Soil & Water Res.

, 2, 2007 (3): 104111

On the Determination of the Stable Bed Slope of a Channel

Using Mathematical Model

Frantiek Kovk

Department of Land Use and Improvement, Faculty of Forestry and Environment,

Czech University of Life Sciences in Prague, Prague, Czech Republic

Abstract: The paper deals with an analysis of selected equations used for the determination of a stable longwise
slope calculation of torrential rivers. Irregularity of the gradient, accompanied by heavy bed-load experiencing
abrupt changes of the flow as a result of heavy rainfalls of short duration and high intensity, these are typical
features impacting the behaviour and characteristics of torrential rivers. The determination of the stable bottom
slope, when the river bed is kept unpaved but still provides resistence against harmfull effects of rapids, becomes
an essential objective of the study. Three methods are used to determine the stable slope: the first is based on
tangent tension (shear stress theory), the second observes a (critical) non-scouring cross-sectional velocity
(critical mean channel velocities), and the third applies the bottom layer velocity, (the critical bed velocities).
The mathematical hydraulic model HEC-RAS v. 3.1.3 has been used for the verification of the methods in the
Jindichovick creek case study in the Krun hory.

Keywords: stable bed slope; shear stress theory; critical mean channel velocities; critical bed velocities; HEC-RAS
model

Typically, the hydraulics of torrential rivers is the sediment deposition during the decrease of the
quite different if compared with those of lowlands. driving force becomes an unavoidable consequence.
Irregularities of the lengthwise river-bed gradient One of the basic objectives of the respective studies
and a significantly varying grain-size distribution is the determination of the stable bed slope of the
of the bed-load are specific features of such rivers channel that would resist the driving force during the
(Novk 1981). design floods. The creation of a sustainable bottom
The bed-load is usually blended and it consists slope depends not only on the sediment grain-size
of sandy, gravelly, and cobble-formed grain par- distribution, but also on the saturation with water of
ticles. Boulders can also appear in some of the the bed-load. The process is quite complex and not
mountainous creeks (Novk 1988). yet formulated in a comprehensive manner (Dvok
Sudden changes in the flow rate triggered by & Novk 1994; Gordon et al. 1996).
flash rainfall of short duration and high inten- The theoretical scope of the study aims mainly
sity usually hit solely small drainage areas; this is at three methods of the stable bed slope analysis.
also a typical feature of such channels behaviour The methods are based on the shear stress theory,
(Kovk & Kov 2005). on the critical mean channel velocities distribu-
High flow rate results in losses of the bed-load from tion, and on the critical bed velocity that is based
the channel bottom and from the river banks. Thus, on the bottom velocities.

104
 Soil & Water Res., 2, 2007 (3): 104111

The hydraulic model HEC-RAS v. 3.1.3 has a the constant in Sticklers equation related to the
been used for the method verification in the value of de (Dvok & Novk 1994)
Jindichovick creek in Krun hory. C characteristic of the sediment load (after Novk
1988)
MATERIAL AND METHODS The hydraulic model HEC-RAS has been used
to quantitatively analyse the above equations.
The choice of the methods listed below is not
accidental, only those have been applied where Structure of the model
adequate data were available. With respect to the
The mathematical hydraulic model HEC-RAS
restricted extent of the paper, the complete formula-
(Hydrologic Engineering Centers River Analysis
tion of the governing equations and their theoretical
System) came into consideration as an optimal tool
reasoning were omitted. The reader is therefore
for the solution of the sediment transport within
referred to the respective references, e.g. Novk
the river studied. The program HEC-RAS uses
(1988) or Lopez (1993). Similarly, the methods of
an integrated environment of MS Windows and
the standard (effective) grain size determination
provides an outstanding graphical users interface
are excluded from further considerations.
(GUI) in the detailed hydraulic analysis of the open
(1) Methods based on critical shear stress (after
channel flow and of the flow control measures in
Shields)
both artificial and natural channels. It works both
0.06 (m ) de in steady and unsteady conditions. The calcula-
is = (1)
R tion requires the setting up of three principal data
categories: the river-bed and control-structures
(2) Methods based on critical average channel geometry, the hydraulic losses, and the boundary
velocities (Manning-Strickler) conditions. It is with advantage to use the links to
v v2 CAD and GIS systems also in 3D display already
is = (2) built into the modelling system. The channel ca-
ks2 R4/3
pacity can be investigated by two principle means
(3) Methods based on critical bed velocities (No- of solutions for hydraulic open channels analyses,
vk 1988) flow control structures included:
(1) To solve the passage of the design flood wave
de
is = 0.0035 C2 (3) by hydraulic model based on numerical solution
R of unsteady flow process. Such technique requires
hydrographs input at upper boundary profile of
The following equation has been used for the
the reach and, similarly to the further, the detailed
conversion of kand n coefficients,
description of geometric and hydraulic river-bed
R1/6 parameters. This is rather demanding compu-
n= (4)
tationally and usually out of interest for lesser
18 log10 (12.2 R )
k streams analyses.
(2) To use the methods of steady flow hydraulics
where: in the determination of longitudinal water surface
is stable bed slope (m/m)
profiles, corresponding to N-year design floods.
m bed-load material density (kg/m3)
Indeed, such a method does not provide for solution
water density (kg/m3)
of an unsteady mode; utilities for a more detailed
de effective grain diameter (m)
R hydraulic radius (m)
formulation of flow through control structures are
vv critical mean channel velocities (m/s) sometimes preferred as an advantage.
k coefficient, bottom roughness (m) In the steady flow option the program can solve
n Mannings roughness coefficient both flow regimens: the sub- and the super-critical.
ks mean velocity coefficient of wetted perimeter, The mixed flow option interfaces mathematical
unpaved channel bed (after Strickler) formulation of both.

105
Soil & Water Res., 2, 2007 (3): 104111

Detailed analysis of the flow through control The state of the creek prior to the river
structures including various hydraulic regimens training
is also possible, making the model a suitable tool
for solving various hydraulic options. It guaran- The upper part of the trained river reaches lines
tees reliable analyses, mainly in locations where up to the channel lining by a stony pavement at
the flow control structures affect hydraulics of lower parts of the creek. Several steps of almost 2m
the open channel flow. This is also the case of the height are located in the investigated river reach.
Jindichovick creek. The river bed is simply trapezoidal, the width of
The steady flow model provides higher values the bed is 11.5 m, and the depth 0.61m. The bed
in the solution of the water surface profiles; thus suffers from devastation by extensive stream-side
its results are on the side of safety. and bottom scouring in several reaches. Several
Therefore, as already stated, the modelling sys- bottom breakdowns and outcrops appear in the
tem HEC-RAS has been used in this study. The channel created by scouring more than 3 m in
detailed program description, the users manual depth. In these locations, the bed is laid bare down
and hydraulic formulations are available from to the bedrock. The river-bed is stony, granular-
(HEC-RAS 2005). ity of carpeting 510cm, and granularity of the
boulder strata 2025 cm. These characteristics
Description of the basin were identified by a detailed field investigation
of all river reaches. Sizable sandy benches appear
The Jindichovick creek is a sinistral tributary in patches of lower flow velocity. The flood plain
of the Rotava River at its fluvial kilometer 2.0. is overgrown by a mixed forest, the spruce cover
Its torrential character is remarkable. The mean dominates. Herbal vegetation with marshland-
slope of the channel is 4%. The catchment area is and ruderal species dominating makes up the
F <35km 2, the other factors are: mean altitude sub-canopy of the plain (Kovk 2002).
H > 200 m MSL, the stream slope is rather vari-
able, the flow rates also vary significantly induc- Hydraulic structures and proposal of water
ing enormous erosion, transport- and bed-load management measures
sedimentation, stony up to boulder river bed,
fluvial lee wards and hide-outs, trout fish zone Being led along the main stream of the basin,
(Kovk 2002). most of the proposed thalweg will copy the exist-
ing centerline. Due to the considerable lengthwise
Hydrologic ID number: 1-13-01-114 slope, the channel has only a little tendency to
Total catchment area: 5.964 km 2 meander. Therefore, only few corrections will be
Catchment area to US section of river training: necessary to apply in some river reaches where
1.33 km 2 the irregular trace of the bed will be replaced by
Forest coverage: 47% free-curvature bedding. The present hollow-ranges
Length of watershed: 1.62 km of breakdowns will be utilized for the creation of
Length of watershed divide: 4.35 km pools. The stabilizing of the pools will be effected
Watershed shape factor: A = 0.653 using transversal structures of stony riprap and
Type of the watershed: Fan-shaped, stone packing (rockfill). The modified bed will
hydrographic network not developed keep the trapezoidal shape, width in bed 1 m,
Torrential coefficient: KB = 0.118 cross sectional slope 1:1.5 up to the bank lines,
Table 1 lists the N-year discharges. stretches with pools excluded. The river bed at

Table 1. Design discharges

N (years) 1 2 5 10 20 50 100

Q (m3/s) 0.9 1.2 2.2 2.9 3.7 5.4 6.9

106
 Soil & Water Res., 2, 2007 (3): 104111

the foot and banks up to the height of 0.250.30 Hydrologic input data
m will be protected by stone packing (stony rock-
fill). Rest of the banks above the stony pavement The above-mentioned data set-ups were com-
will be sown. puted for 2 flow rates of Q 100 = 6.9 m 3 /s and
In effect to considerable gradient in design, Q 5=2.2m 3/s.
the vertical alignment will take into account a
number of structures for the slope control of the Geometric input data of the river bed
channel-bed. These will be constructed as wooden and of the flow control structures
sills (shoals), stony weirs (steps) and rock chutes
(overruns). Constructional modifications of the All river reaches including hydraulic structures
structures will provide for fish migration in both on the trained stream were calculated as the re-
directions. This means that the constructional sults of the detailed survey of longitudinal- and
height should not exceed 0.4 m at any cross sec- transverse profiles. The river profiles are notably
tion and the cross section of the overrun plane impacted by these measures.
should provide for continuous water jet. Natural
materials will be exclusively used for all hydraulic Hydraulic input data of the river
measures and their form will cope with the natural
formation of the torrent. The principle hydraulic characteristics of the
model are as follows: the roughness coefficient n
Computation scenarios devised by Manning, and coefficient k according
to Strickler. With respect to the original soil of the
The evaluation of the channel capacity, the flow river bed, to gravel materials, and to building ma-
velocity, and the proposed stable slope of the riv- terial of the structures, the values of the hydraulic
er-bed covers 3 choices of the river reaches, all parameters were selected according to HEC-RAS
characterized by the effective grain size of d e = users manual, determined by the methods of soil
0.06 m (d 6), 0.10 m (d 10), and 0.16 m (d 16). mechanics or from the field survey individually for

Figure 1. Axonometry of the selected reach of the channel, flooded by Q100

107
Soil & Water Res., 2, 2007 (3): 104111

RS = 0.1591 * velocity distribution

WS Q100

WS Q5

1 m/s
Elevation (m)

2 m/s

3 m/s

Ground

Bank Sta

Distance (m)

Figure 2. Velocity distribution in cross section, flooded by Q5 and Q100

each cross sectional profile. Granulometric analy- torrential bed-load of granite geological forma-
sis was applied to determine the effective grain tions. Sediment particles are rounded, the shapes
sized e . The C coefficient after Novak came in for of sphere and ellipsoid dominate.
the torrential rivers Rotava and Rolava (Novk Table 2 shows well-arranged hydraulic charac-
1988). The sediment mixture matches the blended teristics of the considered river.

Shear Chan Q100

Shear LOB, Shear Chan, Shear ROB (N/m2)

Shear Chan Q5

Shear LOB Q100

Shear ROB Q100

Shear ROB Q5

Shear LOB Q5

Distance (m)

Figure 3. Shear stress distribution in lengthwise profile, flooded by Q5 and Q100

108
 Soil & Water Res., 2, 2007 (3): 104111

Table 2. Hydraulic characteristics

Reach No. de a n k m C R

1 0.06 18.29 tables tables 2650 1000 5.58 calculated

2 0.10 16.78 tables tables 2650 1000 5.58 calculated

3 0.16 15.31 tables tables 2650 1000 5.58 calculated

RESULTS OF CALCULATIONS Novak indicates somewhat higher values of the

slope stability. On the other hand, it reflects the
Steady-state calculations under non-uniform features and the origin of the bed-load motion. The
flow conditions were performed for three selected difference between the extreme results amounts
river-reaches, all characterized by the effective to 10% (Figures 4 and 5). Such an agreement is
grain sized e . The discharge, mean flow veloc- influenced both by the selection of the equation
ity, and geometric characteristics of the cross and, primarily, by detailed analysis of the sediment
sectional profiles were identified by virtue of the load granularity and by a careful evaluation of the
model HEC-RAS. (Figures13) To specify realistic roughness coefficients. Such procedure is usu-
values of the roughness coefficients, the pictorial ally laborious and thus ineffective in engineering
description from www.hec.usace.army.mil/n-val- practices. Consequently, for the torrent training
ues.index was used. it is hardly applicable.
On the basis of these data the individual equa- Due to the complexity of the processes involved
tions have been verified and the results summarized in the water-induced sediment load, the problems
into the following pictures and graphs (Figures of water saturation by the bed-load were totally
4 and 5). omitted in this analysis. Until today, the process
has not yet been completely examined and thus
CONCLUSIONS its use in field is not easily applicable. At a high
rate of saturation, part of energy entering the
The basis for the calculation of the stable bottom bed-load transport and the scouring may be miss-
slope became the classical Shields equation(1) ing. Unexpectedly, this may become a paradox
based on the shear stress philosophy. Surprisingly, if applied at higher flow rates. In addition, the
a good agreement was also shown with the calcu- design flow rate determination brings further
lation after Manning-Strickler (2). The Eq. (3) of difficulties. The uncertainty in its definition is

Stable bed slope Q5 = 2.2 m3/s

0.060 Schields
Manning-Strickler
0.050
Novk
0.040
is (m/m)

0.030

0.020

0.010

Figure 4. Comparison of stable bed slope;

0.000
equations of Schields, Manning-Strickler
1 2 3
de (m) 1 = 0.06 2 = 0.10 3 = 0.16
and Novak, discharge Q5
de (m) 1=0.06 2=0.10 3=0.16

109
 Soil & Water Res., 2, 2007 (3): 104111

Stable bed slope Q100 = 6.9 m3/s Figure 5. Comparison of stable bed
0.035
slope; equations of Schields, Manning-
Schields
0.030 Strickler and Novak, discharge Q100
Mannig-Strickler
0.025 Novk

0.020
is (m/m)

0.015

0.010

0.005

0.000
1 2 3
de (m) 1 = 0.06 2 = 0.10 3 = 0.16
de (m) 1=0.06 2=0.10 3=0.16

rather complex, particularly in the selection of Comment at the end

its true value needed for the determination of
the optimum stable slope. For the Jindichovick The last version HEC-RAS 4.0 Beta (2006) in-
creek, the centenary flood from 1 km 2 amounts cludes two new features related to the sediment
to, e.g., 5.2 m 3/s. transport.
In general, it can be stated that the model accu- Sediment transport/movable bed analyses.
racy increases with the reliability of granulometric This component of the modelling system is in-
analysis of the bed-load. The more detailed and tended for the simulation of one-dimensional
reliable data exist, the better results of the model sediment transport/movable boundary calcula-
can be expected. The optimal determination is tions resulting from scour and deposition over
questionable from the practical point of view moderate time periods (typically years, although
and also from the economic aspects. Therefore, applications to single flood events are possible).
it is with advantage to practice the use of the The sediment transport potential is computed by
model categories, where cross section profiles grain size fraction, thereby allowing the simulation
can be easily subdivided in horizontal and in of hydraulic sorting and armouring. The model
vertical directions. Then spatial distribution of is designed to simulate long-term trends of scour
the values of n or of k in any river section can be and deposition in a stream channel that might
investigated. Since no comprehensive method still result from modifying the frequency and duration
exists for the channel roughness determination, of the water discharge and stage, or modifying
torrent-training designer is forced to use its own the channel geometry. This system can be used
experience. The freely distributed documentation to evaluate the deposition in reservoirs, design
on HEC-RAS Web pages has been made available channel contractions required to maintain naviga-
and it becomes a useful starting point for initial tion depths, to predict the influence of dredging
estimates of both parameters. In these materials, on the rate of deposition, to estimate maximum
one can find well verified roughness values for possible scour during large flood events, and to
natural rivers, with respect to detailed geom- evaluate sedimentation in fixed channels.
etry of the river-bed and to drainage area. The Sediment Impact Analysis Methods (SIAM).
photo-documentation of the representative cases SIAM is a sediment budget tool that compares
in the United States is also available. It would annualized sediment reach transport capacities
be therefore highly recommended to work out to supplies and indicates the reaches of overall
similar guidelines, at least for the representative sediment surplus or deficit. SIAM is a screening
Bohemian torrential rivers. Professor Novak has level tool to compute rough, relative responses
initiated such an effort in the past and it remains to a range of alternatives, in order to identify the
just to follow up. most promising alternatives (which should then

110
 Soil & Water Res., 2, 2007 (3): 104111

be modelled in more detail). The algorithms in Gordon N.D., McMahon T.A., Finlayson B.L. (1996):
SIAM evaluate the sediment impact caused by Stream Hydrology An Introduction for Ecologist.
local changes in the system from the sediment J.Wiley, Sussex.
continuity perspective. The results map potential HEC-RAS (Hydrologic Engineering Centers River Ana-
imbalances and instabilities in a channel network lysis System), Version 3.1.3, (May 2005): Computer
Program. US Army Corps of Engineers, USA.
and provide the first step in designing or refin-
HEC-RAS Version 4.0.0 Beta (Nov 2006): Computer
ing remediation. Users can begin with the exist-
Program. US Army Corps of Engineers, USA.
ing geometry and flow data and develop a set
Lopez L. (1993): Torrent control and streambed stabi-
of sediment reaches with unique sediment and lization. FAO, Rome.
hydraulic characteristics. The SIAM program Kovk F. (2002): Adaptation of Jindichovick creek.
will then perform sediment transport capacity Design, LR OST, Plze. (in Czech)
computations to determine potential imbalances Kovk F., Kov P. (2005): Discharge modelling on
and instabilities in a channel network. SIAM does forest torrent sections. Zprvy lesnickho vzkumu.
not predict intermediate or final morphological Krajina, les a lesn hospodstv. Svazek 49, . 14/2004.
patterns and does not update cross sections, but VLHM Jlovit-Strnady, 5154. (in Czech)
rather indicates trends of locations in the system Novk L. (1981): Protection and Creation of Landscape
for potential sediment surpluses or deficits. The II. VZ, Praha. (in Czech)
results can be used to design or refine remediation Novk L. (1988): Torrent Bed Stabilization. VZ, Praha.
(in Czech)
efforts in the system.
www.hec.usace.army.mil

References Received for publication March 13, 2007

Accepted after corrections May 19, 2007
Dvok J., Novk L. (1994): Soil Conservation and
Silviculture. Development in Soil Science 23. Elsevier,
Brzda, Prague

Corresponding author:
Ing. Frantiek Kovk, CSc., esk zemdlsk univerzita v Praze, Fakulta lesnick a environmentln,
katedra biotechnickch prav krajiny, Kamck 129, 165 21 Praha 6-Suchdol, esk republika
tel.: + 420 224 382 138, + 420 224 382 148, e-mail: [email protected]

111