Flood Routing

Rain clouds

Rock
Deep percolation

1
2
 Flood Routing

Contents Introduction 4

Basic concepts 4

(a) Definitions ................................................................................................ 4

(b) Hydrologic and hydraulic flood routing methods ....................................... 5

(c) NRCS computer programs ....................................................................... 8

Reservoir flood routing method 8

(a) Elevation-storage relations for reservoirs ................................................. 8

(b) Elevation-discharge relations for reservoirs ............................................. 9

(c) Elevation-storage and elevation-discharge relations for ......................... 10

channel reaches

Channel flood routing methods 14

(a) Hydraulic characteristics of channel reaches ......................................... 14

(b) Muskingum method................................................................................ 15

(c) Muskingum-Cunge method .................................................................... 16

(d) Compare and contrast the Muskingum and Muskingum-Cunge ............... 16

routing methods

Discussion 17

3
Introduction:

Flood routing and hydrograph generation (see Part 630 Hydrology, chapter 16, 2007) form the key
elements of a watershed hydrologic model. Stream channels, flood- plains, and reservoirs can have a
significant impact on the delivery of water to any location along a stream network. Flood routing impacts
the magnitude of the peak discharge, the time of the peak discharge, depth and extent of flooding, and
environmental factors such as stream bank erosion, floodplain scour, sediment transport, and
deposition.

The examples in this chapter contain many tables generated by computer and spreadsheet. Some
table values may differ from values calculated by other methods. Numerical accuracy is a function of
the number of significant digits and the algorithms used in data processing, so some slight differences
in numbers may be found if the examples are checked by other means.

Basic concepts:

(a) Definitions:

Flood routing is variously defined as follows:

Routing, flood—The procedure that determines the timing and magnitude of a flood wave at a point on a
stream from the known or assumed data at one or more points upstream (Chow 1964).

Routing, flow—A mathematical procedure that predicts the changing magnitude, speed, and shape of a
flood wave as a function of time at one or more points along a watercourse (Maidment 1993).

Routing, stream channel—Mathematical relations that calculate outflow from a stream channel once in-
flow, lateral contributions, and channel characteristics are known (Ponce 1989).

4
These definitions relate to flood routing in streams and rivers. Definitions related to reservoirs and
breaching dams are:

Routing, reservoir—This procedure derives the out- flow hydrograph from a reservoir from the inflow
hydrograph into the reservoir with consideration of elevation, storage, and discharge characteristics of
the reservoir and spillways. The conservation of mass equation is solved with the assumption that outflow
discharge and volume of storage are directly related.

Routing, breach—This procedure derives the hydrograph at downstream locations from the hydrograph
generated by a breach or failure of a dam. This routing may include hydrographs resulting from precipitation
in addition to the breach hydrograph.

Throughout this document, when a variable first ap- pears in an equation, it is defined in the list of variables
just below the equation. It is also listed in appendix D with its definition and a list of the equations in which it
is found. After the first occurrence in an equation, the variables are not defined again with each usage; the
reader should refer to appendix D.

(b) Hydrologic and hydraulic flood routing methods:

Many textbooks provide detailed descriptions, derivations, and examples of various flood routing methods.
Two general divisions or classifications of flood routing methods are hydrologic and hydraulic. Hydrologic
methods are generally based on the solution of the conservation of mass equation and a relation of
storage and discharge in a stream reach or reservoir.

Hydraulic methods are based on solutions of the conservation of mass and the conservation of momentum
equations. These methods have many formulations and solution techniques, many of which were
developed before computers were available. Computers have saved much time and also allowed more
robust solutions to the flood routing problem. It is still beneficial to understand how flood routing can be
done manually so the engineer can analyze the impact of the various assumptions, limits of application,
sensitivity of results, and adequacy of computer program results. Data requirements vary for each type of
flood routing method. Choice of a method involves understanding the flow characteristics being modeled;
available physical data such as flow data, stream and floodplain topography, roughness, etc.; and the
purpose of the hydrologic analysis with associated risk.
5
Hydrologic flood routing methods are described in this chapter because of their inclusion in U.S.
Department of Agriculture (USDA) Natural Resources Conservation Service (NRCS) computer programs.
Hydraulic methods are available from several sources including the U.S. Army Corps of Engineers
Hydrologic Engineering Centers River Analysis System (HEC–RAS 2010), U.S. Geological Survey (1981),
and the National Weather Service (Fread and Lewis 1988).

Hydrologic methods have been used for many years and have proven their utility in analyzing flood
conditions along streams and rivers. In general, hydrologic flood routing methods involve simplified
numerical techniques, conservation of mass, and steady flow hydraulics. The hydraulic flood routing
methods involve complex numerical solutions of partial differential equations and the theory of unsteady
flow hydraulics. Hydrologic routing methods require a relation of dis- charge and storage, which can be
derived from water surface profiles. Hydraulic methods do not require such a relation. Typical input data
for hydraulic routing methods consists of topographic data such as channel geometry and reach
lengths, roughness, down- stream boundary conditions, and inflow hydrographs.

The movement of flood waves in streams is complex. Linsley, Kohler, and Paulhus (1982) describe two
general classes of natural flood waves. The first includes waves in which the forces of momentum and
acceleration control. These waves are fast rising and occur frequently on streams of steep gradient.
The second class includes flood waves in which friction is the pre- dominant force. Slow rising flood
waves on streams of mild gradients are examples of this class.

The movement of a flood wave down a stream system may be described in terms of translation and
storage effects. These represent characteristics of the two general classes of natural flood waves.
Translation involves maintaining the same hydrograph shape as the flood wave moves downstream
the storage or attenuation effect involves use of valley storage to reduce the peak flow and change the
shape of the hydrograph. These are not necessarily mutually exclusive conditions. Most flood waves
exhibit both characteristics, though one may be more evident than the other based on the specific local
situation. These two characteristics describe the upper and lower limiting conditions of the flood routing
in a channel reach when considered individually. For example, the upper limit of flood routing is
translation of the peak discharge with no attenuation as the flood wave moves downstream. If storage
effects alone act on a flood wave, the result is maximum attenuation, so the peak discharge may not
be further attenuated. This concept represents the lower limit to the flood routing of a channel reach.
These two characteristics of natural flood waves, translation and storage, are represented by the
6
conservation of mass and conservation of momentum equations. The conservation of mass equation,
sometimes referred to as the “continuity equation,” is a mass balance and represents storage effects.
It states that inflow minus outflow is equal to change in storage between two locations and between
two moments in time. The conservation of momentum equation derives from the basic equation of
physics, which states that force equals mass times acceleration, arranged in terms related to movement
of water in a stream such as discharge, end-area, depth, top width, etc. This equation also applies
between two locations and between two moments in time. The solution of these two equations
represents the hydraulic routing method.

Solving these equations for the extreme conditions of pure storage and pure translation provide the
lower and upper limits of downstream peak discharge, respectively. Most flood routing situations are a
com- bination of translation and storage effects. By solving the conservation of mass and conservation
of momentum equations, the most physically based solution may be reached.

Hydrologic flood routing methods use the conservation of mass equation with simplifying assumptions
to reduce the conservation of momentum equation to a relation of discharge to storage. The primary
assumption is that acceleration terms are negligible with respect to other terms in the equation and are
ignored. In many flood routing situations, this is a reasonable assumption. The acceleration terms are
relatively small for slow-rising flood waves on moderate gradient streams. This is not a good
assumption when the stream gradient is very small, which makes the acceleration terms more
important even when they them- selves are small, or for fast rising hydrographs such as those from a
breaching dam, where the acceleration terms are relatively large.

For reservoirs of the general size considered by the NRCS, the assumption of a level pool is reasonable
and the conservation of mass equation along with a relation of discharge and storage is solved. The
de- scription on reservoir routing describes the storage- indication routing method, which is appropriate
for most NRCS applications.

For channel and valley flood routing applications, the Muskingum-Cunge method of flood routing is de-
scribed. The Muskingum-Cunge method is classified as a hydrologic method because it requires a
relation of discharge and storage. This relation is normally developed from steady flow water surface
profiles, but may also be developed from a solution of Manning’s equation for a cross section. This
chapter has derivations, application strategies, and examples for applying the Muskingum-Cunge.
7
 (c) NRCS computer programs

Routing of reservoirs and channels may be done with computer programs as part of a comprehensive
hydro- logic analysis of a watershed. Reservoir routing with the Storage-Indication method and channel
routing with the Muskingum-Cunge method are integrated into the WinTR–20 (2010), SITES (2005),
and WinTR–55 (2011) computer programs. Manual versions of both methods are described in this
chapter with their data requirements, data preparation, assumptions, and limitations.

Reservoir flood routing method

(a) Elevation-storage relations for reservoirs.

One of the important data items needed to complete a reservoir routing is a relation of elevation or stage
with volume of storage. If a reservoir is being constructed, elevation-storage information is needed from the
lowest elevation at the centerline of dam up to an elevation slightly higher than the expected top of dam.
For an existing reservoir, elevation-storage information is generally needed from the elevation of the
permanent pool/principal spillway up to an elevation slightly higher than the top of dam. The presence of
residential or rural development may influence the ac- curacy needed in development of the elevation-storage
relation.

The relation is developed from a contour map or its equivalent of the reservoir area. A table is generally

the easiest way to organize and document the calculations. Table 17–1 is a working table that shows data
and computed results.

Once the contour map of a site is available,

Step 1: Select elevation increments that define the topography with reasonable accuracy, and tabulate
them in column 1.

Step 2: Determine the reservoir surface area at each elevation. For this table the areas were determined
in acres and tabulated in column 2.

Step 3: Compute average surface areas in column 3.

Step 4: Tabulate the increments of depth in column 4.

8
Step 5: Compute the increments of storage for column 5 by multiplying an average area in column 3
by its appropriate depth increment in column 4.

Step 6: Accumulate the storage increments of column 5 in column 6 for each elevation of column 1.

Table 17–1 Elevation, surface area, and storage at a reservoir site

Elevation Surface area Average Depth Storage Accumulated

(ft) (acres) surface increment increment storage
area (acres) (ft) (acre-ft) (acre-ft)
(1) (2) (3) (4) (5) (6)

570 0 0.00
4.82 2 9.64
572 9.64 9.64
18.37 2 36.73
574 27.09 46.37
40.80 2 81.59
576 54.5 127.96
71.63 4 286.50
580 88.75 414.46
106.67 5 533.35
585 124.59 947.81
153.60 5 767.98
590 182.6 1,715.79
205.64 5 1,028.18
595 228.67 2,743.96
250.01 5 1,250.05
600 271.35 3,994.01

The relation of data in columns 1 and 6 is plotted in figure 17–3 as an elevation-storage curve. English units
are used in this example. If SI units are used, elevations would be in meters, areas in hectares, and
storage values in hectare-meters.

(b) Elevation-discharge relations for reservoirs.

The second important data item needed for reservoir routing is a relation of elevation or stage with
discharge through the principal spillway, auxiliary spillway, and over the top of the dam if overtopping is
anticipated during the routing. Select a set of elevations starting at the elevation of the lowest outlet and
ending at the expected maximum routed water level, and calculate discharge at each elevation through the
9
 single or multiple spillways. The elevations selected should be the same as those for which the elevation-
storage relation has been developed in the previous step to eliminate interpolations needed to make
elevations consistent in both tables. The end result will be a table of elevations with the appropriate
storage and discharge.

The wide variety of spillway types, combinations of spillways, and the quantity of hydraulic references
and software used to calculate these spillway ratings precludes their inclusion in this chapter. The
procedures used to develop the discharges in this chapter are based on spillway capacity charts
available in other references (such as Brater and King 1982).

English units of feet and cubic feet per second are commonly used. If SI units are used, elevations
would be in meters and discharge values in cubic meters per second.

Figure 17–3 Plot of elevation-storage data from table 17–1

600

595

590
Elevation (ft)

585

580

575

570
0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000

Storage (acre-ft)

(c) Elevation-storage and elevation-dis- charge relations for channel reaches

Reservoir routing procedures may be used for channel reaches under certain circumstances. An
example is a floodplain where a road or railroad fill backs up the water and forces flow to pass through
a restrictive bridge or culvert. The decision to treat such a flood- plain area as a reservoir requires
judgment based on each case’s physical characteristics:
10
• Is the floodplain relatively wide and the bridge or culvert restrictive?

• Does the history of flooding in the area indicate significant storage of water?

• Could the situation be treated as a reservoir with- out overtopping occurring?

• Is the slope of the stream and floodplain flat enough that the backwater would extend a significant
distance upstream?

• Does enough storage in a combination of depth, width, and length exist to significantly impact the
routing of the inflow hydrograph?

A rule of thumb is that if the volume of storage is more than 10 percent of the volume of the inflow
hydrograph, the area should be treated as a reservoir.

The answers to many of these questions may be de- rived from water surface profiles and preliminary
estimates of peak discharge and volume of runoff at the location. In a typical case, the 100-year flood
may cause backwater with a reservoir-like situation and smaller floods will not. Depending on the
purpose of the hydrologic analysis, the decision to treat the area as a reservoir is generally applied to
all floods consistently.

Development of an elevation-discharge curve and water surface profiles for a stream reach is
covered in NEH630.14. Typically, elevation-discharge curves are developed using a water surface
profile computer program, such as HEC–RAS or solutions of Manning’s equation. In this chapter, it is
assumed that water surface profiles exist. Water surface profiles are generally computed for multiple
discharges ranging from base flow to the expected peak flow. In the example of a road fill that restricts
flow, the downstream end of the reach is at the road. The upstream end of the reach is where the water
surface profile gradually changes from backwater with a nearly flat surface to normal flow where the
slope of the water surface is about the same as the slope of the channel bottom. This exact lo- cation
may be difficult to determine precisely because the position where the flow transition occurs usually
depends on discharge. Typically, low discharges return to normal flow closer to the road than do high
dis- charges. In making the selection of the upper limit of the reach, other factors may need
consideration such as total length of reach, location of tributaries, and roads or other structures
upstream requiring individual routings.

Elevation-discharge curves for a stream reach can be prepared using the procedure for reservoirs,
11
 but ordinarily a modified approach is used, and the storage-discharge curve is prepared directly. If the
reach in question has several cross sections, they can be weighted according to the distances
between cross sections and the total length of reach.

If the reach has three or more cross sections, a weighting method is needed. With only one or two
cross sections the weighting is eliminated, but the reach storage is not defined as accurately. The
weights are then used in a working table to develop an elevation-discharge curve. The calculations in
the working table are typically based on the assumption that steady flow occurs in the reach at all
depths of flow.

Development of the storage-discharge curve goes as follows:

• Select a series of discharges from zero to a dis- charge greater than any to be routed. Run water
surface profiles with constant discharges, and display the cross-sectional end-area in the computer
output.
• Record the cross-sectional end-area, or flow area, for the discharge for each of the cross sections.
• Compute weighted end-areas for each profile.
• Sum the weighted end-areas for each discharge to obtain an average end-area for each profile.
• Compute storage for each profile using this equation:

𝐴
𝑆 = 𝑟𝑒𝑎𝑐ℎ 𝑙𝑒𝑛𝑔𝑡ℎ 𝑥
3,600

where:

S= storage in ft3/s-h

reach length = ft

A= the average end-area

The unit conversion factor to convert seconds to hours is 3,600.

• Compute reach storage in acre-feet for each profile by multiplying the storage values by the
reach length in feet and dividing by 43,560 square feet per acre. The units of cubic feet per
second-hour are generally used for storage- indication routing for convenience.

A hypothetical storage-discharge curve is plotted in figure 17–4. Data in this plot can be used to
prepare a working curve for routing.

12
 If steady flow water surface profiles are computed using the HEC–RAS computer program,
accumulated storage (HEC–RAS output variable Volume) may be requested in an output table.
The storage for each reach is tabulated in acre-feet or hectare-meters for each profile and
represents storage from the lower end of the reach up to any particular cross section. The storage
in the area of concern may be computed as the difference in accumulated storage from the
upstream cross section minus the accumulated storage at the downstream cross section for each
profile.

Figure 17–4 Plot of hypothetical discharge versus storage curve for a channel reach

10,000

9,000

8,000

7,000
Discharge (ft3/s)

6,000

5,000

4,000

3,000

2,000

1,000

0
0 1,000 2,000 3,000 4,000 5000 6,000 7,000 8,000

Storage (ft3/s-h)

13
Channel flood routing methods

(a) Hydraulic characteristics of channel reaches

NRCS hydrology computer programs previously used the Convex (SCS 1965) and Modified Att-
Kin (Comer, Theurer, and Richardson 1982) channel flood routing methods. The discharge and
end-area relation were used to compute one routing coefficient. The use of discharge and end-
area values by the Convex and Modified Att-Kin methods allowed the engineer to base the channel
routing on as little data as a velocity and reach length. To increase accuracy somewhat, the
engineer could enter an elevation-discharge-end-area relation for a reach. This allowed the routing
coefficient to be based on a particular discharge. Thus, for a given reach, if the hydrograph to be
routed remains within the channel, the routing coefficient is based on a channel velocity. If the
hydrograph to be routed extends into the floodplain, the routing coefficient is based on an average
cross section velocity.

The Muskingum-Cunge (Ponce 1989) method of flood routing uses discharge and end-area along
with top width and slope in developing its two routing coefficients. An actual cross section, either
surveyed or approximated, is needed for the Muskingum-Cunge method. As a trade-off for
requiring more information, greater accuracy and range of applicability are avail- able with the
Muskingum-Cunge method. This section outlines the development of hydraulic characteristics of
cross sections and channel reaches needed by the Muskingum-Cunge method.

Development of hydraulic information for cross sections and stream reaches is described in detail
in NEH630.14, Stage Discharge Relations. Depending on the amount of data available and the
purpose of the project, stage discharge relations are generally developed based on water surface
profiles for stream reaches using a series of cross sections along a stream or river or by solving
Manning’s equation for a series of depths at a single cross section. Both of these procedures are
automated in the HEC–RAS computer pro- gram (U.S. Army Corps of Engineers 2010).

The NRCS hydrologic models WinTR–55, SITES, and WinTR–20 are set up to assign a single
cross section to represent the hydraulic characteristics of a particular reach.

Two important judgments are needed by the engineer when developing data for channel reach

14
routing. One is the selection of a representative cross section.
This cross section should represent an average flow velocity through the reach. If there are several
cross sections available for selection, the engineer should select the section most typical of the
reach conditions.
The second important judgment is the selection of a reach length. To properly represent reach
storage characteristics, the reach length should be an average length over the routing reach. HEC–
RAS allows for left overbank, right overbank, and channel reach lengths. WinTR–20 allows for only
floodplain length and channel length. WinTR–55 is limited to only a single channel length.

Storage in a reach is often underestimated because backwater storage in tributaries is usually not
considered in developing water surface profiles using HEC–RAS. If this type of storage is
significant, it should be estimated. The simplest way to account for this would be to increase the
floodplain length. A more complex analysis would involve adjusting HEC–RAS cross sections to
include an ineffective flow area for tributary backwater.

(b) Muskingum method

A brief review of the Muskingum method precedes the description of the Muskingum-Cunge
method. The routing equation is similar for the methods. The Muskingum channel routing method
is based on two equations (Linsley, Kohler, and Paulhus 1982). The first is the continuity equation
or conservation of mass.

(𝐼1+𝐼2) (𝑂1+𝑂2)
𝛥𝑡 − 𝛥𝑡 = 𝑆2 − 𝑆1 (Equation 17–5)
2 2

Equation 17–5 states that inflow to the reach minus outflow from the reach is equal to the change
in storage. This is the same basic equation used in routing of reservoirs. Where the reservoir
routing method assumes a relation of storage and outflow discharge, the Muskingum method
assumes the amount of storage is related to both inflow and outflow discharge.

The reservoir routing method assumes a level pool, and the Muskingum method assumes a sloping
water surface.

𝑆 = 𝐾(𝑋1 + (1 − 𝑋)𝑂) (Equation 17–6)

15
where:

S = reach storage, ft3

K = storage constant, s
X = weighting factor, dimensionless I = inflow discharge, ft3/s
O = outflow discharge, ft3/s

When X equals zero in equation 17–6, the equation reduces to a simple relation of storage and
outflow discharge: S is KO (reservoir routing assumption). The values in the equations are in units
of feet and seconds. They may also be defined in any units of length and time as long as all values
in the equation are consistent.

(c) Muskingum-Cunge method

Cunge (1969) developed equations to estimate K and X from hydraulic properties of the reach.
These are based on the linking of the convection diffusion model and the Muskingum routing
equation.

(d) Compare and contrast the Muskingum and Muskingum-Cunge routing methods

The Muskingum-Cunge method is more applicable for use on ungagged streams. It uses the same
routing equation as the original Muskingum method. The weighting factor X is defined
mathematically in the Muskingum- Cunge method and is based on hydraulic properties of the
reach. The solution technique is somewhat different in that the Muskingum method treats the reach
as a single routing distance step and the Muskingum- Cunge method defines a routing distance
step length and sub-steps a reach if needed.

The storage constant K is defined differently in the two methods. The Muskingum method defines
K in the relation of discharge and storage, and the Muskingum- Cunge method defines K as the
routing distance step length divided by the wave celerity. K has the units of time in both procedures
and X is nondimensional. The value of X in the Muskingum method may vary from 0.0 to 0.5. The
value of X for the Muskingum-Cunge may be positive or negative.

16
Discussion

The channel tests had a correlation coefficient of 0.97 but slightly underestimated the peak
discharge (the regression line is slightly below a 1:1 relation). The floodplain tests had a correlation
coefficient of 0.70 and slightly overestimated the peak discharge (the regression line is slightly
above a 1:1 relation). The plot of all 77 tests showed a correlation coefficient of 0.86 and a
regression line on the 1:1 relation; meaning the total of all results are not biased on the high or low
side.

The error in outflow peak time was computed based on the travel time of the peak discharge. The
error is the time to peak outflow (Muskingum-Cunge) minus the time to the peak outflow (dynamic
wave) divided by the time to peak outflow minus the time to the peak inflow of the dynamic wave
routing. For the 77 tests, the average error in this timing for the Muskingum- Cunge was 6.8 percent
with most values (60) ranging from plus to minus 20 percent. Errors in timing seemed to be random
and not specifically related to any cross section or flow conditions. The minimum travel time in any
routing distance step (and this included the solution of the dynamic wave equations) was one-time
step. Of the 77 tests run, 48 had a timing error of one-time step or less; 19 had a timing error of
two-time steps. This left 10 tests with a timing error of more than two-time steps. The conclusion is
that tim- ing of routings is adequate.

Since the floodplain tests had more variability, there seems to be a problem modeling the
interaction of the channel and floodplain or representing how the cross-section stage-discharge-
area relation is used to calculate the Muskingum-Cunge routing coefficients. This points to the need
for further testing and development.
These results also point to the need to identify under what conditions the Muskingum-Cunge
procedure pro- duces a significantly higher or lower peak discharge than the dynamic wave model.

17