3.2.4 Campbell River 25

3.2.5 Coquihalla River 27

3.2.6 lllecillewaet River 28

3.2.7 Jordan River 30

3.2.8 Littie Swift River 31

3.2.9 Naver Creek 32

3.2.10 Stitt Creek 33

3.2.11 Tabor Creek 34

3.2.12 Watching Creek 35

4. MODEL CALIBRATION 51

4.1 The Watershed Model 51

4.1.1 The conceptual design 52

4.2 Calibration of the Watershed Model to Study Watersheds 54

4.2.1 Introduction 54

4.2.2 Sensitivity of precipitation parameters 57

4.2.3 Fraction of impermeable area 66

iv 4.2.4 Parameters affecting the time distribution of runoff 76

4.3 Calibration Results 77

5. AVERAGING OF THE PARAMETERS 84

5.1 Parameters Variability Among the Study Watersheds 84

5.2 Reliability of the Results Obtained by Averaged Set of Parameters 95

6. CONCLUSIONS 98

6.1 General Conclusions 98

6.2 Implications for the Watershed Behavior 102

REFERENCES 104

APPENDIX - Calibration Statistics and Verification 107

v LIST OF TABLES

3.1 Physical characteristics of the study watersheds 38

4.1 Calibrated periods for the study watersheds 56

4.2 Analysis of sensitivity of impermeable fraction for the Bone Creek watershed 72

5.1 Calibration values for the parameters affecting the time distribution of runoff 84

5.2 Averaged values for the parameters affecting the time distribution of runoff 88

5.3 Statistical measures of the model performance for both runs (Illecillewaet River) .... 95

5.4 Average statistical differences between flowscalculate d with calibrated and those

with averaged parameters 96

vi LIST OF FIGURES

2.1 Derivation of the time-area diagram 10

2.2 Channel reach storage 12

2.3 The relation between drainage area and peakedness index for Scottish basins 18

3.1 Locations of the study watersheds 37

3.2 Barlow Creek watershed and its area-elevation distribution curve 39

3.3 Bone Creek watershed and its area-elevation distribution curve 40

3.4 Bridge River watershed and its area-elevation distribution curve 41

3.5 Campbell River watershed and its area-elevation distribution curve 42

3.6 Coquihalla River watershed and its area-elevation distribution curve 43

3.7 Illecillewaet River watershed and its area-elevation distribution curve 44

3.8 Jordan River watershed and its area-elevation distribution curve 45

3.9 Little Swift River watershed and its area-elevation distribution curve 46

3.10 Naver Creek watershed and its area-elevation distribution curve 47

3.11 Stitt Creek watershed and its area-elevation distribution curve 48

3.12 Tabor Creek watershed and its area-elevation distribution curve 49

3.13 Watching Creek watershed and its area-elevation distribution curve 50

4.1 Relationship between mean annual flood and drainage area for 12 study watersheds 57

vii 4.2 Relationship between mean annual flood and mean annual precipitation for 12 study

watersheds 58

4.3 Mean annual water yield versus mean annual precipitation for 12 study watersheds . 58

4.4 Sensitivity of the precipitation adjustment factors for study watersheds 60

4.5 Sensitivity of the fractiono f impermeable area for study watersheds 66

4.6 Starting estimates of impermeable fraction versus calibrated values 73

4.7 Surficial geology of the Watching Creek watershed near Kamloops, B.C 75

4.8 Improved estimates of impermeable fraction versus calibrated values 76

4.9 Observed and calculated hydrographs for the 12 studied watersheds 78

5.1 Routing time constants versus drainage area for 12 studied watersheds 85

5.2 Flows calculated with calibrated (Qcal) and averaged parameters (Qcal(avr)) 89

viii LIST OF SYMBOLS

a = coefficient in Eq. 4.6

A = drainage area of the watershed b = coefficient in Eq. 4.6

B = channel width c = wave celerity

C = Chezy's resistance coefficient

Dj = drainage density

D! = coefficient of determination

E! = coefficient of efficiency g = gravitational acceleration i = local inflow into the reach

/ = inflow into the reservoir

K = storage time constant

L = length of the channel reach

L0 = length of overland flow

L5 = sum of the stream lengths for the watershed n = Manning's resistance coefficient

= number of days for daily runs or hours for hourly runs

O = outflow from the reservoir

ix PODZSH = deep zone share

PODZTK = deep zone groundwater time constant

POFRTK = rainfall fast runoff time constant

POFSTK = snowmelt fast runoff time constant

POIRTK = rainfall interflow runoff time constant

POISTK = snowmelt interflow time constant

POPERC = groundwater percolation

POUGTK = upper groundwater time constant

Q = discharge

R = hydraulic radius s2 = variance

S = storage

Sf = friction slope

S0 = bed slope

/ = time v = flow velocity

W = water input to the first linear reservoir in Eq. 2.24 x = distance in the flow direction

X = weighting factor in Eq. 2.9 y = flow depth r = Gamma function cW - volume error

X = ratio of the stream equilibrium time to the catchment equilibrium time ACKNOWLEDGEMENTS

I would like to express my sincere appreciation and gratitude to my research supervisor, Dr.

Michael Quick, who has a very positive impact on my academic as well as personal attitude.

I am particularly grateful to my wife Sandra who contributed her time and knowledge in helping me to set up all watershed files.

I appreciate the assistance I received fromMr . Edmond Yu in the computer works.

I am also thankful to Drs Barbara Lence and Dennis Russell for their comments and suggestions which improved the quality of this thesis.

This research has been partly funded by the Walter C. Koerner Graduate Fellowship.

Finally, I would like to thank my daughter Natasa for being a constant source of joy in my life and Don Pigone for being my role-model.

xi CHAPTER 1

INTRODUCTION

1.1 General Introduction

Water resource management implies a good knowledge of the hydrological regimes of the basins

and understanding of the physical processes involved in streamflow generation, especially for

low-flow or flood periods. Problems for which design floods must be determined include flood

protection works, dams, river crossings, urban drainage, floodplain delineation, culverts, channel

restoration, river diversions and impoundments. Design low flow is often used as the minimum

flow required for favorable water quality conditions and preservation of fish and wild life habitat.

The response of a watershed to precipitation is a very significant relationship in the field of water

resources engineering. However, because of its variability in both space and time, the

characterization of the hydrological regime can be very difficult. The conversion of precipitation

into streamflow is the result of complex interactions between the hydrologic processes and

numerous climatic and physiographic factors. Thus, in the solution of practical problems such as

determination of the design flows, we must make some idealization and simplifying assumptions

hoping that close imitation of nature is achieved.

Two basic approaches to the determination of design flows can be distinguished - statistical frequency analysis and streamflow simulation.

1 The statistical frequency analysis gives us the future probabilities of occurrence by interpreting the past record of streamflow events. We start with a data set of a certain streamflow variable.

That can be any variable of interest (maximum daily discharge, 7-day low flow,...). For a given data set, we fit a theoretical probability distribution, which actually replaces the limited data set we have with the infinite data set we would like to have. The question is how reliable this replacement is. Or can we predict an event with a return period of 200 years by fitting theoretically 25 years of data? This is the main limitation of statistical frequency analysis of streamflow.

Another way to determine design flows is to simulate streamflow from the given basin by using a hydrologic model to transform precipitation input into hydrologic output, i.e. streamflow. This hydrological transformation is a complicated physical process and to design a perfect model is impossible. The development of a hydrologic model involves many starting assumptions and simplifications because of data and computing limitations. Therefore, a compromise between theory and reality in hydrologic modeling is inevitable. Model efficiency can be determined by comparing simulated streamflow with existing streamflow record for the same period.

1.2 Regional Flow Estimates

Most of the streams in the province of British Columbia are either ungauged or have a short period of gauging record. However, for various engineering applications, estimates of design flows on these sites are usually required. These estimates are usually obtained through regional flood frequency analysis. Within the homogenous region, regression equations are developed by regressing flood quantiles estimated at gauged basins against measured basin characteristics. It is

2 then assumed that these equations can be applied to any basin in that region. Although useful, regional regression equations have serious shortcomings. Standard errors of the predicted quantiles are sometimes too high. Often these equations neglect some important physical factors.

At the other times they incorporate physically irrelevant basin characteristics or two or more

strongly correlated basin characteristics ( e.g. basin area and main channel length ).

A regional analysis can also be done using hydrologic ( precipitation - runoff ) models. A model

can be calibrated on a gauged basin in a hydrologically homogeneous region and then applied to an ungauged basin within the same region. But what if regional boundaries are neglected ? Is it

possible to use the same model parameters to estimate flows in hydrologically, climatically and

geographically different regions ?

1.3 Objectives

This study develops a general hydrologic representation for the whole region which does not

have to be homogeneous. Therefore, by inputting different climatic information, the model will

automatically generate the flow estimates. The UBC Watershed Model will be used for that. The

model transforms meteorological input (precipitation and temperatures) into hydrologic output

(streamflow).

Ideally, a hydrologic model would be a true physical representation of the watershed processes.

Using all necessary data as input, such a model could then generate true flow estimates.

However, all the necessary data required to run such an ideal hydrologic model are not available

in real-life applications. In addition, such a model would be so big in detail that it would be

impossible to manage the interrelationships among processes and parameters and their temporal

3 and spatial variabilities. Thus, at the present time, a hydrologic model must be some kind of a compromise between theory and reality.

The UBC Watershed Model was developed to be as simple as possible, yet physically as realistic as possible. This implies that although certain processes are not completely realistically represented within the model, it should be designed to overcome the ignorances of processes and provide the flow estimates close to those historically observed. For example, in the UBC model, each component of the runoff (fast, medium, slow and very slow) is subject to storage routing through a single or a series of linear reservoirs. The use of linear reservoirs guarantees

conservation of mass and accurate and simple balance of the water budget, but is also not physically true since the response of a natural watershed is non-linear. This non-linearity of the watershed response is handled by the soil moisture sub-model which accounts for the

evapotranspiration losses and divides precipitation input into the four previously mentioned

components of runoff. The soil moisture sub-model calculates the soil moisture deficit which

determines the non-linear response of a watershed by distributing rain and snowmelt into the

runoff components. In addition, when rain exceeds a specified threshold, the flash runoff

mechanism is activated, which converts all further rain into the fast runoff. It is clear that runoff

distribution in a natural watershed is more complex than that assumed by the UBC model and,

also, a linear reservoir is a fiction.However , application of the UBC Watershed Model to many

different watersheds has shown that despite these simplifications, the model provides rather

realistic representation of the watershed behavior.

Complicated processes of streamflow generation are simulated through interaction of various

model parameters. Only some of the parameters can be accurately measured. The rest of them

must be determined through the calibration procedure. In addition, there are several key question

4

r that should be answered - are those parameters stable and do they have any physical reality, especially can they be identified from watershed characteristics or are they regionally similar?

Therefore, the objectives of this study are summarized as follows :

• To calibrate the UBC Watershed Model for various watersheds with different physical

characteristics.

• To examine the variability of model parameters among the watersheds.

• To identify parameters with small variability and average their values.

• To run the model with this fixed set of parameters for all the watersheds and assess the

reliability of such streamflow estimates.

• To identify implications which these results have for watershed behavior.

By assigning constant values to some parameters we minimize the number of remaining parameters which have to be calibrated. If this considerably simplified model still produces reliable results for physically different watersheds, it can universally be applied for ungauged basins regardless of their location.

1.4 Study Outline

This thesis is divided into six chapters. The Chapter 2 offers a literature review on theoretical developments in flow routing as well as some previous studies focused on travel times of water through land and channel phase within a watershed. Detailed physical description of each of the

12 studied watersheds is given in Chapter 3. Chapter 4 deals with all aspects of the calibration of the UBC Watershed Model to study watersheds and provides the final results of that calibration.

The simplified approach to model calibration that uses a fixed set of averaged parameters is

5 introduced in Chapter 5. The reliability of the flow estimates obtained by this approach is also analyzed in this chapter. The concluding remarks are given in the sixth and last chapter.

6 CHAPTER 2

LITERATURE REVIEW

2.1 Introduction

Part of the precipitation is intercepted by vegetation and other forms of cover on the watershed and is returned to the atmosphere by evaporation. Precipitation that reaches the ground will infiltrate the soil, unless the ground is impermeable. Part of infiltrated water flows parallel to the soil surface as interflow until reaching the channel, part of it percolates to the groundwater, and one part fills depressions in the ground and eventually evaporates. When precipitation input exceeds the local infiltration rate, and also in impermeable parts of a watershed, water will reach the channel in the form of overland flow.

Therefore, the runoff process of the given watershed is composed of the channel element and the slope element which is attached to the channel element and consists of four zones. These zones can be classified by the time distribution of runoff as fast (overland flow), medium (interflow), slow (upper groundwater) and very slow (deep groundwater) zones. The process in the slope element through which precipitation is transformed into the inflow to the channel element can be called the land phase, while the process of water flowing down in the channel element can be called the channel phase. The finalhydrograp h at the watershed outlet is obtained by routing precipitation input through both the land and channel phase.

7 2.2 Land Phase Routing

* Unit hydrograph

The use in practice of the unit hydrograph method is based on the implicit assumption that the

rainfall-runoff response behavior of the watersheds is time-invariant and linear. The Unit

Hydrograph method was first introduced by Sherman (1932), and assumed that for a given

duration of effective rainfall, the hydrograph time base should remain constant. Sherman defined

"the unit graph" as follows:

If a given one-day rainfall produces a 1 inch depth of runoff over the given drainage area, the

hydrograph showing the rates at which the runoff occurred can be considered a unit graph for

that watershed

Implicit in deriving the unit hydrograph is the assumption that the effective rainfall is uniformly

distributed both in time and space.

The principle of time-invariance assumes that the hydrograph of direct runoff from a watershed,

due to a given pattern of effective rainfall at whatever time it may occur, is invariable. In other

words, it is assumed that physical characteristics of a watershed do not change with seasons.

The principle of linearity assumes that the ordinates of the direct runoff hydrograph of a given

base time are directly proportional to rainfall excess volumes. Thus, application of a unit

hydrograph to rainfall excess volumes other than 1 inch is undertaken by multiplying those

rainfall excess volumes by the unit hydrograph ordinates.

8 * Linear reservoirs

All four previously mentioned runoff zones can be thought of as linear reservoirs, with water

storage determined only by outflow from the reservoir as follows,

S = KO (2.1) where S = storage [m3], K = storage time constant [s], O = outflow from the reservoir [m3/s].

From the principle of continuity,

1-0 = — (2.2) dt

where / = inflow into the reservoir [m3/s] and t = time.

The conceptual model by Nash (1957) in which the flow is routed through a series of linear

reservoirs will be discussed later in this chapter.

* Time-area diagram

The concept of a time-area diagram can also be used to describe the watershed hydrograph. The

method is known as Clark's method after R. H. Clark who developed it in the 1940's. The

effective rainfall is considered to be inflow and the watershed hydrograph to be outflow. Inflow is

lagged by dividing the watershed into zones by isochrones of travel time from the watershed

outlet. The areas between adjacent isochrones are measured and a time-area diagram is plotted

(Fig. 2.1). It should be mentioned that in this method, the channel phase is neglected and travel

time is assumed to be uniform across a watershed.

9 (A)

0 At 2At 3At 4At 5 At (time)

Isochrones

Figure 2.1: Derivation of the time-area diagram

For a storm of duration equal to the interval between isochrones (At), the average runoff can be calculated for each time step and the time-runoff diagram can be plotted. This diagram is actually the inflow hydrograph, which is then routed through storage to give the actual outflow hydrograph from the watershed. If rain lasts more than one time period (At), the individual time- runoff diagrams are lagged and superimposed and the summation is routed. Therefore, unlike the

Unit Hydrograph, this method accounts for temporal and spatial variability of rainfall over a watershed.

10 2.3 Channel Phase Routing

In the channel reach, non-uniform and unsteady flow is described by the Continuity (2.3) and

Momentum Equation (2.4) as follows,

& + B& = i (2.3) dx dt where B - channel width [m], y = flow depth [m] and / = local inflow into the reach [m3/s/m]. s'-k (24)

where S/= friction slope, v = flow velocity [m/s], C = Chezy coefficient, R = hydraulic radius [m]

Friction slope can also be expressed as,

Sf=St-%-l».i% (2.5) dx g dx g dt where So = bed slope, x = distance in flow direction [m] and g = gravitational acceleration [m/s2].

Combining Equations (2.4) and (2.5),

f^ dy v dv 1 dv" v = C\R (2.6) \ 0 dx g dx g dt)

Thus, discharge through the reach is defined as,

V Q= CAJSQ-^.- -^.1^\ (2.7) \ dx gdx gdt)

The flood wave motion within a channel can be approximated by so called "kinematic wave"

explained by Lighthill & Whitham (1955). It is assumed that disturbance is propagated only in the

downstream direction. The wave celerity of such a wave is defined as,

c (2.8) -i« K B dy '

11 Henderson (1966) noted that term —=- is a total derivative and has unique meaning only if Q is dy function ofy alone. This requirement is met in the case of uniform flow. Therefore, the kinematic wave assumption requires that all slope terms other than So be equal to zero in Equation (2.5), which is essentially uniform flow where Sf = So. The other terms than So in the Friction Slope

Equation (2.5) cause dynamic disturbance which propagate upstream as well as downstream.

Henderson (1966) stated that these terms can be assumed negligible in natural rivers with bed

slopes greater or equal to 0.002. Also, in the Continuity Equation (2.3) / is usually ignored and added at the downstream end of each routing step.

The fact that Q depends only onj> implies that the storage in the channel reach is determined only by inflow and outflow. This idea is incorporated in the so-called Muskingum method presented by McCarthy (1938). Such a reach is shown on Fig. 2.2.

i

prism storage: KO

Figure 2.2: Channel reach storage

In this case, unlike a linear reservoir, the storage depends not only on outflow but also on inflow

to the reach. Thus,

S = KO + KX(I-0) = K[XI + (\-X)0] (2.9)

where X is a weighting factor which assigns relative importance of / and O in determination of £.

12 The attenuation effect (reduction in peak discharge) in the channel reach is described by Equation

(2.9). To describe the flow translation effect - that is the time lag of the flood wave, the relationship between flood wave velocity and channel flow velocity has to be determined.

Assumptions are that the channel reach is a wide rectangle in cross-section (R & y) and the

Manning coefficient n is constant along the reach. In this case,

Q = -By%sY2 (2.10) n and,

f = > (2.H) ay 3 Combining Equations (2.11) and (2.8),

c = |v (2.12) which is a relationship between flood wave speed and flow velocity.

For known length of the channel reach (L), travel time of the flood wave along the reach is equal to L/c.

2.4 Previous Studies

For hydrological applications where input data are often scarce, the simple approaches to watershed routing are usually quite satisfactory. Generally, if the land phase is the controlling one, the channel phase can be neglected in a routing procedure and the watershed can be treated as one or a series of linear or non-linear reservoirs.

Nash (1957) proposed a conceptual model in which the watershed is considered as n serially arranged reservoirs with a linear relationship between storage and outflow (Equation 2.1). The

13 first reservoir is assumed to be instantaneously full and thus its inflow is equal to zero (7/ = 0).

The outflow from the first reservoir is the inflow into the second and so forth.

From Equations (2.1) and (2.2),

(2.13) dt

For the first reservoir (Ij = 0), Equation (2.13) becomes,

.dOx Q-Ox=K- (2.14) dt

After solving this differential equation, outflow from the first reservoir is,

(2.15) a K

For the second reservoir (J2 = Oi),

d0 0i-02=K i (2.16) dt

The solution to Equation (2.16) is,

l'-2 (2.20) o2=- 2 K For the n-th reservoir (/„ = 0„.i) the solution is,

0 = (2.21) " KT(n) KKJ where T(n) is the gamma function of n.

If n is integer (usually it is),

r(«) = (n-i)i (2.22)

So, Equation (2.21) becomes,

14 1 r""1 -t/ 0=—J-—e/K (2.23) " K"(n-l)\ where t is the time after the water input has occurred.

The principles of this method are employed in the UBC Watershed Model routing procedure for fast runoff and interflow. After the water input (precipitation) in the first reservoir and the reservoir time constant have been determined, outflow can be calculated. Instead of solving these differential equations, the UBC model uses a stepwise calculation of the outflow from each

successive reservoir.

Thus, for the first reservoir, which has water input W,

W-Oin fl = fl'" + (2-24)

For the n-th reservoir,

O - Oin fl, = Qn + Ji" (2-25) 1 T 1 J\. where superscript in signifies initial values of outflows from the respective reservoirs.

Many hydrologists examined the land and channel phases in attempts to determine which phase is

controlling in formation of the watershed hydrograph.

Kirkby (1976), in his theoretical analysis of the hydrologic response to network topology argued that with increasing drainage area, the travel time of runoff on hillslopes becomes negligible in comparison with the travel time through channel network.

Later, Kirkby (1988) suggested that watersheds may be thought of as a sequence of moisture

stores. Flow from a point on the watershed to its outlet may pass through stores for surface detention, infiltration, unsaturated vertical percolation, saturated downslope flow and channel

15 flow. For any such sequence of storage in series, the behavior is dominated by that of the stores with the longest residence times. The residence times are on the order of:

Surface detention 0.1 -1 h

Infiltration 1 - 20 h

Percolation 1 - 50 h

Downslope flow 1 -12 h

Channel flow 0.5 h (1 km2); 7 h (100 km2) ; 100 h (10000 km2)

Based on this, Kirkby proposed that for a small catchment, it may be efficient to make a model which takes account of infiltration, percolation and downslope flow. For the large catchment, he

suggested taking account of percolation alone and to combine it with the channel routing procedure. The residence times mentioned earlier can be estimated empirically or measured

experimentally. Horton (1945) wrote that the average length of overland flow (L0) can be

estimated as,

4= — (2-26) ° 2Dd

Where Dj is the drainage density defined as,

(2.27)

where Ls = sum of the stream lengths for the watershed and A = drainage area of the watershed.

Ree (1963) used Equations (2.26) and (2.27) and estimated the average length of overland flow to be 28 m for a 0.83 km2 grassed watershed in Oklahoma. Velocity of the overland flow is then

calculated using the Manning equation.

16 However, Dunne (1978) argued that application of the Manning equation to overland flow is questionable, because the mat of vegetation in runoff-producing areas is very thick. On a hillslope with a gradient of 0.4, he measured velocities of overland flow ranging from 0.03 - 0.15 m/s. On hillslopes with lower gradients and thick vegetation covers, however, Dunne measured these velocities to be only 0.001 m/s. In the same paper, he reported average velocities of subsurface stormflow through a permeable forest topsoil to be from maximum value of 1.25-10"4 m/s down to much smaller values.

Ishihara & Kobatake (1975) measured velocities through every moisture store in an experimental sub-basin in Japan. The sub-basin had an area of 0.18 km2 and was covered densely with trees.

The gradients of the mountain slopes and the stream reaches ranged from 0.47 to 0.84 and from

0.09 to 0.27, respectively. The measured velocities were:

Overland flow 0.033 m/s

Prompt interflow 4.1 • 10"4 m/s

Delayed interflow 2.3-IO"4 m/s

Groundwater flow 4.1 • 10'8 m/s

Channel flow 0.47 m/s

Klein (1976) analyzed formation of a watershed hydrograph by introducing a new parameter called the"peakedness index" of the watershed which is defined as mean flow as a percentage of the highest flow. Klein's analysis comprised watersheds of various sizes in Yorkshire, Scotland and Ohio, with similar climate, geology and hydrologic regime. Klein concluded that a break occurs at about 300 km2 between small watersheds with peakier responses and larger watersheds

17 with more attenuated hydrographs, although the data scatter is considerable (Figure 2.3).

However, according to data points on Figure 2.3, a single line would fit just as well as existing two lines.

Figure 2.3: The relation between drainage area and peakedness index for Scottish basins

The break point of the regression line is explained in terms of the contribution of runoff to the channel from overland flow (low gradient regression line) and subsurface flow (steeper line).

Thus, in small watersheds, the hydrograph shape will be dominated by headwater area overland flow, subsurface flow being too highly delayed to contribute to the peak flows, and channel flow being too fast to have a large influence on the hydrograph shape. For a constant channel velocity of 1 m/s and a delay to peak subsurface flow of 8h, runoff from headwater source will flow 28.8 km along the channel network, before the subsurface flow peak reaches the channel banks. Using the empirical relationship between length of channels and basin area,

18 L *A06 (2.28)

for L = 28.8 km, it can be calculated A = 270.6 km2, which is very close to the break point on

Figure 2.3. It is suggested that in watersheds larger than this size, routing times through the channel network may be long enough, resulting in a hydrograph shape dominated by subsurface flow contribution.

However, when I increased channel velocity to 1.5 m/s and delay to peak subsurface flow to 10 h break point in basin area took value of A = 771.5 km2.

Wooding (1965, 66) presented a hydraulic model for the catchment-stream problem. The model is composed of a V-shaped catchment draining into a stream located in the apex of the V. It is assumed that the catchment length is constant everywhere, and the slope and roughness are uniform so that the same depth-discharge relation may be applied at any point. The addition of a stream (channel phase) introduces one further parameter (X), that is the ratio of the stream equilibrium time to the catchment equilibrium time. The equilibrium time is the time required by the flow regime to achieve steady state. If this ratio (X) is very small, the channel effect is negligible. If X is very large, the land phase residence time is small compared with channel residence time and channel phase is controlling one. Finally, if X is of order of unity, the catchment-stream model leads to improvement in the representation of the runoff process.

Wooding tested his catchment-stream model on several natural watersheds in order to obtain values for X,. For the Warragamba watershed in N. S. W., Australia with drainage area of 8758 km2, X was found to be 0.5, which is in disagreement with Kirkbys (1976) argument that the channel travel time is dominant for the watersheds larger than 230 km2 in area.

19 2.5. Summary

Within a watershed, water flows through the hillslopes (land phase) and stream network (channel phase). It is very difficult to present a realistic method of analysis of runoff under these condition due to many interrelated physical processes involved. The travel times through the system depend on many controlling factors - physical properties and depths of the soils, basin topography and climate being most important. However considering previously mentioned velocity measurements in different zones of the land phase, the following can be concluded - the water velocity in the channel network is several orders of magnitude greater than that of the land phase. Therefore, the effect of travel time of the flow in a channel phase seems very small when compared with the observed runoff delays and can be neglected in watershed routing procedure. If this is true, then the land phase of runoff is the main process and a watershed can be reasonably well represented by one or more serially arranged reservoirs with linear or non-linear relationship between storage and outflow. This represents one simple model of the process and appears to give reasonable answers, because it guaranties conservation of mass and accuracy of a water budget balance. An assumption of linearity is commonly considered acceptable for larger watersheds due to dampening of variation in input parameters. However, Loukas & Quick (1992) showed that a small forested, mountainous watershed (3 km2) also has a linear hydrological response to water input (precipitation) due to formation of subsurface pipe flow. A non-linearity in watershed response can be accounted for through soil moisture balance, as mentioned previously on the

UBC Watershed Model example.

20 CHAPTER 3

STUDY WATERSHEDS

3.1 General Information

In this study, hydrological behavior of 12 watersheds in British Columbia was analyzed.

Locations of the watersheds are shown on Figure 3.1. They are chosen to be heterogeneous, with different sizes of drainage area, climate, topography, soil types, geology and hydrologic regime.

Some key factors are summarized in Table 3.1. Following is detailed description of each particular watershed.

3.2 Physical Description of the Watersheds

3.2.1 Barlow Creek

* Location, size and topography

Barlow Creek is located 3 km north of Quesnel city limits. The Barlow Creek watershed is intersected by Highway 97, and covers an area of 71 km2, 87% of which is covered with Cariboo aspen/lodgepole pine forest. It has a southwestern orientation, and an average slope of 5.03%.

This watershed spans elevations from 580 to 975 m, with mean elevation of 745 m.

21 * Climate

This watershed falls on the border of three climatic regions of B.C. - Central Interior, Southwest

Interior and Southeast Interior. Mean annual precipitation is approximately 500 mm, 40% of which falls as snow. Temperatures range from an average of-9.1 °C in January to 16.6 °C in July.

* Soil

The soils are dominantly Luvisolic. Their main feature is a clay accumulation in the subsoil as a result of leaching from above. Sometimes an impervious layer ("hardpan") that restricts root penetration is produced. A small southwestern part of the watershed has shallow Regosolic soils with recent parent materials (alluvium).

* Geology

Surficial geology is a mix of undivided glacio-lacrustine deposits. The watershed is underlain by stratified Mesozoic rocks (Cretaceous and early Tertiary) mainly sandstones and some volcanic rocks.

* Hydrologic regime

The snowmelt season in this watershed generally continues from early March to May. The annual floods are generated mainly from seasonal snowmelt and sometimes from spring rainstorms and therefore, always occur in spring (late March to late April). The mean annual flood is 0.069 m3/s/km2, with a very high coefficient of variation of 0.641. Such a high variation is due to rain- generated floods which are typically much greater than snowmelt floods.

22 3.2.2 Bone Creek

* Location, size and topography

Bone Creek is an eastern tributary of North Thompson River. The Bone Creek watershed is situated on western slopes of the 20 km northeast of Blue River and 30 km northwest of Mica dam on . It is bounded by glaciers and icefields on the north, east and south and North Thompson River valley on the west. This is a rugged mountainous watershed with 268 km2 of drainage area, 53% of which is covered with Subalpine Engelmann spruce and Interior Douglas fir forest and 11.5% with glaciers or icefields. It has western to southwestern orientation, and an average slope of 35.76%, with elevation range from 700 to

2980 m and mean elevation of 1889 m.

* Climate

This watershed belongs to the Southeast Interior climatic region of B.C. but is close to the border with Central Interior Region. Mean annual precipitation is 1420 mm, 50% of which falls as snow. Temperatures range from an average of -9.2 °C in January to 16.5 °C in July.

* Soil

The soils are Podzolic of Humo-Ferric type. These soils are well drained but infertile with iron and aluminum leached from the topsoil into the subsoil. They are associated with dense coniferous forest.

* Geology

Geology of the Bone watershed is characterized with very old, stratified, mainly sedimentary

Proterozoic rocks and minor volcanic rocks.

23 * Hydrologic regime

The snowmelt season in the Bone watershed generally continues from late April to the end of

August followed by the glacier melt season from mid June to mid September. The annual floods are generated from snowmelt or a mix of summer rainfall and glacier melt and always occur in summer (middle of June to late July). Mean annual flood is 0.184 m3/s/km2, and the coefficient of variation of annual floodsi s 0.25.

3.2.3 Bridge River

* Location, size and topography

This watershed consists of that part of the Bridge River drainage area fromit s headwaters below

Bridge Glacier to Lajoie Dam which forms Downton Lake. The watershed is located in Coast

Mountains 60 km northwest of Pemberton. It is bounded by Bridge Glacier on the west, Dickson

Range icefields on the north, several smaller glaciers and icefields on the south, and Carpenter

Lake valley on the east. This typical mountainous watershed covers an area of 984 km2, 53% of which is covered with Subalpine Engelmann spruce/Subalpine firfores t and 18% with glaciers or icefields. The watershed has an eastern orientation, and an average slope of 25.00% with an elevation range from 750 to 2936 m and a mean elevation of 1845 m.

* Climate

The Bridge watershed is on the border of the Southwest Interior and West Coast (sub-region

Head of Fjord) climatic regions. Mean annual precipitation is 1380 mm, 76% of which falls as snow. Temperatures in the lower part of the watershed range froma n average of -5 °C in January to 17.7 °C in August.

24 *Soil

The soils are dominantly Lithic with significant soil inclusions of Ferro-Humic Podzols and

Folisols.

* Geology

The watershed is underlain mostly by granitic intrusive Mesozoic rocks (Cretaceous and Tertiary) except of small portion in the southwestern part which has metamorphic rocks of undetermined age (mainly schist and gneisses).

* Hydrologic regime

The snowmelt season continues from early May to late August followed by the glacier melt season which lasts from early July to late September or early October. The mean annual flood is

0.223 m3/s/km2 and the coefficient of variation of annual floods is 0.319.

3.2.4 Campbell River

* Location, size and topography

The Campbell River watershed is located in the middle part of Vancouver Island, 45 km east of

Georgia Strait. It drains an area upstream of a hydro dam which forms Upper Campbell Lake which is further upstream called Buttle Lake. The main part of this watershed is actually

Strathcona Provincial Park. It is bounded by Vancouver Island ranges on the east and Stratchona

Provincial Park mountains on the south and west. The watershed covers an area of 1194 km2,

72% of which is covered with Coastal western hemlock and Subalpine mountain hemlock forest.

It has a northern orientation, spans elevation from 215 to 2065 m and has an average slope of

29.34% with a mean elevation of 977 m.

25 * Climate

This watershed belongs to the West Coast climatic region of B.C. (sub-region Inner Coast).

Being on the lee side of Vancouver Island the Inner Coast sub-region is afforded some protection from Pacific influences. Consequently, precipitation and cloudiness is reduced and the ranges of temperatures somewhat increased compared to Outer Coast climatic sub-region. The winter weather in the Campbell River watershed is dominated by heavy rainstorms and some snowfall.

Mean annual precipitation is 2430 mm, 33% of which falls as snow. An average temperatures range from 0.9 °C in January to 16.7 °C in July and August.

* Soil

The soils are predominantly Podzolic of Humo-Ferric type.

* Geology

Geology of this watershed is a combination of two types of stratified rocks : mainly sedimentary

Paleozoic rocks (only on the southwest part) and volcanic and sedimentary Mesozoic rocks from

Triassic period.

* Hydrologic regime

The heavy winter rainstorms produce a fast watershed response, and typically spiky annual hydrographs. The annual floods are generated mainly from fall or winter rainstorms, and very rarely from spring rainstorms on shallow snowpack. The annual floods generally occur between late October and early March except of spring rain-on-snow events which occur in April-May period. This watershed is characterized with very high mean annual flood of 0.670 m3/s/km2 and the coefficient of variation of annual floods of 0.405.

26 3.2.5 Coquihalla River

* Location, size and topography

Coquihalla River is an eastern tributary of Fraser River with confluence at Hope. The Coquihalla

River watershed is situated on western slopes of Cascade Mountains. From headwater area in

Cascade Mountains, it slopes down in south west direction to Fraser River valley. The watershed is intersected by Highway #5. It covers an area of 742 km2, 93% of which is covered with combination of Coastal western hemlock, Subalpine mountain hemlock, Interior Douglas-fir and

Subalpine Engelmann spruce forest. The Coquihalla River watershed spans elevation from 90 to

2195 m and has an average slope of 35.01% and a mean elevation of 1203 m.

* Climate

Although this watershed is located in transition between the Southwest Interior and the West

Coast (sub-region Inner Coast) climatic region of B.C., its climate is typical for coastal mountainous watersheds. Winter frontal systems on their way from Pacific coast inland hit the

Cascade Mountains. Warm, moist air is then lifted along the western slopes of these mountains which causes heavy rainfall. As a result, Coquihalla River watershed receives mean annual precipitation of 1800 mm, 63% of which falls as snow. Temperatures at the lower elevation range from an average of 0.7 °C in January to 18.7 °C in August.

* Soils

The soils are dominantly Podzolic of Humo-Ferric type.

* Geology

This watershed has a very complex geology. It is a mix of intrusive and stratified rocks. The intrusive rocks are granitic intrusions from Triassic/Jurassic and Cretaceous/Tertiary periods,

27 respectively. The stratified rocks are sedimentary Paleozoic rocks and Mesozoic sandstones from

Cretaceous/early Tertiary period and volcanic rocks from Triassic period.

* Hydrologic regime

The annual floods in the Coquihalla River watershed can be separated according to generating mechanism on spring snowmelt floods and autumn and winter rainfall floods (with possible sub• group being rain-on-snow floods). Of 36 analyzed years, snowmelt floods occurred in 16 years and rainfall or rain-on-snow floods occurred in 20 years. Snowmelt floods are generated by sudden increase in air temperature in spring which causes melting of accumulated snowpack.

Rain-on-snow floods are most severe and they occur in late autumn or early winter always after cold spell which causes snowpack to extend down to low elevations. The arrival of warm, saturated frontal system from Pacific produces heavy rain and also moves freezing level to higher elevation triggering melting of shallow snowpack. This combination generated the highest floods in the Coquihalla River watershed. Therefore, for this watershed, two values for the mean annual flood are determined. For snowmelt floods it is 0.242 m3/s/km2 with the coefficient of variation

of 0.233 whereas for rainfall and rain-on-snow floods it is 0.385 m3/s/km2 with the coefficient of variation of 0.398.

3.2.6 Ulecillewaet River

* Location, size and topography

Ulecillewaet River is an eastern tributary to Columbia River with confluence at Revelstoke. The

upper part of the Ulecillewaet River watershed is located in , while lower part

slopes down to Columbia River valley. The watershed is bounded by high glaciers and icefields,

28 the most significant being Albert Glacier on the south, Ulecillewaet Glacier on the east and

Dismal and Durrand Glacier on the northwest. Trans-Canada highway intersects the watershed on two almost equal halves. This is rugged mountainous watershed with drainage area of 1150 km2, 74% of which is covered with Interior western hemlock, Subalpine Engelmann spruce and

Subalpine fir forest and 7% with glaciers. It has southwestern orientation and an average slope of

37.32% with elevation range from 520 to 3260 m and a mean elevation of 1717 m.

* Climate

This watershed belongs to the Southeast Interior climatic region of B.C. usually referred to as the

Interior Wet Belt because it receives more precipitation than the Central or Southwest Interior.

Mean annual precipitation is 1715 mm (it ranges from 950 mm at lower elevations to over 2500 mm at higher elevations). In an average year more than 66% of the annual precipitation falls as snow. Temperatures range from an average of -5.6 °C on January to 18.2 °C in July at lower elevations and from -10 °C in January to 12 °C in July at higher elevations.

* Soil

The soils are predominantly Podzolic of Humo-Ferric type

* Geology

Most of the watershed is underlain by stratified Paleozoic rocks. Small part in the southwestern part of the watershed consists of metamorphic rocks of undetermined age.

* Hydrologic regime

The snowmelt season continues from early April to late July, followed by a glacier melt season, once glaciers become snowfree. The annual floods are generated almost exclusively from seasonal snowmelt and occur in the mid May - early July period. The mean annual flood is 0.233 m3/s/km2, and the coefficient of variation of annual floods is 0.199.

29 3.2.7 Jordan River

* Location, size and topography

Jordan River is a western tributary to Columbia River with confluence just north of Revelstoke.

The watershed is placed between Monashee and Selkirk Mountains. Its headwaters originate from Jordan Range, from where it slopes down in southeastern direction towards the Columbia

River valley. The Jordan watershed covers an area of 272 km2, 61% of which is covered with

Interior western hemlock, Subalpine Engelmann spruce and Subalpine fir forest and 2% with icefields. It spans elevations from 540 to 2600 m with a mean elevation of 1593 m, and has an average slope of 37.76%

* Climate

The Jordan watershed belongs to the Southeast Interior climatic region of B.C. Mean annual precipitation is 1925 mm, 70% of which falls as snow. An average temperatures at the lowest part of watershed range from -5.6 °C in January to 18.2 °C in July. At the upper part of the basin these temperatures are much lower due to orographic temperature gradients.

* Soil

The soils are Podzolic of Humo-Ferric type.

* Geology

Geology of this watershed is rather uniform and consists of metamorphic rocks (mainly schists and gneisses) of undetermined age.

* Hydrologic regime

The snowmelt season continues from April to late July. The annual floods are snowmelt generated and occur in the May-July period. The mean annual flood is 0.404 m3/s/km2 with the coefficient of variation of 0.285.

30 3.2.8 Little Swift River

* Location, size and topography

Little Swift River is located in the Interior Plateau 45 km northwest of Quesnel Lake. This is typical watershed for this region, with drainage area of 133 km2, 87% of which is covered with

Interior western hemlock, Subalpine Engelmann spruce and Subalpine fir forest. It has a southwestern orientation, and spans elevations from 1065 to 1980 m, with a mean elevation of

1466 m and an average slope of 15.09%.

* Climate

The watershed belongs to Southeast Interior climatic region of B.C. Mean annual precipitation is

900 mm, 57% of which is snow. Temperatures range from an average of -9.2 °C in January to

12.2 °C in July.

* Soil

The soils are Podzolic of Humo-Ferric type.

* Geology

The watershed is underlain by stratified, mainly sedimentary Proterozoic rocks.

* Hydrologic regime

The snowmelt season continues from early April to late June. The annual floods are mainly result of either snowmelt or combination of summer rainstorms and snowmelt, and generally occur in the May-June period. The mean annual flood is 0.188 m3/s/km2 and the coefficient of variation of annual floods is 0.347.

31 3.2.9 Naver Creek

* Location, size and topography

Naver Creek is a tributary of Fraser River with confluence at Hixon. The watershed is located in the Interior Plateau 60 km south of Prince George. It is intersected by Cariboo Highway and many smaller roads. The Naver Creek watershed has a drainage area of 658 km2, 95% of which is covered with Cariboo aspen and lodgepole pine forest. It is rather difficult to determine orientation of this basin, because Naver Creek flows down to the south from its headwaters and then suddenly makes 180° turn and flows to the north to its confluence with Fraser River. This is a flat watershed, with an average slope of 7.54%. The elevations range from 600 to 1676 m, with a mean elevation of 961 m.

* Climate

The watershed is on the border between the Central and Southeast Interior climatic regions of

B.C. Mean annual precipitation is 800 mm, 48% of which is snow. An average temperatures range from -8.2 °C in January to 16.3 °C in July.

* Soil

The soils are dominantly Luvisolic (Gray Luvisol).

* Geology

Surficial geology consists mainly of ground moraine deposits and some proglacial stratified drift.

This watershed has a very complex bedrock geology. The north part of the watershed is underlain by granitic intrusive Mesozoic rocks (Cretaceous and Tertiary), while the rest of the watershed consists of mix of stratified Proterozoic and Mesozoic rocks. Later are the combination of Cretaceous sandstone and Triassic volcanic and sedimentary rocks.

32 * Hydrologic regime

The floods in this watershed are generally generated by two different mechanisms - the combination of spring rainfall and snowmelt and summer or autumn rainstorms. The later are most severe although they are not frequent. The mean annual flood is 0.122 m3/s/km2 and the coefficient of variation of annual floods is 0.290.

3.2.10 Stitt Creek

* Location, size and topography

This is a small, very steep mountainous watershed located on western slopes of Selkirk

Mountains, 35 km east of Columbia River and 60 km north of Revelstoke. The Stitt Creek watershed is completely surrounded by glaciers - Remillard on the west, OK on the north,

Austerity on the east and Goldstream and Sir Sandford on the south. This watershed covers an area of 139 km2, 41% of which is covered with Subalpine Engelmann spruce and Subalpine fir forest and 13% with glaciers. It has a southwestern orientation, and an average slope of 40.34%, with elevation range from 790 to 2925 m and a mean elevation of 1947 m.

* Climate

This watershed is located in the Southwest Interior climatic region of B.C., but very close to the

Central Interior region which results in somewhat reduced precipitation compared with those in

Ulecillewaet and Jordan watersheds. Mean annual precipitation is 1530 mm, 79% of which is

snow. An average temperatures range from -10.3 °C in January to 12.9 °C in July.

* Soil

The soils are mostly Podzolic of Humo-Ferric type

33 * Geology

The geology of this watershed is combination of stratified, mainly sedimentary Proterozoic and

Paleozoic rocks.

* Hydrologic regime

The snowmelt season begins in late April and ends by late July. In this watershed, the glacier melt plays significant role in annual water yield (more than 20%). The glacier melt season continues from mid June to early October. The annual floods are snowmelt generated and occur in the late

May-early July period. The mean annual flood is 0.273 m3/s/km2 with the coefficient of variation of 0.283.

3.2.11 Tabor Creek

* Location, size and topography

Tabor Creek is an eastern tributary of Fraser River with confluence just south of Prince George.

The Tabor Creek watershed is located 5 km southeast of the Prince George city limits and is rather urbanized. It covers an area of 113 km2, 85% of which is covered with Sub-boreal spruce forest. In addition, there is 50 km of roads in this watershed. The Tabor Creek watershed has a southern orientation and an average slope of 14.50%. It spans elevations from 645 to 1270 m, with a mean elevation of 823 m.

* Climate

The watershed is located on the border of the Central Interior and the Southeast Interior climatic regions of B.C. Mean annual precipitation is 595 mm. In an average year 40% of the annual

34 precipitation falls as snow. Temperature range from an average of-9.9 °C in January to 15.3 °C in July.

* Soils

The soils are dominantly Luvisolic.

* Geology

Surficial geology is characterized by undivided glacio-lacrustine deposits. Bedrock geology consists of stratified, mainly sedimentary Proterozoic rocks and in very small portion of stratified

Mesozoic rocks (mainly sandstone and shale fromCretaceou s and early Tertiary period)

* Hydrologic regime

Annual floods are the result of spring rainstorms combined with snowmelt. They typically occur in the April-May period. The mean annual flood is 0.084 m3/s/km2, with the coefficient of variation of 0.447.

3.2.12 Watching Creek

* Location, size and topography

This watershed is located 20 km northwest of the city of Kamloops, in the Tranquille Plateau. It has a drainage area of 80 km2, and is completely covered with forest (Interior Douglas-fir and

Subalpine fir). It has a southern orientation and an average slope of 11.83%. The Watching

Creek watershed spans elevations from 900 to 1830 m with a mean elevation of 1384 m.

* Climate

This watershed belongs to the Southwest Interior Climatic region of B.C., which is the driest and hottest part of the province. Mean annual precipitation is 560 mm, 72% of which falls as snow.

35 Temperatures in the lower part of the watershed range from an average of-4.8 °C in January to

20.8 °C in July.

*Soil

The soils are predominantly Luvisolic.

* Geology

Surficial geology is a mix of undivided moranial deposits with some rock outcrop or near-surface rock areas. The watershed is underlain by two types of stratified rocks - mainly sedimentary

Paleozoic rocks on the northeast and Cenozoic shales and sandstones on the southwest.

* Hydrologic regime

The snowmelt season continues from mid April to late June. The annual floods are always generated from snowmelt and occur in the April-May period. The mean annual flood is 0.062 m3/s/km2 and the coefficient of variation of annual floods is 0.423.

36 1 - Barlow Creek (70.7 km2) 7 - Jordan River (272 km2 )

2 - Bone Creek (268 km2) 8 - Little Swift River (133 km2)

3 - Bridge River (984 km2) 9 - Naver Creek (658 km2)

4 - Campbell River (1194 km2) 10 - Stitt Creek (139 km2)

5 - Coquihalla River (742 km2) 11 - Tabor Creek (113 km2)

6 - Ulecillewaet River (1150 km2) 12 - Watching Creek (80 km2)

Figure 3.1: Locations of the study watersheds

37 Fraction of Drainage Average Glaciated Forested Mean Mean Mean snow in the Minimal Elevation Mean Impermeable Latitude Longitude area surface fraction fraction annual annual annual mean elevation range elevation fraction slope flood precip. annual

yield precip.

i

Jo Jo n s 6- I I UIU a E mm B t- o\ r-~ Tt m CN Tt »-H in o o ro >n o 0 Tt o o o 0 Barlow 53/01 to 122/19 to 70.7 0.0503 0.0689 Creek 53/08 122/30 Tt t— m o f- o o >n ro fN VO 0 Bone 52/11 to 118/47 to 0.3576 11.5 0.1840 1104 1420 2280 1889 Creek 52/24 119/11 fN Tt r-~ m o ro ON 0 Tt Bridge 50/41 to 122/46 to 0.2500 18.29 0.2232 1189 1380 2186 1845 River 51/00 123/44 0 ON r» CN ro ro CN in o CN Campbel 49/27 to 125/18 to 1194 0.2934 0.6697 1803 2430 2065 River 50/00 125/56 ON CN VO ro ON © O ON ro t-» Tt CN Coquihalla 49/15 to 121/01 to 0.3501 0.3212 1216 1800 2105 1203 River 49/39 121/26 VN CN VO VO V) CN o r» Tt Illecillewaet 50/56 to 117/24 to 1150 0.3732 6.61 0.2333 1374 1715 2740 1717 River 51/26 118/05 in Tt © Tt o VO CN t- CN Jordan 51/01 to 118/15 to 0.3776 1.54 0.4042 1701 1925 2060 1593 River 51/14 118/34 Tt >n © ON vo in 0 Naver 53/13 to 122/07 to 0.0754 0.1217 1076 Creek 53/33 122/38 tN r~- ON r- ON o Tt ro ON Stitt 51/37 to 117/55 to 0.4034 13.31 0.2725 1641 1530 2135 1947 Creek 51/44 118/10 to fN ON m VO CN m O fN in ON in Tt NO Tt *—i 0 ON 0 in ro © Tabor 53/47 to 122/27 to 0.1450 0.0837 Creek 53/57 122/39 CN O O ON ro o in VO o CN ON CN CN VO o o o o o Watching 50/51 to 120/25 to 0.1183 0.0623 1384 Creek 50/58 120/35 T3 •a H ro PH I CO tn O tn O o ti 38 N

5km

Barlow Creek (70.7 km )

3500

3250 -|

3000

2750

2500 4

2250 4

£ 2000 -f 2 2 1750 -1 1500 -

1250 -

1000 - 9?5

750 4 580

500 -+- -+- —I— —(— 10 20 30 40 50 60 70 80 90 100 Fraction of ana (%)

Figure 3.2: Barlow Creek watershed and its area-elevation distribution curve

39 Glaciers

Bone Creek (268 km2) {GI. 30.9 km2}

3500

3250

3000

2750

2500 1 2250 | 2000 2 2 1750

1500

1250

1000

750

500 0 10 20 30 40 50 60 70 80 90 1 00 Fraction of ana (%)

Figure 3.3: Bone Creek watershed and its area-elevation distribution curve

40 Figure 3.4: Bridge River watershed and its area-elevation distribution curve 3500

3250

3000

2750

2500

2250 g 2000 J2065 I 1750 2 ig 1500 1250 1000 750 500 250 0 0 10 20 30 40 50 60 70 80 90 100 Fraction of area (%)

Figure 3.5: Campbell River watershed and its area-elevation distribution curve

42 Figure 3.6: Coquihalla River watershed and its area-elevation distribution curve Ulecillewaet River (1150 km2) {GI. 76 km2}

3500

0 10 20 30 40 50 60 70 80 90 100 Fraction of area (%)

Figure 3.7: Ulecillewaet River watershed and its area-elevation distribution curve 10 km

Jordan River (272 km2) {GI. 4.2 km 2)

3500 -E

3250 -

3000 -

2750 -

Fraction of area (%)

Figure 3.8: Jordan River watershed and its area-elevation distribution curve

45 Figure 3.9: Little Swift River watershed and its area-elevation distribution curve

46 Naver Creek (658 km2)

3500

3250

3000

2750

2500 -

•g- 2250 -

| 2000-f 2 i§ 1750 4 1676 1500

1250 -

1000 -

750 -

500 -+- 10 20 30 40 50 60 70 80 90 100 Fraction of area (%)

Figure 3.10: Naver Creek watershed and its area-elevation distribution curve Stitt Creek (139 km2) {GI. 18.5 km2}

Figure 3.11: Stitt Creek watershed and its area-elevation distribution curve

48 5km

Tabor Creek (113 km2)

3500 j

3250 -

3000 -

2750 -

2500 -

| 2250-

| 2000-

500 4 1 1 1 1 1 1 1 1 1 1 0 10 20 30 40 50 60 70 80 90 100

Fraction of area (%)

Figure 3.12: Tabor Creek watershed and its area-elevation distribution curve

49 Watching Creek (80 km2)

3500 3250 -I 3000 2750 2500 12250 4 | 2000 -j i 1830 B 1750 4" 1500 1250 4 1000 900

750 -+- 500 4- 10 20 30 40 50 60 70 80 90 100 Fraction of area (%)

Figure 3.13: Watching Creek watershed and its area-elevation distribution curve CHAPTER 4

MODEL CALIBRATION

4.1 The Watershed Model

The UBC Watershed model was presented by Quick and Pipes (1977), although the earliest ideas date back to 1963 and, thus, a very early version of the model was used for the Fraser River flood forecasting in 1964. The UBC Watershed model calculates daily watershed outflow using precipitation and maximum and minimum temperatures as an input. The model was designed primarily for streamflow simulation from mountainous watershed where streamflow consists of snowmelt, rain and glacier outflow. Since the hydrological behavior of the mountainous watershed is a function of elevation, the model uses the area-elevation bands concept. This concept accounts for orographic gradients of precipitation and temperatures which are assumed to behave similarly for each storm and are dominant gradients of behavior in mountainous areas.

Besides the daily streamflow estimates, the UBC Watershed Model provides information on area of snow cover, snowpack water equivalent, energy available for snowmelt, evapotranspiration and interception losses, soil moisture, groundwater storage and surface and sub-surface components of runoff. All this information is available for each elevation band separately as well as for the whole watershed (average values). This is a continuous hydrologic model, which means that it will provide streamflow output as long as meteorological input is available. The physical description of a watershed is given for each elevation band separately in the form of different

51 variables such as area of the band, forested fraction and forest density, glaciated fraction, band orientation and fraction of impermeable area.

4.1.1 The conceptual design

The UBC Watershed Model was designed to run from a minimum of meteorological and flow data, because these data are often sparse in the mountainous regions. In addition, most of these sparse data are from the valley stations. As a result of these constraints, an important aspect of the model is the elevation distribution of data. The model is made up of several sub-models.

Meteorological sub-model: distributes the point values of precipitation and temperatures to all elevation zones of a watershed. Both precipitation and temperature are assumed to vary with elevation. The variation of precipitation with elevation controls the total amount of water which is input to the model. The variation of temperature with elevation controls whether precipitation falls as rain or snow and also controls the melting of the snowpacks and glaciers.

The model uses variable temperature lapse rates based on temperature range.

Soil moisture sub-model: represents the non-linear behavior of a watershed. The impermeable area is the fast responding region of a watershed and is assumed to be adjacent to a well developed stream channel system. This area changes as a function of soil moisture deficit

(variable source area). Rain or snowmelt from this fast runoff region is then subjected to soil moisture losses and routed through the fast runoff routing process.

The remaining permeable area is subjected to a soil moisture accounting. Both rain and snowmelt are treated equally. The first demand is to satisfy the soil moisture deficit. An actual

52 evapotranspiration rate is computed as a function of this soil moisture deficit. The second demand is to satisfy the groundwater abstraction rate which is assumed to have a fixed maximum value. If there is still an excess of water input, the residual is assumed to become a medium speed runoff, often thought of as interflow. The groundwater abstraction is sub-divided into fixed percentages of upper (slow) and deep (very slow) groundwater.

Therefore, all the non-linearity of the watershed behavior is concentrated into this soil moisture sub-model which sub-divides the water input (rain and snowmelt) into these four components of runoff. It is considered that these different runoff components can be recognized in the outflow hydrograph because of their characteristically different time delays, which can vary by the order of magnitude.

One can argue that there are more than four runoff components (6 or 7), each of them with slightly different time delay. However, the goal in the design of this model was to go with a reasonable minimum of four runoff components.

Routing sub-model: routing is linear which leads to great simplifications of model structure. It guarantees conservation of mass and simple and accurate water budget balance.

Since the non-linear watershed response is handled by the soil moisture sub-model, this linear routing works very well. The total status of the runoff process can be defined by the current outflow value of each runoff component. Because the process is linear, the runoff value is also a measure of the water in storage.

This process avoids convolution, because it automatically accounts for all preceding runoff events. Therefore, only the current values have to be stored, avoiding complex multi-time period

storage and convolution.

53 4.2 Calibration of Watershed Model to Study Watersheds

4.2.1 Introduction

The successful application of a hydrologic model depends on how well the model is calibrated.

Therefore, values for the various model parameters have to be selected so that the model closely simulates the behavior of the study site. In addition, for calibration and verification purposes, historical streamflow records must exist, otherwise there can be no calibration. Both the volumes and patterns of runoff simulated by the model have to show good agreement with those from historical record. This is usually a very tedious trial-and-error procedure, because of many interactions between various interdependent parameters. A model performance is measured visually and statistically. The visual criterion involves plotting the calculated hydrograph on top of the historical one and comparing goodness of fit. The statistical criterion used in the UBC model is the model efficiency suggested by Nash & Sutcliffe (1970), and based on the difference between variances of observed and calculated flow. The variances of calculated flow [^(Qcai)] and observed flow [^(QobJ] can t>e expressed as follows:

n

(4.1)

n

(4.2) where,

n •obs •obs (4.3) n

54 and, n = the number of days for daily runs or hours for hourly runs

Qobs= tne observed flow on day (hour) i

Q'cai= trie calculated flow on day (hour) /

The model efficiency is then given as:

ElsA(Li-AoJi (4.4)

Combining Equation (4.4) with Equations (4.1) and (4.2),

Z(QL-QLY E\ = l--f— —— (4.5) Qobs ~ Qobs) 2i= l (

El relates how well the calculated hydrograph compares in volume and shape to the observed one.

A negative value of El means that the observed mean flows are closer to observed flows than the flows calculated by model. The coefficient of determination (Df) relates how well the calculated hydrograph compares in shape to the observed one. Therefore, it depends only on timing, but not on volume. The coefficient of determination is calculated as follows:

D\ = l-*—H — (4.6) 2(floiw ~ Qobs) i=l

where,

1 f « (4.7) a = cal n ^ 1=1

55 1 " Z Qobs ' Qcal 2J Qobs Z catQc, b = -& n i=l i=l (4.8)

;=i " ^ i=l >

The volume error is given as:

Q s-Q, SV = ob cat (4.9) Q,obs

For a successfully calibrated model, the values of El, D!, and 1 - bV should be close to 1.

The study watersheds described in Chapter 3 were calibrated depending on availability of historical streamflow and meteorological records. In addition, due to computer constraints, the maximum number of years for model to be run was 8. This can be avoided by using the Pentium processor which is capable of running an unlimited number of years. Table 4.1 shows the calibrated period for each of the 12 study watersheds. The calibration statistics are shown in the

Appendix. Where data was available, the calibration was verified for a different period of years

(also shown in the Appendix).

Table 4.1: Calibrated periods for the study watersheds

Watershed Period of calibration Watershed Period of calibration

Barlow Creek 1970 - 1974 Jordan River 1983 - 1988

Bone Creek 1979 - 1982 Little Swift River 1988 - 1991

Bridge River 1988 - 1994 Naver Creek 1971 - 1975

Campbell River 1983 - 1990 Stitt Creek 1987 - 1991

Coquihalla River 1971 - 1976 Tabor Creek 1974 - 1979

Ulecillewaet River 1981 - 1989 Watching Creek 1967 - 1972

56 4.2.2 Sensitivity of precipitation parameters

From the Figure 4.1, it is obvious that mean annual flood per unit area does not depend on watershed size.

200 400 600 800 1000 1200 Drainage area (km2)

Figure 4.1: Relationship between mean annual flood and drainage area for 12 study watersheds

However, the mean annual flood per unit area shows strong correlation with precipitation as shown on Figure 4.2.

57 0 4 500 1000 1500 2000 2500 Mean annual precipitation (mm)

I with glaciers • without glaciers

Figure 4.2: Relationship between mean annual flood and mean annual precipitation for 12 study watersheds

The mean annual water yield is also highly correlated with precipitation (Figure 4.3)

2500

500 1000 1500 2000 2500 Mean annual precipitation (mm)

I with glaciers > without glaciers

Figure 4.3: Mean annual water yield versus mean annual precipitation for 12 study watersheds

58 These graphs indicate that precipitation strongly influences watershed hydrographs. Therefore, special attention should be paid to calibration of precipitation parameters within the model.

The precipitation data can be obtained from the Atmospheric Environment Service (AES) for the various stations throughout the Canada. However, these data are point measurements and thus, have to be distributed over the watershed. Particularly important parameters are the snowfall and rainfall adjustment factors (POSREP and PORREP). They are used to adjust precipitation at each

AES station. If the data from AES stations are reliable for the watershed, POSREP and PORREP are equal to 0, but this usually is not the case. Therefore, precipitation amounts applied to watershed are increased or decreased by certain percentages compared to those recorded at AES station, until a good efficiency is achieved. For example, a POSREP value of -0.15 means that snowfall is reduced for 15%. Figure 4.4 shows that the precipitation adjustment factors showed very high sensitivity for all 12 studied watersheds. Usually, there is a strong correlation between

El and bV. In some cases, the big difference between D! and El (or SV) indicates that even for the rather poor volume estimate, the shape of the hydrograph could still be very good.

The second important group of precipitation parameters are previously mentioned (Chapter 4.1) orographic gradients of precipitation. These gradients are generally the highest on lower mountain

slopes, and get smaller close to the top of the mountain. This applies to the weather side of the mountain range. On the lee side, however, an inverse situation may occur. The best values for the orographic gradients of precipitation are obtained by a trial-and-error procedure until an improved

efficiency is achieved. For the 12 studied watersheds, precipitation gradients took values from 1 to 5, which is a 1 to 5% increase in precipitation per every 100 meters.

In summary, a good calibration of the precipitation parameters is the first and most important step

in model calibration, because the amount of precipitation is highly correlated with the watershed

outflow. Therefore, a high sensitivity of these parameters is not surprising.

59 Barlow Creek

100

Snowfall adjustment factor

Bone Creek

Snowfall adjustment factor

Figure 4.4: Sensitivity of the precipitation adjustment factors for study watersheds

60 Bridge River

Campbell River

100n

-1 -0.5 0 0.5 1 Rainfall adjustment factor

Figure 4.4 (continued): Sensitivity of the precipitation adjustment factors for study watersheds

61 Coquihalla River

Snowfall adjustment factor

Ulecillewaet River

Snowfall adjustment factor

Figure 4.4 (continued): Sensitivity of the precipitation adjustment factors for study watersheds

62 Jordan River

Little Swift River

Snowfall adjustment factor

Figure 4.4 (continued): Sensitivity of the precipitation adjustment factors for study watersheds

63 Naver Creek

100 -i

-1 -0.5 0 0.5 1 Rainfall adjustment factor

64 Tabor Creek

Rainfall adjustment factor

Watching Creek

100

Snowfall adjustment factor

Figure 4.4 (continued): Sensitivity of the precipitation adjustment factors for study watersheds

65 4.2.3 Fraction of impermeable area in the watershed

After the watershed is divided into area-elevation bands, certain physical characteristics have to be defined. Physical characteristics such as forested and glaciated fractions, orientation and density of the forest canopy can be easily measured from the topographic maps and aerial photographs. A rather difficult problem is to determine the fraction of impermeable area in the each elevation band. This parameter is the main subdivision between the water entering the subsurface and water flowing as fast runoff and is a very important variable for each watershed.

Therefore, this study has examined whether this parameter can be determined from watershed information such as topographic, surficial geology and soil maps. The analysis of sensitivity of this parameter is conducted for all 12 study watershed and results are shown on Fig. 4.5. The sample calculation procedure for one of the watersheds (Bone Creek) is given in Table 4.2.

Barlow Creek

100 -i

Fraction of impermeable area (%)

Figure 4.5: Sensitivity of the fractiono f impermeable area for study watersheds

66 Bone Creek

Bridge River

40 -

30 -

20

10 -

0 -I 1 1 1 1 1 I I 1 1 1 0 10 20 30 40 50 60 70 80 90 100 Fraction of impermeable area (%)

Figure 4.5 (continued): Sensitivity of the fraction of impermeable area for study watersheds

67 Campbell River

E! DI 100-dV I

10 20 30 40 50 60 70 100 Fraction of impermeable area (%)

Coquihalla River

El DI — — — 100-dV

10 20 30 40 50 60 70 Fraction of impermeable area (%)

Figure 4.5 (continued): Sensitivity of the fraction of impermeable area for study watersheds

68 Illecillewaet River

o -I 1 1 1 1 1 1 1 1 1 1 0 10 20 30 40 50 60 70 80 90 100 Fraction of impermeable area (%)

69 Little Swift River

40

30

20 --

10

0 1 1 1 1 1 1 1 1 1 1 1 0 10 20 30 40 50 60 70 80 90 100 Fraction of impermeable area (%)

Naver Creek

100 i

Fraction of impermeable area (%) Figure 4.5 (continued): Sensitivity of the fraction of impermeable area for study watersheds

70 Stitt Creek

40

30 -

20

10 ••

0 - 0 10 20 30 40 50 60 70 80 90 100 Fraction of impermeable area (%)

Tabor Creek

71 Watching Creek

100 i

Fraction of impermeable area (%)

Figure 4.5 (continued): Sensitivity of the fraction of impermeable area for study watersheds

Assumed va ues of impermeable area (%) Area Trial Trial Trial Trial Trial Trial Trial Trial Trial Trial (km2) 1 2 3 4 5 6 7 8 9 10 Band 1 22.6 0 0 0 0 0* 10 20 32 48 100 Band 2 64.4 0 0 0 0 5* 17 27 40 63 100 Band 3 96.3 0 0 11 28 42* 56 75 91 100 100 Band 4 66.5 0 50 70 83 95* 100 100 100 100 100 Band 5 18.2 0 63 78 90 100* 100 100 100 100 100 Whole 268 0 16.7 26.7 36.7 46.7* 56.7 66.7 76.7 86.7 100 basin E! (%) 33.59 69.57 80.89 86.95 88.46 86.65 81.65 74.25 64.73 50.49 D! (%) 58.48 83.61 87.40 88.73 88,94 88.64 87.71 86.63 85.55 83.9 100-8V ( 55.31 70.05 78.29 86.38 94.17 98.51 91.52 84.79 77.54 68.32 * Estimates based on lack of vegetation from the topographic map

Table 4.2: Analysis of sensitivity of impermeable fraction for the Bone Creek watershed

72 An accurate determination of impermeable fraction of the watershed is not possible without detailed field measurements, which are in most cases unavailable. In this study, starting values for the fraction of impermeable area were estimated from available data sources such as topographic, surficial geology and soil maps. When watershed hydrographs calculated with these starting values did not show good fit with observed hydrographs, trial-and-error method was used to find the impermeable fractionwhic h provides the best model efficiency. These final values of the impermeable fractions are called "calibrated values" and are plotted on Figure 4.6 against the starting estimates based on natural lack of vegetation from the topographic maps.

Percentage of impermeable area in the watershed

calibrated

Figure 4.6: Starting estimates of impermeable fraction versus calibrated values

It is clear fromFigur e 4.6 that in only some of the watersheds these two values are almost the

same, which means that a natural lack of vegetation estimated from a topographic map is

sometimes a good indicator of impermeable area in the watershed. In addition, for some

73 watersheds (eg. Stitt Creek), the topographic map indicates a presence of "moraine or scree" deposits on the unvegetated areas. The areas with these deposits are permeable and, thus, estimated values of impermeable fraction based on a lack of vegetation should be reduced in these cases.

For some watersheds, estimates of impermeable area based on topographic maps are far from calibrated (i.e. optimal) values. In those cases, the surficial geology map can help estimate the fraction of impermeable area in the watershed. For example, the Watching Cr. watershed is 100% covered with forest. Therefore, an estimate of impermeable fraction based on lack of vegetation would be 0%, which is hardly possible in a natural watershed. According to the surficial geology map for this region, about 18% of the watershed is an area of rock outcrop or near-surface rock and 7% is covered with glacio-lacrustine deposits with ridged or kettled topographic expression

(Fig. 4.7). Both of these areas can be considered impermeable. Therefore, the percentage of impermeable area based on a surficial geology map for this watershed is about 25%, which is very close to the calibrated value of 27%. Unfortunately, surficial geology maps are available for only a small portion of B.C. They were available for four out of twelve watersheds studied here.

A percentage of impermeable area can also be estimated from the soil maps, because certain types of soil (eg. Luvisolic soils) develop an impervious layer as a result of clay accumulation in the subsoil.

Estimation of impermeable fraction for 12 watersheds in this study can be summarized as follows:

• For five watersheds, estimates of impermeable area based on topographic maps (lack of

vegetation and moraine or scree deposits) were good enough.

• For the next four watersheds starting estimates were improved considerably after use of

surficial geology maps. These improvements are shown on Figure 4.8.

74 For the remaining three watersheds, fraction of unvegetated area in the watershed does not correspond with the fraction of impermeable area. The use of surficial geology maps for those regions would help improve estimates of impervious areas, but they are unavailable at the present time.

Figure 4.7: Surficial geology of the Watching Creek watershed near Kamloops, B.C.

75 Percentage of impermeable area in the watershed

30 calibrated

• already fair estimates improved estimates

• sources for improvement unavailable — • line of equality

Figure 4.8: Improved estimates of impermeable fraction versus calibrated values

4.2.4 Parameters affecting the time distribution of runoff

After a good agreement between calculated runoff volumes and observed ones is obtained, the time distribution of runoff has to be simulated. The time distribution of runoff determines the hydrograph shape, and is controlled by two groups of parameters.

The parameters from the first group allocate water input at fast runoff, interflow, upper groundwater and deep groundwater reservoirs respectively. It is already mentioned (Chapter

4.2.3) that the fraction of impermeable area divides water input into the part that flows as fast overland flow and the part that infiltrates to subsurface flow. Infiltrated water must first satisfy

76 soil moisture deficit. A majority of the infiltrated water is then stored in the groundwater, and the rest goes to the interflow. This is determined by "groundwater percolation" parameter

(POPERC). The groundwater is further divided into an upper groundwater and a deep groundwater reservoirs by "deep zone share" parameter (P0DZSFI).

After the amounts of water are allocated by the soil moisture sub-model in each of the watershed altitude zones, they have to be routed to the watershed outlet. This is accomplished by the second group of parameters - the routing time constants for each of the four runoff mechanisms.

These parameters are:

• Snowmelt and rainfall fast runoff time constants (POFSTK and POFRTK)

• Snowmelt and rainfall interflow time constants (POISTK and POIRTK)

• Upper groundwater time constant (POUGTK)

• Deep zone groundwater time constant (PODZTK)

The only way to calibrate parameters affecting the time distribution of runoff is by trial-and-error procedure. However, the starting values of these recession constants can be determined from the observed hydrograph, because of the different time delay signatures of fast, medium, slow and very slow runoff components. For example, the recession constants for groundwater can be analyzed by selecting a recession flow period, such as the end of summer when snowmelt has ceased, or after a large rain event when recessional flowsdominate .

4.3 Calibration Results

All 12 studied watersheds were calibrated until maximum possible efficiency was obtained. The results of the calibration in the form of hydrographs are shown on Figure 4.9.

77 78 Bridge

200

89/Oct.01 89/Dec.31 90/Apr.01 90/Jul.01 90/Sep.30

-Qobs -Qcal

Campbell

600

85/Oct.01 85/Dec.31 86/Apr.01 86/Jul.01 86/Sep.30

-Qobs -Qcal

Figure 4.9 (continued): Observed and calculated hydrographs for the 12 studied watersheds

79 Coquihalla

200

180 - El = 87.07% 160 - dV= 17.83% 140 - Dl = 89.80%

74/Oct.01 74/Dec.31 75/Apr.01 75/Jul.01 75/Sep.30 -Qobs Qcal

Illecillewaet

350

86/OCI.01 86/Dec.31 87/Apr.01 87/Jul.01 87/Sep.30 -Qobs -Qcal

Figure 4.9 (continued): Observed and calculated hydrographs for the 12 studied watersheds

80 Jordan

120

84/Oct.01 84/Dec.31 85/Apr.01 85/Jul.01 85/Sep.30

•Qobs -Qcal

Little Swift

20

18 - El = 92.78% 16 - 6V= 1.01%

14 - Dl = 93.28%

<| 12 + Si 10

.2 8 •a

6 -

4 --

2

1 88/Oct.01 88/Dec.31 89/Apr.OI 89/Jul.OI 89/Sep.30

-Qobs -Qcal

Figure 4.9 (continued): Observed and calculated hydrographs for the 12 studied watersheds

81 Naver

80

70 - £/ = 72.19% SV= 3.75% 60 - Dl = 73.87%

«• 50 +

68/Oct.01 68/Dec.31 69/Apr.OI 69/Jul.01 69/Sep.30

Qobs Qcal

Stitt

40

90/Oct.01 90/Dec.31 91/Apr.01 91/Jul.01 91/Sep.30

Qobs Qcal

Figure 4.9 (continued): Observed and calculated hydrographs for the 12 studied watersheds

82 Tabor

77/Oct.01 77/Dec.31 78/Apr.01 78/Jul.OI 78/Sep.30

-Qobs -Qcal

Watching

69/Dec.31 70/Apr.01 70/Jul.OI 70/Sep.30

-Qobs -Qcal

Figure 4.9 (continued): Observed and calculated hydrographs for the 12 studied watersheds

83 CHAPTER 5

AVERAGING OF THE PARAMETERS

5.1 Parameters Variability Among the Study Watersheds

As already mentioned in Chapter 4.2.4, the values for the parameters affecting time distribution of runoff for all 12 watersheds are obtained through trial-and-error procedure. Those values are shown in Table 5.1.

POPERC PODZSH P0FRTK P0FSTK P0IRTK P0ISTK P0UGTK P0DZTK

mm/day % days days days days days days

Barlow Cr. 7 0.23 0.55 0.80 2 4 18 122 Bone Cr. 38 0.79 1.18 1.30 4 3 29 196

Bridge R. 37 0.04 0.57 1.10 3 8 36 184

Campbell R. 18 0.46 0.38 0.40 2 2 22 72

Coquihalla R. 10 0.20 0.50 0.50 2 2 8 150

Illecillewaet R. 31 0.25 0.78 1.00 2 3 17 168

Jordan R. 39 0.38 0.67 0.70 4 4 44 139

Little Swift R. 34 0.40 0.89 1.00 5 5 8 190

Naver Cr. 29 0.15 0.68 0.70 3 5 7 133 Stitt Cr. 43 0.61 0.94 1.5 5 7 24 133

Tabor Cr. 30 0.20 1.04 1.10 3 4 8 123 Watching Cr. 23 0.29 1.63 1.9 6 6 39 226

Table 5.1: Calibrated values for the parameters affecting time distribution of runoff

84 From Table 5.1, it can be noticed that these parameters show rather low variability among the watersheds, even though the watersheds are very heterogeneous in many aspects. The reason for this is the difference in time delays between channel phase and land phase. The main delay in runoff is in the land phase. Once this land phase runoff reaches the channel, it is routed relatively quickly to the watershed outlet, because the channel flow is very fast and of much shorter duration than the previous land phase flow. Lack of variability of the parameters from Table 5.1 among the studied watersheds supports the idea that the land phase controls the runoff process and the channel phase is secondary and appears almost negligible.

The routing time constants for fast runoff, interflow and groundwater, for all 12 studied watersheds were found to be independent of watershed size as shown on Figure 5.1.

200 400 600 800 1000 1200 Drainage area (km2)

Figure 5.1: Routing time constants versus drainage area for 12 studied watersheds

85 * 20 !

200 400 600 800 1000 1200

Drainage area (km)

«r so i

200 400 600 800 1000 1200

Drainage area (km2)

200 400 600 800 1000 1200 Drainage area (km2) Figure 5.1 (continued): Routing time constants versus drainage area for 12 studied watersheds

86 200 400 600 800 1000 1200

Drainage area (km2)

300

250 +

200 +

>> ra 150 + TJ

100

200 400 600 800 1000 1200

Drainage area (km)

Figure 5.1 (continued): Routing time constants versus drainage area for 12 studied watersheds

The low variability of these parameters and the fact that they do not depend on drainage area indicates the possibility of assuming that these parameters can be taken to be constant for all watersheds in the studied region which in this case is the province of British Columbia. Reliability of the streamflow estimates based on this assumption can then be evaluated.

87 Thus, parameters from Table 5.1 were averaged and results were rounded to the nearest reasonable number. The set of averaged parameters is shown in Table 5.2.

POPERC PODZSH POFRTK POFSTK POIRTK POISTK POUGTK PODZTK

(mm/day) (%) (days) (days) (days) (days) (days) (days)

25 0.30 0.60 1.00 3 4 20 150

Table 5.2: Averaged values for the parameters affecting time distribution of runoff

This means that all parameters in the model were held constant except the precipitation distribution parameters and the fraction of impermeable area. It should be mentioned that the number of parameters in the UBC Watershed Model is pre-calibrated and is usually kept constant all the time. These parameters include all snowmelt, temperature, evapotranspiration and interception parameters.

The model was then rerun for each of the twelve watersheds using fixed set of parameters from

Table 5.2, instead of original parameter values from the Table 5.1. The results of these reruns in the form of hydrographs are plotted on Figure 5.2. The thin lines (Qcai) represent the model results with original parameters and are already shown in the Chapter 4 on Figure 4.9. The thick lines (Qcai(avr)) are obtained by running the model with the set of averaged parameters from the

Table 5.2.

88 Barlow

70/OCI.01 70/Dec.31 71/Apr.01 71/Jul.01 71/Sep.30

Qcal Qcal(avr)

Bone

80 j

70 -

60

Figure 5.2: Flows calculated with calibrated (Qcal) and averaged parameters (Qcal(avr)) Bridge

89/Oct.01 89/Dec.31 90/Apr.01 90/Jul.OI 90/Sep.30

-Qcal •Qcal(avr)

Campbell

600

85/Oct.01 85/Dec.31 86/Apr.01 86/Jul.01 86/Sep.30

-Qcal •Qcal(avr)

Fig.5.2 (continued): Flows calculated with calibrated (Qcal) and averaged parameters (Qcal(avr))

90 Coquihalla

180

74/Oct.01 74/Dec.31 75/Apr.01 75/Jul.01 75/Sep.30

Qcal Qcal(avr)

Ulecillewaet

350 T"

300

250 -

86/Oct.01 86/Dec.31 87/Apr.01 87/Jul.01 87/Sep.30

Qcal Qcal(avr)

Fig.5.2 (continued): Flows calculated with calibrated (Qcal) and averaged parameters (Qcal(avr))

91 Jordan

120 i

84/Oct.01 84/Dec.31 85/Apr.01 85/Jul.01 85/Sep.30

Qcal Qcal(avr)

Little Swift

18 i

88/Oct.01 88/Dec.31 89/Apr.01 89/Jul.01 89/Sep.30

Qcal Qcal(avr)

Fig.5.2 (continued): Flows calculated with calibrated (Qcal) and averaged parameters (Qcal(avr))

92 Naver

68/Oct.01 68/Dec.31 69/Apr.01 69/Jul.01 69/Sep.30

Qcal Qcal(avr)

Stitt

40 -

35

90/Oct.01 90/Dec.31 91/Apr.01 91/Jul.OI 91/Sep.30

Qcal Qcal(avr)

Fig.5.2 (continued): Flows calculated with calibrated (Qcal) and averaged parameters (Qcal(avr))

93 Tabor

77/Oct.01 77/Dec.31 78/Apr.01 78/Jul.01 78/Sep.30

Qcal Qcal(avr)

Watching

o -I I i 1 ' 69/Oct.01 69/Dec.31 70/Apr.01 70/Jul.01 70/Sep.30

Qcal Qcal(avr)

Fig.5.2 (continued): Flows calculated with calibrated (Qcal) and averaged parameters (Qcal(avr))

94 It is clear from Figure 5.2 that model results obtained through the full calibration procedure do not differ significantly from those obtained through much a simpler process (i.e., with fixed set of parameters for all 12 watersheds). However, the reliability of the model results obtained by this simplified method can be determined statistically.

5.2 Reliability of the Results Obtained by Averaged Set of Parameters

For each of the 12 studied watersheds, the model was run twice for the same period of years.

First time with the parameters calibrated separately for each watershed to produce the maximum efficiency and second time with the fixed set of averaged parameters from Table 5.2. The model performance in both runs was measured statistically using statistical measures defined in Chapter

4.2.1 (coefficients of efficiency and determination and volume error). A sample calculation for the Ulecillewaet River watershed is shown in Table 5.3.

E! E!(cal)-E!(avr) D! D!(cal)-D!(avr) 5V 5V'Cal)-oV'avr) year method (%) (%) (%) (%) (%) (%) cal. 91.24 92.27 5.80 81/82 0.23 0.24 -0.89 avr. 91.01 92.03 6.69 cal. 92.02 92.30 5.84 82/83 1.08 1.11 0.27 avr. 90.94 91.19 5.57 cal. 89.48 91.31 14.52 83/84 0.86 0.80 -0.19 avr. 88.62 90.51 14.71 cal. 94.28 94.54 3.03 84/85 0.25 0.32 -0.01 avr. 94.03 94.22 3.04 cal. 91.70 92.44 9.42 85/86 0.07 0.10 -0.14 avr. 91.63 92.34 9.56 cal. 91.28 91.76 0.53 86/87 0.54 0.51 -0.34 avr. 90.74 91.25 0.87 cal. 89.38 90.89 6.60 87/88 0.67 0.75 0.14 avr. 88.71 90.14 6.46 cal. 89.99 93.54 6.99 88/89 -0.33 -0.13 0.08 avr. 90.32 93.67 • 6.91 Average differences 0.42% - 0.46% -0.136% Table 5.3: Statistical measures of the model performance for both runs (Ulecillewaet River)

95 It can be seen that coefficients of efficiency and determination (E! and Df) for the run with the averaged set of parameters are only slightly less than those for the run with the calibrated parameters. In addition, the volume errors (ST) are only a little greater.

The same analysis is done for the all 12 watersheds and results are summarized in Table 5.4. In most cases, the differences between two approaches appear insignificant.

Average difference in El Average difference in Dl Average difference in ST

(%) (%) (%) Barlow Cr. 2.98 0.27 1.09 Bone Cr. 8.26 2.74 2.41 Bridge R. 5.55 4.38 -0.57 Campbell R. 1.71 2.57 0.35 Coquihalla R. 5.85 7.95 0.21 Illecillewaet R. 0.42 0.46 -0.14 Jordan R. 5.37 1.27 0.38

Little Swift R. 0.94 1.22 0.02 Naver Cr. 7.81 9.63 1.27

Stitt Cr. 5.84 1.03 -1.55

Tabor Cr. 4.14 3.31 -0.16 Watching Cr. 11.28 10.56 -1.20

Table 5.4: Average statistical differences between flows calculated with calibrated and those with averaged parameters

These results indicate that reasonably reliable flow estimates can be obtained by assuming

constant values for the parameters affecting the time distribution of runoff regardless of size, soil

type or geology of the watershed. It means that after defining topography, precipitation and

96 fraction of impermeable area, the model can be run using a fixed set of parameters for any watershed. This considerably simplifies model calibration and is an excellent firstste p in obtaining a full calibration, or is very useful for estimating runoff from an ungauged catchment.

97 CHAPTER 6

CONCLUSIONS

6.1 General Conclusions

Hydrological behavior of the twelve watersheds was analyzed using the UBC Watershed Model.

The watersheds are heterogeneous in terms of drainage area, climate, topography, soil type, vegetation, geology and hydrologic regime. The coefficients of variation of mean annual flood for these twelve watersheds range from 0.199 to 0.641, which indicates that any attempt at a statistical regionalization of streamflow characteristics for these watersheds would be highly unreliable.

It is found that mean annual floods per unit area do not depend on the watershed size and are highly correlated with mean annual precipitation. This applies for both watersheds with glaciers and those without them.

Therefore, accurate precipitation data are the first and most important requirement for reliable model results. This was confirmed through analysis of sensitivity of precipitation parameters. In some cases, 10% change in precipitation input caused a drop in the model efficiency of 25%.

The physical characteristics of the watersheds such as forested and glaciated fractions, density of the forest canopy and watershed orientation were measured from the topographic maps and aerial photographs. However, a rather difficult problem encountered in this study was the determination of the fractiono f impermeable area for each watershed. This parameter controls the fractiono f

98 water entering the subsurface and the fraction flowing as fast runoff. The sensitivity analysis for all twelve watersheds showed that the fraction of impermeable area is also quite sensitive parameter, and thus should be determined accurately. In fact, given good precipitation data, the fraction of impermeable area is the next most important factor. Therefore, various methods for estimating impermeable fraction of a watershed were examined.

For some watersheds, the fraction of impermeable area was found to correspond with the fraction of unvegetated area determined from a topographic map. However, this is not always the case. In some cases, a certain portion of unvegetated area is permeable or some of the area covered with vegetation is still impervious. This fraction cannot be determined from the topographic maps and, thus, some other data sources have to be used in those cases.

The surficial geology maps provide information on impermeable areas such as areas of rock outcrop or near-surface rock within the vegetated areas of a watershed. Unfortunately, while the entire area of British Columbia is covered with the bedrock geology maps, only a small part of it is covered with surficial geology maps. In this study, the surficial geology maps were available for only four watersheds.

The soil maps were found to be useful too. The main feature of certain types of soils such as

Luvisolic soils is a clay accumulation in the subsoil as a result of leaching from above. As a result,

an impervious layer ("hardpan") is produced. Therefore, presence of the Luvisolic soils on a

certain part of a watershed indicates impermeability even though that part of a watershed is

covered with vegetation.

The accuracy of the estimated values of impermeable fraction was then verified by comparing the

streamflow calculated with this value and historically observed streamflow for that period.

99 After precipitation and fraction of impermeable area have been determined, values for the parameters affecting the time distribution of runoff should be estimated. The optimal values of these parameters for each watershed are determined by trial-and-error procedure. The starting values of the parameters affecting the time distribution of runoff can be estimated from the information in the observed hydrograph, because of characteristically different time delays of fast, medium, slow and very slow runoff components. For example, the recession constants for groundwater can be analyzed by selecting a recession flow period which usually begins at the end of summer when snowmelt has ceased or after a large rain event when recessional flows dominate.

Examination of the parameters for each watershed revealed that there was quite a consistent set of parameters for everything except the fraction of impermeable area and the precipitation gradients. The precipitation gradients may be judged from snowcourse measurements to some extent, and also, in any sub-region may have to be assessed from the nearest similar watershed.

Besides the snowmelt, temperature, evapotranspiration and interception parameters which are pre-calibrated and held constant all the time, the calibrated values of parameters affecting the time distribution of runoff, along with groundwater percolation and deep zone share, showed relatively low variability among the studied watersheds. Because of this, these parameters were assigned constant values and each of the twelve study watersheds was rerun with the fixed set of parameters listed below:

Groundwater percolation 25 mm/day

Deep zone share 0.30%

Fast runoff routing time constant for the rainfall 0.60 days

100 Fast runoff routing time constants for the snowmelt 1 day

Interflow routing time constant for the rainfall 3 days

Interflow routing time constant for the snowmelt 4 days

Upper groundwater routing time constant 20 days

Deep zone groundwater routing time constant 5 months

This simplified approach produced quite reliable results. The reduction in the coefficient of efficiency compared to full calibration procedure for each watershed ranged from 0 to 11%, and stayed within 6% for nine of the twelve studied watersheds.

The reduction in the coefficient of determination ranged from 0 to 10% and stayed within 4% for nine of twelve watersheds.

The differences in the volume errors between two approaches were very small and appeared negligible.

Therefore, a simplified approach presented in this study enables us to obtain reasonably reliable flow estimates by inputting precipitation distribution, fraction of impermeable area and measurable physical characteristics of a given watershed and running the model with the fixed set of parameters listed above. This method was tested on twelve heterogeneous watersheds and gave excellent results. The implication is that the same parameters may be applicable to any watershed within the studied area regardless of size, topography, climate, soil, geology or hydrological regime, provided that impermeable fractiono f the watershed can be estimated and precipitation input is adequate.

101 6.1 Implications for the Watershed Behavior

The pathways for water flow in a watershed consist of a hillslope land phase and a stream network, channel phase. The travel time of the water through the system depends on many controlling factors - basin topography and physical properties and depths of the soils being the most important.

The velocity measurements mentioned in Chapter 2 indicate that water velocity in the channel network is several orders of magnitude greater than the velocity of water in the land phase. The main delay in runoff is in the land phase. Once this land phase runoff reaches the channel, it is routed relatively quickly to the watershed outlet, because the channel flow is fast and of much shorter duration than the previous land phase flow. The results of this study conducted on the watersheds ranging from 71 to 1194 km2 in area, show that routing time constants for all four runoff components (fast runoff, interflow, upper and deep zone groundwater) do not increase with drainage area of the watershed. This finding supports the idea that the land phase controls the runoff process and the length of flow paths in the land phase is independent of watershed size. The length of channel network, however, increases with watershed size (Eqn. 2.28).

Lack of variability of the routing time constants with a watershed size indicates that the land phase is the controlling phase in the hydrological behavior of a watershed and the channel phase is secondary and appears to be almost negligible.

This finding is in direct contradiction with some of the previous theories which suggest that with increasing drainage area, the travel time of runoff on hillslopes becomes negligible in comparison with the travel time through the channel network. These previous studies indicated that the break point between small and large watersheds with the dominance of land and channel phase

102 respectively, was approximately 200 km2. However, the present study indicates that even in the watersheds larger than 1000 km2, the land phase is dominant and is the controlling factor in the

streamflow generating process.

103 REFERENCES

Chow, V. T., 1964. Handbook of Applied Hydrology. McGraw-Hill, Inc. USA.

Dunne, T., 1978. "Field studies of hillslope flow processes." In Hillslope Hydrology, (ed. by

M. J. Kirkby), John Wiley & Sons, Chichester, Great Britain, 227-293.

Farley, A. L., 1979. Adas of British Columbia. University of B. C. Press, Vancouver, B. C,

Canada.

Henderson, F. M., 1966. Open Channel Flow. Macmillan Publ. Co., New York, USA.

Horton, R. E., 1945. "Erosional development of streams and their drainage basins; hydrophysical

approach to quantitative morphology." Bull. Geol.Soc. Am., 56, 275-370.

Hydrology of Floods in Canada: A Guide to Planning and Design., 1989, (editor-in-chief W. E.

Watt), National Research Council of Canada.

Ishihara, Y. and Kobatake, S., 1975. "The roles of slope and channel processes in storm runoff in

the Ara experimental basin." International Symposium on the Characteristics of River

Basins, December 1-8, 1975, Tokyo, Japan, IAHS Publ. No. 117, 75-86.

Kirkby, M. J., 1976. "Tests of the random network model and its application to basin hydrology"

Earth Surface Processes, 1, 197-212.

Kirkby, M. J., 1988. "Hillslope runoff processes and models." Journal of Hydrology, 100,

315-339.

Klein, M., 1976. "Hydrograph peakedness and basin area." Earth Surface Processes, 1, 27-30.

104 Lighthill, M. J. and Whitham, G. B., 1955. "On kinematic waves: I - Flood movement in long

rivers." Proc. Roy. Soc, London, Vol. 229, No. 1178.

Linsley, R. K. Jr., Kohler, M. A. and Paulhus, J. L. H., 1975. Hydrology for Engineers.

McGraw-Hill, Inc. USA.

Loukas, A. and Quick, M. C, 1993. "Hydrologic behaviour of a mountainous watershed." Can.

Journal of Civil Engineering, 20(1), 1-8.

McCarthy, G. T., 1938. "The unit hydrograph and flow routing." (unpublished manuscript).,

Presented at U. S. Corps of Engineers Conference, North Atlantic Division, June 1938.

Micovic, Z., 1998. "Effects of forest removal on streamflow in a large mountainous watershed: a

model approach." CWRA 51st Annual Conference Proceedings, June 1998, Victoria, B. C,

Canada, 60-66.

Nash, J. E., 1957. "The form of the Instantaneous Unit Hydrograph." Int. Assoc. Sci. Hydrol.,

Publ. 45, Vol. 3, 14-121.

Nash, J. E. and Sutcliffe, J. V., 1970. "River flow forecasting through conceptual models. Part I -

a discussion of principles." Journal of Hydrology, 10, 282-290.

Quick, M. C, 1995. "The UBC Watershed Model." In Computer Models of Watershed

Hydrology, (ed. by V. J. Singh), Water Resources Publications, P.O. Box 260026,

Highlands Ranch, Colorado 80126-0026, USA, 233-280.

Quick, M. C. and Pipes, A., 1975. "Nonlinear channel routing by computer." J. of Hydraulics

Division, ASCE, 101(HY6), 651-664.

Quick, M. C. and Pipes, A., 1977. "UBC Watershed Model." Hydrological Sciences Bull. XXI,

1, 3, 285-295.

105 Ree, W. O., 1963. "A progress report on overland flow studies." Soil Conserv. Serv. Hydraulic

Engrs. Meeting, August 12-16, 1963, New York, USA, 18 pp.

Sherman, L. K., 1932. "Stream-flow from rainfall by the Unit-Graph method." Eng. News-Rec,

108, 501-505.

UBC Watershed Model Manual, Version 4.0, February 27, 1995, Mountain Hydrology Group,

Department of Civil Engineering, Vancouver, B. C, V6T 1Z4.

Wooding, R. A., 1965/1966. "A hydraulic model for the catchment-stream problem." Journal of

Hydrology, 3, 254-267 -1: Kinematic wave theory; 3, 268-282 - II: Numerical solution;

4, 21-37 - ni: Comparison with runoff observations.

106 APPENDIX

CALIBRATION STATISTICS AND VERIFICATION

The calibration was verified for some of the watersheds (where data was available) by running already calibrated watershed for different period of years and comparing statistics for both runs.

The results were very satisfactory. For some watersheds, verification statistics were even better than those of the original calibration.

107 Barlow Creek (ORIGINAL CALIBRATION: 1970 -1974)

STATISTICS FOR THE OCT 1, 1970 - SEP 30 , 1974 WATER YEAR(S)

MeanQobs MeanQ^ TotQobs TotQea TotQobs Coeff.of Coeff.of 3 3 3 irr7s/d m/s/d m/s m/s -TotQest Eff. Det.

YEAR (701001-710930) YEAR 0.3 0.3 99.4 101.1 -1.7 0.8345 0.8435 YEAR (711001-720930) YEAR 0.3 0.3 114.0 114.4 -0.4 0.6444 0.6459 YEAR (721001-730930) YEAR 0.2 0.2 77.5 82.5 -5.0 0.7494 0.7660 YEAR (731001-740930) YEAR 0.4 0.4 141.5 130.1 11.4 0.8155 0.8385

Barlow Creek (VERIFICATION: 1965 -1967)

STATISTICS FOR THE OCT 1, 1965 - SEP 30 , 1967 WATER YEAR(S)

MeanQobs MeanQest TotQobs TotQrat TotQobs Coeff.of Coeff.of 3 3 3 m/s/d m/s/d nrVs m/s -TotQest Eff. Det.

YEAR (651001-660930) YEAR 0.2 0.2 90.1 89.0 1.1 0.8385 0.8388 YEAR (661001-670930) YEAR 0.3 0.2 110.8 74.0 36.7 0.7468 0.8342

108 Bone Creek (ORIGINAL CALIBRATION: 1979 - 1982)

STATISTICS FOR THE OCT 1, 1979 - SEP 30 , 1982 WATER YEAR(S)

Mean Qobs Mean Qest Tot Qobs TotQe* Tot Qobs Coeff.of Coeff.of m3/s/d m3/s/d m3/s m3/s -TotQert Eff. Det.

YEAR (791001-800930) YEAR 9.0 8.7 3288.2 3194.2 94.1 0.8469 0.8534 YEAR (801001-810930) YEAR 12.3 10.1 4482.7 3695.5 787.2 0.7834 0.8360 YEAR (811001-820930) YEAR 10.8 10.5 3949.4 3825.3 124.1 0.8777 0.8955 Bridge River (ORIGINAL CALIBRATION: 1988 -1994)

STATISTICS FOR THE OCT 1, 1988 - SEP 30 , 1994 WATER YEAR(S)

MeanQob, Mean TotQobs TotQ^t TotQobs Coeff.of Coeff.of 3 3 3 3 m/s/d m/s/d m/s m/s -TotQest Eff. Det.

YEAR (881001-890930) YEAR 39.8 37.8 14512.2 13782.1 730.1 0.8040 0.8389 YEAR (891001-900930) YEAR 43.2 39.0 15752.0 14233.9 1518.1 0.8935 0.9038 YEAR (901001-910930) YEAR 55.3 52.7 20192.7 19236.4 956.3 0.8717 0.8758 YEAR (911001-920930) YEAR 48.2 42.3 17652.0 15482.7 2169.3 0.8631 0.8994 YEAR (921001-930930) YEAR 40.9 36.0 14941.5 13132.5 1809.0 0.7963 0.8104 YEAR (931001-940930) YEAR 42.1 32.9 15348.8 12009.0 3339.8 0.8014 0.8472 Campbell River (ORIGINAL CALIBRATION: 1983 -1990)

STATISTICS FOR THE OCT 1, 1983 -SEP 30 , 1990 WATER YEAR(S)

Mean Qobs Mean Qest Tot Qobs Tot Qest Tot Qobs Coeff.of Coeff.of 3 3 3 3 m/s/d m/s/d m/s m/s -TOt Qest Eff. Det.

YEAR (831001-840930) YEAR 75.5 79.2 27631.1 28998.6 -1367.6 0.6745 0.6812 YEAR (841001-850930) YEAR 59.7 55.6 21773.0 20284.1 1488.9 0.7220 0.8002 YEAR (851001-860930) YEAR 75.5 70.6 27546.7 25779.7 1767.0 0.8216 0.8262 YEAR (861001-870930) YEAR 89.6 92.5 32701.3 33765.3 -1064.0 0.7262 0.7307 YEAR (871001-880930) YEAR 70.2 69.0 25705.9 25249.3 456.6 0.5603 0.6117 YEAR (881001-890930) YEAR 63.6 64.5 23203.1 23551.8 -348.7 0.6600 0.6602 YEAR (891001-900930) YEAR 65.5 66.9 23893.7 24418.1 -524.4 0.7526 0.7855 Coquihalla River (ORIGINAL CALIBRATION: 1971 -1976)

STATISTICS FOR THE OCT 1, 1971 - SEP 30 , 1976 WATER YEAR(S)

Mean Qob, Mean Qe8t Tot Qob, T0t Q.st Tot Qobs Coeff.of Coeff.of m3/s/d m3/s/d m3/s m3/s -TotQes, Eff. Det.

YEAR (711001-720930) YEAR 45.8 30.5 16756.7 11157.8 5598.8 0.7552 0.8719 YEAR (721001-730930) YEAR 21.3 15.3 7776.8 5571.2 2205.6 0.7419 0.8284 YEAR (731001-740930) YEAR 39.1 40.6 14259.9 14802.0 -542.2 0.7381 0.7959 YEAR (741001-750930) YEAR 28.9 23.7 10531.1 8653.3 1877.8 0.8707 0.8980 YEAR (751001-760930) YEAR 41.8 40.7 15296.2 14899.2 397.0 0.7028 0.7601

Coquihalla River (VERIFICATION: 1974 -1979)

STATISTICS FOR THE OCT 1, 1974 - SEP 30 , 1979 WATER YEAR(S)

Mean Qobs Mean Qest Tot Qobs T0t Qest Tot Qobs Coeff.of Coeff.of m3/s/d m3/s/d m3/s m3/s -TotQes, Eff. Det.

YEAR (741001-750930) YEAR 28.9 23.7 10531.1 8648.7 1882.4 0.8628 0.8908 YEAR (751001-760930) YEAR 41.8 40.7 15296.2 14888.5 407.7 0.7024 0.7599 YEAR (761001-770930) YEAR 22.8 21.3 8308.4 764.2 544.2 0.4895 0.6019 YEAR (771001-780930) YEAR 29.3 23.8 10694.7 8693.2 2001.6 0.6338 0.7252 YEAR (781001-790930) YEAR 22.3 17.8 8142.7 6488.0 654.7 0.7951 0.8481 Illecillewaet River (ORIGINAL CALIBRATION: 1981 -1989)

STATISTICS FOR THE OCT 1, 1981 - SEP 30 , 1989 WATER YEAR(S)

Coeff.of Coeff.of Mean Q0bS Mean Qest Tot Qob, TotQrat Tot Qob, 3 3 3 3 m/s/d m/s/d m/s m/s -TotQest Eff. Det.

YEAR (811001-820930) YEAR 57.0 53.6 20787.7 19579.3 1208.4 0.9124 0.9227 YEAR (821001-830930) YEAR 52.3 49.2 19083.0 17960.6 1122.4 0.9200 0.9228 YEAR (831001-840930) YEAR 52.9 45.3 19376.0 16561.5 2814.5 0.8948 0.9131 YEAR (841001-850930) YEAR 49.4 47.9 18045.5 17490.7 554.8 0.9429 0.9454 YEAR (851001-860930) YEAR 54.3 49.1 19804.5 17924.7 1879.8 0.9172 0.9246 YEAR (861001-870930) YEAR 51.9 52.2 18934.2 19045.2 -111.0 0.9125 0.9173 YEAR (871001-880930) YEAR 49.4 52.6 18065.4 19255.2 -1189.8 0.8940 0.9089 YEAR (881001-890930) YEAR 47.3 50.6 17255.7 18451.2 -1195.6 0.9003 0.9354

Illecillewaet River (VERIFICATION: 1971 -1979)

STATISTICS FOR THE OCT 1, 1971 - SEP 30 , 1979 WATER YEAR(S)

Qob, Coeff.of Coeff.of Mean Qob, Mean Qest Tot Qob, TotQct Tot 3 3 3 3 m/s/d m/s/d m/s m/s -Tot Qe,t Eff. Det.

YEAR (711001-720930) YEAR 63.1 59.2 23096.9 21653.9 1442.9 0.9491 0.9518 YEAR (721001-730930) YEAR 43.0 4L6 15681.8 15179.4 502.5 0.9308 0.9316 YEAR (731001-740930) YEAR 59.6 51.5 21768.6 18803.7 2964.9 0.9487 0.9649 YEAR (741001-750930) YEAR 47.5 49.3 17327.2 18000.9 -673.7 0.9252 0.9283 YEAR (751001-760930) YEAR 65.5 59.5 23978.2 21786.4 2191.7 0.9176 0.9264 YEAR (761001-770930) YEAR 44.7 47.2 16303.4 17236.5 -933.2 0.9327 0.9388 YEAR (771001-780930) YEAR 50.3 44.8 18352.2 16335.5 2016.7 0.9048 0.9163 YEAR (781001-790930) YEAR 49.9 52.4 18218.3 19131.7 -913.4 0.9380 0.9533 Jordan River (ORIGINAL CALIBRATION: 1983 -1988)

STATISTICS FOR THE OCT 1, 1983 - SEP 30 , 1988 WATER YEAR(S)

Mean Qobs Mean QMt Tot QODs TotQct Tot Qobs Coeff.of Coeff.of m3/s/d m3/s/d m3/s m3/s -TOt Qest Eff. Det.

YEAR (831001-840930) YEAR 17.2 15.1 6299.3 5536.0 763.4 0.8064 0.8441 YEAR (841001-850930) YEAR 14.9 14.3 5451.2 5217.2 234.0 0.8662 0.8739 YEAR (851001-860930) YEAR 16.2 15.9 5918.4 5814.1 104.3 0.7924 0.8501 YEAR (861001-870930) YEAR 16.2 14.4 5899.4 5272.4 627.1 0.8381 0.8460 YEAR (871001-880930) YEAR 16.1 15.2 5876.1 5580.5 295.6 0.8548 0.8691

Jordan River (VERIFICATION: 1970 - 1975)

STATISTICS FOR THE OCT 1, 1970 - SEP 30 , 1975 WATER YEAR(S)

Mean Qobs Mean Qset Tot Qobs TotQ^ TotQobs Coeff.of Coeff.of m3/s/d m3/s/d m3/s m3/s -TOt Qest Eff. Det.

YEAR (701001-710930) YEAR 17.3 17.0 6306.0 6221.3 84.7 0.7717 0.8469 YEAR (711001-720930) YEAR 22.5 20.1 8219.0 7359.7 859.3 0.8205 0.8318 YEAR (721001-730930) YEAR 14.7 11.4 5362.1 4152.4 1209.7 0.8040 0.8448 YEAR (731001-740930) YEAR 21.0 16.6 7663.9 6044.8 1619.1 0.8923 0.9212 YEAR (741001-750930) YEAR 16.2 13.7 5905.0 4989.0 916.1 0.8588 0.8739 Little Swift River (ORIGINAL CALIBRATION: 1988 - 1991)

STATISTICS FOR THE OCT 1, 1988 - SEP 30 , 1991 WATER YEAR(S)

Mean Qobs Mean Q^ TotQob, TotQct TotQob, Coeff.of Coeff.of 3 3 3 3 m/s/d m/s/d m/s m/s -TOtQest Eff. Det.

YEAR (881001-890930) YEAR 2.5 2.5 897.1 906.2 -9.1 0.9278 0.9328 YEAR (891001-900930) YEAR 3.4 3.2 1236.1 1160.5 75.6 0.8552 0.8575 YEAR (901001-910930) YEAR 3.1 2.6 1129.0 950.8 178.3 0.7648 0.8090

Little Swift River (VERIFICATION: 1972 -1975)

STATISTICS FOR THE OCT 1, 1972 - SEP 30 , 1975 WATER YEAR(S)

Mean Qob, Mean Qe3t TotQob, TotQrat TotQob, Coeff.of Coeff.of m3/s/d m3/s/d m3/s m3/s -TOtQest Eff. Det.

YEAR (721001-730930) YEAR 2.3 2.5 821.5 927.9 -106.4 0.8490 0.8900 YEAR (731001-740930) YEAR 2.7 3.6 989.3 1320.2 -330.9 0.7055 0.9099 YEAR (741001-750930) YEAR 2.7 2.8 992.5 1008.0 -15.4 0.8791 0.8792 Naver Creek (ORIGINAL CALIBRATION: 1971 - 1975)

STATISTICS FOR THE OCT 1, 1971 - SEP 30 , 1975 WATER YEAR(S)

Mean Qobs MeanQe,t TotQob, TotQes, TotQob» Coeff.of Coeff.of 3 3 3 3 m/s/d m/s/d m/s m/s -TotQrat Eff. Det.

YEAR (711001-720930) YEAR 9.2 6.9 3358.2 2508.7 849.5 0.6868 0.7317 YEAR (721001-730930) YEAR 6.3 6.0 2286.5 2200.9 85.7 0.7219 0.7387 YEAR (731001-740930) YEAR 10.8 7.6 3935.7 2780.5 1155.2 0.6691 0.7132 YEAR (741001-750930) YEAR 6.2 5.0 2250.3 1824.6 425.7 0.5682 0.6195

Naver Creek (VERIFICATION: 1965 -1969)

STATISTICS FOR THE OCT 1, 1965 - SEP 30 , 1969 WATER YEAR(S)

Mean Qobs MeanQest TotQob, TotQrat TotQob, Coeff.of Coeff.of 3 3 3 3 m/s/d m/s/d m/s m/s -Tot QMt Eff. Det.

YEAR (651001-660930) YEAR 8.7 5.9 3176.6 2166.9 1009.7 0.6034 0.6684 YEAR (661001-670930) YEAR 8.4 4.8 3077.6 1759.0 1318.6 0.7504 0.8781 YEAR (671001-680930) YEAR 8.8 7.5 3211.7 2748.1 463.6 0.6044 0.7130 YEAR (681001-690930) YEAR 6.1 4.9 2229.8 1782.2 447.6 0.6444 0.7442

116 Stitt Creek (ORIGINAL CALIBRATION: 1987 - 1991)

STATISTICS FOR THE OCT 1, 1987 - SEP 30 , 1991 WATER YEAR(S)

Mean Qobs Mean Qest TotQobs TotQ.,t TotQob, Coeff.of Coeff.of 3 3 3 3 m/s/d m/s/d m/s m/s -TotQe,t Eff. Det.

YEAR (871001-880930) YEAR 7.0 6.9 2549.7 2531.1 18.6 0.7042 0.7093 YEAR (881001-890930) YEAR 6.3 6.9 2306.3 2511.6 -205.3 0.7236 0.7664 YEAR (891001-900930) YEAR 7.6 7.3 2756.1 2678.9 77.2 0.8821 0.8847 YEAR (901001-910930) YEAR 8.0 7.8 2936.8 2843.1 93.7 0.8997 0.9006

Stitt Creek (VERIFICATION: 1981-1985 )

STATISTICS FOR THE OCT 1, 1981 -SEP 30 , 1985 WATER YEAR(S)

Mean Qob, Mean Qest Tot Qob, TotQrat TotQob, Coeff.of Coeff.of 3 3 3 3

m/s/d m/s/d m/s m/s -TotQe,t Eff. Det.

YEAR (811001-820930) YEAR 7.6 6.8 2773.4 2480.6 292.8 0.7806 0.7951 YEAR (821001-830930) YEAR 7.0 6.6 2545.1 2423.3 121.8 0.7946 0.7972 YEAR (831001-840930) YEAR 6.4 5.9 2335.5 2153.5 182.0 0.8266 0.8403 YEAR (841001-850930) YEAR 4.9 6.9 1790.5 2525.8 -735.3 0.4261 0.9055 Tabor Creek (ORIGINAL CALIBRATION: 1974 -1979)

STATISTICS FOR THE OCT 1, 1974 - SEP 30 , 1979 WATER YEAR(S)

Mean Qobs Mean Qest TotQobs TotQo,t Tot Qob, Coeff.of Coeff.of m3/s/d nrVs/d m3/s m3/s -Tot Q.,t Eff. Det.

YEAR (741001-750930) YEAR 0.6 0.8 218.7 284.1 -65.4 0.7916 0.8434 YEAR (751001-760930) YEAR 1.5 1.2 540.7 450.0 90.7 0.7418 0.7766 YEAR (761001-770930) YEAR 0.8 0.7 302.0 253.5 48.5 0.7068 0.7393 YEAR (771001-780930) YEAR 0.5 0.5 172.6 179.9 -7.3 0.8135 0.8141 YEAR (781001-790930) YEAR 0.8 0.9 304.2 312.6 -8.3 0.7770 0.7772

118 Watching Creek (ORIGINAL CALIBRATION: 1966 - 1972)

STATISTICS FOR THE OCT 1, 1966 - SEP 30 , 1972 WATER YEAR(S)

Mean Qobs Mean Q08t Tot Qob, TotQe* Tot Qob, Coeff.of Coeff.of m3/s/d m3/s/d m3/s m3/s "TotQert Eff. Det.

YEAR (671001-680930) YEAR 0.5 0.5 183.5 179.8 3.7 0.9144 0.9403 YEAR (681001-690930) YEAR 0.7 0.8 257.3 282.0 -24.7 0.8443 0.8669 YEAR (691001-700930) YEAR 0.4 0.4 146.3 155.8 -9.5 0.8777 0.8948 YEAR (701001-710930) YEAR 0.7 0.5 242.1 192.3 49.8 0.8558 0.9019 YEAR (711001-720930) YEAR 0.7 0.7 253.6 270.2 -16.6 0.9038 0.9085