Singh, Kinematic Wave

HYDROLOGICAL PROCESSES
Hydrol. Process. 16, 667716 (2002)

DOI: 10.1002/hyp.306
Is hydrology kinematic?
V. P. Singh*
Department of Civil and Environmental Engineering, Louisiana State University, Baton Rouge, LA 70803-6405, USA
Abstract:
A wide range of phenomena, natural as well as man-made, in physical, chemical and biological hydrology exhibit
characteristics similar to those of kinematic waves. The question we ask is: can these phenomena be described using
the theory of kinematic waves? Since the range of phenomena is wide, another question we ask is: how prevalent
are kinematic waves? If they are widely pervasive, does that mean hydrology is kinematic or close to it? This paper
addresses these issues, which are perceived to be fundamental to advancing the state-of-the-art of water science and
engineering. Copyright 2002 John Wiley & Sons, Ltd.
KEY WORDS biological hydrology; chemical hydrology; physical hydrology; kinematic wave theory; kinematics; ux
laws
INTRODUCTION
There is a wide range of natural and man-made physical, chemical and biological ow phenomena that exhibit
wave characteristics. The term wave implies a disturbance travelling upstream, downstream or remaining
stationary. We can visualize a water wave propagating where the water itself stays very much where it was
before the wave was produced. We witness other waves which travel as well, such as heat waves, pressure
waves, sound waves, etc. There is obviously the motion of matter, but there can also be the motion of form
and other properties of the matter.
The ow phenomena, according to the nature of particles composing them, can be distinguished into two
categories: (1) ows of discrete noncoherent particles and (2) ows of continuous coherent particles. Examples
of ows in the rst category include trafc ow; transport of sand and gravel in pipes, umes and rivers;
evolution of ripples, dunes and bars on a river bed; settlement of sediment particles in a tank; and movement of
microbial organisms. Exemplifying ows in the second category are overland ow, ood movement, baseow,
snowmelt, movement of glaciers, inltration, evaporation, solute transport, ion exchange and chromatographic
transport. Because of their wavelike behaviour these ow phenomena can be described by the hydrodynamic
wave theory or one of its variants, including the kinematic wave theory, diffusion wave theory, gravity wave
theory, linear wave theory, and so on.
A survey of geophysical literature reveals that the kinematic wave theory has been applied to a wide
spectrum of problems, including watershed runoff modelling, ood routing in rivers and channels, movement
of soil moisture, macropore ow, subsurface storm ow, erosion and sediment transport, debris ow, solute
transport, ion exchange, chromatography, sedimentation, glacial motion, movement of snowmelt water, ow
over porous beds in irrigation borders and furrows, river-ice motion, vertical mixing of coarse particles in
gravel-bed rivers, to name but a few. Other elds, outside the realm of geophysics, applying the kinematic
wave theory include trafc ow, movement of agricultural grains in a bin (Cowins and Comfort, 1982) and
blood circulation. By discussing a number of illustrative hydrologic problems, it is reasoned that within certain
limitations, which may vary with the problem, the hydrologic nature can indeed be approximated quite closely
* Correspondence to: Professor V. P. Singh, Department of Civil and Environmental Engineering, Louisiana State University, Baton Rouge,
LA 70803-6405, USA. E-mail: [email protected]
Received 17 September 1997
Copyright 2002 John Wiley & Sons, Ltd. Accepted 21 February 2001
668 V. P. SINGH
by kinematic wave theory. Outside these limitations, the theory provides only a crude approximation and the
full hydrodynamic wave theory or one of its higher-order variants may be needed to mimic nature.
Before proceeding further, the question we ask is: what is meant by kinematic? Although papers and
textbooks in hydrology dene kinematic waves mostly from an approximation of the momentum equation and
occasionally from dening ux laws, a comprehensive account of the term kinematic is lacking. Therefore,
a short discussion to that end is in order.
WHAT IS KINEMATICS?
The term kinematic involves consideration of such macroscopic properties of a ow phenomenon as velocity,
discharge, concentration, time of travel, etc. It does not involve consideration of forces and mass. Thus, the
connotation of kinematic is that it is phenomenological and less physical, but its representation may be
derived from or related to dynamical considerations involving forces, mass, momentum and energy.
Although studies based on kinematics had existed long before the development of Greek science, it is
Ampere (1834) who is credited with having introduced the term kinematics for the study of motion in
his famous essay on the classication of sciences. His intent was to create a discipline for the study of
motion without regard to the forces involved. In mechanical engineering kinematics is viewed as the study
of the motion of particles, and it is about the motion of rigid bodies or chains of rigid bodies in engineering
mechanics. Thus, kinematics is dened as the science that deals with the study of motion in the most general
way. The motion can be translatory, dilatational or rotational (angular) (Singh, 1996a).
It may be instructive to view kinematics in the context of geometry and dynamics. From the standpoint
of dimensions, geometry implies only length, whereas kinematics involves length and time, and dynamics
considers length, mass and time. Thus, kinematics is somewhere between geometry and dynamics, perhaps
closer to the latter. Although kinematics is much related to geometry, it can also be studied without geometry
by means of numerical description of the functions as the Babylonians seem to have donea precursor to
curve tting.
Theoretically, kinematics is about
x
i
D x
i
X, t 1
where x
i
is the position (coordinates) of a material point i, X is the position vector that identies each point
and t is time. X can be the initial position vector. Equation (1) is applicable to discrete as well as continuous
systems, and to all kinds of modelsdeterministic as well as stochastic. This can be a representation of a
particle that follows Newtons law or one that performs a random walk. It can also represent any discontinuity
in the kinematical model.
Consider the motion of a single point representing either a small particle or that of a rigid body undergoing
pure translation. The point could be in curvilinear motion either free or subject to some constraints as
exemplied by a projectile in one case and by a pendulum in another. In general
x
1
D Ft, x
2
D Gt 2
where F and G are some functions. Theoretically, Equation (2) can be expressed as
x
2
D fx
1
3
where f is some function. Equation (3) is less informative than Equation (2). Once the kinematics of a point
is developed, the kinematical theory of a rigid body and that of chains of rigid bodies or mechanisms may
readily be developed.
In kinematics not only are we concerned with the position of a point at a certain time, but also with
the displacement accomplished and the distance traversed in a certain time interval. Other functions of time
Copyright 2002 John Wiley & Sons, Ltd. Hydrol. Process. 16, 667716 (2002)
IS HYDROLOGY KINEMATIC? 669
such as those of velocity and acceleration can be introduced. Higher derivatives as the rate of change of
acceleration, called jerk, and even higher-order derivatives (Hartenberg and Denavit, 1964) can be included.
A set of points (of a rigid body or uid) undergoes translation if the velocity vectors of all points are
identical. This however does not exclude the existence of acceleration and higher derivatives of velocity. The
pathlines of a body undergoing a translatory motion do not necessarily have to be straight lines.
It is difcult to nd a science where motion does not play a major role. One might even conjecture that the
history of kinematics is indeed the history of motion. Heavily dependent on kinematics are different branches
of engineering including hydrology, hydraulics, snow and ice engineering, river engineering, irrigation and
drainage engineering, environmental engineering, geological engineering, aeronautics, naval engineering,
biomechanical engineering, mechanical engineering, and so on. As is the case in kinematics of particles
and of bodies in mechanics, our interest is in the velocity and eventual acceleration. Leonardo da Vinci
studied the motion of particles of all kinds of uids (Macagno, 1991). He invented the technique of ow
visualization in water. Kinematics also nds an interesting application in chemistry where once a reaction
starts, the rate of production of a new compound or velocity of reaction is of interest. There are chemical
processes that move at different speeds, as in electrophoresis and chromatography. Table I lists a number of
elds applying kinematics.
The motions of the human body have been of much interest in the health science arena. Rehabilitation
professionals need to know the kinematics of the human body. This is also needed to increase the productivity
of workers (Rabinach, 1991). In human organs motion at different scales is important, as in the case of
contractions along the digestive system (Macagno, 1980). In biology, studies of the motions of the heart and
its valves and those of the eye, and ow in circulatory, respiratory and lymphatic systems, involve kinematics.
A multitude of interesting aspects of kinematics arise in the study of wide ranging phenomena: from the study
of Brownian motion to the ows in nebular stellar dynamics, from trafc ow to forest re propagation, from
Table I. Fields applying kinematics (adapted from Macagno, 1991.
Reproduced by permission of lowa Institute of Hydraulic Research)
Sciences Engineering
Astronomy Materials engineering
Stellar dynamics Mechanical engineering
Physics Mechanisms and machines
Geography Transportation
Geophysics Hydraulics
Geology Hydrology
Soil mechanics Environmental engineering
Meteorology Geological engineering
Oceanography Irrigation and drainage engineering
Chemistry Hydronautics
Biological ows Aeronautics and astronautics
Biological motions Transport phenomena
Bio-engineering
Snow and ice engineering
Trafc engineering
Naval engineering
Art Exercises and sports
Compositional ow Individual exercises
Kinetic art Games
Animation in movies Exercise machines
Mobiles Human body kinematics
670 V. P. SINGH
laminar to turbulent ow, from sediment transport to blood ow, and so on. A comprehensive history of
kinematics is summarized by Macagno (1991).
It should be noted that the introduction of mass does not necessarily make some motions strictly dynamical;
they may remain strongly and functionally kinematical. For instance, the formulation of equations for
conservation of volume in uid mechanics is very much in the realm of kinematics. When the equations
for conservation of mass are formulated, they are still in the same domain; the two treatments are very
similar. Even the discussions of momentum ux and kinetic energy ux are fundamentally kinematical. In
general, there are terms in equations of dynamics that are essentially kinematical. The difference between
kinematics and dynamics thus requires close scrutiny.
In a similar vein, in the study of kinetic theories of different states (solid, liquid, gas) and mass transport
processes, kinematics plays an important role. In the theory of mass transport by uid ows, the notion of
velocity of a mixture of different species is introduced, taking into account different densities (Bird, et al.,
1960). In such a case, either the common density or the molar concentration is used, and the motion of mass
ux, a kinematical concept, is invoked. Thus, mass becomes necessary to dene velocity in this case. This is
done in a purely geometrical manner. A question arises: when can kinematics be decoupled from dynamics
and when can it not ? One aspect of kinematics is the development of the theory of kinematic waves, whose
discussion forms the basis of much of the discussion that follows. It should be noted that much kinematics
existed before the kinematic wave theory was developed.
KINEMATIC WAVE THEORY
In a seminal contribution Lighthill and Whitham (1955a,b) developed in 1955 the kinematic wave theory
and gave a full account of the theory for describing ood movement in long rivers and trafc ow on long
crowded roads. The theory is comprised of three components: (1) theory of kinematic continuous waves;
(2) theory of shock waves; and (3) theory of formation of shock waves out of continuous waves. Iwagaki
(1955) developed an approximate method of characteristics for routing steady ow in open channels of any
cross-sectional shape with nearly uniform lateral ow and proposed that the method would be applicable
to hydraulic analysis of runoff estimation in river basins. Implicit in his method was the kinematic wave
assumption. Thus, Iwagaki (1955) can also be credited to have independently conceived the kinematic wave
concept and developed the method of kinematic wave routing.
Following Lighthill and Whitham (1955a,b), kinematic waves exist if it can be assumed with sufcient
accuracy that there is a functional relationship between ux, concentration and position. Thus, the kinematic
wave theory is mathematically expressed by the law of conservation of mass through the continuity equation
and a uxconcentration relation. The coupling of these two equations leads to a rst-order partial differential
equation having only one system of wave characteristics. The essence of this is that the wave property is
derived from the continuity equation alone. Since the governing equation (or continuity equation) is of
rst order only, the kinematic waves possess only one system of characteristics. This would also imply that
kinematic waves travel in the downstream direction only, and the kinematic wave theory cannot accommodate
waves that travel in the upstream direction as in the case of backwater ow. In the absence of any lateral
inow and/or outow, the ux along the characteristics is constant and the kinematic waves are nondispersive
and nondiffusive. However, the waves are dispersive and diffusive when there is lateral inow and/or outow,
as for example in the case of rainfall runoff.
Continuous waves
If the volume of a quantity passing a given point x in unit time is denoted by ux, F, and the concentration is
dened as the volume of the quantity per unit distance by C, then for a one-dimensional ow system without
any lateral sources and sinks, the continuity equation or the mass conservation law is
C
t
C
F
x
D 0 4
Equation (4) is of rst order and states that the volume of the quantity in a small element of length changes
at a rate equal to the difference between the rates of inow and outow.
Fundamental to the development of the kinematic wave theory is the development of a uxconcentration
relation. A general uxconcentration relation can be assumed as
F D FC, x 5
Equations (4) and (5) have two unknowns, F and C, and can therefore be combined into one equation having
one unknown. Multiplying Equation (4) by
c D
F
C
x constant
D cC, x 6
one obtains
F
t
Cc
F
x
D 0 7
Equation (7) is the kinematic wave equation. The quantity c is described in various ways. In hydraulics it is
called wave velocity or celerity, and it is referred to as mobility in vadose zone hydrology. It is the slope of
the uxconcentration relation at a xed x. This is not the same as the mean velocity u, which is expressed as
u D F/C 8a
The relation between c and u is thus obvious:
c D
d
dC
uC D u CC
du
dC
8b
Equation (9) shows that c is greater than u if u increases with C as in the case of river ow, it is less than
u if u decreases with C as in the case of sedimentation, and it is equal to u if u does not change with C.
Equation (7) states that F is constant on waves travelling past the point with celerity c given by Equation (6).
The kinematic wave Equation (7) has one system of characteristics travelling in the downstream direction
only, given by dx D cdt, and along each of these characteristics F is xed.
Although Equation (5) constitutes the basic building block of the kinematic wave theory, a more popular,
but more restrictive, derivation of the theory in hydrology is obtained by assuming the local acceleration,
convective acceleration and pressure gradient (or concentration gradient) in the momentum equation to be
negligible. This assumption states an equivalence between friction and gravity forces. When only the local
and convective acceleration terms are assumed negligible, the remainder of the momentum equation (i.e. the
pressure gradient equalling the difference between gravity and frictional slopes) leads, in conjunction with
the continuity equation, to diffusion wave theory. When the full momentum equation is used jointly with
the continuity equation, the result is the dynamic wave theory. In general, the local acceleration and the
convective acceleration are of the same order of magnitude, but are of opposite sign, thus counteracting each
other. In a wide range of problems in hydrology, the friction and gravity terms are dominant and this is one
of the reasons for the popularity of the kinematic wave theory.
Formation of kinematic shocks
A kinematic shock is a discontinuity representing a sudden rise or surge in the ow depth. For example,
during wave movement faster-moving waves overtake slower-moving waves and there will, at a xed position,
be an increase in ux and concentration as functions of time, leading to shock formation. Thus, ood waves
have an intrinsic, nonlinear tendency to steepen as they propagate downstream, eventually forming a shock. In
the characteristic plane, shock formation results from the intersection of characteristics. After some time the
shock weakens and dissolves into a region of uniform ow. Shocks arise in a variety of hydrologic processes,
672 V. P. SINGH
and what these shocks mean and their practical relevance will be discussed in different sections of the paper.
A short but general discussion of shocks is relevant however and is given here. The shocks or discontinuities
are of rst and second orders. A discontinuity of the rst kind in the concentration is dened by a sudden
change of concentration at a certain level. In such a case the differential equation of continuity [Equation (4)]
no longer applies; instead it is replaced by an equation stating that the ow of mass into one side of the
shock is equal to that on the other side of the shock. Denoting the velocity of discontinuity by U, the mass
conservation leads to
U D
F
2
F
2
C
2
C
1
9
where subscripts 2 and 1 denote the quantities ahead and behind the shock. Equation (9) shows that in general
the discontinuity is not at rest but moves with velocity U, which is the slope of the chord joining the points
(F
1
, C
1
) and (F
2
, C
2
) on the FC diagram.
A discontinuity of the second kind is represented by a very small change in concentration. If C
2
C
1
D dC
is small, the expression for U becomes dF/dC D c, the celerity. In this case the velocity of a discontinuity
between concentrations C and C CdC is the same as the celerity c. A small change, if maintained, is
propagated through the medium of concentration C with velocity c in a manner similar to the propagation of
sound through air with a denite velocity. A line of constant concentration therefore describes the motion of
the boundary between media of concentrations C and C CdC so that its slope is necessarily equal to c. The
complete modication of concentration at a position, as well as its prole, can be characterized as a series of
small discontinuities propagated through the medium.
Determination of kinematic shocks
Many factors affect formation of shocks and these factors can broadly be distinguished to be of three types:
(1) initial and boundary conditions; (2) lateral inow and outow; and (3) watershed geometric characteristics.
A full account of the theory of shock formation was given by Lighthill and Whitham (1955a,b) and a
mathematical treatment of the formation and decay of shocks was given by Lax (1972).
Carrier and Pearson (1976) developed a method for determining the enveloping curve for any region
of intersecting characteristics. Croley and Hunt (1981) determined analytically boundary-dependent shocks
in planar ow. Hairsine and Paralange (1987) dealt with kinematic shocks on curved surfaces. Kibler
and Woolhiser (1970) investigated geometry-dependent shocks using a kinematic cascade. They derived
mathematical properties of kinematic cascades and developed a criterion, based on the properties of adjacent
owplane pairs, to predict when a shock would occur in a cascade. They also developed a numerical
procedure for shock tting. General properties of shock waves along with continuous kinematic waves were
discussed. Full equations of motion were employed to investigate the structure of the kinematic shock.
Building on the work of Kibler and Woolhiser (1970), Schmid (1990) investigated the effect of lateral
inow and outow as well as geometry on development of shocks. He derived a generalized criterion for
shock formation on an inltrating cascade that also encompasses cascades with time-dependent rates of
effective rainfall. Using simulated conditions, Ponce and Windingland (1985) determined ow and channel
characteristics that either tend to promote or inhibit development of kinematic shocks. Borah (1989) and
Borah et al. (1980) developed approximate but efcient numerical methods for determining the shock path
and shock tting.
FLUX LAWS
The term ux is dened in two ways in hydraulics. First, ux denotes a volume of any quantity per unit
area per unit time. Thus, its dimensions are L/T if the quantity is, say, runoff. Second, ux is dened as any
quantity per unit time. For example, the volume ux is the volume of a quantity per unit time as exemplied
by discharge, mass ux is mass per unit time, momentum ux (it has the same dimensions as force) is
momentum per unit time, energy ux (the kinetic energy ux has the same dimensions as power) is energy
per unit time, and so on. In water and environmental engineering both denitions are used.
Flux laws are fundamental to the development of transport theories. The ux laws, most common in
environmental and water sciences, are of two types: (1) power laws and (2) gradient laws. The kinematic wave
ux laws are perhaps the most popular ux laws of power type, and Darcys law and the DarcyBuckingham
law are the most popular gradient-type ux laws. Generalized ux laws which specialize into power and
gradient laws are also used. An example of a generalized ux law is the Burgers law or a generalized
version thereof (Singh and Prasana, 1999). Power-type ux laws lead to kinematic wave equations (see
Table II), whereas the gradient-type ux laws lead to diffusion waves governed by elliptic or parabolic partial
differential equations.
Algebraic laws
A general expression for the uxconcentration relation is given by Equation (5). One of its special forms
is the popular kinematic wave ux law expressed as
F D C
n
10
where is a parameter and n is an exponent. The meaning and interpretation of and n may vary with the
problem to which Equation (10) is applied. Equation (10) leads to the famous Chezy and Manning equations
which are popular in studies on overland ow and ood routing in open channels and rivers. At a xed
location x, (F/C) denes the wave celerity c : c D F/Cj
xxed
which is not the same as the average
velocity of ow. When n D 1, Equation (10) becomes
F D C 11
This is a linear uxconcentration law, used for describing the movement of meltwater runoff from snowpack.
Beven (1979) employed two somewhat uncommon three-parameter forms of the uxconcentration relation
for his channel network routing model:
F D
Ca Cbk k
1 bC
12
and
F
a1 expkF CbF
D C 13
where a, b and k are parameters.
Another ux law is dened as
F D C
2
C
0
C 14
where C
0
is the maximum value of C and is a parameter. Equation (14) is used for describing the settlement
of particles in a dispersion. Still another ux law used for describing the movement of sediment in umes
and pipes is
F D u
0
C
1
C
C
0
15
where u
0
is the velocity of a single particle when there are no other particles in ow. All of the above ux
laws lead to kinematic waves.
Gradient laws
A gradient-type ux law is expressed as
F D GC/x 16
674 V. P. SINGH
Table II. Kinematic ux laws applied to different branches of hydrology
Branch of
hydrology
Flux law Symbols and denitions Problems to which
ux law is applied
Surface water
hydrology;
river and
coastal
hydrology
Q D h
n
Q D discharge per unit width, h D ow
depth, n D exponent, D friction
parameter
Overland ow; channel
ow; depthdischarge
rating curve,
storagedischarge
relation, dam break ow,
sediment transport in
upland areas, etc.
Irrigation
hydrology
Q D A
n
A D ow cross-sectional area, Q D discharge
per unit width, n D exponent,
D kinematic wave friction parameter
Furrow irrigation and
border irrigation
Vadose zone
hydrology
q D K; q D bw
a
K D unsaturated hydraulic conductivity,
D moisture content, q D volumetric
moisture ux, a D exponent,
b D conductance parameter,
w D macropore water content
Soil water ow,
unsaturated ow and
macropore ow
River and coastal
hydrology
Q
s
D kv
0
1
k
k
0
Q
s
D sediment ux (rate of transport),
v
0
D speed of a single particle on the bed
of a ume of owing water, k D mean
linear concentration or density,
k
0
D maximum linear concentration
River channel bars and
dunes
T D v
0
w
1
w
w
0
T D transport rate, v
0
D velocity of a single
particle, w D areal concentration,
w
0
D areal concentration when transport
ceases
Transport of sand by water
Snow and ice
hydrology
q D kS
n
q D snowmelt water ux, k D permeability,
S
D saturation, n D exponent,
D constant
Vertical movement of water
in snowpacks
q D D constant, D saturated hydraulic
conductivity
Basal runoff from
snowmelt
q D
g
A
m
h
mC1
sin
m
s q D volume of ice passing a given point in
unit time per unit width, s D slope,
m D exponent, h D glacial thickness,
A D a measure of roughness of the bed,
g D acceleration due to gravity,
D density
Glacial motion
Q D
h
nC2
n C2
Q D glacial ux, h D glacial thickness,
n D exponent
Surface wave motion in
glacial movement
Water quality
hydrology
S D v D number of particles crossing a horizontal
section per unit area per unit time (local
concentration), v D velocity of a particle,
S D particle ux
Sedimentation and
settlement of particles
S D a
2
0
S D ux, D local concentration,
0
D maximum concentration
Sedimentation and
settlement of particles
Q D MfC C D concentration, M D amount of aborbant
per unit length of the column,
fC D isotherm, and Q D ux
Chromatography
where G is some function. As an example
F D C/x
m
17
where is a parameter and m is an exponent. Equation (17) is a nonlinear version of Darcys law. When
m D 1, Equation (17) becomes
F D C/x 18
Equation (18) is Darcys law with describing the negative of the saturated hydraulic conductivity and C
the hydraulic head.
A more general form of Equation (18) is when depends on C:
F D CC/x 19
Equation (19) is the DarcyBuckingham law with C describing the hydraulic head and the negative of
the unsaturated hydraulic conductivity which varies with the moisture content. When a gradient ux law is
coupled with the continuity equation, the resulting partial differential equation turns out to be of parabolic
type under transient ow conditions and of elliptic type under steady state conditions.
Generalized ux laws
A generalized ux law is obtained by combining the power and gradient-type ux laws. In general
F D GC, C/x, x 20
where G is some function with its argument dened by C, C/x and x. One form of Equation (20) is
F D C
n
CC/x
m
21
where m and n are exponents. Equation (21) specializes into (Singh and Prasana, 1999)
F D C
n
CC/x 22
and
F D C CC/x 23
Equation (23) is the Burgers ux law (Burgers, 1948) used in turbulence modelling and when used in the
continuity equation the resulting partial differential equation is the Burgers equation. Equations (21) and (22)
have been employed for routing of ows in open channels (Singh and Prasana, 1999).
APPLICATION OF KINEMATIC WAVE THEORY TO SURFACE WATER HYDROLOGY
Depthdischarge relation: rating curve
The depthdischarge relation, which forms the basis of much of the hydrologic literature on overland and
channel ow, is a special case of Equation (10) and can be expressed as
q D h
n
24
676 V. P. SINGH
619
G3=617.58
617
615
613
611
G
a
g
e

a
t

T
e
h
r
i

g
i
r
d
e
r

b
r
i
d
g
e

(
m
e
t
e
r
)
G2=610.18
G1=608.28
609
607
0.4
Q
1
=
4
0
Q
2
=
2
4
0
Q
3
=
1
4
4
0
2.0 4.0 6.0 8.0
Discharge at Tehri girder bridge ( 10
2
cumec)
10.0 12.0 14.0 16.0
U
P
P
E
R
E
N
V
E
L
O
P
IN
G
M
E
A
N
E
N
V
E
L
O
P
IN
G
L
O
W
E
R
E
N
V
E
L
O
P
IN
G
Figure 1. Stagedischarge curve at Tehri Girder bridge in Uttar Pradesh, India, within range of observations
where q is the discharge per unit width, h is the depth of ow, is a resistance parameter and n is an exponent
indicating the regime of ow. A typical curve for a river is shown in Figure 1. The value of n varies from 1
(highly turbulant) to 3 (laminar). The ow is transient when n is greater than 167 but less than 3. Thus, the
n exponent reects the scale of turbulance in ow from the state of no turbulance (when n D 3) to one of
extreme turbulance (when n D 1). Geometrically, n is a measure of nonlinearity. When n D 1, Equation (24)
becomes linear and when n D 3, it is highly nonlinear. These two limits of n correspond to the two extremes
of nonlinearity. Relating the degree of nonlinearity to the scale of turbulence, it can be inferred that turbulence
tends to linearize the ow, i.e. the higher the degree of turbulence the less the degree of nonlinearity and vice
versa. On the other hand, laminarity represents a case of extreme nonlinearity. This interpretation is important
in that surface water hydrologic processes are nonlinear but their degree of nonlinearity is conducive to
applicability of kinematic wave theory.
Parameter reects the interaction between the uid particles themselves and between uid and the conduit
boundaries within which the uid is owing. It therefore varies in space accounting for spatial heterogeneities
and in time reecting the temporal variability of ow and the conduit. Normally, is space-dependent but is
taken as constant in practice.
The origin of Equation (24) can be traced to the works of Manning and Chezy, but perhaps a more
meaningful and generalized formulation was advanced by Lighthill and Whitham (1955a). In hydraulics,
Equation (24) is expressed as
q D Kh
a
S
1/2
e
25
where S
e
is the slope of the energy line or frictional slope and K and a are empirical constants. Equation (25)
specializes into the Manning equation when the exponent a takes on the value of 5/3 and K equals 1/n
m
,
with n
m
being the Manning roughness coefcient. Likewise, Equation (25) specializes into the Chezy equation
when exponent a equals 3/2 and K equals C
z
, the Chezy coefcient.
In Equation (25) the exponent of h is variable and the justication for its variability can be found in
Kouwen et al. (1981), Maheshwari (1994) and Singh (1996a), when one balances the force causing the ow
(i.e. gravity force) with that resisting the ow (i.e. drag force) in a section of ow. This means that the
Manning roughness coefcient in Equation (25) is allowed to vary with the depth of ow as
n
m
D 1/Kh
5/3a
26a
Similarly, the Chezy coefcient follows
C
z
D Kh
a15
26b
Because the ow depth varies in space and time, so do the coefcients n
m
and C
z
.
Interaction between parameters and n
By denition, is a resistance parameter and varies in time and space, reecting the spatial and temporal
variability of ow and the hydraulic system. Furthermore, depends on n. In an experimental study, Singh
(1976) found, as shown in Figure 2, for a converging section:
D a10
n
or log D log a Cn 27
where a is an empirical constant. Equation (27) shows that is highly sensitive to exponent n. Therefore,
an accurate determination of n is essential for accurately determining . Because n is reective of the ow
regime which changes with the evolution of ow, its dependence on the ow depth or Reynolds number needs
to be quantied. The value of a varies with the nature of the surface. Some typical values of parameters are
shown in Table III.
In hydrologic modelling it is usually assumed that parameters and n are independent of each other, that
parameter may or may not be a constant, and that the exponent n is taken as 15 or 167. The reason is that
Equation (24) is then expressible either as Chezys equation n D 15 or Mannings equation n D 167.
Forcing n to take on a xed value translates Equation (24) into a one-parameter kinematic wave ux equation.
The advantages of so doing are simplicity, reduced need for data and familiarity with the use of Chezys
or Mannings equation. The above assumption is clearly an approximation, not a true representation of ow
characteristics.
Storagedischarge relation
In systems hydrology a fundamental relation is the storagedischarge relation which is frequently invoked
for routing excess rainfall. This relation is represented as
S D kQ
m
28
where S is storage, Q is discharge of runoff and k and m are empirical parameters. If m is assumed to equal
one, k becomes the lag time. This assumption implies that the watershed is linear and forms the basis of
the unit hydrograph theory. The value of k varies with watershed characteristics and m varies with the ow
regime, namely laminar, turbulent or mixed (called mixed ow). The value of m is not constant and may have
a pronounced effect on the predicted hydrograph (Weeks, 1980; Bates and Pilgrim, 1983; Bates and Townley,
1985; Pilgrim, 1986; Singh, 1988; Kuczera et al., 1989; Maheshwari, 1994). Equation (28) is a variant of the
kinematic wave ux law given by Equation (24). As in the case of Equation (24), parameters k and m in
Equation (28) are also interdependent. If m is not equal to one, the meaning of k changes and it no longer
remains the lag time.
678 V. P. SINGH
1000
500
100
50
10
5
1
1.0 1.2 1.4 1.6 1.8
n
2.0 2.2 2.4

Figure 2. Relation between and n for a converging section on an experimental facility (after Singh, 1976)
Overland ow
Overland ow is dominant on small watersheds, parking lots, rooftops, highways, airport runways,
articially constructed wetlands, agricultural plots, lawns, and so on. Such basins are bereft of a channel
network and their surface ow is essentially overland ow. Woolhiser and Liggett (1967) derived a criterion
for judging the goodness of kinematic wave approximation in modelling ow over a sloping plane subject to
rainfall or lateral inow. By expressing the St. Venant equations of continuity and momentum they derived
what is now referred to as the kinematic wave number, K, which reects the effects of length and slope of
the plane as well as normal ow depth and velocity. The kinematic wave number is expressed as
K D SoLo/hoFo
2
29
where So is the slope of the plane, Lo is the length of the plane, ho is the normal ow and Fo is the Froude
number for normal ow at x D Lo at a discharge of i
max
Lo, i
max
D maximum rainfall intensity.
For K !1, the dimensionless momentum equation reduces to the Chezy relation. Based on numerical
experimentation on rising hydrographs at the downstream end of the plane, Woolhiser and Liggett (1967)
Table III. Values of n and for butyl, gravel and
butyl plus gravel surfaces (after Singh, 1976)
Surface n (ft
2n
/s)
Butyl 154 1192
Butyl 1952 13826
Butyl 2243 114787
Butyl 1835 4929
Gravel 1707 1435
Gravel 1947 4049
Gravel 1828 1819
Butyl plus gravel 2149 16554
showed that for K > 20 the kinematic wave approximation would be very good. For highly subcritical ows,
Morris and Woolhiser (1980) showed that for KFo
2
5 the kinematic wave approximation would be good.
For many natural slopes K is usually greater than 50, and the kinematic wave theory applies. In smooth
urban surfaces the value of K is normally between 5 and 20, the kinematic wave approximation would be a
crude approximation. Since Woolhiser and Liggett (1967) showed that for large values of the kinematic wave
number (greater than 20) the kinematic wave approximation was sufciently accurate for overland ow, there
have been numerous studies applying the kinematic wave theory to modelling overland ow (Eagleson, 1970;
Singh, 1996a).
For overland ow on a wide rectangular plane the uxconcentration relation is expressed as
q D h
n
30
where q is the discharge per unit width of the plane, h is the depth of ow and and n are parameters.
Equation (30) is the same as Equation (10), where the ow depth is a measure of ow concentration and the
discharge per unit width a measure of ux. A one-dimensional continuity equation taking into account the
lateral inow to and outow from the plane can be written as
h
t
C
q
x
D ix, t fx, t 31
where ix, t is the intensity of rain falling on the plane and fx, t is the rate of inltration of water into the
plane bed. Combining Equations (30) and (31), the kinematic wave equation for overland ow on a plane is
obtained:
h
t
Cc
h
x
D ix, t fx, t 32
where
c D q/h
x constant
D nh
n1
D nu u D q/h 33
where u is the average velocity and c is the wave celerity.
Wooding (1965a,b, 1966) was probably one of the rst to apply the kinematic wave theory to model surface
runoff. There has since been a proliferation of its application in watershed hydrology, and there is a general
consensus that the kinematic wave theory is reasonably accurate for modelling overland ow (Smith and
Woolhiser, 1971; Kibler and Woolhiser, 1972; Singh and Woolhiser, 1996; Woolhiser, 1996; Singh, 1996a).
The question arises: is overland ow (or surface runoff) kinematic? The answer to the question depends
on the nature of the surface and the spacetime distribution of rainfall. In most cases, overland ow is
subcritical (Froude number less than one) and turbulent. In rare cases it is laminar as, for example, it may
680 V. P. SINGH
develop for a short time on urban and highway pavements and in wetlands. The discussion in the preceding
section shows that the kinematic wave is the dominant wave in surface runoff, especially during the rising
part of the hydrograph and during much of the recession part. The dynamic and diffusion waves do exist but
are short lived and play a minor role. Thus, overland ow can essentially be construed as kinematic, with full
recognition that diffusion and dynamic waves may dominate in some cases.
Channel ow
Flow routing in channels involves calculation of a downstream hydrograph from a known upstream
hydrograph. The use of kinematic wave theory is quite popular for ow routing in storm sewers for urban
drainage and in channels. The movement of ood waves in a river is dominated by a balance between the
friction of the bottom and the component of gravity in a direction which is downstream and parallel to the
free surface (Lighthill and Whitham, 1955a). Denoting the downward slope of the free surface by S, the
gravitational force per unit length of the river is gSC, where C is the concentration (volume of water per
unit length), g is the acceleration due to gravity and is the density. The frictional force per unit length can
be expressed as fu
2
P, where P is the wetted perimeter of the ow cross-section and f is the coefcient of
friction. Equating these forces one obtains
u D
g
f
05
C
P
05
S
05
34
The quantity (C/P) is equal to the hydraulic radius R. If g/f
05
is equated to Chezys coefcient,
Equation (34) is the famous Chezy formula. Equation (34) can be written as
u D R
05
D g/f
05
S
05
35
The coefcient of friction f varies with the ratio of the size of typical roughness elements in the bed to
the hydraulic mean depth R. Taking f as f D R
1/3
, D constant, Equation (34) becomes for constant S
and P:
u D R
2/3
D g/f
05
S
05
36
which is the famous Manning relation. Thus, the uxconcentration relation for ood movement in rivers is
kinematical.
In his fundamental study Seddon (1900) emphasized that great rivers, unlike man-made conduits, do not
have a uniformly sloping bed, nor do they in any way approximate to this condition. The bed slope exhibits
enormous variations (including changes of sign) across the width of the river as well as in the downstream
direction over distances comparable to the width. Furthermore, the large-scale variation of the bed frequently
resembles a series of pools and bars. The ow control may thus vary with ow depth. At low ow, the
ow from one pool to the next is more like the ow over a submerged weirthat is one of the bars. In the
neighbourhood of the bars the value of S would be large. At high ow, the part of a pool may become the
narrowest section of the river and control ow like an orice with vertical walls.
In the case of alluvial rivers the bed constantly changes with time. Variations of 3 m in depth about its
mean are common on the lower Mississippi. At the same time the elevation of the water surface changes only
slightly, say from 5 cm to 10 cm, for the same value of specic discharge q. The elevation of the surface
varies far more smoothly in time as well as in space than the depth of the bottom. Denoting the elevation of
the free surface at each point above a certain datum by stage h, the ow q can be regarded as a function of
h rather than the cross-section area A. For constant x
dA D Bdh 37
where B is the local width of the river. Thus, Equation (6) becomes
c D
1
B
q
h
x xed
38
Equation (38) is less susceptible to variation in time. Thus, in general one can write
q D qh, x 39
With the knowledge of the width
B D Bh, x 40
as a function of stage and position, the wave celerity c can be predicted. According to Seddon (1900), rivers
conform to Equation (38).
In ow in channels, dynamic waves always occur. The friction and slope terms modify the amplitude of
waves. In some cases, slope is the dominant controlling factor, as for example ow over a spillway, ow
in a steeply sloping gutter, ow on an embankment surface, to name but a few. Friction dominates in such
cases as at to gently sloping forest, grass and cropped lands. On the other hand, geometric features, such
as bars, dunes, antidunes, pools, rifes, meandering, etc., exercise local control on ow and modify wave
characteristics. Then, there are man-made controls, such as dams, bridges, levees, etc. which modify ow
characteristics. These modications are not conducive to kinematic waves.
In oods, the wave modication is made to such a degree that dynamic waves rapidly become negligible
and the kinematic waves assume the dominant role. When the Froude number is less than 1, dynamic waves
decay exponentially and are readily damped out. For F D 2, the kinematic waves merge with dynamic waves
and for F > 2, the kinematic waves no longer exist. For 1 < F < 2, dynamic waves dominate kinematic
waves. In channel ow, all waveskinematic, diffusion and dynamic as well as their variantsexist. At a
given position, the relative signicance of these waves changes with the changing nature of ow in time. For
most river ow without articial structures, F < 1, kinematic waves are dominant and hence the kinematic
wave theory applies.
For ood movement in rivers the uxconcentration Equation (39) is of the type given by Equation (24).
Its special forms, given by Equations (35) and (36), are commonly used in channel ow routing. Coupling it
with the one-dimensional continuity equation, the kinematic wave equation becomes
h
t
Cnh
n1
h
x
D 0 41
where the wave celerity is
c D nh
n1
D nu 42
which is the same as Equation (33). Since the pioneering work of Lighthill and Whitham (1955a), the kinematic
wave theory has been popular for routing ows in channels and channel networks (Singh, 1996a).
Dam break ow
Dams can break instantaneously or over a nite period of time. A ood wave generated by dam rupture
propagates downstream. Such oods are usually large and cause huge damage in terms of property, animal and
human life and environment. A large number of approaches have been employed to determine the movement
of a dam-break ood wave down a river. Hunt (1982, 1983, 1984a,b, 1987) pioneered the use of kinematic
wave theory for modelling ood wave propagation under different conditions: (1) on a dry sloping bed, (2) on
a wet sloping channel, and (3) based on a reservoir, dam breach and downstream channel, all of different
widths.
682 V. P. SINGH
The equations for dam-break ood wave propagation are the same as Equations (7) and (24). In this case,
however, when water is released upon dam break, a front wall of water advances down the channel. This wall
of water is the shock front, that is, the interface between (newly) water-covered and dry (or previously existing
water) parts of the channel. Let x D st or inversely t D wx be the time history of that advancing front; this
gives the spacetime history of the shock front. The front is a free boundary that must be determined along
with the solution. The equation for the free boundary is obtained from the expression for the ow velocity
ux, t by observing that the shock front moves with the speed of water immediately behind it. Thus, the
shock velocity takes the form of ust, t or ux, wx. This yields
dst
dt
D h
n1
st, t 43
A comparison with previously published solutions and experimental results showed the kinematic wave
solution was asymptotically valid after the advance of the ood wave downstream by about four reservoir
lengths. Katapodes and Schamber (1983) found in a comparative study of ve dam-break ood wave models
ranging from the complete dynamic equations to a simple normal depth kinematic wave equation that the
kinematic wave model was sufciently accurate.
The ow prole at a given time upon dam break has three regions from the upstream to the downstream:
region I is close to the dam where the dynamic wave is the dominant wave; region III is the region close to
the right end of the prole or shock front where the dynamic wave is also dominant; and region II is between
these two regions, which actually occupies much of the channel longitudinally and where the kinematic wave
is the dominant wave. Thus, dam-break ood waves can be reasonably approximated by kinematic waves.
With progression of time, the dominance of kinematic waves grows and dynamic waves dissipate.
Comparison of kinematic wave theory with diffusion wave and dynamic wave theories
Lighthill and Whitham (1955a) have shown that for Froude numbers below 1 (appropriate to ood waves)
the dynamic waves are rapidly attenuated and the kinematic waves become dominant. Using a dimensionless
form of the St. Venant (SV) equations, Woolhiser and Liggett (1967) obtained the kinematic wave number
K as a criterion for evaluating the adequacy of the kinematic wave approximation. For K greater than 20,
the kinematic wave approximation was considered to be an accurate representation of the SV equations in
modelling overland ow. Morris and Woolhiser (1980) modied this criterion with an explicit inclusion of
Froude number corresponding to normal ow, Fo, and showed, based on numerical experimentation, that
Fo
2
K 5 was a better indicator of the adequacy of the kinematic wave approximation.
Using a linear perturbation analysis, Ponce and Simons (1977) derived properties of the kinematic wave,
diffusion wave and dynamic wave representations in modelling open channel ow. Menendez and Norscini
(1982) extended the work of Ponce and Simons (1977) by including the phase lag between the depth and
velocity of ow. Based on propagation characteristics of a sinusoidal perturbation, they derived criteria to
evaluate the adequacy of kinematic wave and dynamic wave approximations. Daluz Vieira (1983) compared
solutions of the SV equations with those of the kinematic wave and dynamic wave approximations for a
range of Fo and K, and dened regions of validity of these approximations in the KFo space. Fread (1985)
developed criteria for dening the range application of the kinematic wave and dynamic wave approximations.
Ferrick (1985) dened a group of dimensionless scale parameters to establish the spectrum of river waves,
with continuous transitions between wave types and subtypes.
The comparative studies, cited above, show that many cases satisfy the conditions for the validity of
kinematic wave theory and that in surface water hydrology kinematic waves dominate in such cases. Other
wave types may exist but either they are short lived or play a minor role. Thus, it may be concluded that
surface hydrology is kinematic under certain restrictions. This is further elaborated by considering two special
cases: (1) uniform, unsteady ow and (2) steady, non-uniform ow.
Unsteady uniform ow. In a series of papers Singh (1993, 1994ac, 1995, 1996a,b) derived, under simplied
conditions, error equations specifying error as a function of time for kinematic wave approximation for space-
independent ows. He considered four different types of scenarios depending upon the presence of lateral
inow or rainfall and inltration: (1) lateral inow is constant and there is no inltration and if there is, it is
included in lateral inow; (2) both lateral inow and inltration are considered as constant; (3) both lateral
inow and inltration are included but their difference is zero; (4) there is no lateral inow but inltration
is included. These scenarios were analysed under two types of initial conditions: (1) the plane or channel is
initially wet and (2) the plane or channel is initially dry. In all 18 cases were analysed.
For space-independent ow the continuity equation takes the form
dh
dt
D i f 44
and the momentum equation takes the form
du
dt
D gS
0
S
f

iu
h
45
where g is acceleration due to gravity, i is lateral inow (or rainfall intensity), f is inltration rate, S
0
is bed
slope, S
f
is slope of the energy line. For kinematic wave approximation, Equation (45) becomes
S
0
D S
f
46
which can be expressed as
u D
S
0
05
h
05
47
where is a parameter. Singh (1994a,b, 1995, 1996a,b) derived for all cases error equations which turned
out to be Riccati equations of the form
dE
d
D C
0
CC
1
, E CC
2
, E
2
48
where E is error dened as (kinematic wave solution dynamic wave solution)/dynamic wave solution, is
dimensionless time, is a dimensionless parameter and C
i
, i D 1, 2, 3 are Riccati coefcients. The parameter
is analogous to the kinematic wave parameter and is dened as
D
4g
2
S
0
h
0
i
2
0
49
or
D 4gS
0
50
depending upon whether lateral inow (i
0
) is included or not. In general, the kinematic wave approximation
is very good if is equal to or greater than 10. As time progresses, i.e. is greater than or equal to 10,
the kinematic wave approximation converges to the dynamic wave solution, as shown for a typical case in
Figure 3.
Steady non-uniform ow. Steady non-uniform ows are encountered in a variety of natural situations.
In overland ow, the steady state is attained for constant rainfall after the ow depth at the outlet has
reached equilibrium. This same is true for channel ow subject to constant lateral inow. For a channel
receiving a constant inow of long duration at the upstream boundary, the ow at the downstream end would
reach equilibrium. The steady state solution aids in understanding the nature of water surface prole. It may
684 V. P. SINGH
0.53
0.48
0.43
0.38
0.33
0.28
E
r
r
o
r
0.23
0.18
0.13
0.08
0.03
0 10
= 0.5
= 1.0
= 1.5
= 2.0
= 2.5
= 3.0
= 3.5
= 4.0
20 30
Dimensionless time
40 50 60
Figure 3. Error in kinematic wave approximation as a function of dimensionless time for a space-independent case for different values of
, with initial condition dened by constant inow, constant velocity or constant ow depth and zero inltration
help dene the condition for use of zero depth in place of zero inux at the upstream boundary. When
rainfall duration is much longer than the time of equilibrium, the steady state water surface proles are very
useful. Pearson (1989) examined the criteria for using the kinematic wave approximation to the SV equations
for shallow water ow. For steady state one-dimensional ow over a plane he derived a new criterion as
K 3 C5/Fo
2
where K is the kinematic wave number and Fo is the Froude number corresponding to
normal ow.
Parlange et al. (1990) investigated errors in the KW and DW approximations by comparing their predictions
with the numerical solution of the SV equations under steady state conditions. They suggested splitting the
solution in two regions, one near the downstream end of the plane and the other covering most of the plane.
Singh and Aravamuthan (1995a,b, 1996, 1997) derived errors in the kinematic wave approximations under
four conditions: (1) zero ow at the upstream boundary, (2) nite depth at the upstream boundary, (3) critical
ow depth at the downstream end, and (4) zero-depth gradient at the downstream boundary. Depending on the
inclusion of lateral inow and inltration, 21 cases were analysed for the four different scenarios mentioned
K=3.0
10.00 0.05 0.10 0.15
100.043
88.366
76.689
65.012
53.335
41.658
29.981
18.304
6.627
5.050
16.727
P
e
r
c
e
n
t
a
g
e

e
r
r
o
r

i
n

h
e
a
d
0.20 0.25 0.30
Dimensionless distance X
0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00
K=5.0 K=10.0 K=30.0
Figure 4. Percentage error in kinematic wave approximation as a function of dimensionless distance for different values of kinematic wave
number (K), with critical depth downstream boundary condition, constant inow at the upstream boundary q D 1, zero inltration and
Froude number as 1
above. By comparing the kinematic wave solution with the dynamic wave solution, error equations were
derived for all cases.
For time-independent ow, the governing equations are
dh
dx
D i f 51
and
d
dt
1
2
u
2
Cgh
D gS
0
S
f

iu
h
52
The error equations were generalized Riccati-type equations:
dE
dx
D C
0
E CC
1
E
2
CC
2
E
3
CC
3
E
4
CC
4
E
5
53
where the parameters C
i
, i D 1, 2, 3, 4 are nonlinear functions of x, h, K and Fo which are as dened before.
In most cases it was found that the downstream boundary exercised signicant inuence on the adequacy of
the kinematic wave approximation. Away from the boundaries, i.e. 01 x/L 09, where L is the length
of the plane, the kinematic wave approximation was a good approximation, as shown for a typical case in
Figure 4.
APPLICATION OF KINEMATIC WAVE THEORY TO IRRIGATION HYDROLOGY
Surface irrigation involves movement of water as shallow ow over planes or in channels. When an inow
stream is introduced at the upstream end of the plane, water advances with a sharply dened wetting front
686 V. P. SINGH
down the slope toward the downstream end in what is referred to as the advance phase of the irrigation
ow process. The wetting front is a shock and constitutes a free boundary. This phase is characterized by
downeld movement of the advancing water front and continues until the water reaches the lower end of the
eld. Assuming continued inow after water has advanced to the downstream end of the eld, water will, if
there is no downstream dam, ow out at the end of the eld and continue to accumulate in the eld in the
storage phase. In this phase, water exists on the entire eld, no other boundary moves and inow continues at
the upper end of the eld. The storage phase ends and the depletion phase begins when the inow ceases. The
depletion phase continues until the depth of surface water at the upstream end is reduced to zero. This phase
differs from the storage phase only in the absence of inow into the eld. The recession phase begins when
the depth of surface water at the upstream end decreases to zero. This marks the formation of the recession
or drying front. The downeld movement of the drying front characterizes the recession phase of the inow.
This phase continues until no water remains in the eld and irrigation is complete. This is the schematic of
the irrigation cycle (Sherman and Singh, 1978; Maheshwari, 1988).
It should be pointed out that irrigation hydrology has many similarities with surface water hydrology of
the preceding section. Thus, only those aspects that are peculiar to irrigation hydrology will be dealt with
here. In a study on errors in the kinematic wave solution by comparison with full dynamic wave solution
in irrigation borders and channels, Reddy and Singh (1994) and Singh (1994ac) found the kinematic wave
approximation to be reasonably accurate. Irrigation hydrology is not kinematic in a strict sense, but as will
be shown in what follows kinematic wave theory is a good approximation.
Flow in irrigation borders
For ow in irrigation borders, the uxconcentration relation of Equation (30) holds. However, because
irrigation borders have porous beds, the continuity equation requires modication and can be expressed as
h
t
C
q
x
D f 54a
where f is the inltration rate (volume per unit area) and is the elapsed time after the front wall of water
covers a position x and is dened as
D t x 54b
where t D x is the time history of the advancing front and constitutes a free boundary of the solution
domain. The equation for the free boundary x D st or t D x can be obtained by replacing x by st and
observing that ust, t is the velocity of the front wall of water, that is
ust, t D
dst
dt
D h
n1
st, t s0 D 0 55
Chen (1970), Smith (1972) and Cunge and Woolhiser (1975) were among the rst to apply the kinematic
wave theory to model some aspects of ow in irrigation borders. Sherman and Singh (1978, 1982) and Singh
and Ram (1983) presented a complete mathematical treatment of the irrigation cycle based on the kinematic
wave theory and showed that the theory was sufciently accurate for irrigation modelling.
Flow in irrigation borders is similar to ow in inltrating channels or over porous beds. The kinematic wave
theory is a good approximation but is not capable of simulating the entire irrigation cycle, as exemplied by
its inability to simulate the depletion phase. This is because the kinematic wave assumption gives zero ow
depth at the upstream end as soon as the ow is terminated, whereas in reality the depth becomes zero after
some time. However, in most irrigation cases this limitation is not severe, for the depletion phase is minor
compared with other phases of the irrigation cycle. There is another crude approximation that kinematic wave
theory makes, i.e. theory yields that the shock front moves as a vertical wall whereas in actuality it moves
as a front with nite slope (h/x < 0). These limitations notwithstanding, the border irrigation ow, for the
most part, does satisfy the conditions for validity of the kinematic wave approximation.
Flow in irrigation furrows
Irrigation with furrows is similar to that with borders but its hydraulics and intake phenomena are
complicated by the geometry of the furrow cross-section and by the occurrence of unsteady and non-uniform
ow. Walker and Humpherys (1983) discussed three modications that are required to develop border irrigation
models into furrow irrigation models. First, the geometry of the furrow cross-section must be prescribed.
Although furrows are not prismatic, their hydraulic parameters, such as the depth of ow, wetted perimeter,
cross-sectional area, etc. can be described using simple power functions. Second, inltration must be described
by a simple function that incorporates both a time-dependent rate and a basic or steady rate term (such as the
KostiakovLewis equation) rather than a simple time-dependent rate term (such as the Kostiakov equation).
Elliott and Walker (1982) found that the advance could be simulated reasonably well without the addition
of the basic rate but not runoff. Third, inltration is affected by wetted perimeter and thereby ow rate
(Fangmeier and Ramsey, 1978). A commonly used assumption is that inltration is only a function of intake
opportunity time and can therefore be expressed as a function of time. However, inltration depends on
discharge, thereby making it a function of time as well as ow rate. The furrow geometry is expressed using
rigid perimeter models, non-rigid perimeter models or constant shape models.
The kinematic wave equations for furrow irrigations are similar to those for border irrigation. Walker and
Humpherys (1983) applied the kinematic wave approximation to furrow irrigation and solved the equations
using rst-order Eulerian integration. Rayej and Wallender (1987, 1988) reformulated the model of Walker
and Humpherys (1983) by using distance as input and solving for advance times. This permitted inltration
characteristics to vary at any point along the furrow. Wallender and Yokokura (1991) developed an explicit
kinematic wave model in which the time step was adjusted to have computational nodes to fall at specied
locations along the furrow. In all of these modelling efforts the kinematic wave approximation was found to
be sufciently accurate.
In furrow irrigation, the kinematic wave theory is not as good an approximation as it is in border irrigation.
One reason is that the depletion phase may be signicant and lasts much longer. Because furrows are temporary
channels, there may be strong interaction between ow and furrow geometry. Furrow geometry changes in time
due to erosion and sediment transport. Consequently, the hydraulic gradient of ow depth may be signicant
and cannot be neglected. Thus, waves tend to be more diffusive. Nonetheless, as a rst approximation the
kinematic wave theory is a good approximation.
APPLICATION OF KINEMATIC WAVE THEORY TO VADOSE ZONE HYDROLOGY
Much of the mathematical treatment of ow in unsaturated porous media has dealt with capillary-induced
ow (Smith, 1983). However, there exists a multitude of cases where gravity dominates vertical movement
of soil moisture. Some examples include drainage following inltration, water percolation deeper into the
soil, vertical movement of moisture in relatively porous soils when rainfall or surface uxes are typically of
the order of, or less than, the soil-saturated hydraulic conductivity, to name but a few. For treatment of such
cases, the kinematic wave theory is simple yet reasonably accurate.
Unsaturated ow
For gravity-dominated unsaturated ow, it can be postulated that the moisture ux is a function of only the
moisture content . This postulate permits application of kinematic wave theory. The moisture volume ux
(or Darcy ux) q can be expressed as
q D K 56
where K is the unsaturated hydraulic conductivity. A popular model relating K to the moisture content is
(Brooks and Corey, 1964)
q D K D K
s

0

0
a
D K
s
S
a
e
S
e
D
S S
0
1 S
0
, S D

57
688 V. P. SINGH
where K
s
is the saturated hydraulic conductivity,
0
is the moisture content at eld capacity, S
e
is the
effective saturation, S
0
is the saturation at D
0
, is the porosity and a is an exponent related to the pore
size distribution index. Equation (57) is a kinematic wave ux law. One can also write Equation (57) as
D
0
C
0
q
K
s
1/a
58
In the kinematic wave approximation the dissipative terms (second-order spatial derivatives) are neglected and
hence the character of the transport equation is changed from parabolic to hyperbolic. In many applications,
the role of dissipation decreases with time while that of advection remains constant, thus validating the use
of the kinematic wave approximation.
The one-dimensional continuity equation can be written as
t
C
q
z
D ez, 59
where z is the vertical distance positive in the downstream direction, e (z, ) is the evaporation rate, D t sz
is the time history of the wetting front and sz is the spacetime history of the advance front. Coupling
Equations (56) and (59) one gets the kinematic wave equation for soil moisture movement in the vadose zone:
t
CM
z
D ez, 60
where MD q/
z constant
is referred to as the mobility of water in soil (Irmay, 1956).
Although Sisson et al. (1980) applied the kinematic wave theory to internal drainage, Smith (1983) was
probably the rst to apply the theory to develop a complete kinematic wave model for soil moisture movement.
Charbeneau (1984) extended Smiths work. In a series of papers, Germann and coworkers (Germann, 1987;
Germann and Beven, 1985, 1986; Germann et al., 1987) extended the application of the theory to inltration
and drainage into and from soil macropores, as well as microbial transport. Yamada and Kobayashi (1988)
discussed the kinematic wave characteristics of vertical inltration and soil moisture, with the aid of eld
observations on tracers. They concluded that the vertical inltration of soil moisture had the characteristics of
kinematic waves. Singh and Joseph (1994) extended the theory to model soil moisture movement with plant
root extraction.
The question arises: is unsaturated ow kinematic? The answer to the question depends on the status of soil
moisture. At low moisture contents, capillary forces dominate and ow is dominated by the diffusion wave,
although the kinematic wave also exists. With increasing moisture content, the diffusion wave diminishes in
value and the kinematic wave begins to dominate. Thus, at a xed position in soil both waves exist at a time
but their relative magnitudes are governed by the moisture content at that time.
Macropore ow
Structural soil pores constitute important pathways for water and may carry water before the ner pores of
the soil matrix are fully saturated. Macropores may be characterized as animal burrows, such as worm and
insect tunnels or structures built by small mammals; as channels formed by plant roots; as desiccation cracks;
or as cracks and joints formed by various physical and chemical geological processes. The permeability of a
soil is controlled by the presence of large pores that do not exert capillary forces on water owing in them.
Water is more efciently transmitted through structural pores and not all smaller pores are lled with air
prior to the inltration event participating in the ow process. Flow may occur in two domains: (1) structural
pores and (2) soil matrix. In structural pores, referred to as macropores or the macropore system, the ow of
water is primarily driven by gravity and is only faintly impeded by capillary forces. In the soil matrix or soil
micropores, the ow of water is subject to capillary forces. Micropores are often not completely saturated
with water before structural pores can actively transport water, because the time required for saturation of
micropores may exceed that required to establish ow in macropores (Germann, 1990a,b). When the minimum
width is 3 mm, macropores are assumed to exert negligible capillarity on water owing in them and the
capillary potential is assumed greater than 001 m (Beven and Germann, 1981). Macropore systems may
also include channels of considerably smaller diameter, as small as 003 mm, provided the channels extend
markedly in the general direction of ow.
To deal with the ow in macropores, the entire soil or rock volume, not the individual ow path, is
considered. This is a macroscopic denition of the ow system and does not require a detailed description of
the actual ow paths, including morphological descriptions and detailed geometrical and spatial denitions of
the macropore systems. Fundamental to this approach is the uxconcentration relation dened as
q D bw
a
61
where q (m/s) is the volume ux density, dened as the volume of water in the macropore system owing
across a unit area of the entire porous medium per unit time; w is the macropore water content (m
3
/m
3
)
dened as the volume of water in the macropore per unit volume of the entire porous medium; and a and
b are parameters. Parameter b (m/s) is called conductance and parameter a is a dimensionless exponent. An
appropriate value of parameter a is 25. Germann (1990a,b) has presented solutions for ow in macropores
under different conditions: (1) Flow in a homogeneous macropore system that is surrounded by a water-
saturated matrix; (2) ow in a homogeneous macropore system that is surrounded by a water-sorbing matrix;
and (3) ow in a heterogeneous macropore system that is surrounded by the water matrix. He found the value
of parameter a to vary between 22 and 376.
The one-dimensional continuity equation is
w
t
C
q
z
D 0 62
Coupling Equations (61) and (62), one obtains the kinematic wave equation for ow in a homogeneous
macropore system surrounded by a water-saturated matrix as
q
t
Cc
q
z
D 0 63
where c is the wave celerity dened by
c D
dq
dw
D ab
1/a
q
a1/a
64
If the macropore system is surrounded by a water-sorbing matrix then a sink term has to be added to
Equation (62) to account for the loss of water into the matrix. Thus, the kinematic wave equation becomes
q
t
Cc
q
z
Ccrw D 0 65
where r (l/s) is the sorbancethe rate of water loss per unit time and per unit w such that
w
t
D rwt 66
Germann (1987, 1990a,b), Germann et al. (1987), DiPietro and Lafolie (1991) among others have demonstrated
that the kinematic wave theory ts the experimental data reasonably. When rain intensities are low, however,
there is high dispersion and the kinematic wave theory does not produce ow hydrographs accurately. The
question arises: is macropore ow kinematic? In macropores ow is governed by gravity and capillarity
plays a minor role, unless rainfall intensities are very low. Thus, it can be concluded that macropore ow is
kinematic under a broad range of conditions but is diffusive outside this range.
690 V. P. SINGH
APPLICATION OF KINEMATIC WAVE THEORY TO SUBSURFACE STORM FLOW
Interow
In many parts of the world storm runoff is produced by subsurface runoff (Dunne, 1978). Subsurface
ow, throughow, saturated interow and interow are other designations used somewhat interchangeably
in hydrological literature. Beven (1989) dened interow as the near-surface ow of water within the soil
prole resulting in seepage to a stream channel. It may involve both saturated and unsaturated ows in both
vertical and downslope directions, and may encompass all surface contributions to the storm hydrograph.
When the seepage face extends onto the soil surface the inow process may transform into or encompass the
process of return ow by which the subsurface ow can contribute to overland ow. In some watersheds,
there may exist a steep hydraulic gradient either on account of the slope of the hillslope (Beven, 1981) or
the buildup of groundwater mounds on shallow slopes (Sklash and Farvolden, 1979). Under such conditions,
the watershed or hillslope response to rainfall may be dominated by subsurface stormows involving both
saturated and unsaturated ows. The characteristics of such a response depend on the depth and hydraulic
conditions of the unsaturated zone existing at the time of rainfall storm (Dunne, 1978). The kinematic wave
theory can be applied to model stormow, provided certain requirements relative to the physical nature of the
watershed or hillslope are met. According to de Vries (1987) these requirements are a thin saturated zone, a
steep bed and a low recharge rate. If the zone of saturation is thin, water table and ow line can be assumed
to be approximately parallel to the bed; this assumption, however, requires a high value of saturated hydraulic
conductivity in this zone.
On a sloping soil mantle with constant hydraulic conductivity Beven (1981) found that the kinematic wave
approximation was a good representation of the extended Dupuit Forchhmier theory in terms of predicting
both the water table proles and subsurface stormow hydrographs, if the nondimensional parameter was
less than 075. This parameter is dened as
D
4i cos S
0
Ksin S
0
67
where i is the input per unit area parallel to the impermeable bed, S
0
is slope of the hillslope and K is
the hydraulic conductivity. By using the critical value of , a range of values of slope angle and saturated
hydraulic conductivity for which the kinematic wave approximation is valid was specied. Beven (1981)
found the kinematic wave approximation to be useful in cases of practical interest. Beven (1982a) examined
the response times of unsaturated and saturated ows on hillslopes using the kinematic wave approximation.
He showed that for many combinations of hillslope and soil parameters subsurface response times were too
long to be considered as stormow and the combinations of high hydraulic conductivities, shallow soils and
wet initial conditions leading to fast response were quite reasonable and are expected in eld conditions. For
such cases the kinematic wave approximation yields useful results.
The question arises: is subsurface ow kinematic? The answer is not always so, depending on the rate
of recharge, hydraulic conductivity of the soil, bed slope and the thickness of the saturated zone. One case
where the kinematic wave theory does not do well is where the porous medium is cut by the bed of a stream
or lake. The kinematic wave assumption does not accommodate the lower boundary condition. As a result it
will always overpredict the length of the seepage face at the downstream boundary, since it takes no account
of downstream effects in the vicinity of the boundary (Beven, 1981). Away from the seepage face to the
upstream, the theory would yield acceptable results, provided is less than 075. The farther the distance
from the lower boundary, the more accurate the theory, since the effect of the downstream boundary will
decline.
Hillslope response
Three types of approach can be distinguished for the study of storm runoff generation: antecedent moisture-
based, runoff type-based and generation mechanism-based. The generation mechanism-based approach forms
the basis of the most widely accepted theories of storm runoff generation. This approach examines the manner
in which the runoff water travels over the last several tens of metres to the stream. It accounts for both the rapid
response of the stream to runoff-producing events and the observed increase in stream discharge. The dominant
mechanisms in streamow generation are Hortonian overland ow, partial area-overland ow, variable source
area-overland ow, variable source area-subsurface storm ow (Wigmosta et al., 1994) and deep aquifer or
groundwater ow. Freeze and Harlan (1969) sketched a blueprint for investigating the dynamic response of a
hillslope. Smith and Hebbert (1983) described the interaction of surface and subsurface hydrologic processes.
They modelled surface ow using the kinematic wave approximation, the subsurface ow by a simplied but
analytical approach and the subsurface saturated ow by the Dupuit approximation. Beven (1982b) employed
the kinematic wave approximation to model ow from both the unsaturated and saturated zones and showed
that the kinematic wave model produced results that were in good agreement with eld observations. Sloan
and Moore (1984) obtained similar results for subsurface stormow on steeply sloping forested watersheds.
In hillslope hydrology, many complications arise from variable rainfall intensities, variable initial conditions
downslope, changes in soil depth and slope angle along the slope, topographic convergence and divergence,
existence of a saturated zone prior to the onset of rainfall, macropore channelling through unsaturated zone,
etc. The hillslope hydrology is not kinematic but, as argued by Beven (1987), the kinematic wave theory
can be considered as a good approximation to a more rigorous, physically based description. Thus, there is
a strong argument in favour of the kinematic wave theory capable of mimicking the important features of
streamow generation.
APPLICATION OF KINEMATIC WAVE THEORY TO RIVER AND COASTAL HYDROLOGY
In river and coastal hydraulics, kinematic wave theory has received much less attention. That may be because
historically laboratory experimentation has played a signicant role in development of hydraulics and there is
a good deal of both laboratory and eld data on hydraulic phenomena for model parameter calibration. The
spatial scales at which hydraulic problems, in general, are dealt with are much smaller than those in hydrology.
As a result, there occur much less averaging and lumping. Nevertheless, many hydraulic phenomena can indeed
be approximated by the kinematic wave theory as shown in what follows.
Sediment transport in upland areas
Erosion and transport of soil by water from upland areas represents a complex interaction between the
kinematics of the falling rain, the hydrodynamics of ow and the dynamics of granular materials. Sediment
transport involves: (1) erosion by rainfall input; (2) erosion by owing water; (3) transport by rainfall splash;
(4) transport by owing water; and (5) deposition during ow. The amount of sediment available for transport
is the sum of the sediment inow from contributing ow units, initially loose soil left from previous storms,
the amount of soil detached by raindrop impact and the amount of soil detached by ow of water. The erosion
and sediment ux due to rainfall impact, E
I
, is expressed as
E
I
D BI
a
68
where B is a coefcient and I is rainfall intensity; a is an exponent; and E
I
represents interrill erosion or
sheet erosion and varies with rainfall intensity, soil characteristics, vegetative cover and slope. Usually, the
exponent a is taken as unity. Interestingly, Equation (68) is a kinematic relation.
The sediment ux, E
R
, representing rill erosion and accounting for deposition can be expressed as
E
R
D Bh
n
CQ 69
where and B are coefcients, Q is water discharge, h is the ow depth and C is sediment concentration.
Equation (68) is a kinematic relation, meaning that both aggradation and degradation are kinematic.
692 V. P. SINGH
The sediment discharge Q
s
can be expressed as
Q
s
D CQ 70
where Q is water discharge and is obtained from the kinematic wave Equation (32).
A one-dimensional form of sediment continuity equation is
Ch
t
C
Q
s
x
D E
I
CE
R
71
Equation (71) is the kinematic wave equation and has been the basis of a number of studies on soil erosion.
Smith (1981), Singh (1980, 1983), Singh and Prasad (1982), Singh and Regl (1983), Hjelmfelt et al. (1975),
Hjelmfelt (1976), Croley (1982), Croley and Foster (1984), Laguna and Giraldez (1993), Rose et al. (1983a,b)
among others applied the kinematic wave theory to model soil erosion by water from watersheds and found
reasonable agreement between theory and eld or experimental data. It is concluded that soil erosion and
sediment transport on upland watersheds are kinematic for the most part. This is also because on upland
watersheds overland ow, which is dominant, is kinematic and is primarily responsible for erosion of soil
and subsequent transport of eroded soil. Therefore, there is little surprise that upland watershed erosion is
dominantly kinematic.
Bed load transport
The transport of larger and denser sediments, such as sand and aggregates, takes place as bed load. This
is usually the case in furrows, rills and other ow concentrations which exist on intensively cropped land.
Sediment transport occurs in the form of a densely packed moving layer near the bottom of the channel. In
developing a hydrodynamic model for sediment transport in rill furrows, Prasad and Singh (1982) employed
continuum mechanics, taking into account the interaction between suspension region and bed layer. When
sediment transport is assumed to occur mainly due to the action of shear stress, the resulting equation turns
out to be the kinematic wave equation:
b
t
CKb
b
x
D
I
C
b
72
where b is the bed layer thickness, x is the distance, K is a nonlinear function of shear stress at the bed, I is the
net sediment deposited on the channel bed which is transported in the bed layer and C
b
is the concentration
of sediment in the bed.
Using measurements in the East Fork River in Wyoming, Weir (1983) showed that the wave equation for
the bed was the kinematic wave equation which adequately described the gravel wave translation:
z
t
Cc
z
x
CcS D 0 73
where z is the bed thickness above some xed datum, c is characteristic velocity of bed waves found previously
by Gradowczyk (1968), x is horizontal distance and S is friction slope. Both c and S can be functions of x and
t. The term c is associated with the net rate of gravel ow. Weir (1983) extended his analysis to predict annual
bedload transport and annual variations in riverbed cross-sectional area. Song (1983) extended the kinematic
wave theory to explain various aspects of bed form phenomena, including identication of the forces that are
related to the migration speed and the growth or decay of sand waves. He suggested that the migration speed
of sand waves was determined by the mean hydrodynamic condition that caused the bed load to be affected
by bed elevation changes.
Bed load, under certain conditions, is kinematic. Outside those conditions, it is not. For example, the
exchange of sediment back and forth between suspension zone and bed load zone does not occur. However,
the conditions for the validity of the kinematic wave theory have not yet been delineated.
River channel bars and dunes
River beds are made up of grains or cobbles and exhibit wavelike undulations. They are neither smooth
nor regular in form. The irregularity of a river bed may be either patterned or textured. Sand beds are
typically characterized by ripples and dunes except when there is high ow. A majority of river beds have
alternating deeps and shallows, called pools and rifes. These bed forms are caused by accumulation of gravel
in bars along the stream at distances equal to ve to seven times the channel width. Forming all of the bed
formsripples, dunes and gravel barsare groups of noncoherent particles which are deposited in some
characteristic manner. Langbein and Leopold (1968) recognized that ripples, dunes and bars of a river exhibit
characteristics similar to those of kinematic waves wherein these waves are composed of bed particles that
concentrate temporarily in bed form and are later transported downstream. River channel bars and dunes are
not kinematic but do exhibit kinematic wavelike properties.
Flow of beads in a ume. To establish a relation between the speed of discrete particles moved by water
and the distance between them, Langbein and Leopold (1968) derived a uxconcentration relation for this
kind of transport in a ume study. For glass spheres of 0185 inch diameter introduced at the upper end of
the ume, they found a relation
v
v
0
D 1
C
C
0
74
where v is the mean speed of beads, v
0
is the speed of a single particle on the bed of ume of owing water,
C is the mean linear concentration or density (beads per unit length, say an inch) and C
0
is the maximum
linear concentration or density. This relation shows that the mean speed decreases linearly from v
0
, when a
single bead is moving, to zero, when beads are in contact, which is the case when C/C
0
D 10. According
to Langbein and Leopold (1968) the particles do not move uniformly but form and move in groups rather
quickly. Different beads move at different speeds depending on the size of their group as they leave one group
and move ahead to overtake another. When the rate of feed at the upstream end is increased, the size of the
groups increases. This leads to decrease in the average speed and ultimately a point is reached when a group
becomes so large that a jam is formed and all motion is halted. This situation is analogous to a trafc jam
and constitutes a shock. The close spacing of particles has the effect of retarding the motion by shielding one
particle from its neighbours downstream. Taking account of Equation (74) the uxconcentration relation is
expressed as
Q
s
D Cv
0
1
C
C
0
75
Viewing the uxconcentration relation, the peak represents the maximum rate of transport. The decreasing
ux represents a kind of statistical average between two cases: (1) when the concentration of beads breaks
up or becomes more open and transport takes place at the maximum speed and (2) when the beads jam and
transport is zero. The greater spatial concentration results in greater frequency of jamming. The effect of close
spacing does not become very large until the particles are about seven diameters apart. Equation (75), based
on laboratory experiments, is clearly kinematic and therefore the ow of beads in a ume can be inferred to
be kinematic.
Transport of sand in pipes. Considering the movement of sand in pipes and umes, it is observed that sand
particles start, stop, exchange positions or readily pass one another in a complex manner. Thus, rather than
considering the movement of individual particles, it is advisable to consider bulks of particles. In this case
concentration should be interpreted in terms of the weight of particles per unit space rather than the number
of particles per unit space and includes moving sediment as well as sediment deposited temporarily at rest.
This denition is different from the one used in sediment transport where concentration is dened as the mass
of sediment per unit mass of water. Grains grouped together move less readily than those widely spaced, and
694 V. P. SINGH
one type of group may include grains lying in a ripple or dune and may thus be covered temporarily by other
grains. When the pipe is clogged, v D 0 and when sand transport is light, v D v
0
. For medium sand, Langbein
and Leopold (1968) obtained linear (or spatial) concentration as
C D
07
1
v
v
0
CC
a
v
v
0
Aw 76
where C
a
is the per cent of sand divided by 100, A is the cross-sectional area of pipe, v
0
is the mean velocity
for clear water and w is the unit weight of sediment (sand) material.
For data from Blatchs experiment (Blatch, 1906) on transport of 06 mm sand in a 1-inch pipe for a
head loss of 30 ft per 100 ft of pipe, the value of v
0
was 925 ft/s for clear water (C
a
). The concentration
was measured. In turbulent suspension C D C
p
Aw, where C
p
is the proportion of sand in the efuent per
unit volume of the sediment material (mixture). At the other extreme of the pipe being clogged, sediment
concentration is 1 Aw, where D porosity ' 03 for sand particles. Within these limits concentration
varies linearly with (1 v/v
0
) where v is the mean velocity and v
0
is the mean velocity for the same rate of
head loss for clear water. The average speed of particle movement decreases with an increase in concentration.
The ux is obtained as the product of per cent sand (volume of sand transported in ratio to the total mixture
discharge) times mean velocity times cross-sectional area of pipe:
Q
s
D C
p
vA 77
The uxconcentration relation takes the form
Q
s
D v
C
w
D
07
1
v
v
0
CC
a
v
v
0
Av 78
This relation is shown in Figure 5. According to Blatch (1906), the rising limb of the curve corresponds to
complete suspension of the transported sand. The falling limb indicates that sand is being dragged along the
bottom of the pipe. A layer of sand is formed on the bottom and transport of sand occurs in much the same
manner as in an open ume. With increasing load, there is spasmodically a blockage and transport occurs in
an average sense. As spatial concentration approaches its maximum of 063 lb/ft, transport ceases entirely.
The maximum transport occurs when concentration is at about 45% of its maximum value.
Equation (78) is kinematic, for Equations (76) and (77) are kinematic, and is based on experimental data.
Thus, transport of sand in pipes can be referred to as kinematic. This is also because ow is kinematic for
the most part.
Transport of sand in umes. For transport of sand in umes, one can invoke Equation (74) where v
0
is
the velocity of a single particle, k is linear concentration or particles per unit length and k
0
is the reciprocal
of the linear dimension of a single particle. Since the mean rate of transport T equals the product of mean
speed and mean linear concentration, we obtain
T D vk 79
where v is given by Equation (74).
For transport of sand by owing water in an open ume, the transport rate and concentration are dened
a little differently. T is in pounds of sand carried per second per foot of width and W, in terms of areal
concentration, is in pounds of sand in motion per square foot. Thus, transport is effectively the product of
two factorsmean particle velocity and mean areal concentration. Thus the uxconcentration relation is
dened as
T D v
0
W
1
W
W
0
80
0.4
0.3
0.2
0
0 0.2
Linear concentration, in pounds per foot
T
r
a
n
s
p
o
r
t
,

i
n

c
u
b
i
c

f
e
e
t

p
e
r

m
i
n
u
t
e
0.4 0.6 0.8
Head loss, 30 ft
per 100 ft
0.1
Figure 5. Fluxconcentration relation for movement of sand in a pipe (data from Blatch, 1906)
where v
0
is the velocity of particles as the concentration approaches zero (meaning clear water) and is a
function of the velocity of water, W is the areal concentration (weight of sand per square foot of channel)
and W
0
is the areal concentration when transport ceases T D 0.
There is a suggested maximum for a value of W of about 25 lb/ft
2
and W
0
may be estimated at 50 lb/ft
2
. The
quantity (1 W/W
0
) decreases as W increases, reaching zero when WD W
0
. Conceptually, W
0
corresponds
to a state represented by dunes whose height is so great that the ow in the ume is blocked. As in the case
of pipe ow, concentration is the weight of the sediment in motion and the sediment deposited temporarily
at rest. It includes sediment moving in suspension or in bed forms. In umes, moving bed forms account
for nearly all transport. An estimate of W can be made from the reported height of dunes or sand waves
on the ume bed. Taking the average depth of sand in motion as about half the dune height and the weight
volume ratio as about 100 lb/ft
3
, W in pounds per square foot numerically equals 50 times the dune height
in feet. Equation (80) shows that transport would be zero when WD 0 and when WD W
0
D 50. Transport
for a given depth and size of sediment varies with the water velocity and with areal concentration. The mean
particle speed T/W decreases linearly with W to a value of zero when WD W
0
. It should be noted that in
a sand-channel ume, water velocities are different at troughs and crests of dunes. This means that different
uxconcentration curves apply at such points. In these cases the ow is unsteady and uniform, and the
transport equation takes the form
C
t
C
Q
s
x
D 0 81
which is the kinematic wave equation. Thus, the kinematic wave theory, based on laboratory data, provides
an adequate qualitative description of the transport processes pertaining to sand in umes. The transport of
sand in umes is approximately kinematic, for the ow in umes is kinematic.
696 V. P. SINGH
Vertical mixing of coarse particles in gravel bed rivers
Vertical mixing of particles leads to uvial dispersion of sediment and is part of the evolution of the
streambed. A proportion of active particles deposited on the bed surface gets buried by the sediment deposited
later. The vertical location of particles in the streambed is controlled by a layer, called the active layer. This
is the layer of episodically mobilized material of the streambed. The thickness of this layer is determined by
the ow conditions, sediment characteristics, and bed and channel morphology, as well as interactions among
these factors (Hassan and Church, 1994). Buried particles begin moving again when all particles covering
them have been removed. The movement of particles depends on their relative location within the scouring
layer. The particles buried more deeply are less mobile than those on the surface. The particle exchange is
concentrated primarily in top layers. The movement of particles on river beds was studied by Hassan and
Church (1994). In sand-bed rivers vertical mixing is dominated by wavelike bedforms, such as dunes and
ripples, whereas in gravel-bed rivers the movement is sporadic, and the mixing results primarily from local
scour and ll.
Hassan and Church (1994) modelled vertical mixing of coarse particles in gravel-bed rivers using a
kinematic model. They examined distributions of burial depth of particles by considering observed burial
depth of individual size groups of all mobile particles and of all tagged particles (including static ones).
When a group of tracer particles was released simultaneously on the bed surface of the stream channel, the
tracer material moved and dispersed downstream from the source area and vertically from the bed surface
to the lower limit of the mobile layer. Initially the vertical distribution of the tracers in the active layer was
not uniform because of the progressively limited exposure of the bed at increasing depth. After the initial
dispersion, further movement of a particle depended on, among other factors, its vertical location in the bed.
To account for the variations of frequency of burial depth of particles, the model consisted of m layers of
constant thickness equal to the median size of the bed material. The vertical distribution of the particles after
the rst event was taken as
fy D k expky 82
where fy is the frequency of the proportion of particles in a given layer, y is the layer number and k is the
mixing parameter equal to the reciprocal of the mean scaled burial depth. It should be noted that Equation (82)
is essentially a relation between concentration and position. The concentration is expressed in terms of the
proportion of particles. The proportion of particles which changed vertical position from a given layer was
assumed constant from event to event for a series of discrete ow events of equal magnitudes. This is a
key assumption, meaning a constant ux of particles in the vertical plane. Although crude, this assumption
appears to be supported by data.
Using the same distribution for both the movement and the vertical mixing, Hassan and Church (1994)
derived the proportion of moved particles in layer y
m
, fy
m
, in the nth step as
f
n
y
m
D
m
iD0
f
n1
f
1
y 83
where the operator asterisk denotes convolution. The stationary particles, fy
s
, are considered to remain in
the same layer and their proportion may be expressed as
fy
s
D f
n1
y [1 f
1
y] 84
This calculation is based on the assumption that there is no net change in the elevation of the bed surface over
the entire area and hence in layer assignment for the stationary particles. The kinematic model was applied
to the Carnation Creek in British Columbia, Canada and the model results were consistent with the overall
distributions obtained from several events.
Thus, it can be concluded that vertical mixing of coarse particles in gravel-bed rivers approaches the
kinematic wave behaviour if there is no change in bed elevation over the entire area and the proportion of
particles changing vertical position from a given layer is constant. However, vertical mixing is complicated
by bed features and other geometric complexities resulting in nonkinematic behaviour of vertical mixing of
sediment in general.
Debris and mud ow
Rainfall or snowfall may induce debris ow. Indeed every canyon mouth in the Wasatch Front range
in northcentral Utah has an alluvial fan built by debris ows (Anderson et al., 1984). A prolonged late
snowmelt can cause slope failure. The failure can mobilize into a debris ow which can accumulate and
deposit debris and bury houses. The slope failure in a canyon may be caused by saturated conditions resulting
from heavy rains and heavy snowfall followed by heavy snowmelt. The mass of material moves rapidly
downslope, mobilizing great amounts of material such that debris ow can increase in volume 50-fold by the
time it reaches the canyon mouth.
Santi (1989) summarized mobilization (erosion), transport and depositional characteristics of debris ow.
By plotting the total volume of debris as a function of distance from the failure scar in the Lightning Canyon in
northcentral Utah, he delineated four geomorphic regions: scar, upper, middle and lower canyon areas. The
scar area (average slope 23
) is transport dominated; the upper canyon area (average slope 20
) is dominated
by mobilization of colluvium; the middle portion (average slope 16
) is a transition between a deep, debris-

conning canyon above and a wider, more gently sloping canyon below and is mobilization dominant but to
a less pronounced degree; and the lower canyon area (average slope 12
) is transport-dominant and exhibits

the rst occurrence of deposition in the form of levees.
Santis data showed two thresholds which may be used to describe debris ow behaviour. The rst threshold
divided the scar area and the upper canyon area and corresponded to a rapid increase in the rate of scour of
colluvium from the channel. The second threshold divided the middle and lower canyon areas and corresponded
to the rst instance of debris deposition. Santi (1989) identied thresholds between the mobilization (erosion),
transport and deposition characteristics of debris ows.
In the scar area, debris is not sufciently thick to cause signicant colluvium mobilization because the ow
is not well channellized and the total amount of debris is small. This area can be conjectured to be kinematic
to a large degree. In the upper canyon area, the amount of debris increases signicantly and is sufciently
concentrated by a deepening channel such that a large amount of debris is possible. This area is kinematic to a
signicant degree. In the middle canyon area the mobilization is still dominant but is less pronounced. In this
area, other waves may also be signicant. The lower canyon area shows the rst occurrence of deposition in
the form of levees. The debris is quite thick (3 m approximately) in the centre of the channel. Nevertheless the
wide, at morphology of the channel permits debris to further thin out in the banks, causing deposition. A few
metres further down, transport is entirely dominant and a few metres further down still a signicant amount
of mobilization takes place. In this area, the kinematic wave is still signicant but may not be dominant.
In summarizing the kinematics of debris ow down a canyon, Santi (1989) identied thresholds between
the mobilization (erosion), transport and deposition characteristics of debris ows. He derived empirically an
expression for the critical minimum debris thickness, d
m
, necessary to cause mobilization of the underlying
colluvium:
d
m
D
h cos S
0
tan S
f
CA
sin S
0
h 85
where h is the thickness of the colluvium, S
0
is the slope of the base, S
f
is the internal friction slope, is
the unit weight of the colluvium and C is a parameter. The values of A and S
f
are chosen for drained and
undrained conditions. Santi (1989) also derived, using the rigid plug model of Johnson (1970), a factor of
safety, FS:
FS D
A
d
d
sin S
0
86
698 V. P. SINGH
where S
0
is channel slope and d
d
is the thickness of the plug. Deposition will occur when the factor of safety
is greater than one, made possible by thin debris (small d
d
) or low channel slope (small S
0
). Deposition
usually occurs in the form of mounded levees high up on the gently sloping channel banks. Lang and Dent
(1987) summarized kinematic properties of mudows on Mt. St. Helens which erupted on May 18, 1980. The
properties were derived from application of a computer model, SMAC, which is an incompressible, linear
viscous model. By calibrating the viscosity term, the model was veried by simulating mudow processes.
In a discussion of SMAC, Bradley (1981) emphasized that mudows are non-Newtonian in nature and must
be modelled accordingly.
In debris and mud ow, the kinematic wave behaviour is weak and dynamic waves are more dominant.
Nevertheless, at certain scales the debris ow is kinematic. Indeed, the debris behaviour in mobilization,
transport and deposition can be modelled kinematically to a rst degree reasonably accurately. However,
because of the non-Newtonian nature of mud ows a more complete dynamic description is needed.
Wave phenomena near coastal structures
Physical processes near coastal structures are inuenced by the hydrodynamics of ow due to a combination
of incident and partially reected wave trains. Wave kinematics are developed as a summation of many irregu-
lar incident and reected wave velocity components. Observations show that extreme single wavesreferred
to as freak waveshave become increasingly substantial during the last decade. These waves occur in both
deep water as well as in shallow water. The waves are too high, too asymptotic and too steep. An analysis
by Klinting and Sand (1987) of some of the freak waves from a 13-h storm in September 1983 at the Gorm
eld in the Danish sector of the North Sea concluded that asymmetry, steepness, crest height and directional
composition were unique.
Hughes and Fowler (1995) estimated wave-induced kinematics, including root mean squared (RMS) velocity
parameters associated with irregular waves that have been reected by coastal structures. Wave kinematics
were developed as a summation of many regular incident and reected wave velocity components, and the
resulting RMS horizontal and vertical velocity expressions were functions of the incident wave spectrum,
water depth, location in the water column relative to the structure toe and reection coefcient and reection
phase angle associated with each wave component. For most tested conditions, good correspondence was
found between velocity measurements and estimates obtained using measured wave spectra as input.
To conclude, hydrodynamics of wave phenomena is complicated and is not kinematic but certain wave
features are when incident and reected wave velocity components are summed. There is a high degree of
scale dependence. At certain frequencies, the kinematic wave approximation is reasonably accurate but at
other frequencies a more dynamic description is needed.
APPLICATION OF KINEMATIC WAVE THEORY TO SNOW AND ICE HYDROLOGY
Movement of meltwater in deep snowpacks
A snow cover is deposited by a sequence of discrete snowstorms and is usually a layered medium. After
each snowfall event, the snow surface may be rearranged by snow drifting which tends to break down the
snow grains and repack them into a higher density wind crust. Snowpacks in temperate regions generally
experience freezethaw cycles, which lead to formation of stratigraphic features, known as ice layers. These
layers form preferentially at windcrust and have more ice content and are much less permeable than the
surrounding snow.
Before any meltwater can leave a snowpack, the snow temperature must be at the melting point, and the
snow in a vertical drain must be saturated to at least the liquid water content held by the snowpack after
free drainage. For a homogeneous snowpack the whole snowpack must be saturated to this irreducible liquid
content, before any meltwater can reach the base of the snowpack.
Gerdel (1965) discussed vertical ow in channels in the snowcover. These channels appear to form from an
accelerated melt metamorphosis in the presence of liquid. The channels consist of coarse grains and behave
as drains, conducting away a disproportionate share of water.
To consider percolation of water in a snowpack, one can rst assume the snowpack to be homogeneous.
Then the effects of ice layers and time variations can be superimposed. Wet snow in many ways is analogous
to water bearing soils, sands or glass beads. In case of snow, small ice grains are quickly eliminated by
metamorphosis and rapid growth continues until grains of diameter 1 to 2 mm are achieved (Wakahama,
1968). In mature snow there is a narrow distribution of grain sizes and grains are large enough to provide
high permeability and reduced capillary potential. Thus, water moves rapidly through a mature snowcover
with gravity forces dominating ow. This suggests applicability of the kinematic wave theory.
When water ows into a snow cover for the rst time, small grains increase capillary attraction of the
snow matrix. Once the small grains disappear, the liquid water held immobile by capillarity is released. This
storage mechanism is essentially responsible for delay of water runoff during early stages of snow melt.
Fundamental to describing the movement of snowmelt water in a snowpack is the law relating the water ux
to concentration. Colbeck and Davidson (1973) formulated this ux law as
q D kS
n
87
where for water is 547 10
6
/m-s, k is permeability, S
is effective saturation D S S
i
/1 S
i
, S is
saturation, S
i
is irreducible saturation, n is an exponent and q is volume ux. The value of n depends on
the type of snow with different stages of metamorphosis (Denoth et al., 1978): type A, surface snow after
snowfall, aged for two days of fair weather, single crystals of about 05 mm in diameter; type B, snow
several weeks old from a depth of about 1 m, after a long period of fair weather, coarse grained, 1 to 2 mm
in diameter. For type A, n D 15 and for type B, n D 37. For 17 experiments, depending upon the porosity
and grain size at the beginning of the experiment, and k and grain size, the calculated value of n was found
to vary from 14 to 46.
The one-dimensional continuity equation for meltwater ow can be written as
t
C
q
z
D 0 D S 88
Combining Equations (87) and (88), one gets
1 S
i
k
1/n

t
q
1/n
C
q
z
D 0 89
which is the kinematic wave equation. Here is porosity assumed constant.
The theory of water percolation in snow based on Equations (87) and (88) or Equation (89) was developed
by Colbeck and is reported in a series of papers (Colbeck, 1972, 1974, 1977, 1978; Colbeck and Davidson,
1973). Colbeck (1972) has shown that under normal conditions of gravity drainage in snow, the ow occurs
under the inuence of gravity with little capillary inuence. Jordan (1983a,b) applied the theory to a watershed
in British Columbia, Canada. Marsh and Woo (1984a,b, 1985) investigated the movement of meltwater in
snowpacks. Using eld observations they showed that in a naturally stratied snow cover the movement of
meltwater was complicated by the interaction of the wetting front with stratigraphic horizons. Wankiewicz
(1978ac) considered Equation (88) to incorporate the effect of pressure gradient in snow. Singh et al. (1997a)
provided a comprehensive mathematical treatment of vertical movement of snowmelt water in snow.
The question we ask is: is ow of meltwater in snowpacks kinematic? The answer to this question is
straightforward. In homogeneous snowpacks the ow is close to being kinematic. However, in stratied
snowpacks the ow is complicated by the presence of stratigraphic horizons and is not kinematic. Nevertheless,
even under these conditions, the kinematic wave theory provides a good rst-order approximation and this is
borne by eld data.
700 V. P. SINGH
Runoff from snowmelt
A snowpack can be idealized to be comprised of two layers over an impermeable boundary. The vertical
ow occurs through the unsaturated layer and ow along the boundary occurs in the boundary layer. The
vertical ow is described in the foregoing section. In the basal layer, the intrinsic permeability is much greater
than in the unsaturated upper layers. Colbeck (1978) has shown that ow in the saturated layer (horizontal)
is directly amenable to description by kinematic wave theory. Colbeck (1974) has shown that the ux is
q D 90a
where is the moisture content. Coupling Equation (90a) with the continuity equation
h
t
C
uh
x
D Ix, t 90b
results in the kinematic wave equation. Here is the effective porosity of snow, Ix, t is the net input, i.e.
the inux at the top minus the seepage into the underlying ground, h is the thickness of the saturated zone,
u is the volume ux in the saturated layer (per unit cross-sectional area) and x is the coordinate along the
base of the snowpack. The experiments conducted by Colbeck (1978) conrm the validity of the theory for
modelling saturated basal ow in a snowpack. Singh et al. (1997b) extended the Colbeck theory by including
inltration at the snowpack base.
The question arises: is runoff from a snowpack kinematic? If the meltwater is moving horizontally over an
impervious barrier, the ow is denitely kinematic and this is borne out by eld data.
Glacial motion
Glaciers are extremely sensitive to climatic variations. Even a slight climatic variation is sufcient to cause
a signicant advance or retreat of ice. This suggests that glacial advances and retreats form a record of
climatic changes of the past. This record can be interpreted by knowing how a glacier responds to a change in
climate. To that end, a glacier can be treated as an essentially one-dimensional ow system which continuously
throughout its length either receives new material by snowfall or loses material by melting or evaporation.
When there is a sudden change in the rate of accumulation (snowfall), all parts of the glacier thicken, but the
lower parts thicken unstably until the wave of ice arrives to restore stability. The thickening of lower parts
and the advance of glaciers can be very large even for only a small change in accumulation.
Glacial motion is of three types: surface waves, seasonal waves and surges. Surface waves are undulations
of the glacial surface prole which travel down glaciers at speeds (typically) of three to four times the
surface speed, which itself is of the order of 100 m/year. These are analogous to ordinary ood waves and
are dominated by the kinematic wave. Fundamental to describing the glacier motion is the relation between
glacial ow q and glacial thickness h. The glacial ow is the volume of ice per unit width passing a given
point in unit time. The glacial thickness varies in time and space and in a manner such that h/x
t
1.
Thus the surface slope sx, t, downhill in the x-direction, is given by
s D b h/x 91
where b is the slope of inclination upon which the glacier rests and ows down in the x-direction.
A one-dimensional continuity equation per unit width for glacial motion is
h
t
C
q
x
D ax, t 92
where ax, t is the rate of accumulation at the surface to the upper surface by snowfall and avalanching
measured as thickness of ice per unit time. A negative value of a implies melting or evaporation of ice
(ablation) from the upper (or lower) surface of the glacier.
If there is a unique relation between q and h at a xed x, then
c D q/h
x
93
where c is the celerity of the glacier wave. If this relation is unique then coupling it with the continuity equation
gives the kinematic wave equation. This approach was pioneered by Nye (1960, 1963a,b) and Palmer (1972),
amongst others.
Many relations have been proposed to relate glacial thickness to glacial ow. For small perturbations, Nye
(1960) took
q
1
D C
0
h
1
CD
0
s 94
C
0
D q/h
0
D
0
D q/s 95
where
q D q
0
Cq
1
, h D h
0
Ch
1
, s D s
0
Cs
1
96
Then Equation (92) becomes
h
1
t
C
q
1
x
D a
1
97
The derivatives are evaluated at the steady values h
0
, a
0
, and s
0
. The function qx, h, s was considered by
Weertman (1958). Accordingly
q D
g
A
m
h
mC1
sin
m
s 98
c
0
D
q
h
0
D m C1
q
0
h
0
D m C1u
0
' 3u
0
99
Thus
c
0
D
q
h
D x 0 x 1/2; C
0
D 1 x, 1/2 x 1 100
where A is some constant. Assuming that a is such that q is small, the kinematic wave velocity is xed at
each point by the steady state conguration.
Another relation is
q D Fh
nC1
sin
n
s CGh
mC1
sin
m
s 101
where F and G are constant. The rst term arises from differential motion and the second term arises from
sliding over the bed; n is about 3 or 4 and m is about 2. Thus
c D n C2u
d
Cm C1u
b
102
where u
d
D mean velocity due to shear motion and u
b
is mean sliding velocity. The mean forward velocity
of the ice u D u
d
Cu
b
. Thus c always lies between m C1u and n C2u or between 3u and 6u, depending
on the two forms of motion.
Seasonal waves manifest themselves as uctuations in the surface velocity which propagate down-glacier
at speeds in the range 20 to 150 times the surface speed. They can then travel at speeds of the order of
magnitude of 20 km/year. The speed can uctuate greatly but there is virtually no depth perturbation. For
surface waves, the uxconcentration relation (Fowler, 1982) can be approximated as
Q D
H
nC2
n C2
103
where H is the glacial thickness.
702 V. P. SINGH
Glacier surging is a manifestation of instability and is characterized by large-scale relaxation oscillations
of a large portion of glacier ice mass. Surging is a self-regulating mechanism due to internal glacier ow
dynamics. Meir and Post (1969) described a surge on the Tikke Glacier in the Alsek range of British Columbia,
and summarized the main characteristics of this type of motion. A surge-type glacier exhibits a quiescent phase
lasting 10 to 100 years (typically 10 to 20 years) during which the ice mass increases owing to accumulation
of ice on its upper portions (the accumulation area). The thickness of the ice increases in the reservoir area and
when it reaches a critical depth, a surge is initiated. In surging mode, the ice in the reservoir area moves very
quickly and there is substantial displacement of ice into the receiving area. This can lead to rapid advances
of the glacier and velocities up to 10 km/year have been recorded. Field data show that the rapid ow zone is
preceded by an acceleration zone, so that as a wave passes a point on the ice the velocity of the point increases
by a factor of four. The zone of rapid ow is followed by a deceleration wave moving at about the same speed
as the accelerating wave and as it passes the ice returns to its original velocity. The velocity of acceleration
and deceleration waves, called surge velocity, is about ve times the velocity behind the acceleration wave.
Since glacial surge is a self-induced instability, it can be explained by the kinematic wave theory. Indeed
this can be explained, as discussed by Palmer (1972), by considering a relation between the ow of ice and
the ice depth, which is different when ice is accelerating from when ice is decelerating. This then takes care
of the apparent hysteresis in the owconcentration relation. Undeniably there is diffusion, but its effect is
qualitatively similar to that of the two-way ux concentration relation. Diffusion can be expected to cause
sharp shock fronts to decay rapidly, and exerts a generally stabilizing inuence. As reasoned by Palmer
(1972), surges exhibit some of the features of kinematic waves.
The above ux relations point to the use of kinematic wave theory for modelling glacial ow. Nye (1960,
1963a,b) pioneered kinematic wave analysis to model the response of glaciers and ice-sheets to seasonal and
climatic changes as well as to changes in the rate of nourishment and wastage. Meier and Johnson (1962)
described the kinematic wave on Nisqually Glacier in Washington. Palmer (1972) described a kinematic
wave model of glacier surges. Lliboutry (1971) presented a comprehensive treatment of the glacier theory
and discussed the various aspects under which the kinematic wave theory would be applicable. Fowler and
coworkers (Fowler and Larson, 1978, 1980; Fowler, 1982; Fowler and Walder, 1993) presented comprehensive
mathematical treatments of glacial motion including the use of the kinematic wave theory. Hutter (1983)
presented a comprehensive discussion of theoretical glaciology including kinematic wave modelling.
The question we ask is: is glacial motion kinematic? Based on eld evidence, it appears that surface
waves and seasonal waves are not exactly kinematic but are so to a high degree. Although glacial surging
is not kinematic, the kinematic wave theory does provide a good rst-order approximation to the otherwise
complicated phenomenon of glacial surging.
River ice motion
When there occurs an ice break upon a river, complex interactions between the ow of water and ice
motion take place in both the movement of water and ice. The interactions cause rapid changes. Ferrick et al.
(1993) recognized an analogy between ice motion during dynamic breakup and the ow of trafc on crowded
roads and proposed to describe the river ice and ow processes using the kinematic wave theory. Treating the
dynamics of ice motion as kinematic, they identied four frontsbreaking front, convergence front, stoppage
front, and release front and related their speeds with ice discharge and volume per unit surface area on
either side of each front.
Different fronts move through the ice eld in response to changing conditions. The breaking front travels
downstream and separates the stationary ice ahead and the moving ice behind. The convergence front moves
with the upstream limit of an ice accumulation, demarcating it from thinner ice upstream. The stoppage front
moves upstream following the cessation of ice motion and separates the moving ice upstream from ice at rest
downstream. The release front moves upstream through the ice eld, triggering motion in response to an ice
release downstream.
The conservation of mass can be expressed for ice near each front that connects the speed of the front to
the change in ice thickness and accumulation porosity across the front. The ice discharge on either side of
the front can also be obtained. Since ice does not accumulate in any arbitrary moving control volume, there
is no timerate-of-change term to consider. By balancing the uxes, the ice continuity equation is
c u
i1
D Rc u
i2
104
where c is the front speed in the downstream direction, u
i1
is the velocity of ice moving in, u
i2
is the velocity
of ice moving out and R is the dimensionless ratio of effective control volume surface areas dened as
R D
B
2
t
i2
1 E
c2
B
1
t
i1
1 E
c1
105
where subscripts 1 and 2 correspond to entry downstream surface and exit upstream control surface,
respectively, E
c
is the porosity of ice accumulation, t
i
is the thickness of the ice sheet or accumulation,
B is the river width and Bt
i
is the area of the control surface. The sign convention yields negative speed, c,
for fronts moving upstream. The ice discharge at a point p, Qp, can be expressed as
Qp D Bpu
ip
t
ip
1 E
cp
106
Each of the fronts is now considered.
Breaking front. For the breaking front, we consider a reach of broken ice jammed upstream of intact ice
cover. A ow surge moving downstream arrives at the accumulation and ice motion is triggered. The ice
movement downstream keeps adding ice to the front of the ice pack, and the speed of the breaking front c
b
is always greater than the local ice velocity u
i2
. With appropriate substitutions the dimensionless speed of the
breaking front is obtained from Equation (104) as
c
b
u
i2
D
R
b
R
b
1
107
where R
b
is R at the breaking front, R
b
1. If the breaking front comes across an unbroken ice sheet of
thickness t
is
in place of ice rubble, t
il
D t
i2
. If the sheet has negligible open water area, the incoming sheet
porosity E
c1
D 0.
The ice discharge upstream of the breaking front, Q
b
, is obtained from Equations (105) and (107) as
Q
b
D c
b
D
b
108
where D
b
is the value of D associated with the breaking front:
D D B
2
t
i2
1 E
c2
B
1
t
i1
1 E
c1
109
Ferrick et al. (1992) analysed river ice motion near a breaking point in detail.
Convergence front. The movement of a breaking front produces ice convergence which decreases the surface
area occupied by ice. In the ice accumulation immediately upstream of the breaking front ice velocities are
lower due to high resistance. Farther upstream, the ice velocities are typically higher due to lower resistance.
This difference in velocities causes ice convergence at the upstream limit of a moving accumulation. All ice
converges to the convergence front, located at the upstream limit of the accumulation. The dimensionless
speed of the ice convergence front is obtained from Equation (104) as
c
c
u
i2
D
u
i1
/u
i2
R
c
1 R
c
110
704 V. P. SINGH
where c
c
is the speed of the convergence front, R
c
is R at the convergence front, u
i1
is the accumulation
velocity and u
i2
is the velocity of the ice immediately upstream. R
c
< 1 indicates convergence of thinner
or more porous ice from upstream of the accumulation and c
c
is positive for front moving downstream and
negative for front moving upstream.
The ice discharge upstream of the convergence front, Q
c
, is obtained from Equations (105) and (110) as
Q
c
D Q
b
D c
c
D
c
111
where D
c
is the value of D as in Equation (111).
Stoppage front. The initial stoppage occurs upstream of the breaking front. Stoppage of ice motion
progresses upstream following the arrest of a breaking front or of the moving ice somewhere behind this
front. The dimensionless speed of the stoppage front is obtained from Equation (104) as
c
s
u
i2
D
R
s
1 R
s
112
where R
s
is R at the stoppage front and R
s
1, indicating that the unit ice volume downstream of the front
is greater than that upstream. As R
s
approaches 1, the front progresses upstream very rapidly. The breaking
and stoppage fronts are opposites. A stoppage front is the limiting case of a convergence front with a velocity
ratio of zero when the speed of ice accumulation approaches zero, c
c
becomes negative and the convergence
front becomes a stoppage front. The stoppage front travels upstream through the rigid body accumulation.
The ice discharge upstream of the stoppage front, Q
s
, is obtained from Equations (105) and (112):
Q
s
D c
s
D
s
113
where D
s
is the value of D associated with the stoppage front.
Release front. When there occurs a failure of the resistance at a point, the ice motion is initiated. With
the release of ice and initial motion, the release front moves upstream. The dimensionless speed of the ice
release front is obtained from Equation (104) as
c
i
u
i1
D
1
R
i
1
114
where R
i
is R at the release front, with R
i
1. The ice release causes divergence because the water surface area
available to an ice accumulation is increased. From the point of release a breaking front travels downstream
and a release front travels upstream. Both fronts initiate ice motion, with convergence occurring upstream of
the breaking front and divergence occurring downstream of the release front.
The ice discharge downstream of a release front, Q
i
, is obtained from Equations (106) and (113):
Q
i
D C
i
D
i
115
where D
i
is the value of D associated with the ice release front.
Ferrick et al. (1993) simulated the ice and front motion through time for a reach of the Connecticut River,
with the assumption that accumulation thickness and porosity are uniform and changes in ice conditions and
motion occur only at a front. The kinematic wave model simulated results consistent with observations.
The question we ask is: is river ice motion kinematic? Based on the work pioneered by Ferrick et al. (1993),
the kinematic wave theory provides a good approximation to ice motion, provided the following assumptions
are valid: the changes in ice motion and conditions occur only at the front and these changes are small.
The accumulating thickness and porosity are uniform. Otherwise, the full dynamic wave theory would be the
logical theory for describing the river ice motion.
APPLICATION OF KINEMATIC WAVE THEORY TO WATER QUALITY HYDROLOGY
Solute transport in surface runoff
Non-point source pollution often has low strength but high volume and causes signicant environmental
degradation in both urban and rural watersheds. During certain storm events runoff from street surfaces
contributes signicantly more biochemical oxygen demand to the areas waters than the efuent from
sewage treatment plants. While point source pollutants originate from waste products, many non-point source
pollutants (nutrients, pesticides, de-icing salts, etc.) result from chemicals deliberately applied for benecial
use, as for example from debris and contaminants from streets, contaminants from open land areas, publicly
used chemicals, and dirt and contaminants washed from vehicles. In most cases, overland ow, rather than
surface ow is the primary transport of non-point source pollutants. Thus, the rate and amount of pollution
abatement are closely tied to the rate and volume of surface runoff.
Postulating the analogy between the bulk transport of conservative soluble pollutants and the convective
movement of water particles in overland ow, the pollutant transport in surface water becomes amenable to
the use of the kinematic wave theory (Brazil et al., 1979; Buchberger and Sanders, 1982; Akan, 1987). The
implication of this analogy is that the fundamental principles of overland ow may be employed to estimate
the discharge of soluble non-point source pollutants.
Pollutant transport involves four mechanisms: (1) dilution, (2) convection, (3) diffusion and dispersion,
and (4) chemical reactions. During a rainfall event the depth of ow increases in the direction of ow.
Dilution must occur as the conservative soluble pollutant mass is transported downstream into progressively
deeper ow. The effect of convection can be visualized by considering the movement of two pollutant particles
at different distances from the watershed outlet. The non-uniform convection spreads the pollutant mass as it
moves down the watershed, because the downstream particle moves faster than the upstream particle. Diffusion
occurs because the solute particles moving from the upstream boundary of the watershed are followed by
unpolluted runoff. This causes a concentration gradient which then causes a diffusive mass ux in the upstream
direction.
When pollutants are transported in shallow overland ow over a short period, the effects of dispersion and
biochemical reactions can be neglected (Nakamura, 1984). Thus the transport becomes convective and the
kinematic wave theory applies. The fundamental premise is that the velocity of a soluble particle equals the
velocity of a water particle at the same location. Thus, everywhere the pollutant particles are transported at
the local convective velocity. Hence, the pollutant mass discharge per unit width is equal to the product of the
convective particle velocity and pollutant mass per unit area. The rate is given by the pollutant concentration
and runoff per unit width.
Dening concentration C as the mass of solute M per unit volume of solution V, C D M/V, the mean
ow velocity u D q/h, q D discharge per unit width, h D depth of ow, mass per unit area w D M/A,
A D watershed area and MD total initial pollutant mass, the pollutant discharge ux is
uw D Cq 116
where q is obtained using the kinematic wave theory.
The one-dimensional continuity equation for solute transport can be written as
Ch
t
C
CQ
x
D Gx, t 117
where G is some function depending upon the injection of solute in ow. For example, for instantaneous
mixing of soluble pollutant in runoff water
Gx, t D wt 118
where t is the instantaneous unit ux of the solute.
706 V. P. SINGH
If the pollutant mixes in a nite period of time
Gx, t D C
r
i
w
t
119
where C
r
is the mass of pollutant per unit volume of rainwater (concentration in rainwater) and i is rainfall
intensity. The quantity w/t species the wash rate and depends on the depth of ow and slope of the plane.
If the plane is inltrating, Gx, t takes the form
Gx, t D Kh Ci f CC
r
i Cf 120
where f is the rate of inltration and K is a constant. Buchberger and Sanders (1982), Brazil et al. (1979),
Ingram and Woolhiser (1980), Akan (1987), Havis et al. (1992), among others, have applied the kinematic
wave theory to model solute transport by surface runoff and found that the theory-based predictions were in
good agreement with laboratory as well as eld data.
To conclude, solute transport by surface water is kinematic for the most part. This is because the surface
runoff is kinematic.
Solute transport in unsaturated ow
Solute transport closely parallels soil moisture transport in unsaturated or vadose zone. In case of solute
transport a complication arises, however, because many solutes interact strongly with soil. As a result, there
occurs a partitioning of the solute between the mobile solution phase and the immobile soil surface phase. To
that end, frequently an equilibrium sorption is used which with a general nonlinear sorption isotherm can be
expressed as
s D FC 121
where s is the sorbed phase concentration in mass sorbed per unit mass of the soil and C is the solute
concentration in mass per unit volume of soil water.
The solute mass ux, q
s
, can be expressed as
q
s
D Cq 122
With as the bulk density of the soil and the solute mass ux, the solute mass continuity is
C
t
C
s
t
C
q
s
z
D 0 123
The mass ux includes both the effects of advection and mixing. Using Equation (122) with the right side
set to zero, Equation (123) reduces to
R
C
t
Cv
C
z
D 0 124
where v D q/ is the seepage velocity, R D 1 CF
0
/ is the retardation factor and F
0
is the slope of the
sorption isotherm at concentration C. Equation (124) is the governing equation for solute transport in saturated
ow. With knowledge of v given by the kinematic wave equation for ow of water this equation can be solved
for C. Charbeneau (1984) was probably the rst to have applied the kinematic wave theory to model solute
transport in soils. Charbeneau et al. (1989) extended the theory to model multiphase solute transport in the
vadose zone. They included transient hydrologic phenomena such as evaporation and inltration and a model
for stochastic generation of rainfall. Based on laboratory experimentation they suggested that the kinematic
rainfall and oily pollutant transport model would be preferred for most applications. Weaver et al. (1994)
used GreenAmpt and kinematic wave theory to develop a screening model for non-aqueous phase liquid
transport in the vadose zone. They veried their model using results from a simple column experiment and
concluded that the model was capable of capturing the qualitative behaviour of the experimental system.
Solute transport in unsaturated zone is not kinematic under all conditions but is so under a good range
of conditions. This is borne out by both eld and experimental data. These conditions are related to the
conditions under which moisture transport is kinematic.
Solute transport in macropores
In undisturbed soils with macropores, there is only partial displacement of resident water and solutes by
incoming water and solutes. This means that the interaction between incoming water due to rain or irrigation
and the relatively immobile soil water is limited. This behaviour has important bearing on nitrate leaching,
desalinization and contamination of groundwater and streams. Because the effect of soil macropores is to
increase or even control the maximum rate at which water inltrates, contaminants and fertilizers move
more deeply into the soil prole and at an accelerated rate (German, 1988; Levy and Germann, 1988).
Solute loss from the macropore system into various solute storages due to mixing can be considered in a
manner similar to water sorbance. Germann (1990a,b) considered solute transport in macropores surrounded
by (1) water-saturated matrix and (2) water-sorbing matrix. In the rst case, the solute transport equation is
Cq
t
Cc
Cq
z
CcrwCCcswCC
s
D 0 125
where C and C
s
(kg/m
3
) are the solute concentrations in the water moving as kinematic wave and in the
storage zone prior to the passage of the wave, respectively, s (1/s) is the exchange coefcient of the solute,
(1/s) is sorbance D loss of water per unit time and per unit w (macropore water content). With q given by
the kinematic wave equation for soil water ow, Equation (125) is the governing equation for solute transport
in macropores and is the kinematic wave equation.
Germann (1987) modelled the transport of Escherichia coli in the vadose zone taking into account the
macropore ow. Germann (1988) discussed restrictions imposed by convectivedispersive approaches to
water ow and solute transport in porous media. Levy and Germann (1988) applied the theory to model
solute transport in soils, accounting for the loss of a conservative solute tracer from preferred paths during
macropore ow. Hornberger et al. (1990) conducted experiments under steady conditions on ow and solute
transport in macroporous forest soils. Hornberger et al. (1991) investigated throughow and solute transport
in an isolated sloping soil block in a forested catchment. There is general consensus that under certain
limitations the theory is an adequate approximation to simulate solute transport in macropore ow. These
limitations fortunately are not severe and solute transport in macropores can be concluded to be kinematic.
Solute transport in snowpacks
Isotopes, salts and dyes are used extensively for identifying particle ow paths in wet snow, determining
atmospheric conditions at the time of deposition of dry snow. Snowmelt water contains some or all of the
constituents normally found in surface water and ground water. Because snow cover is inuenced by ground
level air masses, it is regarded as a temporary depot of air pollutant. Furthermore, there are interactions between
air pollution and the chemical composition of a snow cover. Snowpacks receive pollutants from natural sources,
atmospheric as well as man-made sources. As a result, snowpacks contain a range of impurities. The pollutants
whether coming from precipitation or deposited later into the snow surface are accumulated over a period
of time during winter and transmitted to water channels in the spring when snowpack melts. Colbeck (1977)
noted that early in the melt season, melt freeze cycles concentrated soluble pollutants in the lower portion
of the snowpack, thus preparing the contaminants for rapid removal. These pollutants are quickly removed
at high concentrations within the rst fraction of the snowmelt runoff. Colbeck (1977) was probably the
rst to have applied the kinematic wave theory to model tracer movement. The percolation of melt water is
not a homogeneous process; meltwater often ows in preferred channels through the snowpack rather than
708 V. P. SINGH
percolating as a uniform front of water. The effect of such inhomogeneities is to cause pollutant and melt
water to penetrate the snowpack more quickly within the preferred channels, resulting in an earlier and less
abrupt start to the meltwater runoff. Hibberd (1984) extended the theory for pollutant concentrations during
snow melt and found the results of the model to be in agreement with experimental studies.
The pollutant transport through snow involves four components: movement of meltwater through snowpack,
(2) movement of pollutant in meltwater through snowpack, (3) movement of meltwater at the base of snowpack
and (4) movement of pollutant in snowmelt runoff. The movement of meltwater through snow and at the base
of a snowpack was already described in an earlier section. The movement of pollutant can be described using
the kinematic wave equation
C
t
Cu
C
z
D 0 126
where C is concentration of pollutant in snowmelt water, u is snowmelt water ow velocity and z is distance
measured positive downward from the top of the snow surface. In a similar fashion, the movement of pollutant
in melt water runoff can be expressed as
C
t
C
KS
s
e
C
x
D
1
e
[IC
I
C fC
f
C] 127
where C
I
is pollutant concentration in the lateral inow of meltwater (I), C
f
is the pollutant concentration in
the melt water entering the soil,
e
is effective porosity of the snowpack, f is the rate of inltration, S
s
is
the slope angle of the base and K is the saturated hydraulic conductivity.
Solute transport in snowpacks is kinematic for a good range of conditions, especially early in the melt
season. As noted by Colbeck (1977), snowmelt water ows in preferred channels and follows kinematic wave
theory reasonably accurately. This means that solute transport in snowpacks can be accurately described by
the theory.
Microbial transport
Germann et al. (1987) applied kinematic wave theory to establish threshold values for macropore ow
parameters critical to transport of microbial suspensions through macroporous soils. These thresholds were
used to assess the maximum depth of microbial translocation under various conditions of input rates and
durations. It was found that hardly any microbes were transported deeper than 23 m in soils under frequently
occurring natural precipitation intensities and durations. However, prolonged or intensied water input to the
soil surface was capable of efciently carrying microbes deeper than 100 m into the vadose zone. The sizes
and shapes of micro-organisms in suspension and their afnity to the soil particles need, however, to be
included in such an analysis.
Much of the work on microbial transport in soils has been based on advectiondispersion theory and there
is little work reported on microbial transport in surface water systems. In vadose zone, it can be concluded
that if the ow of water is conducive to description by kinematic wave theory, as for example in macropores
and saturated or near-saturated soils, it is reasonable to surmise that the kinematic wave theory would be
a good approximation. The theory would not be accurate under all conditions but would be under certain
conditions which are quite prevalent.
Sedimentation
Kynch (1952) developed a theory of sedimentation assuming that the fall of particles in a dispersion is
determined by the local particle density only. The theory has implications in design of biological clariers,
settling basins, naval ships, etc. For settling of a dispersion of similar particles, the velocity of any particle is
a function of the local concentration of particles, , in its immediate vicinity. Concentration is dened as the
number of particles per unit volume of dispersion. For particles of the same shape and size, it is proportional
to the volume fraction. The uxconcentration relation is
S D v 128
where is the volume concentration, S is the number of particles crossing a horizontal section per unit area
per unit time or the particle ux and v is the velocity of the fall. The concentration is assumed to be the same
everywhere across any horizontal layer. An example of a uxconcentration relation is
S D a
2
0
129
where a is constant and
0
is the maximum value of . Equation (129) states that ux increases with
concentration up to a certain maximum value and then decreases with increasing concentration.
Assuming that at any point in a dispersion the velocity of fall of a particle depends only on the local
concentration of particles, the settling process can then be determined from the continuity equation alone,
without knowing the details of the forces on the particles. Changes in particle density are propagated through
a dispersion just as sound is propagated through air. A one-dimensional continuity equation is
t

S
x
D 0 130
Combining Equations (130) and (128), one gets
t
CV
x
D 0 131
where
V D
dS
d
132
which is nothing but the kinematic wave equation. Thus this theory of sedimentation is based on the kinematic
wave theory.
Schneider (1982) extended the kinematic wave theory to investigate two-phase ow in settling basins
with walls that are inclined to the vertical. The theory predicted sedimentation processes with centred waves
which emerge from the bottom if the initial concentration was such that a simple concentration jump from
the suspension to the sediment was not possible. Although interparticle forces, that are important in the
high-concentration region near the bottom, are ignored by the theory, centred waves have been observed
in one-dimensional ow in vessels with vertical walls and good agreement has been found between the
experimental observations and the theory-predicted results.
Is sedimentation kinematic? It is surprising that Kynchs original formulation more than four decades ago
has not been followed up and therefore, experimental evidence backing the kinematic wave theory is lacking
and it is difcult to argue about its validity. However, based on the work in chemical and naval engineering,
the theory appears to hold great promise. From a conceptual viewpoint, the theory should provide a good rst
approximation to the process of settlement of particles in dispersion.
Chromatography
When a solution containing a mixture of coloured solutes is allowed to run through a vertical glass tube
lled with a suitable powdered adsorbing material, the material adsorbed in the tube appears as a series of
coloured bands, exhibiting that a partial separation of components has occurred. This series of coloured bands
is referred to as a chromatogram. The separation can be accomplished by a procedure known as development
of the chromatogram in which a suitable solvent is poured through the tube, which washes the coloured bands
down the tube at different rates. The lowest lying band moves the fastest. The chromatographic methods have
been popular methods for separation, purication and identication of small amounts of complex, naturally
occurring organic compounds.
710 V. P. SINGH
Wilson (1940) proposed a theory of chromatography which is based on two assumptions: (1) there exists
instantaneous equilibrium between the solution and adsorbant; (2) the effects of diffusion are neglected.
DeVault (1943) and Weiss (1943) extended Wilsons theory. The ux law assumed was
Q D MfC 133
q/m D fC 134
where q is the number of millimoles of solute adsorbed on m grams of adsorbant in equilibrium with a solute
whose concentration is C moles per litre, Q is adsorbed millimoles of solute per unit length (cm) of the
column and M is the amount of adsorbant per unit length of the column in grams. fC is the isotherm which
represents the adsorption of the given solute on the adsorbant. Thus the uxconcentration relation is through
the isotherm. Q is a function of time; C, as a function of time and position in the column, is the amount of
solute in solution per unit volume of solution.
A one-dimensional mass conservation equation for a single solute is
Q
V
C
C
x
D 0 135
where V is the volume of solution. Combining Equations (135) and (133), the differential equation for
chromatography is obtained:
V
MfC C
C
x
D 0 136
which is essentially the kinematic wave equation.
When the amount of solute in the solution in the column is taken into account the continuity equation
becomes
C
x
C
C
V
C
Q
V
D 0 137
where is the pore volume per unit length of the column. When D 0, Equation (137) reduces to
Equation (135). Combining equations (137) and (133), the kinematic wave equation becomes
C
x
C[ CMf
0
C]
C
V
D 0 138
Cassidy and Wood (1941) compared the Wilson theory with data obtained in laboratory and found, on
the whole, quite good agreement between experimental results and theory-based predictions. Bolt (1979)
summarized the principles of adsorption and exchange chromatography for movement of solutes in soil. He
also reported on the applicability of the kinematic wave theory, and concluded that in certain cases unsteady
ow of water was likely to exercise a far greater inuence on the movement and spreading of a solute front
than a diffusion/dispersion term as used in steady state solutions.
The question is: is chromatography kinematic? The answer is partly yes and partly no. As noted by Bolt
(1979), under certain conditions unsteady ow (convection of water) plays a far greater role in the movement
and spreading of the solute front than do diffusion and dispersion. Under those conditions, the kinematic wave
theory holds. This is borne out by experimental data. Under steady conditions, the diffusion and dispersion
processes exercise greater inuence than they do under unsteady conditions and the theory would not be a
good approximation. Fortunately, chromatography is usually unsteady and therefore the theory would be a
good rst-order approximation.
Ion exchange
When a solid is immersed in a solution, there will be an exchange of ions. The solid substance, carbonaceous,
resinous or mineral, capable of exchanging ions with a solution is dened as a zeolite. The performance of
a mass of zeolite can be evaluated if the law governing the exchange is known. Thomas (1944) described
heterogeneous ion exchange in a owing system, assuming that the ion exchange was governed by a law of
concentration product type. The process of ion exchange may be applicable to the problem of water softening.
Let Cx, t and px, t be the concentrations of the cation in the solution and in the zeolite, respectively, at
time t after the entrance of the solution into the column and at distance x from the input end of the column.
Let C
0
be the initial exchangeable cation concentration of the solution. The ion exchange can be characterized
by a law explicit in concentrations only of ions in solution and of exchangeable atoms or ions in the zeolite.
The net rate of exchange of the cation into zeolite is given by
p
t
D FC, p 139
where F is any function. Various forms of F may be expressed reecting various theories of interaction
between zeolite and solute. Equation (139) holds if the exchange processes are slow compared to any diffusion
processes in the solution or in the zeolite particles. When the zeolite rst approaches saturation, this type of
law is expected to hold. The form of F may vary with the interaction between zeolite and solute. For an
exchange between univalent ions a simple law is given by
p
t
D k
1
Ca p k
2
pC
0
C 140
where k
1
and k
2
are velocity constants associated with two opposing second-order reactions presupposed in
Equation (140), and a is the initial exchange capacity of the zeolite. The quantities p and a are measured in
milliequivalents per gram of zeolite and C and C
0
are measured in milliequivalents per millilitre of solution.
The coefcients k
1
and k
2
are estimated by tting to breakthrough curve data on a column run under known
conditions.
The continuity equation expressing the conservation of exchanging cation can be expressed as
R
C
x
C
C
t
Cm
p
t
D 0 141
where m D M/v
0
, M is overall density of zeolite as packed in the column, R D V/v
0
x D linear rate of
ow of the solution, v
0
is fractional free space in the zeolite column, x is the distance from the input end
of the column, t is the time after entrance of the solution into the column and V is the volumetric ow
of the solution. Equations (141) and (140) constitute the mathematical formulation for heterogeneous ion
exchange in a owing system. Thomas (1944) elaborated on the feasibility of application of the above theory
to experimental results. The theory can be used for description of the performance of packed ion exchange
columns used for water softening.
The question arises: is heterogeneous ion exchange kinematic? The answer to this question is difcult for
the following reason. Since the work of Thomas more than ve decades ago, no follow-up work on applying
the kinematic wave theory appears to have been reported. Consequently, experimental evidence supporting
the applicability of the theory to ion exchange is lacking. However, conceptually it appears that theory would
provide a good approximation if the ion exchange process is slow compared to the diffusion process in the
solution or zeolite particles. Clearly this is the case when the zeolite approaches saturation. Thus, at least
under certain conditions the theory holds.
SUMMARY AND CONCLUSIONS
The foregoing discussion shows that there is a full spectrum of waves exhibited in nature in general and in
hydrology in particular. However, in a wide range of hydrologic processes kinematic waves occupy a large
portion of this spectrum. In some extreme cases the spectrum may be approximately enveloped by kinematic
712 V. P. SINGH
waves and in others kinematic waves may occupy little or no part of the spectrum. These are two extreme
limits. In a signicant number of cases, however, the spectrum is overwhelmed by kinematic waves. Thus, the
answer to the question raised in the title of the paper is very denitely no, but within certain constraints and
spacetime scales the answer becomes yes. There is another point to be noted about the general physically
based descriptions. Our knowledge about the validity of these descriptions and the physical meaning and
measurability of the parameters contained in them is woefully inadequate. A close examination of these
descriptions suggests that the so-called physical descriptions are not really physical after all, for we cannot
determine their parameters beforehand and therefore a lot of tting is to be undertaken. Thus, the kinematic
wave theory is attractive for the reasons of being simple, exible and conceptually appealing.
The following conclusions are drawn from this study. (1) The kinematic wave theory is a good approxi-
mation of the full dynamic wave theory for a variety of ow phenomena. (2) The kinematic wave theory has
wide-ranging application in a variety of environmental and water science areas. (3) Under certain conditions,
many ow phenomena are indeed kinematic and must so be dealt with. (4) The potential of the kinematic
wave theory should be fully utilized in hydrologic investigations.
REFERENCES
Akan AO. 1987. Pollutant washoff by overland ow. Journal of Environmental Engineering 113(4): 811823.
Ampere AM. 1834. Essai sur la philosophie de sciences, or exposition analytique d une classication naturelle de toutes les connaissances
humaines. Bachelier: Paris.
Anderson LR, Keaton JR, Saarinen TF, Wells WG. 1984. The Utah Landslides, Debris Flows, and Floods of May and June 1983. National
Academy Press: Washington, DC.
Bates BC, Pilgrim DH. 1983. Investigation of storagedischarge relations for river reaches and runoff routing models. Institution of
Engineers, Australia, Civil Engineering Transactions CE25: 153161.
Bates BC, Townley LR. 1985. Estimation of parameters and uncertainty in a runoff routing model. National Conference Publication No. 85/2,
Institution of Engineers: Australia; 4852.
Beven K. 1979. On the generalized kinematic routing method. Water Resources Research 15(5): 12381242.
Beven K. 1981. Kinematic subsurface stormow. Water Resources Research 17(5): 14191424.
Beven K. 1982a. On subsurface stormow: an analysis of response times. Hydrological Sciences Journal 4: 505521.
Beven K. 1982b. On subsurface stormow: predictions with simple kinematic theory for saturated and unsaturated ows. Water Resources
Research 18(6): 16271633.
Beven K. 1987. Reply. Water Resources Research 23(4): 749.
Beven K. 1989. Interow. In Unsaturated Flow in Hydrologic Modelling: Theory and Practice, Morel-Seytoux HJ (ed.). Kluwer Academic:
Dordrecht.
Beven K, Germann P. 1981. Water ow in soil macropores: 2. A combined ow model. Journal of Soil Science 32: 1529.
Bird RB, Stewart WE, Lightfoot EN. 1960. Transport Phenomena. Wiley: New York.
Blatch NS. 1906. Works for the purication of the water supply of Washington, DC: discussion. Transactions, American Society of Civil
Engineers 57: 400408.
Bolt GH. 1979. Movement of solutes in soil: principles of adsorption/exchange chromatography. In Soil Chemistry, B. Physico-Chemical
Models, Bolt GH (ed.). Elsevier Science: New York; 285348.
Borah DK. 1989. Runoff simulation model for small watersheds. Transactions of the ASAE 32(3): 881886.
Borah DK, Prasad SN, Alonso CV. 1980. Kinematic wave routing incorporating shock tting. Water Resources Research 16(3): 529541.
Bradley JB. 1981. Discussion of Kinematic properties of mudows on Mt. St. Helens, T. E. Lang and J. D. Dent. Journal of Hydraulic
Engineering: ASCE, 114(12): 15381539.
Brazil LE, Sanders TG, Woolhiser DA. 1979. Kinematic parameter estimation for transport of pollutant in overland ow. In Surface and
Subsurface Hydrology, Morel-Seytoux HJ (ed.). Water Resources Publications: Fort Collins, CO; 555568.
Brooks RH, Corey AT. 1964. Hydraulic properties of porous media. Hydrology Paper 3, Colorado State University: Fort Collins, CO.
Buchberger SG, Sanders TG. 1982. A kinematic model for pollutant concentrations during the rising hydrograph. In Modelling Components
of Hydrologic Cycle, Singh VP (ed.). Water Resources Publications: Fort Collins, CO; 405427.
Burgers JM. 1948. A mathematical model illustrating the theory of turbulence. Advances in Applied Mechanics 1: 171199.
Carrier GF, Pearson CE. 1976. Partial Differential Equations: Theory and Practice. Academic Press: New York.
Cassidy HG, Wood SE. 1941. Chromatography of solutions containing a single solute. Journal of the American Chemical Society 63:
26282630.
Charbeneau RJ. 1984. Kinematic models for soil moisture and solute transport. Water Resources Research 20(6): 699706.
Charbeneau RJ, Weaver JW, Smith VJ. 1989. Kinematic modelling of multiphase solute transport in the vadose zone. EPA Report EPA/600/2-
89/035, R.S.K. Environmental Research Laboratory, US Environmental Protection Agency: Ada, OK; 1588 pp.
Chen CL. 1970. Surface irrigation using kinematic wave method. Journal of Irrigation and Drainage Division, ASCE 96(IR-1): 3948.
Colbeck SC. 1972. A theory of water percolation in snow. Journal of Glaciology 11: 369385.
Colbeck SC. 1974. Water ow through snow overlying an impermeable boundary. Water Resources Research 10(1): 119123.
Colbeck SC. 1977. Tracer movement through snow. IAHS Publications 118: 255262.
Colbeck SC. 1978. The physical aspects of water ow through snow. In Advances in Hydroscience, Vol. 11, Chow VT (ed.). Academic
Press: New York; 165206.
Colbeck SC, Davidson G. 1973. Water percolation through homogeneous snow. In Proceedings of the Banff Symposium on the Role of Snow
and Ice in Hydrology, Vol. 1. UNESCO: Geneva; 242257.
Cowins SC, Comfort III WJ. 1982. Gravity-induced density discontinuity waves in sand columns. Journal of Applied Mechanics 49:
497500.
Croley TE. 1982. Unsteady overland sedimentation. Journal of Hydrology 56: 325346.
Croley TE, Foster GR. 1984. Unsteady sedimentation in nonuniform rills. Journal of Hydrology 70: 101122.
Croley TE, Hunt B. 1981. Multiple-valued and non-convergent solutions in kinematic cascade models. Journal of Hydrology 49: 121138.
Cunge JA, Woolhiser DA. 1975. Irrigation systems. In Unsteady Flow in Open Channels, Mahmood K, Yevjevich V (eds). Water Resources
Publications: Fort Collins, CO; 522537.
Daluz Vieira JA. 1983. Conditions governing the use of approximations for the Saint-Venant equations for shallow surface water ow.
Journal of Hydrology 60: 4358.
Denoth A, Seidenbusch W, Blumthaler M, Kirchlechner P. 1978. Some experimental data on water percolation through snow. In Proceedings,
Modelling of Snow Cover Runoff , Colbeck SC, Ray M (eds). US Army Cold Regions Research and Engineering Laboratory: Hanover,
NH; 253256.
DeVault D. 1943. The theory of chromatography. Journal of American Chemical Society 65: 532540.
De Vries J. 1987. Comments on On subsurface stormow: prediction with simple kinematic theory for saturated and unsaturated ows by
K. Beven. Water Resources Research 23(4): 747748.
DiPietro L, Lafolie F. 1991. Water ow characterization and test of akinematic-wave model for macropore ow in a highly contrasted and
irregular double-porosity medium. Journal of Soil Science 42: 551563.
Dunne T. 1978. Field studies of hillslope ow processes. In Hillslope Hydrology, Kirkby MJ (ed.). Wiley: Chichester; 227293.
Eagleson PS. 1970. Dynamic Hydrology. McGraw Hill: New York.
Elliott RL, Walker WR. 1982. Field evaluation of furrow inltration and advance functions. Transactions of the ASAE 25(2): 396400.
Fangmeier DD, Ramsey MK. 1978. Intake characteristics of irrigation furrows. Transactions of the ASAE 21(4): 696700, 705.
Ferrick MJ. 1985. Analysis of river wave types. Water Resources Research 21(2): 209220.
Ferrick MG, Weyrick PB, Hunnewell ST. 1992. Analysis of river ice motion near a breaking front. Canadian Journal of Civil Engineering
19: 105116.
Ferrick MJ, Weyrick PB, Nelson DF. 1993. Kinematic model of river ice motion during dynamic break up. Nordic Hydrology 24: 111134.
Fowler AC. 1982. Waves on glaciers. Journal of Fluid Mechanics 120: 283321.
Fowler AC, Larson DA. 1978. On the ow of polythermal glaciers: 1. Model and preliminary analysis. Proceedings of the Royal Society
London, Series A 363: 217242.
Fowler AC, Larson DA. 1980. On the ow of polythermal glaciers: 2. Surface wave analysis. Proceedings of the Royal Society London,
Series A 370: 155171.
Fowler AC, Walder J. 1993. Creep closure of channels in deforming subglacial till. Proceedings of the Royal Society London, Series A 441:
1731.
Fread DL. 1985. Applicability criteria for kinematic and dynamic routing models. Laboratory of Hydrology, National Weather Service,
NOAA, US Department of Commerce: Silver Spring, MD.
Freeze RA, Harlan RL. 1969. Blueprint for a physically-based, digitally-simulated hydrologic response model. Journal of Hydrology 9:
237258.
Gerdel RW. 1965. The transmission of water through snow. Transactions, American Geophysical Union 35: 475485.
Germann P. 1987. Kinematic wave approximation to the transport of Escherichia coli in the vadose zone. Water Resources Research 23(7):
12811287.
Germann P. 1988. Approaches to rapid and far-reaching hydrologic processes in the vadose zone. Journal of Contaminant Hydrology 3:
115127.
Germann O. 1990a. Macropores and hydrologic hillslope processes. In Processes Studies in Hillslope Hydrology, Anderson MG, Burt TP
(eds). Wiley: New York.
Germann P. 1990b. Preferential ow and generation of runoff: 1. Boundary-layer theory. Water Resources Research 26(12): 30553063.
Germann P, Beven K. 1985. Kinematic wave approximation to inltration into soils with sorbing macropores. Water Resources Research
21: 990996.
Germann P, Beven K. 1986. A distribution function approach to waterow in soil macropores based on kinematic wave theory. Journal of
Hydrology 83: 173183.
Germann P, Smith MS, Thomas GW. 1987. Kinematic wave approximation to the transport of Escherichia Coli in the vadose zone. Water
Resources Research 23: 12811287.
Gradowczyk MH. 1968. Wave propagation and boundary instability in erodible-bed channels. Journal of Fluid Mechanics 33: 93112.
Hairsine SY, Paralange JY. 1987. Kinematic shock waves on curved surfaces and application to the cascade approximation. Journal of
Hartenberg R, Denavit J. 1964. Kinematic Synthesis and Mechanisms. McGraw-Hill: New York.
Hassan MA, Church M. 1994. Vertical mixing of coarse particles in gravel bed rivers: a kinematic model. Water Resources Research 30(4):
11731185.
Havis RN, Smith RE, Adrian DD. 1992. Partitioning solute transport between inltration and overland ow under rainfall. Water Resources
Research 28(10): 25692580.
Hibberd SA. 1984. A model for pollutant concentrations during snowmelt. Journal of Glaciology 30(104): 5865.
714 V. P. SINGH
Hjelmfelt AT. 1976. Modelling of soil movement across a watershed. Completion Report, Project No. A-076-MO, Missouri Water Resources
Center, University of Missouri: Columbia, MO.
Hjelmfelt AT, Piest RP, Saxton KE. 1975. Mathematical modelling of erosion on upland areas. Proceedings, XVIth Congress of IAHR, Vol. 2,
Sao Paolo: Brazil; 4047.
Hornberger GM, Beven K, Germann PF. 1990. Inferences about solute transport in macroporous forest soils from time series models.
Geoderma 46: 249262.
Hornberger GM, Germann PF, Beven KJ. 1991. Throughow and solute transport in an isolated sloping block in a forested catchment.
Journal of Hydrology 124: 8199.
Hughes SA, Fowler JE. 1995. Estimating wave-induced kinematics at sloping structures. Journal of Waterway, Port, Coastal, and Ocean
Engineering 121(4): 209215.
Hunt, B. 1982. Asymptotic solution for dam break problem. Journal of the Hydraulics Division, ASCE 108(HY1): 115126.
Hunt B. 1983. Asymptotic solution for dam break on sloping channel. Journal of Hydraulic Engineering, ASCE 109(12): 16981706.
Hunt B. 1984a. Dam break solution. Journal of Hydraulic Engineering 110(6): 675686.
Hunt B. 1984b. Perturbation solution of dam break oods. Journal of Hydraulic Engineering 110(8): 10581071.
Hunt B. 1987. A perturbation solution of the ood routing problem. Journal of Hydraulic Research 25(2): 215234.
Hutter C. 1983. Theoretical Glaciology. Kluwer Academic: Dordrecht; 582 pp.
Ingram JJ, Woolhiser DA. 1980. Chemical transfer into overland ow. In Proceedings, ASCE Symposium on Watershed Management .
American Society of Civil Engineers: New York; 4053.
Irmay SE, 1956. Extension of Darcy law to unsaturated ow through porous media. In Proceedings of the Darcy Symposium. IASHIUGG:
Dijon, France.
Iwagaki Y. 1955. Fundamental studies on the runoff analysis by characteristics. Bulletin 10, Disaster Prevention Research Institute, Kyoto
University: Kyoto, Japan; 25 pp.
Johnson AM. 1970. Physical Processes in Geology. Freeman and Cooper: San Francisco; 577 pp.
Jordan P. 1983a. Meltwater movement in a deep snowpack: 1. Field observations. Water Resources Research 19(4): 971978.
Jordan P. 1983b. Meltwater movement in a deep snowpack: 2. Simulation model. Water Resources Research 19(4): 979985.
Katapodes ND, Schamber DR. 1983. Applicability of dam-break ood wave models. Journal of Hydraulic Engineering 109(5): 702721.
Kibler DF, Woolhiser DA. 1970. The kinematic cascade as a hydrologic model. Hydrology Paper 39, Colorado State University: Fort Collins,
CO.
Kibler DF, Woolhiser DA. 1972. Mathematical properties of the kinematic cascade. Journal of Hydrology 15: 131145.
Klinting P, Sand SE. 1987. Analysis of prototype freak waves. In Proceedings, Specialty Conference on Nearshore Hydrodynamics. University
of Delaware: Newark.
Kouwen N, Li RM, Simons DB. 1981. Flow resistance in vegetated waterways. Transactions of the ASAE 24: 684690.
Kuczera G, Teo SY, Williams BJ. 1989. A generalized power law for RORB model. In Hydrology and Water Resources Symposium,
Christchurch, New Zealand. National Conference Publication No. 89/19, Institution of Engineers: Canberra, Australia; 124130.
Kynch GJ. 1952. A theory of sedimentation. Transactions, Faraday Society 48: 166176.
Laguna A, Giraldez JV. 1993. The description of soil erosion through a kinematic wave model. Journal of Hydrology 145: 6582.
Lang TE, Dent JD. 1987. Kinematic properties of mudows on Mt. St. Helens. Journal of Hydraulic Engineering 113(5): 6461539.
Langbein WB, Leopold LB, 1968. River channel bars and dunestheory of kinematic waves. Geological Survey Professional Paper 422-L,
US Geological Survey: Washington, DC; 20 pp.
Lax PD. 1972. The formation and decay of shock waves. The American Mathematical Monthly 79: 227241.
Levy BS, Germann PF. 1988. Kinematic wave approximation to solute transport along preferred ow paths in soils. Journal of Contaminant
Lighthill MJ, Whitham GB. 1955a. On kinematic waves: 1. Flood movement in long rivers. Proceedings of the Royal Society London, Series
A 229: 281316.
Lighthill MJ, Whitham GB. 1955b. On kinematic waves: 2. A theory of trafc ow on long crowded roads. Proceedings of the Royal Society
London, Series A 229: 317345.
Lliboutry LA. 1971. The glacier theory. In Advances in Hydroscience, Chow VT (ed.). Academic Press: New York; 81167.
Macagno E. 1980. Fluid mechanics of the duodenum. Annual Reviews of Fluid Mechanics 12: 139158.
Macagno E. 1991. Leonardian uid mechanics: 1. History of kinematics and II. Inceptin of modern kinematics. IIHR Monograph No. 112,
Iowa Institute of Hydraulic Research, The University of Iowa: Iowa City, IA; 128 pp.
Maheshwari BL. 1988. Evaluation of border irrigation models for southeast Australia. PhD thesis, University of Melbourne: Australia.
Maheshwari BL. 1994. Values for the exponent of the storage-discharge equation in runoff-routing models. Journal of Hydrology 163:
95106.
Marsh P, Woo MK. 1984a. Wetting front advance and freezing of meltwater within a snow cover: 1. Observations in the Canadian Arctic.
Water Resources Research 20(12): 18531864.
Marsh P, Woo MK. 1984b. Wetting front advance and freezing of meltwater within a snow cover: 2. A simulation model. Water Resources
Research 20(12): 18651874.
Marsh P, Woo MK. 1985. Meltwater movement in natural heterogeneous snow covers. Water Resources Research 21(11): 17101716.
Meier MF, Johnson JN. 1962. The kinematic wave on Nisqually Glacier, Washington. Journal of Geophysical Research 67(2): 886 (abstract).
Meier MF, Post AS. 1969. What are glacier surges? Canadian Journal of Earth Sciences 6(4 : 2): 807817.
Menendez AN, Norscini R. 1982. Spectrum of shallow water waves: an analysis. Journal of the Hydraulics Division, ASCE 108(HY1):
7593.
Morris EM, Woolhiser DA. 1980. Unsteady one dimensional ow over a plane: partial equilibrium hydrograph. Water Resources Research
16(2): 355360.
Nakamura E. 1984. Factors affecting the removal rate of street surface contaminants by overland ow. Journal of Research, Public Works
Research Institute 24: 147.
Nye JF. 1960. The response of glaciers and ice sheets to seasonal and climatic changes. Proceedings of the Royal Society London, Series A
256: 559584.
Nye JF. 1963a. On the theory of the advance and retreat of glaciers. Geophysical Journal of the Royal Astronomical Society 7: 431456.
Nye JF. 1963b. The response of a glacier to changes in the rate of nourishment and wastage. Proceedings of the Royal Society London,
Series A 275: 87112.
Palmer AC. 1972. A kinematic wave model of glacier surges. Journal of Glaciology 11(61): 6572.
Parlange JY, Hogarth W, Sander G, Rose C, Haverkamp R, Surin A, Brutsaert W. 1990. Asymptotic expansion for steady state overland
ow. Water Resources Research 26: 579583.
Pearson CP. 1989. One-dimensional ow over a plane: criteria for kinematic wave modelling. Journal of Hydrology 111: 3948.
Pilgrim DH. 1986. Estimation of large and extreme design oods. Institution of Engineers, Australia, Civil Engineering Transactions CE28:
6273.
Ponce VM, Simons DB. 1977. Shallow wave propagation in open channel ow. Journal of the Hydraulics Division, ASCE 103(HY12):
14611475.
Ponce VM, Windingland B. 1985. Kinematic shock: sensitivity analysis. Journal of Hydraulic Engineering 111(4): 600611.
Prasad SN, Singh VP. 1982. A hydrodynamic model of sediment transport in rill ows. IAHS Publications 137: 293301.
Rabinach A. 1991. The Human Motor: Energy, Fatigue and the Origins of Modernity. Basic Books: New York.
Rayej M, Wallender WW. 1987. Furrow model with specied space intervals. Journal of Irrigation and Drainage Engineering 1134:
536548.
Rayej M, Wallender WW. 1988. Time solution of kinematic-wave model with stochastic inltration. Journal of Irrigation and Drainage
Engineering 114(4): 605621.
Reddy JM, Singh VP. 1994. Modelling and error analysis of kinematic wave equations of furrow irrigations. Irrigation Science 14: 189198.
Rose CW, Williams JR, Sander GC, Barry DA. 1983a. A mathematical model of soil erosion and deposition processes: 1. Soil Science
Society of America Proceedings 47: 991995.
Rose CW, Williams JR, Sander GC, Barry DA. 1983b. A mathematical model of soil erosion and deposition processes: 2. Soil Science
Society of America Proceedings 47: 9961000.
Santi PM. 1989. The kinematics of debris ow down a canyon. Bulletin of the Association of Engineering Geologists XXVI(1): 59.
Schmid BH. 1990. On kinematic cascades: derivation of generalized shock criterion. Journal of Hydraulic Research 28(3): 331340.
Schneider W. 1982. Kinematic-wave theory of sedimentation beneath inclined walls. Journal of Fluid Mechanics 120: 323346.
Seddon JA. 1900. River hydraulics. Transactions, ASCE XLIII: 179229.
Sherman B, Singh VP. 1978. A kinematic model for surface irrigation. Water Resources Research 14(2): 357384.
Sherman B, Singh VP. 1982. A kinematic model for surface irrigation: an extension. Water Resources Research 18(3): 639667.
Singh VP. 1976. Studies on rainfall-runoff modelling: 3. Converging overland ow. WRRI Report No. 073, New Mexico Water Resources
Research Institute, New Mexico State University: Las Graces, NM; 290 pp.
Singh VP. 1980. Sediment yield equations for upland areas. In Proceedings, International Symposium on Water Resources Systems. University
of Roorkee: Roorkee, India; 11-13-7511-13-81.
Singh VP. 1983. Analytical solutions of kinematic equations for erosion on a plane: II. Rainfall of nite duration. Advances in Water
Resources 6: 8895.
Singh VP. 1988. Hydrologic Systems, Vol. 1, Rainfall-Runoff Modelling. Prentice Hall: Englewood Cliffs, NJ.
Singh VP. 1993. Quantifying the accuracy of hydrodynamic approximations for determination of ood discharges. Journal of IWRS 13(3/4):
172185.
Singh VP. 1994a. Accuracy of kinematic-wave and diffusion-wave approximations for space-independent ows. Hydrological Processes 8:
4562.
Singh VP. 1994b. Accuracy of kinematic-wave and diffusion-wave approximations for space-independent ows with lateral inow neglected
in the momentum equation. Hydrological Processes 8: 311326.
Singh VP. 1994c. Derivation of errors of kinematic-wave and diffusion-wave approximations for space-independent ows. Water Resources
Management 8: 5782.
Singh VP. 1995. Accuracy of kinematic-wave and diffusion-wave approximations for space-independent ows on inltrating surfaces with
lateral inow neglected in the momentum equation. Hydrological Processes 9: 783796.
Singh VP. 1996a. Kinematic Wave Modelling in Water Resources: Surface Water Hydrology. Wiley: New York; 1399 pp.
Singh VP. 1996b. Errors of kinematic wave and diffusion wave approximations for space-independent ows on inltrating surfaces.
Hydrological Processes 10: 955969.
Singh VP. 1996c. Dam Breach Modelling Technology. Kluwer Academic: Dordrecht.
Singh VP, Aravamuthan V. 1995a. Accuracy of kinematic wave and diffusion wave approximations for time-independent ows. Hydrological
Processes 9: 755782.
Singh VP, Aravamuthan V. 1995b. Errors of kinematic wave and diffusion wave approximations for time-independent ows. Water Resources
Management 9: 175202.
Singh VP, Aravamuthan V. 1996. Errors of kinematic wave and diffusion wave approximations for steady state overland ows. Catena 27:
209227.
Singh VP, Aravamuthan V. 1997. Accuracy of kinematic wave and diffusion wave approximations for time-independent ow with momentum
exchange included. Hydrological Processes 11: 511532.
Singh VP, Joseph ES. 1994. Kinematic-wave model for soil-moisture movement with plant-root extraction. Irrigation Science 14: 189198.
Singh VP, Prasad SN. 1982. Explicit solutions to kinematic equations for erosion on an inltrating plane. In Modelling Components of
Hydrologic Cycle, Singh VP (ed.). Water Resources Publications: Fort Collins, CO; 515538.
716 V. P. SINGH
Singh VP, Prasana M. 1999. A generalized ux law, with an application. Hydrological Processes 13: 7387.
Singh VP, Ram RS. 1983. A kinematic model for surface irrigation: verication by experimental data. Water Resources Research 19(6):
15991612.
Singh VPP, Regl RR. 1983. Analytical solutions of kinematic equations for erosion on a plane: 1. Rainfall of innite duration. Advances in
Water Resources 6: 210.
Singh VP, Woolhiser DA. 1996. A nonlinear kinematic wave model for watershed surface runoff. Journal of Hydrology 31: 221243.
Singh VP, Aravamuthan V, Joseph ES. 1994. Errors of kinematic wave and diffusion-wave approximations for time-independent ows in
inltrating channels. Irrigation Science 15: 137145.
Singh VP, Bengtsson L, Westerstrom G. 1997a. Kinematic wave modelling of vertical movement of snowmelt water through a snowpack.
Hydrological Processes 11: 149167.
Singh VP, Bengtsson L, Westerstrom G. 1997b. Kinematic wave modelling of saturated basal ow in a snowpack. Hydrological processes
11: 177187.
Sisson JB, Ferguson AH, van Genuchten MT. 1980. Simple method for predicting drainage from eld plots. Soil Science Society of America
Journal 44: 11471152.
Sklash MG, Farvolden RN. 1979. The role of groundwater in storm runoff. Journal of Hydrology 43: 4565.
Sloan PG, Moore ID. 1984. Modelling subsurface stormow on steeply sloping forested watersheds. Water Resources Research 20:
18151822.
Smith RE. 1972. Border irrigation advance and ephemeral ood waves. Journal of Irrigation and Drainage Division, ASCE 98(IR-2):
289307.
Smith RE. 1981. A kinematic model for surface mine sediment yield. Transactions of the ASAE 24: 15081514.
Smith RE. 1983. Approximate soil water movement by kinematic characteristics. Soil Science Society of America Journal 47: 38.
Smith RE, Hebbert RHB. 1983. Mathematical simulation of interdependent surface and subsurface hydrologic processes. Water Resources
Research 14(3): 533538.
Smith RE, Woolhiser DA. 1971. Overland ow on an inltrating surface. Water Resources Research 7(4): 899913.
Song CCS. 1983. Modied kinematic model: application to bed forms. Journal of Hydraulic Engineering 109(8): 11331151.
Thomas HC. 1944. Heterogeneous ion exchange in a owing system. Journal of American Chemical Society 66: 6466.
Wakahama G. 1968. Inltration of meltwater into snow cover. III. Flowing down speed of meltwater via snow cover. Low Temperature
Science, Series A 26: 8586.
Walker WR, Humpherys AE. 1983. Kinematic wave furrow irrigation model. Journal of Irrigation and Drainage Engineering Division,
ASCE 109(4): 377392.
Wallender WW, Yokokura J. 1991. Space solution of kinematic wave model by time iteration. Journal of Irrigation and Drainage Engineering
116(1): 114122.
Wankiewicz A. 1978a. A review of water movement in snow. In Proceedings, Modelling of Snow Cover Runoff , Colbeck SC, Ray M (eds).
US Army Cold Regions Research and Engineering Laboratory: Hanover, NH; 222252.
Wankiewicz A. 1978b. Water pressure in ripe snow packs. Water Resources Research 14: 593599.
Wankiewicz A. 1978c. Hydraulic characteristics of snow lysimeters. Proceedings Eastern Snow Conference 35: 105116.
Weaver JW, Charbeneau RJ, Lien BK. 1994. A screening model for nonaqueous phase liquid transport in the vadose zone using GreenAmpt
and kinematic wave theory. Water Resources Research 30(1): 92105.
Weeks WD. 1980. Using the Laurenson model: traps for young players. National Conference Publication No. 80/9, Institution of Engineers:
Canberra, Australia; 2933.
Weertmen J. 1958. Travelling waves on glaciers. IAHS Publications 47: 162168.
Weir GJ. 1983. One-dimensional bed wave movement in lowland rivers. Water Resources Research 19(3): 627631.
Weiss J. 1943. On the theory of chromatography. Journal of the Chemical Society 139: 297303.
Wigmosta MS, Vail LW, Lettenmaier DP. 1994. A distributed hydrology-vegetation model for complex terrain. Water Resources Research
30(6): 16651679.
Wilson JN. 1940. A theory of chromatography. Journal of the American Chemical Society 62: 15831591.
Wooding RA. 1965a. A hydraulic model for the catchment-stream problem: 1. Kinematic wave theory. Journal of Hydrology 3(3/4): 268282.
Wooding RA. 1965b. A hydraulic model for the catchment-stream problem: 2. Numerical solutions. Journal of Hydrology 3(3/4): 268282.
Wooding RA. 1966. A hydraulic model for the catchment-stream problem: 3. Comparisons with runoff observations. Journal of Hydrology
4: 2137.
Woolhiser DA. 1996. Search for physically based runoff model a hydrologic El Dorado? Journal of Hydraulic Engineering 122(3):
122129.
Woolhiser DA, Liggett JA. 1967. Unsteady one-dimensional ow over a plane: the rising hydrograph. Water Resources Research 3(3):
753771.
Yamada T, Kobayashi M. 1988. Kinematic wave characteristics and new equations of unsaturated inltration. Journal of Hydrology 102:
257266.

Singh, Kinematic Wave

Uploaded by

Copyright:

Available Formats

Singh, Kinematic Wave

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Singh, Kinematic Wave

Uploaded by

Copyright:

Available Formats

HYDROLOGICAL PROCESSES

Hydrol. Process. 16, 667716 (2002)

2.0 2.2 2.4

) is transport dominated; the upper canyon area (average slope 20

) is a transition between a deep, debris-

) is transport-dominant and exhibits

You might also like