SEDIMENT TRANSPORT, PART III: BED FORMS

AND ALLUVIAL ROUGHNESS

By Leo C. v a n Rijn 1
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

ABSTRACT: A method is presented that makes the classification of bed forms,

the prediction of the bed-form dimensions and the effective hydraulic rough-
ness of the bed forms feasible. The proposed relationships are based on the
analysis of reliable flume and field data. A verification analysis using about
1,500 (alternative) reliable flume and field data shows good results in predicting
the hydraulic roughness (friction factor). For field conditions, the proposed
method yields considerably better results than previously proposed methods,
which are reviewed here. The proposed method has also been used to predict
the flow depth and the total bed-material load.

INTRODUCTION

In two companion papers (see Refs. 37 and 38), bed-load and sus-
pended-load transport are investigated, resulting in new relationships
that have been extensively verified. However, in those papers the pro-
posed relationships are not really used in a predictive sense because the
applied Chezy-coefficient is based on measured variables (flow velocity,
depth and energy gradient). Even in uniform flow conditions the mor-
phological behavior of an alluvial channel is rather complicated. The fun-
damental difficulty is that the channel bed characteristics (bed forms),
and thus, hydraulic roughness, depend on flow conditions (flow veloc-
ity and depth) and sediment transport rate. These flow conditions are,
in turn, strongly dependent on the channel bed configuration and its
hydraulic roughness. Therefore, additional information is required for
morphological predictions. Firstly, the relationship between the flow
conditions and sediment transport rate should be known, while the re-
lationship between hydraulic roughness and the flow conditions should
also be available. The importance of the application of a reliable rough-
ness prediction method for morphological computations has been pointed
out very clearly by De Vries (42).
In the present analysis only the hydraulic roughness due to small-scale
bed forms is considered; other geometrical characteristics such as bread-
ing and meandering are not taken into account. Basically, a one-dimen-
sional morphological system can be described by the following 5 equa-
tions:

fluid continuity: Q = udb (1)

8M 8d bzb u\a\
fluid motion: u \- g h6 ? — = —g
5 —— (2)
8x * 8x 8s C2d
^roj. Engr., Delft Hydraulics Lab., Emmeloord, P.O. Box 152, The Nether-
lands.
Note.—Discussion open until May 1,1985. To extend the closing date one month,
a written request must be filed with the ASCE Manager of Technical and Profes-
sional Publications. The manuscript for this paper was submitted for review and
possible publication on October 25, 1982. This paper is part of the Journal of
Hydraulic Engineering, Vol. 110, No. 12, December, 1984. ©ASCE, ISSN 0733-
9429/84/0012-1733/$01.00. Paper No. 19346.
1733

J. Hydraul. Eng. 1984.110:1733-1754.

 sediment transport: Qt = F(p.,p,ps ,g,D50,CTS ,u,C,d,b) (3)
hz„ 1 8Q,
sediment continuity:
3 1 =0 (4)
H • (1 -p)b bx
alluvial roughness: C = F (p., p, p s , g, D50 ,crs,u,d) (5a)
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

known variables: jx, p, p s , g, D s o , p, <rs, Q, b (5b)

unknown variables: Qf,C,u,d,zb (5c)
in which Q = discharge; Q, = sediment transport rate; C = Chezy-coef-
ficient; u = mean flow velocity; d = mean flow depth; b = mean flow
width; zb = bed level; D50 = particle diameter; <js = geometrical standard
deviation of bed material; p, = dynamic viscosity coefficient; p = density
of fluid; ps = density of sediment; g = acceleration of gravity; p = po-
rosity factor; and t = time.
The system is completely closed if a relationship to predict the bed-
roughness, C, can be specified. In the two companion papers (37,38),
such a relationship was supposed to be unknown and, therefore, the
measured Chezy-coefficient was used to predict the sediment transport
rate, Qt.
In this analysis the attention is focused on the prediction of the bed-
form characteristics and the Chezy-coefficient as a function of the flow
variables (mean flow velocity, depth) and sediment properties (size, gra-
dation). The aim of the analysis is to propose simple functions based on
reliable flume and field data that have a good predicting ability for en-
gineering purposes.
Firstly, a classification diagram for determing the type of bed forms
in the lower and transitional flow regime is proposed. Thereafter, new
relationships for the dune height and length are derived by analyzing
flume and field data. Further, the new relationships for the grain- and
form roughness which have been proposed recently (36) are now veri-
fied and compared with the existing methods of Engelund-Hansen (9),
White et al. (43) and Brownlie (5).
Finally, the proposed method for the hydraulic roughness will be used
in predicting the total bed-material load for field data.

CHARACTERISTIC PARAMETERS

It is assumed that the dimensions of the bed forms are controlled mainly
by the bed-load transport. In one companion paper (37), it is shown that
the bed-load transport can be described by a dimensionless particle pa-
rameter, D*, and a transport stage parameter, T, as follows:
1/3

a. particle diameter, D* = D50 (s - DS (6)

in which Dm = particle diameter of bed material; s = specific density;

and v = kinematic viscosity coefficient. And

( t 4 ) 2 - (»*,cr)2
b. transport stage parameter, T = (7)
("*,cr) 2

1734

J. Hydraul. Eng. 1984.110:1733-1754.

 in which u* = (goti/C)u = bed-shear velocity related to grains; C = 18
log (12Ki,/3D90) = Chezy-coefficient related to grains; Rb = hydraulic ra-
dius related to the bed according to Vanoni-Brooks (32); D90 = particle
diameter of bed material; u = mean flow velocity; and M*,cr = critical bed-
shear velocity according to Shields (37).
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

BED-FORM CLASSIFICATION

Usually, the flow conditions in an alluvial channel are classified into

(26):

1. Lower flow regime with plane bed, ripples and dunes.

2. Transitional flow regime with washed-out dunes.
3. Upper flow regime with plane bed and anti-dunes.

In the literature, roughly two groups of classification methods are de-

scribed. Engelund (9) and Garde-Albertson (11) use the Froude number
as a classification parameter, while Liu (19) and Simons-Richardson (26)
describe the type of bed forms in terms of a suspension parameter and
a particle-related Reynolds number.
In this analysis, attention is focused on the lower and transitional flow
regimes only because these regimes are the most important for field con-
ditions. As will be shown, these regimes can be quite well-defined with-
out the use of the Froude number, since the sediment transport is not
related to the Froude number in this regime (see the previous section).
This may also be indicated by the fact that the transitional stage with
washed-out dunes is generated for a Froude number of about 0.6 in flume
conditions, and of about 0.2-0.3 in field conditions, as observed in the
Missouri River (25). Only in the upper flow regime with anti-dunes is
the Froude number of importance, since the generation of anti-dunes is
mainly governed by free-surface phenomena, as indicated by the fact
that the length of the anti-dunes is equal to the wave length of the free
surface (16,46).
Ripples and dunes have different geometrical characteristics. The rip-
ple height is much smaller than and practically independent of the flow
depth, while the ripple length may be as large as the flow depth. The
generation of ripples seems to depend mainly on the stability of the
granular bed surface under the action of turbulent velocity fluctuations.
The dune height is strongly dependent on the flow depth, with local
maximum values up to the flow depth (26), while the dune length is
much larger than the flow depth. The formation of dunes may be caused
by large-scale eddies as described by Yalin (46). Due to the presence of
large (low frequency) eddies, there will be regions at regular intervals
with decreased and increased bed-shear stresses, resulting in the local
deposition and erosion of sediment particles. For increasing stages of
flow, the particles will go into suspension according to the ratio of the
grain-shear stress and the particle-fall velocity resulting in washed-out
dunes. The washing-out process can be described quite well by a trans-
port stage parameter as defined by Eq. 7. This parameter expresses the
grain-shear velocity, H* , which is an estimate for the average shear ve-

1735

J. Hydraul. Eng. 1984.110:1733-1754.

 n^lfr

%P
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

o'b ° f iDUNESl
<*>

8 i Q,
8>" ITT-
k-o goo

IRIPPLESl IDUNES

D plane bed
(no motion)
A ripples
o dunes
® transition
e plane bed

4 6 8 10 1 2 e 102
—> particle parameter, D*
J
Q2 a 3 0.4 0.5 1.0 2.0 3.0 4.0
—»• D50 x 10"3 (m) at Te = 15°C

FIG. 1.—Diagram for Bed-Form Classification in Lower and Transitional Flow

Regime

locity at the upsloping part of the bed forms, in relation to the critical
Shield's value, «*,„. When u i » w*/Cr the sediment particles will go
directly into suspension and the bed forms will be washed out. Similar
reasoning has been done by Yalin (46). The applicability of the T-param-
eter in characterizing the generation of bed forms in the lower and tran-
sitional flow regime is also shown by Fig. 1, which shows distinct zones
for ripples, dunes and washed-out dunes. Fig. 1 is based on a large
number of reliable flume and field data (1,13,21,26,28-30,40,41,44,48).
Dune-type bed forms are present for T < 15. However, for particles smaller
than about 450 \xm, D* = 10 are ripples generated after initiation of mo-
tion, but which disappear for T > 3. The transitional flow regime with
washed-out dunes is present for 15 < T < 25. For T > 25 a flat bed flow
will be generated. Finally, it is pointed out that various classification
diagrams can be found in the literature (9,11,19,26). According to Si-
mons and Sentiirk (27), the diagrams of Liu (19) and Garde-Albertson
(11) do not give acceptable results for field conditions because relatively
few field data were used. The diagram of Engelund-Hansen (9) is mainly
based on flume data. Simons-Richardson (26) use some field data, but
only of small-scale rivers. Therefore, the writer's diagram is assumed to
be most valid for the lower and transitional flow regime in field con-
1736

J. Hydraul. Eng. 1984.110:1733-1754.

 ditions because a large number of field data with small and large flow
depths have been used.

BED-FORM DIMENSIONS
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

It is assumed that dimensions and migration of the bed forms are mainly
determined by the value of the bed-load transport. In Part I (37) the bed-
load transport was described as
qb = 8(, ub cb (8)
in which qb = bed-load transport rate/unit width; uh = velocity of bed-
load particles; cb = bed-load concentration; and 8;, = thickness of bed-
load layer. Using simple kinematic considerations and the continuity
equation, the bed-load transport rate can also be described as (46):
cjb = (1 - p)aAu d (9)
in which p = porosity factor; a = shape factor of bed forms; A = bed-
form height; and ud = migration velocity of bed forms. From Eqs. 8 and
9 it can be derived that
A Cb Uu Si, Dm
- = l-L-E ( 10 )
d (1 -p)audD50 d
Using the functional relationships as given in Part I (37), it is assumed
that

c b , ^ , ^ = F(D*,T) (11)
Urn Uj
resulting in
A /D 50
J = F[-f' D *' T ) W
Likewise, it is assumed that the ratio of the bed-form height, A, and
length, X, can be expressed by a similar functional relationship:

r F (f' D *' r ) (13)

To determine these functional relationships, a large quantity of experi-
mental data of bed-form dimensions was analyzed. In all, 84 flume ex-
periments with particle diameters in the range of 190-2,300 (Jim were
used (13,28,41,44,48). Only experiments in the lower and transitional
flow regime with dune-type bed forms were considered. Ripple data were
not selected because ripples are supposed to be independent of the flow
depth. In regard to reliable field data, which are extremely scarce, only
22 data with particle diameters in the range 490-3,600 |xm were used
from some rivers in the Netherlands: the Ijssel, Waal and Rijn (40); the
Rio Parana in Argentina (29); the Mississippi River (17); and some Jap-
anese irrigation channels (30,31). In summary, the selection criteria used
were: (1) Dune-type bed forms; (2) width-depth ratio larger than 3; (3)
flow-depth larger than 0.1 m; and (4) transport stage parameter, T, smaller

1737

J. Hydraul. Eng. 1984.110:1733-1754.

 source flow velocity flow depth particle size temperature
u (mfc) d(m) D 5 0 (urn) Te (°C>
o Guy et al 0.34-1.17 0.1 6-0.32 190 8-34
X Guy et al 0.41-0.65 0.14-0.3.4 270 8-34
A Guy et al 0.47-1.15 0.16-0.32 280 8-34
o Guy et al 0.77-0.98 0.16 330 8-34
flume D Guy et al 0.48-1.00 0.10-0.25 450 8-34
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

data ? Guy et al 0.53-1.15 0.12-0.34 930 8-34

® Williams 0.54-1.06 0.15-0.22 1350 25-28
0 Delft Hydr. Lab. 0.45-087 0.26 - 0.49 790 12-18
A Stein 0.52 -0.95 Q24 -0.31 400 20-26
d Znamenskaya 0.53 -0.80 0.11 -0.21 800
-
0 Dutch Rivers 0.85-1.55 4.4 -9.5 490-3600 5-20
field X Rio Parana ; ;• i.o 12.7 400
-
data Japanese Channels 0.53 -0.89 0.25 -0.88 1100-2300
*
m Mississippi River 1.35-1.45 6-16 350-550
-
-

10'

c5o" • y ^ ®~

10 l 0—.

IF "*•" BIT., n^-

w ^—"3y\.'^
10
f(^r-a,i (i-*--)M
i
10 15 20 25
transport stage parameter.T

FIG. 2.—Bed-Form Height

than 25. Where the water temperature was not reported, a value of 15° C
was assumed and, where the D^ of the bed material was not reported,
it was computed from the D50 and the gradation,CTS, of the bed material,
assuming the latter to be equal to 2. The influence of side-wall rough-
ness was eliminated by using the method of Vanoni-Brooks (32).
In analyzing the data, only for the Missouri River (25) was a clear
influence of the particle diameter £}# observed. As reported, a decrease
from a temperature of 25° C in early September to 5° C in December (a
decrease in D# of about 30%) caused a remarkable change in the bed
configuration, from dune-type bed forms to almost flat beds; although
the other hydraulic parameters (discharge, depth, slope and particle size)
were essentially unchanged. However, for the large majority of the data,
a clear correlation between the bed-form dimensions and the D*-param-
eter could not be detected. Therefore, in the present stage of analysis,
the D*-influence was neglected and the bed-form dimensions were re-
lated to the particle diameter, D 5u ; the flow depth, d; and the transport
stage parameter, T. The best agreement was obtained for (35)
A /D 5(
- = 0.11 — (1 - e-° sr )(25 - T). (14)
d \d
1738

J. Hydraul. Eng. 1984.110:1733-1754.

Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

5 10 15 20
> transport stage parameter,T

FIG. 3.—Bed-Form Steepness

A (Dso
- = 0.015 — (1 - e~05r)(25 - T). (15)
X V d
It is assumed that for T s 0 and T a 25, the bed surface is almost flat
(anti-dunes are not considered). Eqs. 14 and 15, as well as the error
range of a factor of 2, are shown in Figs. 2 and 3.
Both functions show maximum values for about T = 5. From Eqs. 14
and 15 an expression for the bed-form length can also be derived:
\ = 7.3d (16)
Eq. 16, based on a large number of reliable flume and field data, indi-
cates that the dune length is related only to the mean flow depth. This
has also been reported by Yalin (see Ref. 46, p. 242). Based on theoretical
analysis, Yalin derived X = 2nd, which is close to Eq. 16. Yalin also pre-
sented experimental data supporting a constant dimensionless dune
length. Physically, it means that the bed-form height is reduced for in-
creasing stages of flow, while the bed-form length remains essentially
unchanged. Note finally that Eqs. 14 and 15 are in agreement with the
functional relationships for dune characteristics as proposed by Yalin (see
Ref. 46, p. 237). To apply Eqs. 14, 15 and 16, the mean flow velocity,
flow depth and particle size must be known. The energy gradient need
not be known, a fact which may be an additional advantage because the
energy gradient may be relatively inaccurate (low flow velocities, rising
or falling stages) and because it may be difficult to determine in (iso-
lated) field conditions. Other investigators who have proposed analytical
or graphical relationships for the bed-form dimensions are Tsubaki-Shi-
nohara (31), Yalin (46,47), Ranga Raju-Soni (24), Allen (3) and Fredsee
(10). The writer (35) has compared the available methods by computing
the bed-form height for a sand bed with a particle diameter of D50 = 600
|im (D90 = 1,500 (Jim); a depth of d = 1 and 10 m; and a mean flow
velocity, u = 0.5, 1, 1, 25 and 1.5 m/s. As the methods of Tsubaki-
Sinohara, Yalin and Ranga Raju-Soni require the specification of the en-

1739

J. Hydraul. Eng. 1984.110:1733-1754.

Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

<|-o

0.7
d = 10m
0.6 D50= 600 .urn
0.5
0.4 _e_
.__.
van Rijn
Tsubaki - Shinohara
Yalin
Rangu Raju-Soni +
z
0.3 _x_
_o—
Allen
FredsuSe
t-
0.2

0.1

Q5 1.0 1.5
-¥• mean flow velocity, u (m/s)

FIG. 4.—Computed Bed-Form Heights

ergy gradient, these methods were used iteratively with an equivalent

roughness height equal to half the bed-form height. The results are shown
in Fig. 4.
As can be observed, the methods of Ranga Raju-Soni and Tsubaki-
Shinohara show an increasing trend. Yalin's method yields values which
approach A/<2 = 0.167 for large flow velocities. Only Eq. 14 and the Freds0e
method predict a decreasing dune height for increasing flow velocities
(washed-out dunes). For a small depth, d = l m , the writer's method
and the Freds0e method show good agreement, while for a large depth,
d = 10 m, the Freds0e method yields larger values. Finally, note that
the method of Ranga Raju-Soni produces remarkably small bed-form
heights (and lengths) for large depths, as present in field conditions,
probably because only flume data were used for calibration.

EQUIVALENT ROUGHNESS OF BED FORMS

The hydraulic roughness of a movable bed surface is caused by grain

roughness, kSjgrain, and by form roughness, ks,fmm. In an earlier study,
the writer (36) has shown that the equivalent or effective grain rough-
ness of a flat bed can be related to the D90 of the bed material. From
about 100 flume and field data it was derived that
= 3 D9I (17)
Similar values have been reported by Kamphuis, ks - 2.5 D^ (15); Gladki,
ks = 2.3 D80 (12); Hey, ks = 3.5 DM (14); and Mahmood, ks = 5.1 D^ (20).
1740

J. Hydraul. Eng. 1984.110:1733-1754.

 source d (m) u (m/s) D50(jjm) d (m) u (m/s) D
50 O ^ '
© Guy Gt al 0.16-0.32 0.24-0.59 190-930 \ Barton- Lin 0.14-0.26 0.41 -0.58 180
6 Znamenskaya 0.08 0.49 800 x Japanese Channels 0.11-043 0.55 - 0.73 1260-1440
p Ackers 0.17-0.20 0.35-043 180 o River Luznic 0.14-0.75 0.35-0.74 2400
o Stein 0.24-030 0.88-1.12 400 & Missouri River 4.5 1.5 240
a Laursen 0.17-0.23 0/12 -0.71 100
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

Idunesi ripples

£ » X « V n—-*•"""
^ ^KI
9
_•*'
> e
» *•< ^ *
*.1.1(-.-*) \ ^"% o
O X
. •r'coo i

_-— -
V
- - _-*- A
6 8 10 -2 2 4 6 8 !Q-1
*• bed-form steepness, Y

FIG. 5.—Equivalent Roughness of Bed Forms

In regard to the form roughness, the following functional relationship,

introduced by Yalin (see Ref. 46, p. 235) is assumed to be valid:

;,form FlA (18)

'I
In an earlier study (36), the following expression based on flume and
field data was derived:
-25+
fc,form = 1.1 A ( l - t (19)
in which i|; = A/X bed-form steepness.
Eq. 19, as well as the data, is shown in Fig. 5. Taking into account
both the grain and form roughness, it is proposed to compute the ef-
fective roughness, ks, of a movable bed surface in the lower, transitional
and upper flow regime (with exception of anti-dunes) by means of
k, = 3 D90 + 1.1 A(l - e'25") (20)
It may be noted that for a bed-form steepness, A/X, equal to zero, the
flat bed value, /cS/giain = 3 D 90 , is obtained. Finally, the Ch£zy-coefficient
can be computed by

m
'-""•(IT)
in which Rb = hydraulic radius of the bed according to Vanoni-Brooks
(32). In summary, the proposed method is as follows:

1. Compute particle diameter, D* , by Eq. 6.

2. Compute critical bed-shear velocity, «*,„ (37).
3. Compute transport stage parameter, T, by Eq. 7.
4. Compute bed-form height, A, by Eq. 14.
5. Compute bed-form length, X, by Eq. 16.
1741

J. Hydraul. Eng. 1984.110:1733-1754.

 90

plane bed
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

40 I ' ' '

0.5 1.0 1.5 2.0
*• mean flow velocity, D (m/s)

FIG. 6.—Computed Chezy-Coefficients According to van Rijn Method

6. Compute equivalent roughness, ks, by Eq. 20.

7. Compute Chezy-coefficient, C, by Eq. 21.
The input data are mean flow velocity, u; mean flow depth, d; mean
flow width, b; particle diameter of bed material, D 50 , D90; fluid arid sed-
iment density, p,p s ; and kinematic viscosity coefficient, v. As an exam-
ple, the Chezy-coefficient has been computed for various flow conditions
assuming a fluid temperature, Te, of 15° C, and a geometric standard
deviation of the bed material, crs equal to 2. The results are shown in
Fig. 6.
Independent support for the validity of the proposed method (Eqs. 20
and 21) has been produced by van Urk (39). He used an automatic
sounding system to determine the dune dimensions in some branches
of the Rhine River, a major river in the Netherlands and Europe. Various
hydraulic roughness prediction methods based on the bed-form dimen-
sions were selected from the literature. Using the measured dune di-
mensions, van Urk computed the Chezy-coefficients for the various
methods and compared those values with the Chezy-coefficients derived
from the overall hydraulic conditions (mean flow velocity, depth and
energy slope). This comparison shows that the writer's prediction method
1742

J. Hydraul. Eng. 1984.110:1733-1754.

 and that of Vanoni-Hwang (33) yield the best results for all data.
Other evidence is presented by van Urk and Klaassen (40). They com-
pared various prediction methods for hydraulic roughness using flume
data. This study also supports the validity of the present method.
Finally, a discussion of the form drag expression (Eq. 19) by Engel is
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

reported (8), which shows that Eq. 19 yields a realistic estimate of the
equivalent roughness for dunes.

VERIFICATION

To verify the proposed method, a large amount of flume and field data
were selected from a compendium of solids transport data (23). In ad-
dition, about 200 field data reported in the literature were used
(4,6,7,18,21,22,25). In all, 786 field data and 758 flume data were selected
using the following criteria: (1) Flow depth larger than 0.1 m; (2) width-
depth ratio larger than 3; and (3) particle diameter in the range of 100-
2,500 p-m. To eliminate the influence of side-wall roughness, the method
of Vanoni-Brooks (32) was used. Where the water temperature was not
reported, a value of 15° C was assumed, and where only the D50 of the
bed material was reported, the other characteristic particle diameters,
D 35 , D65, and D 90 , were computed assuming a lognormal particle size
distribution with a geometric standard deviation of <TS = 2. None of the
data were used by the writer to calibrate his proposed relationships! An
investigation into the accuracy of the "measured" Chezy-coefficient has
shown that errors of up to 20% may be expected (34,35,44,45). These
errors may be caused by faulty measuring, non-equilibrium phenomena
(rising or falling stages), and three-dimensional effects (alternate bars)
in case of a large width-depth ratio. Therefore, it may hardly be possible
to predict the Chezy-coefficient with an inaccuracy of less than 20%.
For comparison with the writer's method, the methods of Engelund-
Hansen (9) and White et al. (43) were also applied to the selected data.
These two methods gave the best results in an extensive verification study
carried out by White et al. (43). It must be stressed that Engelund and
Hansen used about 100 flume data (13) to calibrate their formulas, while
White et al. used about 1,400 flume data and 260 field data for calibra-
tion. For each method the score (as a percentage) of the predicted values
in the following error ranges were determined: CmeasUred ± 10%; Cmeasured
± 20%; and Cmeasured ± 30%. The results are shown in Table 1.
The Engelund-Hansen method for flume conditions, and the writer's
method for field conditions produce the best results. On the other hand,
the Engelund-Hansen method produces rather bad results for field con-
ditions, probably because only flume data were used for calibration. The
results of the method of White et al. are not as good as they reported
in their verification study (43). They gave their results in terms of the
ratio of the measured and computed Darcy-Weisbach friction coefficient.
According to White et al., their method scored (for about 2,500 data)
38% in the 0.8-1.25 range (=10% error range for the Chezy-coefficient)
and 89% in the 0.5-2.0 range (= 40% error range for the Chezy-coeffi-
cient). These results are better than those of this analysis, probably be-
cause data of the upper flow regime with flat bed conditions were se-
lected, for which the method of White et al. may be invalid. In the
1743

J. Hydraul. Eng. 1984.110:1733-1754.

 TABLE 1.—Comparison of Computed and
Tem-
Flow pera-
Number velocity, in Flow Particle ture, in
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

of meters per depth, diameter, in degrees

Source tests second in meters micrometers Celsius
(1) (2) (3) (4) (5) (6)
Field data:
U.S.-rivers 55 0.36-1.26 1.50-4.10 140-420 12-26
(Leopold)
U.S.-rivers 300 0.4-2.4 0.3-16.4 120-560 2-35
(Corps of Engineers)
India canals 31 0.7-1.6 1.3-3.4 90-310 10-30
Pakistan canals 142 0.35-1.3 0.7-4.3 110-290 15-35
Missouri River, U.S. 21 1.35-1.75 2.8-4.3 190-230 3-22
East-Fork River, U.S. 45 0.55-1.4 0.25-2.0 450-1,400 15-20
Elkhorn River, U.S. 43 0.5-2.0 0.4-2.2 240 2-26
Rio Grande, U.S. 79 0.45-2.4 0.3-1.5 180-450 2-29
Mountain Creek, U.S. 47 0.5-1.35 0.15-0.45 290-900 15-25
Japanese channels 23 0.55-0.95 0.10-0.75 1,260-1,440 —
786
Flume data:
Guy et al. 292 0.2-1.9 0.1-0.4 190-930 8-34
Stein 54 0.4-1.65 0.1-0.35 400 20-30
Southampton A 67 0.2-0.8 0.15-0.30 150 15-25
Southampton B 139 0.2-0.7 0.15-0.30 480 21
USWES (sand) 128 0.25-0.55 0.1-0.2 180-950 14-18
Meyer-Peter 22 0.35-0.90 0.1-0.5 400-1,500 —
Barton-Lin 29 0.2-1.1 0.1-0.3 180 15-27
Laursen 14 0.4-1.0 0.1-0.3 110 20-26
Williams 13 0.5-1.15 0.15-0.25 1,350 16-26
758
Total 1,544

verification study of White et al., only data in the lower flow regime
were used. To investigate systematic errors, the ratio of the computed
and measured Ch6zy-coefficient were plotted as a function of the trans-
port stage parameter, T. Figs. 7 and 8 show the present method for the
field data only. No systematic errors can be detected in relation to the
transport stage.
Figs. 9 and 10 show the results for the method of Engelund-Hansen
and White et al. Only field data from U.S. rivers and Pakistan canals
are presented. Both methods show serious systematic errors for these
data. On the average, the predicted values are larger than measured
values. In summary, it could be said that the Engelund-Hansen and White
et al. methods are not very good for field conditions.

PREDICTION OP FLOW DEPTH

An important objective in river engineering is the prediction of flow

depth and flow velocity in uniform flow conditions for a given dis-
1744

J. Hydraul. Eng. 1984.110:1733-1754.

 Measured Chdzy-Coeffieients

SCORES, AS PERCENTAGES, OF PREDICTED

CHEZY-COEFFICIENT IN ERROR RANGES
± 1 0 % Error ± 2 0 % Error ± 3 0 % Error
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

van Rijn Engelund-Hansen White et al. R E-H W R E-H W

(7) (8) (9) (10) (11) (12) (13) (14) (15)

29 25 32 65 34 56 80 45 78

40 20 32 68 38 56 82 50 75

48 29 45 70 38 64 83 45 70
62 24 52 84 42 80 94 66 92
62 43 0 67 62 0 100 100 24
31 22 0 94 67 100 100 100 100
47 23 28 91 42 47 100 72 70
47 29 18 80 44 40 100 58 70
29 27 72 48 91 91 89 91 91
34 34 78 73 91 95 82 100 100
43 25 33 74 47 58 89 62 79

35 30 26 60 61 44 75 70 54
38 9 16 62 35 31 83 51 51
13 52 17 31 82 40 49 86 52
48 45 57 73 73 71 79 80 78
28 50 46 51 81 79 67 91 86
22 22 27 31 36 54 36 63 100
27 27 3 31 51 31 58 62 58
50 35 28 64 50 50 92 64 78
23 46 53 53 61 76 92 69 92
34 37 33 56 65 54 71 75 66
39 31 33 65 56 56 80 68 73

charge, Q; energy gradient, S; width, b; particle diameter, D50; and gra-

dation,CTS, of bed material, and fluid temperature, Te.
The methods used to predict the flow depth are those of van Rijn,
Engelund-Hansen (9) White et al. (45) and Brownlie (5). Only the latter
method is straightforward; for the other methods, an iteration technique
is necessary, applied as follows:

1. Estimate flow depth, d; (start value = 1.5 dmeaSured)-

2. Compute flow velocity, fi;, from the discharge.
3. Compute new flow velocity, H ;+1 , according to the prediction meth-
ods.
4. Compute new flow depth, dj+i, from the discharge.
5. Compare new and old flow depths, if |dj +1 - dj] > dmeasmed/100,
then dj+2 = dj+1 + dj/2, and repeat.
In regard to the selected ata, only those of Guy et al. (13,23), Pakistan
canals (21) and U.S. rivers (23) were considered, using the following
1745

J. Hydraul. Eng. 1984.110:1733-1754.

Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

2 "* ° °10'
- > transport stage parameter, T

FIG. 7.—Ratio of Computed and Measured Chezy-Coefficients According to van

Rijn Method (Small Rivers)

Field Data (large rivers)

o U.S.A. - Rivers
(Corps, Eng.), (120-560 jjm)
o U.S.A. - Rivers
(Leopold), (140-420 jjm)

1.4
6
@ < ,
1.3 6 81,,*
e • •
, ' ®e' 4 G
e
1.2 _B 8 %
€o > ©-
» %4
I
"o
a «1.11 0 if' ®

8 1.0
E
i
« b 6
e 8
i_ *fe i
fl
1

0.9
i j2®
sijp 1 ® ®( >
< >
0.8 w—
© ® ®«
0.7 e
0
» @
® Gt
0.6
10" 6 8 10 1 ^ 6 8in2
-*• transport stage parameter, T

FIG. 8.—Ratio of Computed and Measured Chezy-Coefficients According to van

Rijn Method (Large Rivers)

criteria: (1) Guy et al., d > 0.1 m; (2) Pakistan canals, d a 1.0 m; and
U.S. rivers, d z 5.0 m. None of these data were used by the writer to
calibrate the proposed relationships.
Furthermore, only those data with a width-depth ratio larger than 10
1746

J. Hydraul. Eng. 1984.110:1733-1754.

 2.2

2.1

2.0 it
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

1.9

1.8
( X I
an.
1.7 Field Data
9 U.S.A.-Rivers (Corps)
x Pakistan Canals >\
1.6-

1.5

1.4

E 1.3 £21 ^ —
] ©X-
u A
x «
1.2 5^ # ®

1.1
<*£
XXJO 1 ® i
inrf-
9XCX*
1.0 i^px-%;
9 ^
9 9) | 6 9
l | C„«%
0.9 i J y |g -

« «#|
0.8

- , » 9 •
0.7

0.6
"%r
0.5 6
4 6 81Q1
10" <V
-> transport stage p a r a m e t e r , T

FIG. 9.—Ratio of Computed and Measured Chezy-Coefficients According to

Engelund-Hansen Method (Large Rivers)

(to reduce the influence of the side-wall roughness) were selected. No

correction method was applied. Where the water temperature, Te, and
the geometric standard deviation, <rs, of the bed material were not given,
the values Te = 15° C and o-s = 2 were used. The scores (as percentages)
of the predicted flow depths in the error ranges ±10%, ±20% and ±30%,
are given in Table 2.
In regard to the field data, the method of Brownlie is slightly better
than the writer's method and the method of White et al., while the method
of Engelund-Hansen gives the least satisfactory results. Compared with
the flume data of Guy et al., the method of White et al. does not work
well.
Finally, note that the present verification does not give an indepen-
dent check of the Brownlie method since all data used in the present
investigation were used also by Brownlie to calibrate his method. There-
fore, it merely is a check of Brownlie's calibration. Additional research
with data not used by Brownlie is necessary to verify his method.
1747

J. Hydraul. Eng. 1984.110:1733-1754.

 2.2

2.1 m
e »
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

2.0
e
1.9
Field Data
®
1.8 e US.A.-Rivers (Corps)
© © x Pakistan Canals
1.7 a

i
1.6
8
§ »
1.5
e
* *•
9 e ®
>

J
1.4
< X s
e •
1.3 X
X m
e i e
1.2
^xf ©

1.1
X e
®
<©

lA
'J ft'

1J0
• •firs' h e
8X
iy >
0.9 % ^
£kx£
0.8
1®
®
?3 >
i«95; ©

0.7
e i»
@ © e

0.6
•••

0.5
10" 6 8 10"
1n0
* *4 a6
° 81 m0 '1 *• 6 81n2

*• t r a n s p o r t stage p a r a m e t e r , T

FIG. 10.—Ratio of Computed and Measured Cheiy-Coefficients According to Method

of White et al. (Large Rivers)

TABLE 2.—Comparison of Computed and Measured Flow Depths

SCORES, AS PERCENTAGES, OF PREDICTED FLOW DEPTHS
IN ERROR-RANGES
±10% Error ±20% Error ±30% Error
Name and
number of data van Rijn •Engelund White Brownlie R E W B R E W B
(D (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
USA-rivers 240 53 34 50 60 82 57 83 88 92 72 96. 98
Pakistan
canals 139 79 49 76 82 95 82 94 95 97 87 99 99
Guy et al. 147 57 47 29 60 79 71 47 89 88 78 60 97

1748

J. Hydraul. Eng. 1984.110:1733-1754.

 PREDICTION OF SEDIMENT TRANSPORT

In the two companion papers (37,38), a method to compute the bed

and suspended load transport has been proposed. As stated in the in-
troduction, the proposed method was not used in a predictive sense
because, in addition to the mean flow velocity, flow depth and width,
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

the energy-gradient must also be known to compute the overall bed-

shear velocity. In this Section, the total load transport will be predicted
using as input data only the mean flow velocity, depth and width. This
is important for the one-dimensional mathematical modeling of river
systems with a movable bed, as pointed out by De Vries (42). Usually
only the flow discharge is known and the local flow depth and flow
velocity are predicted by solving Eqs. 1-5.
The writer's method of predicting the total bed-material load is given
in the companion papers, while the Chezy-coefficient is predicted by
Eqs. 14, 16, 20 and 21. For comparison with the writer's method, the
sediment transport theories of Engelund-Hansen (9) and Ackers-White
(2) were applied to the selected data. Basically, the sediment transport
theory of Engelund-Hansen is defined in terms of a given flow depth,
energy gradient and Chezy-coefficient. Therefore, the two former pa-
rameters were assumed to be known, while the Chezy-coefficient was
predicted as proposed by Engelund-Hansen (9). The Ackers-White sed-
iment transport theory is defined in terms of a given flow depth, flow
velocity and energy-gradient. In the present analysis the flow depth and
energy-gradient were assumed to be known while the mean flow ve-
locity was predicted by means of the method of White et al. (43).
In regard to the data, only the 266 field data collected by the Corps
of Engineers (23) from U.S. rivers were used, since it represents a wide
range of flow conditions for rather large rivers. This data set consists of
data from the Rio Grande, Mississippi, Red and Atchafalaya Rivers, which
can be considered reliable field data. The accuracy of the three methods
is given in terms of a discrepancy ratio, r, defined as the ratio of the
computed and measured total load. The scores (as percentages) of the
predicted values in three error ranges are given in Table 3. As can be
seen, the writer's method is superior to both other methods. The rela-
tively poor results of the Engelund-Hansen and Ackers-White methods
stem from the relatively strong dependence of both methods (in their
basic form) on the predicted Chezy-coefficient (or mean flow velocity).
Finally, a simplified method is given which can be used to compute the
total bed-material load when only the mean flow velocity, flow depth

TABLE 3.—Comparison of Computed and Measured Total Load Transport

Scores, as Percentages, of Predicted Total
Load in Discrepancy Ranges
Researcher 0.75 < r < 1.5 0.5 < r < 2 0.33 < r < 3
d) (2) (3) (4)
van Rijn 52 75 88
Engelund-Hansen 24 44 58
Ackers-White 22 44 62

1749

J. Hydraul. Eng. 1984.110:1733-1754.

 and the particle size are known. This simplified method is based on the
computation of the sediment transport as proposed in the two compan-
ion papers in combination with the proposed hydraulic roughness pre-
diction method (38). Using regression analysis, the computational re-
sults for various flow and sediment conditions (d = 1-20 m, u = 0.5-
2.5 m/s and D50 = 100-2,000 |xm, as = 2, Te = 15° C) were represented
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

by simple power functions

*• = 0.005 I *-*« S(-T P2)

ud \[(s-l)gD50f5/ \dj
*-=0.012
^ = 0 . 0 1 2[ ] ^ *-**
— r ( ^ -(^U ] ( D , ) - (23)
5
ud {[(s - l)gD50]°- j \d
in which qh = bed-load transport (volume)/unit width; qs = suspended
load transport (volume) /unit width; and ucr = critical mean flow velocity
based on Shield's criterion.
For particles in the range of 100-2,000 fun, the critical mean flow ve-
locity can be computed by

—-*), for 100 s D50 < 500 \xm (24)

3L>9o/

ficr = 8.5 (D 50 )°°log ( - ) , for 500 < D50 ^ 2,000 |xm (25)

in which D 50 , D90 = particle diameters of bed material (in meters). The

accuracy of Eqs. 22 and 23 is somewhat less than the original method,
as given in the companion papers (37,38). For example, the score of the
predicted total loads for the 266 data of the U.S. rivers in the discrepancy
ratio of 0.5-2 is 70%, somewhat less than that derived from the original
method, which scored 79%. However, the score is considerably better
than those of Engelund-Hansen and Ackers-White, which is 44% for both
methods. From Eqs. 22 and 23, it follows that the bed-load transport,
qh, and the suspended load transport, qs, cannot be expressed as unique
functions of the Froude number. This seems to confirm the Froude num-
ber's lack of importance for the generation of bed forms in the lower
flow regime.

CONCLUSIONS

The aim of this analysis was to determine simple functions which specify
the dimensions of the bed-forms as well as their equivalent roughness.
Such a method is not available in literature. From the results of a veri-
fication analysis, it can be concluded that the proposed method has a
rather good predictive ability for the Chezy-coefficient in the dune and
plane bed regimes.
The study led to the following conclusions:

1. The proposed method for the prediction of the Chezy-coefficient

and the total load transport is superior to the methods of Engelund-
Hansen and Ackers-White et al. in the case of field conditions.
1750

J. Hydraul. Eng. 1984.110:1733-1754.

 2. The dimensions a n d the equivalent (or effective) r o u g h n e s s of b e d
forms in the d u n e a n d plane b e d regimes can be expressed by simple
functions that are valid for particles in the range of 160-3,600 (xm a n d
flow depths of u p to 20 m, as s h o w n b y a verification analysis using
about 1,500 flume a n d field data.
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

3. The p r o p o s e d m e t h o d yields better results in predicting the flow

depths than the m e t h o d s of Ackers-White a n d Engelund-Hansen, b u t
the Brownlie m e t h o d gives the best results. However, this verification
is not an i n d e p e n d e n t check of Brownlie's m e t h o d since all verification
data w e r e also u s e d b y Brownlie for calibration.
4. The type of bed forms in the lower a n d transitional flow regime
can be expressed as a functon of a dimensionless transport stage param-
eter and a particle parameter.

ACKNOWLEDGMENT

H. N . C. Breusers and N . J. van Wijngaarden of the Delft Hydraulics

Laboratory, and M. de Vries of the Delft University of Technology, are
gratefully acknowledged for their theoretical support.

APPENDIX I.—REFERENCES

1. Ackers, P., "Experiments on Small Streams," Journal of the Hydraulics Divi-

sion, ASCE, Vol. 90, No. HY4, 1964.
2. Ackers, P., and White, W., "Sediment Transport: New Approach and Anal-
ysis," Journal of the Hydraulics Division, ASCE, Vol. 99, No. HY11, 1973.
3. Allen, J. R. L., Current Ripples, North Holland Publishing Company, Am-
sterdam, The Netherlands, 1968, p. 139.
4. Beckman, E. W., and Furness, L. W., "Flow Characteristics of Elkhorn River
near Waterloo, Nebraska," Paper 1498-B, Geol. Survey Water Supply, Wash-
ington, D.C., 1962.
5. Browlie, W. R., "Flow Depth in Sand-Bed Channels," W. M. Keck Labora-
tory of Hydraulics and Water Resources, Tech. Memo 81-3, California Insti-
tute of Pasadena, Pasadena, Calif., 1981, pp. 52-54.
6. Culbertson, J. K., and Dawdy, D. R., "A Study of Fluvial Characteristics and
Hydraulic Variables, Middle Rio Grande, New Mexico," Paper 1498-F, Geol.
Survey Water Supply, Washington, D.C., 1964.
7. Culbertson, J. K., Scott, C. H., and Bennett, J. P., "Summary of Alluvial
Channel Data from Rio Grande Conveyance Channel," Prof. Paper 562-J, Geol.
Survey, Washington, D.C., 1972.
8. Engel, P., "Discussion of Equivalent Roughness of Alluvial Bed," Journal of
the Hydraulics Division, ASCE, Vol. 109, No. 9, 1983.
9. Engelund, F., and Hansen, E., A Monograph on Sediment Transport, Technisk
Forlag, Copenhagen, Denmark, 1967.
10. Freds0e, J., "Shape and Dimensions of Stationary Dunes in Rivers," Journal
of the Hydraulics Division, ASCE, Vol. 108, No. HY8, 1982.
11. Garde, R. J., and Albertson, M. L., "Sand Waves and Regimes of Flow in
Alluvial Channels," Paper 28, IAHR-Congress, Montreal, Canada, 1959.
12. Gladki, H., "Discussion of Determination of Sand Roughness for Fixed Beds,"
Journal of Hydraulic Research, Vol. 13, No. 2, 1975.
13. Guy, H. P., Simons, D. B., and Richardson, E. V., " Summary of Alluvial
Channel Data from Flume Experiments, 1956-1961," Prof. Paper 462-1, Geo-
logical Survey, Washington, D.C., 1966.
14. Hey, R. D., "Flow Resistance in Gravel-Bed Rivers," Journal of the Hydraulics
Division, ASCE, Vol. 105, No. HY4, 1979.
1751

J. Hydraul. Eng. 1984.110:1733-1754.

 15. Kamphuis, J. W., "Determination of Sand Roughness for Fixed Beds," Jour-
nal of Hydraulic Research, Vol. 12, No. 2, 1974.
16. Kennedy, J. F., "The Mechanics of Dunes and Anti-Dunes in Erodible Bed
Channels," Journal of Fluid Mechanics, Vol. 16, No. 4, 1963.
17. Lane, E. W., and Eden, E. W., "Sand Waves in the Lower Mississippi River,"
Journal Western Soc. Civ. Eng., Vol. 45, No. 6, 1940.
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

18. Leopold, L. B., and Emmett, W. W., "Bed-Load Measurements, East Fork
River, Wyoming," Proceedings of the National Academy of Sciences, Vol. 73, No.
4, 1976.
19. Liu, H. K., "Mechanics of Sediment Ripple Formation," Journal of the Hy-
draulics Division, ASCE, Vol. 83, No. HY2, 1957.
20. Mahmood, K., "Flow in Sand Bed Channels," Water Management Technical
Report No. 11, Colorado State University, Fort Collins, Colo., 1971.
21. Mahmood, K., Tarar, R. N., Hassan, S. A., Khan, H., and Masood, T., "Se-
lected Equilibrium-State Data from Acop Canals," Report No. ERW-79-2, Civil,
Mechanical and Environmental Engrg. Dept., George Washington Univer-
sity, Washington, D.C., 1979.
22. Nordin, C. F., "Aspects of Flow Resistance and Sediment Transport, Rio
Grande near Bernalillo, New Mexico," Paper 1498-H, Geological Survey Water
Supply, Washington, D.C., 1964.
23. Peterson, A. W., and Howells, R. F., "A Compendium of Solids Transport
Data for Mobile Boundary Channels," Report No. HY-1973-ST3, Dept. of Civ.
Engrg., University of Alberta, Alberta, Canada, 1973.
24. Ranga Raju, K. G., and Soni, J. P., "Geometry of Ripples and Dunes in Al-
luvial Channels," Journal of Hydraulic Research, Vol. 14, No. 3, The Nether-
lands, 1976.
25. Shen, H. W., Mellema, W. J., and Harrison, A. S., "Temperature and Mis-
souri River Stages near Omaha," Journal of the Hydraulics Division, ASCE, Vol.
104, No. HY1, 1978.
26. Simons, D. B., and Richardson, E. V., "Resistance to Flow in Alluvial Chan-
nels," Prof. Paper 422-J, Geological Survey, Washington, D.C., 1966.
27. Simons, D. B., and Sentiirk, F., "Sediment Transport Technology," Water
Resources Publications, Fort Collins, Colo., 1977.
28. Stein, R. A., Laboratory Studies of Total Load and Apparent Bed Load,"
Journal of Geophysical Research, Vol. 70, No. 8, 1965.
29. Stuckrath, F., "Die Bewegung von Grossriffleln an der Sohle des Rio Par-
ana," Mitteilungen Franzius Institut, Heft 32, Hannover, West Germany, 1969.
30. Tsubaki, T., Kawasumi, T., and Yasutomi, T., "On the Influences of Sand
Ripples upon the Sediment Transport in Open Channels," Research Institute
for Applied Mechanics, Kyushu University, Vol. II, No. 8, Fukyoka, Japan,
1953.
31. Tsubaki, T., and Shinohara, K., "On the Characteristics of Sand Waves Formed
upon the Bed of the Open Channels and Rivers," Research Institute for Ap-
plied Mechanics, Kyushu University, Vol. VII, No. 25, Fukyoka, Japan, 1959,
pp. 30-31.
32. Vanoni, V. A., and Brooks, N. H., "Laboratory Studies of the Roughness
and Suspended Load of Alluvial Streams," Report E-68, Sedimentation Lab-
oratory, California Inst, of Tech., Pasadena, Calif., 1957.
33. Vanoni, V. A., and Hwang, L. S., "Relation between Bed Forms and Friction
in Streams," Journal of the Hydraulics Division, ASCE, Vol. 93, No. HY3, 1967.
34. van Rijn, L. C , and Klaassen, G. J., "Experience with Straight Flumes for
Movable Bed Experiments, Publication 255, Delft Hydraulics Laboratory, The
Netherlands, 1981.
35. van Rijn, L. C , "The Prediction of Bed Forms, Alluvial Roughness and Sed-
iment Transport," Report S 487-111, Delft Hydraulics Laboratory, The Neth-
erlands, 1982.
36. van Rijn, L. C , "Equivalent Roughness of Alluvial Bed," Journal of the Hy-
draulics Division, ASCE, Vol. 108, No. HY10, 1982.
37. van Rijn, L. C , "Sediment Transport, Part I: Bed Load Transport," Journal
1752

J. Hydraul. Eng. 1984.110:1733-1754.

 of Hydraulic Engineering, ASCE, Vol. 110, No. 10, 1984, pp. 1431-1456.
38. van Rijn, L. C , "Sediment Transport, Part II: Suspended Load Transport,"
Journal of Hydraulic Engineering, ASCE, Vol. 110, No. 11, 1984, pp. 1613-1641.
39. van Urk, A., "Bed Forms in Relation to Hydraulic Roughness and Unsteady
Flow in the Rhine Branches," (The Netherlands), Paper 33, Euromech Con-
ference, Mechanics of Sediment Transport, University of Istanbul, Istanbul,
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

Turkey, 1982.
40. van Urk, A., and Klaassen, G. J., "Relationships for Bed Forms and Hy-
draulic Roughness" (in Dutch), Report 61.000.04, Directorate Upper Rivers,
Rijkswaterstaat, Arnhem, Netherlands, 1982.
41. "Verification of Flume Tests and Accuracy of Flow Parameters" (in Dutch),
Note R 657-VI, Delft Hydraulics Laboratory, The Netherlands, 1979.
42. de Vries, M., "A Sensitivity Analysis Applied to Morphological Computa-
tions," Third APD-IAHR Congress, Bandung, Indonesia, 1982.
43. White, W., Paris, E., and Beltess, R., "A New General Method for Predicting
the Frictional Characteristics of Alluvial Streams," Report No. IT 187, H. R.
S. Wallingford, England, 1979.
44. Williams, P. G., "Flume Width and Water Depth Effects in Sediment Trans-
port Experiments," Prof. Paper 562-H, Geological Survey, Washington, D.C.,
1970.
45. Wijbenga, J. H. A., and Klaassen, G. J., "Changes in Bed Form Dimensions
under Unsteady Flow Conditions in a Straight Flume," Publication No. 260,
Delft Hydraulics Laboratory, The Netherlands, 1981.
46. Yalin, M. S., Mechanics of Sediment Transport, Pergamon Press, Braunschweig,
Germany, 1972.
47. Yalin, M. S., "Geometrical Properties of Sand Waves," Journal of the Hydraul-
ics Division, ASCE, Vol. 90, No. HY5, 1964.
48. Znamenskaya, N. S., "Experimental Study of the Dune Movement of Sedi-
ment," Soviet Hydrology Selected Papers, No. 3, USSR, 1963.

APPENDIX II.—NOTATION

The following symbols are used in this paper:

a = reference level (L);

b = width (L);
C = Chezy-coefficient (La5T_1);
C = Ch6zy-coefficient related to grains (L 0 ' s r _1 );
c = concentration;
D = particle diameter (L);
D* = particle parameter;
d = depth (L);
g = acceleration of gravity (LT~2);
K = equivalent roughness of Nikuradse (L);
V = porosity factor;
Q = discharge (L3 T _1 );
Qt = total load transport (L 3 r _1 );
<] = discharge per unit width (L2T_1);
<Jb = bed-load transport per unit width (LfT-1);
Is = suspended load transport per unit width (l?^1);
R> = hydraulic radius related to bed (L);
s = energy gradient;
s = specific density;
T = transport stage parameter;
Te = temperature (degrees Celsius);
1753

J. Hydraul. Eng. 1984.110:1733-1754.

 t = time (T);
u = m e a n flow velocity (LT"1);
flcr = critical m e a n flow velocity (LT"1);
M(, = velocity of bed-load particles (LT"1);
ud = migration velocity of bed forms (LT'1);
Downloaded from [Link] by Univ Of South Australia Lib on 11/18/12. Copyright ASCE. For personal use only; all rights reserved.

u* = bed-shear velocity (LT"1);

u'* = bed-shear velocity related to grains (IT -1 );
«*,cr = critical bed-shear velocity for initiation of motion (LT"1);
x = longitudinal coordinate (L);
z = vertical coordinate (L);
a = coefficient;
P = ratio of sediment and fluid diffusion coefficient;
A = bed-form height (L);
§£, = thickness of bed-load layer (L);
\ = bed-form length (L);
(j, = dynamic viscosity coefficient (MIT 1 ! 1 - 1 );
H-i, = bed-form factor;
v = kinematic viscosity coefficient (LfT"1);
p = density of fluid (ML~3);
ps = density of s e d i m e n t (ML"3);
er s = geometric standard deviation of bed material;
Tb = bed-shear stress (ML~1r~2); and
»(/ = bed-form steepness.

1754

J. Hydraul. Eng. 1984.110:1733-1754.