United States

Department of Part 654 Stream Restoration Design

Agriculture
National Engineering Handbook
Natural
Resources
Conservation
Service

Chapter 10 Two-Stage Channel Design

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

Issued August 2007

Cover photo: Low gradient, nonincising channels and ditches may be

modified by creating a narrow flood plain, thereby creat-
ing some ecological benefits, while minimizing the need for
maintenance (clean-outs).

Advisory Note

Techniques and approaches contained in this handbook are not all-inclusive, nor universally applicable. Designing
stream restorations requires appropriate training and experience, especially to identify conditions where various
approaches, tools, and techniques are most applicable, as well as their limitations for design. Note also that prod-
uct names are included only to show type and availability and do not constitute endorsement for their specific use.

The U.S. Department of Agriculture (USDA) prohibits discrimination in all its programs and activities on the basis
of race, color, national origin, age, disability, and where applicable, sex, marital status, familial status, parental
status, religion, sexual orientation, genetic information, political beliefs, reprisal, or because all or a part of an
individual’s income is derived from any public assistance program. (Not all prohibited bases apply to all programs.)
Persons with disabilities who require alternative means for communication of program information (Braille,
large print, audiotape, etc.) should contact USDA’s TARGET Center at (202) 720–2600 (voice and TDD). To file
a complaint of discrimination, write to USDA, Director, Office of Civil Rights, 1400 Independence Avenue, SW.,
Washington, DC 20250–9410, or call (800) 795–3272 (voice) or (202) 720–6382 (TDD). USDA is an equal opportunity
provider and employer.

(210–VI–NEH, August 2007)

Chapter 10 Two-Stage Channel Design

Contents 654.1000 Purpose 10–1

654.1001 Introduction 10–1

654.1002 Background 10–2

(a) Advantages of a two-stage channel.............................................................. 10–3
(b) Design of a two-stage channel...................................................................... 10–4

654.1003 Field measurements 10–4

654.1004 Bankfull channel design 10–5

(a) Regional curve development........................................................................ 10–5
(b) Rapid regional curve development.............................................................. 10–5
(c) Reference reach.............................................................................................. 10–6

654.1005 Flood plain channel design 10–6

654.1006 Flood conveyance 10–7

654.1007 Spreadsheet tools for data analysis and design 10–8

(a) Site selection and reconnaissance............................................................... 10–8
(b) Regional curve................................................................................................ 10–8
(c) Discharge data................................................................................................ 10–9
(d) Discharge versus recurrence interval.......................................................... 10–9
(e) Ditch geometry............................................................................................. 10–12
(f) Bed material and bed-load transport......................................................... 10–13
(g) Discussion..................................................................................................... 10–14

654.1008 Conclusion 10–15

(210–VI–NEH, August 2007) 10–i

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

Tables Table 10–1 Discharge vs. recurrence interval results at the gage 10–11
and the Hillsdale ditch

Table 10–2 Relative bed-load transport for various channel 10–14

conditions

Figures Figure 10–1 Trapezoidal cross section of a constructed drainage 10–1

ditch

Figure 10–2 Drainage ditch constructed in north-central Iowa 10–1

Figure 10–3 Conceptual design for two-stage channel system 10–2

Figure 10–4 Ditch before and after maintenance (MN) 10–3

Figure 10–5 Two-stage ditch geometry with minimum sized 10–3

benches

Figure 10–6 Ditch before construction at location 1,000 to 1,200 ft, 10–9
and after widening of small benches at location 1,600
to 2,100 ft (Hillsdale County, MI)

Figure 10–7 Bankfull discharge channel dimensions for the St. 10–10
Joseph River, OH, upstream of the gage near Newville,
IN

Figure 10–8 Profiles of the bed, bench, and top of ditch 10–12

Figure 10–9 Pre- and postmaintenance geometries at a location 10–13

with a grade break, but weak bench formation at the
elevation where it was determined that a bench would
naturally form. The existing main channel has a similar
geometry to the projected geometry

10–ii (210–VI–NEH, August 2007)

Chapter 10 Two-Stage Channel Design

654.1000 Purpose 654.1001 Introduction

Constructed channels are part of extensive portions Agricultural ditches and channels have long been
of productive agricultural land in the United States. used to provide important drainage and flood control.
These channels provide important drainage and flood Historically, many of these drainageways are designed
control functions. However, these agricultural chan- following threshold design techniques and result in a
nels are often constructed as traditional trapezoidal large, trapezoidal cross section. The primary purpose
ditches using threshold design techniques. While this of the constructed channel is to convey water from
approach is suitable in some areas, channels of this agricultural fields.
design can require frequent and expensive mainte-
nance in other parts of the country. In addition, natural Figure 10–1 illustrates the basic design configura-
ecological functions are normally not a consideration tion for a trapezoidal channel, and figure 10–2 is an
in the design of these channels. This chapter presents example of one in Iowa. In many situations, the water-
an alternative design to the conventional trapezoidal way does behave as a threshold channel, so this is an
drainage channel. This two-stage channel system appropriate approach. However, when the waterway
incorporates benches that function as flood plains and behaves as an alluvial channel, the ditch can be too
attempts to restore or create some natural alluvial entrenched and have overwidened bed widths. While
channel processes. However, these two-stage chan- the large section of a traditional agricultural drain-
nels are not an exact copy of natural streams, as the age channel may provide sufficient flood conveyance,
width of the benches is often small due to the confin- the more frequent discharges may not flow at a depth
ing geometry of the constructed channel. This chapter and velocity sufficient to move sediment through the
outlines measurement and analysis procedures that reach. Deposition results, requiring maintenance to
can be used to size two-stage channel systems that maintain the design flow capacity. As deposition oc-
are more self-sustaining than conventional one-stage curs, bank stability may also become an issue as sedi-
constructed channels. Although this chapter focuses ment deposits may force flows into one bank or the
primarily on an alternative design for constructed other. In addition, baseflows in this wide channel may
ditches, the technique may also have application in have a depth which does not provide adequate aquatic
natural streams that have undergone incision or in habitat.
streams where boundary constraints restrict restora-
tion designs such as in urban or developed areas. A
case study is also presented for a constructed two-
stage ditch in Michigan.
Figure 10–2 Drainage ditch constructed in north-central
This two-stage channel design approach is applicable IA
to low gradient ditches and channels that are not un-
dergoing incision.

Figure 10–1 Trapezoidal cross section of a constructed

drainage ditch

(210–VI–NEH, August 2007) 10–1

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

Fluvial processes at work in agricultural ditches func-

tioning as alluvial channel systems often try to develop
a flood plain that consists of low benches. While this
654.1002 Background
deposition reduces flood capacity, these ditches show
improved stability and improved habitat quality. This Highly modified channels drain extensive portions
chapter presents a two-stage approach to design stable of productive agricultural land in the United States.
agricultural drainage channels (fig. 10–3). Specifically, Headwaters are typically the most modified. In some
the two stages are the: areas, virtually all of the natural channels have been
deepened and straightened to facilitate the flow of wa-
• dominant discharge or channel-forming dis- ter from agricultural subsurface drainage outlets and
charge channel to maximize water conveyance. Work is done periodi-
• flood plain bench or flood plain channel cally to maintain the drainage function, which typically
includes removal of woody vegetation, weeds, and
This two-stage approach provides improved physical, deposited sediment. Ancillary work includes stabiliz-
as well as ecological performance. The channel-form- ing bank slope failures and toe scour. Ditch form is a
ing discharge channel provides the necessary sediment result of not only construction and maintenance, but
conveyance, while the flood plain channel provides for also to verifying degrees, due to fluvial (flowing water)
the design flood conveyance. By nesting the channel- processes.
forming discharge channel within the larger channel,
the entire waterway is more stable. Ditch maintenance typically restores the ditch to a
trapezoidal shape designed to transport large storm
The technique described herein uses bankfull dis- events (fig. 10–4). To facilitate drainage and reduce the
charge as representative of the channel-forming frequency of over bank flows, trapezoidal ditches are
discharge. Therefore, the channel-forming discharge designed to accommodate large flows (5- to 100+- year
channel is referred to as the bankfull channel. The recurrence interval) within the ditch. Also the width
bankfull channel has also been referred to as the effec- of the ditch bottom is constructed wider than the
tive discharge channel. However, this is not necessar- channel bottom that would form by fluvial processes,
ily accurate. There are no calculations made to define thus, making the channel relatively wide and shal-
the effective discharge. Rather, this lower stage is as- low. Therefore, the constructed ditch channel is often
sumed to be the bankfull discharge of a low-flow chan- oversized for small flows and provides no flood plain
nel formed in a typical constructed ditch. The distinc- for large flows.
tions between bankfull, channel-forming, and effective
discharge are addressed in more detail in NEH654.05. In contrast to trapezoidal agricultural drainage ditch-
The differences between alluvial and threshold chan- es, integral parts of many natural stream channels are
nels are addressed in NEH654.07. the flood plains. The flood plains of natural streams
(except for those with steep bed slopes) are charac-
terized by frequent, extensive over-bank flow. In dy-
namic equilibrium, a stream system depends on both
the ability of the flood plain to dissipate the energy of
high flows and to concentrate the energy of low flows
to effectively create a balance in sediment transport,
storage, and supply. In natural alluvial streams, fluvial
Figure 10–3 Conceptual design for two-stage channel processes work to size and maintain the dimensions of
system the bankfull channel based on the effective discharge
(Ward and Trimble 2004).
Channel-forming
discharge channel
In response to the construction of an oversized trap-
Flood plain ezoidal channel, alluvial channel processes often work
bench
to create a small bankfull channel by building a flood
Flood plain
plain or bench within the confines of the ditch (fig.
channel 10–4). If conditions allow, these benches can reach

10–2 (210–VI–NEH, August 2007)

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

a stable size, thickly vegetated with mostly grasses. (a) Advantages of a two-stage channel
This results in a two-stage channel. The small bankfull
channel will often meander slightly within the ditch. Benefits of a two-stage ditch over a conventional trap-
The bankfull channel will usually have steep (1H:1V) ezoidal ditch are potentially both improved drainage
sides and a bed consisting of material coarser than function and ecological function. Drainage benefits
that of adjacent reaches where benches have not may include increased ditch stability and reduced
formed. Further details on fluvial processes in ditches maintenance. Evidence and theory both suggest that
are available form Landwehr and Rhoads (2003) and ditches prone to filling with accumulated sediment
Ward, Mecklenburg, and Brown (2002). It is important may require less frequent dipping out if constructed in
to note that these deposits within a constructed trap- a two-stage form. Second, channel stability may be im-
ezoidal ditch reduce the overall flood conveyance. As a proved by a reduction in the erosive potential of larger
result, the channel may no longer provide the designed flows as they are shallower and spread out across the
flood protection. bench (fig. 10–5).

Figure 10–4 Ditch before maintenance and after maintenance (MN)

(a) Before (b) After

Figure 10–5 Two-stage ditch geometry with minimum sized benches

Design discharge
RI of 5 to 100 years

High flows spread across the bench H

W W W

Effective discharge Bench anchors

RI<2 years the side slope

(210–VI–NEH, August 2007) 10–3

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

Stability of the ditch bank should be improved where

the toe of the ditch bank meets the bench, rather than 654.1003 Field measurements
the ditch bottom. Here the water depth is effectively
reduced, and the shear stress (erosive force) on the
toe of the bank is less. Also, not being in contact with Initial reconnaissance of the site area is recommended
low flow, this bank material will be dryer and can be to establish a base knowledge of the project charac-
stabilized with vegetation using threshold design tech- teristics, surrounding area, and regional environment.
niques as described in NEH654.08. Since a two-stage The unique characteristics of the project site will
channel in an alluvial system will be more likely to generate the criteria for regional measurements. If
retain its design shape, it is easier to predict its flood searching for a reference reach, the watershed area,
protection performance. vegetation, soil, land use, and slope should correspond
to the site in question.
The biggest advantage of these two-stage channels
is the ability to transport sediment more effectively. Where a modification will be made to an existing
However, the two-stage ditch also has the potential to channel, detailed measurements should be made of
create and maintain better habitat than a conventional the channel profile and the dimensions of the bank-
trapezoidal ditch. The narrow, deep bankfull channel full channel and benches that have formed within the
provides better water depth during periods of low channel system. Often, the bankfull channel will be
flow. Grass on the benches can provide some instream overwidened, and the benches will be intermittent and
cover and shade. The substrate in the bankfull channel sloping. Guidance on performing such investigations is
is improved as the two-stage form increases sediment provided in NEH654.03. Conducting an onsite geomor-
conveyance and sorting, with fines deposited on the phology study is a simple and reliable method, but is
benches and coarser material forming the bed. only adequate if:
• a bankfull channel and benches have formed
(b) Design of a two-stage channel • the project length is short; the drainage area is
relatively constant
Design and construction of two-stage channels is
different than that of traditional trapezoidal chan- A detailed survey along the reach of interest, or a
nels. The design of a two-stage agricultural ditch in an reference reach, consists of measuring the profile,
alluvial channel system involves correctly sizing the pattern, and dimension of the channel. The profile is
bankfull channel and minimum bench widths for the the slope of the bed surface including all pools, riffles,
flood plain bench. The dimensions of the bankfull dis- and runs. The undulating elevations of the channel
charge or fluvial channel dictate the two-stage channel bed leads to questions of the true channel profile. To
design. If properly sized, the bankfull channel will be compensate for the bed slope variability, the water
maintained by fluvial processes and will reduce or pos- surface is also measured to represent the slope of the
sibly even eliminate large-scale channel maintenance. channel. The pattern of a reach measures the sinuos-
The flood plain bench serves as a flood plain for the ity of the bankfull channel. This is obtained using a
smaller bankfull channel, but it acts more as a thresh- compass and measuring the azimuth from magnetic
old channel. The upper stage must convey the channel- north. The dimensions of the channel are obtained by
forming discharge and must have an adequate size to surveying cross sections, either at increments along
prevent design flood flows from overtopping the ditch the reach or at representative cross sections. A laser
banks and flooding surrounding land. level and survey tape, or a total station instrument, are
often used. The distance from the left channel bank
and the change in elevation are measured for each
grade break across the channel cross section. A pebble
count should be performed to estimate the mean bed
particle size. Guidance for performing pebble counts is
provided in NEH654 TS13A.

10–4 (210–VI–NEH, August 2007)

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

provide an additional point when creating a regional

654.1004 Bankfull channel curve. An applicable and complete regional curve can
be a valuable tool for two-stage channel design, as well
design as many other stream design activities.

The first step in developing a two-stage design is Care must be taken with the use and development of
determining the probable dimensions of the bankfull the regional curves. The data used to develop a curve
channel. This channel will carry most of the sediment needs to be from physiographically similar basins.
in the channel. The width of the bankfull channel is a Drainage network patterns and the relative location of
key design characteristic. It will determine the success the channel site with respect to uplands are significant
in achieving the intended drainage effects, as well as characteristics. The bed and bank characteristics used
ecological benefits. Channel design dimensions are de- in the regional curve development should be the same
termined by measuring the bankfull discharge features as those at the project site. Issues related to the devel-
or calculating the effective discharge at the project opment and use of regional hydraulic geometry curves
site and then by creating a watershed specific regional are described in more detail in NEH654.09.
curve for the project.
Small watersheds that are drained by agricultural
ditches can present particular challenges in the de-
(a) Regional curve development velopment of traditional regional curve data. In most
parts of the Nation, there are a limited number of
The probable dimensions of the bankfull channel can small-gaged watersheds, and these typically have short
be empirically determined based on regional stud- records or have been discontinued. Some additional
ies similar to those that are conducted for natural difficulties associated with developing regional curves
streams. Typically, for natural streams this knowledge are that gages are often located at road crossings or
is acquired by developing regional curves that relate the reach within the vicinity of the gage is highly modi-
the bankfull channel dimensions to drainage area. fied.
Traditional regional curves are created by perform-
ing numerous profile and cross-sectional surveys at
locations with different drainage areas, which often (b) Rapid regional curve development
include U.S. Geological Survey (USGS) stream gage
sites as described in NEH654.05 and NEH654.09. The For two-stage channel design in many agricultural
regional sites should be selected to provide fluvial watersheds, an abbreviated rapid regional curve may
information over a range of drainage areas that can be be adequate. The method consists of finding ditches or
plotted to show channel dimension relationships to streams with well-developed benches/flood plains and
drainage area. measuring at least the width and depth of the natu-
rally formed bankfull channel. The selected channels
Measurements can be taken at the water surface and must have reached a state of equilibrium and must be
the bankfull fluvial features. Each gage station has a stable. Sites for any regional curve should also have
unique rating curve, which is a relationship between similar characteristics to the project site and should
the gage reading and the streamflow rate. The bankfull come from physiographically similar watersheds.
height at each gage can be obtained by the measure- Several measurements should be taken for each range
ments at bankfull and the waters surface along with of drainage areas to verify that the measured feature is
the USGS real-time gage value. However, it is impor- consistent with those across the watershed. Whenever
tant to note that while each gage station may have a possible, precision surveying instruments should be
unique rating curve, the relationship between gage used to make the elevation and distance measure-
height and discharge is not necessarily unique. The rat- ments.
ing curve may shift over the long term as the cross-sec-
tional shape and/or elevation changes, and it may shift Traditional regional curves are created by performing
over the course of a hydrograph due to the unsteady numerous detailed surveys at locations with differ-
loop effect or changing bedforms. If the rating curve is ent drainage areas. In contrast, a rapid regional curve
applicable, these values, combined with the width, will channel dimension measurement consists of quickly

(210–VI–NEH, August 2007) 10–5

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

measuring visual fluvial features with a 100-foot tape

and a telescoping leveling rod. The channel dimen- 654.1005 Flood plain channel
sions taken at a complementary range of several
drainage areas should provide a sufficient spread for design
each log cycle. The bankfull dimensions of width and
depth are measured where visible fluvial features are The formation of benches in constructed ditches is
noticed. The drainage area for each measurement is the natural result of fluvial processes in most alluvial
acquired from a variety of methods such as calculating systems. The bench acts as a flood plain within the
the area by hand using a planimeter or computer GIS ditch to dissipate energy, reduce the erosive potential
software. The measured dimensions can be the plot- of high-flow volumes, and reduce the shear stress on
ted area versus drainage area, and a power regression the bank toe. In establishing two-stage geometry, it
equation can be fitted to the data. This equation can be is often not cost effective or practical to form a flood
used to estimate the bankfull channel design dimen- plain as wide as fluvial processes would form under
sions for a ditch, given the drainage area. natural conditions. Large, deep agricultural ditches
have often already been constructed to handle dis-
This rapid regional curve approach has been used on charges from subsurface drainage systems. Making
several watersheds in Ohio. Reportedly, this approach these large ditches even wider would result in exten-
typically provides relationships between the bankfull sive earth moving, high cost, and substantial losses in
channel and drainage areas with r2 values of 0.8 or productive agricultural land. Therefore, the ability of
greater (Ward 2005; Ward et al. 2003). these small flood plains (benches) to aid in developing
a self-sustaining system is dependent on the establish-
ment of dense grass cover on the benches and banks
(c) Reference reach
of the ditch. Also, the side slopes and depths of the
ditch above the benches must satisfy geotechnical
Measurements from a reference reach can provide
engineering requirements to provide bank stability.
valuable design guidance for the design of the bankfull
channel. Typically, for natural streams this knowledge
In a designed two-stage channel, the elevations of the
is acquired by conducting detailed surveys along a
flood plain channel benches are dependent on cor-
reach of interest and by conducting a detailed survey
rectly determining the size of the bankfull channel.
of a reference reach along the same stream or a simi-
The flooded width is defined as the total width across
lar nearby stream system. However, finding reference
the ditch at the stage elevation where benches have
reaches can be a time consuming, costly, and frustrat-
formed and/or are anticipated to form. The two-stage
ing activity. The attributes of the local subwatershed,
width ratio is defined as the flooded width divided by
such as the topography, soil and bedrock properties,
the top width of the bankfull channel. Based on visual
vegetation on the banks and adjacent riparian zone,
observations and modeling bed-load transport, two
and size and characteristics of the active flood plain,
rules of thumb have been established:
can result in a variety of different stable channel di-
mensions for similar-sized drainage areas within a wa- • If the total width, when out-of-channel flow is
tershed or region. For a reference reach to be directly initiated, is less than three times the top width
applicable, it must have similar climate, history, drain- of the bankfull channel, the benches might not
age area, and watershed conditions. More information fully develop, the benches are more likely to
on the identification and use of reference reaches is be unstable, and shear stresses on the bed and
provided in NEH654.09 and 654.12. banks of the ditch will be high during large
events.
• If the total width, when out-of-channel flow is
initiated, is more than five times the top width
of the bankfull channel, the channel will be-
gin to exhibit a natural meander pattern that,
at places in the ditch, is likely to cut into the
banks of the ditch.

10–6 (210–VI–NEH, August 2007)

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

Therefore, when out-of-channel flow is initiated, the

designed target total width should be between three 654.1006 Flood conveyance
to five channel widths (total bench sizes that are two
to four times the channel width), if the objective is
to provide adequate conveyance capacity and a more The overall conveyance capacities of the two-stage
self-sustaining system, while maintaining a relatively systems can be sized based on the probability of out-
straight ditch geometry. If the project goals or the sta- of-ditch flooding into adjacent areas. This probability
bility requirements of the channel design require the is based on the recurrence interval storm event that
development of a sinuous channel, a wider bench may the entire ditch can transport. Where possible, stream
be required. However, channel alignment design ele- gage data should be used to determine the discharges
ments, such as are described in NEH654.12, will need associated with a prescribed recurrence interval.
to be addressed so that a stable planform is chosen. However, in most parts of the Nation, there are limited
numbers of small watershed gages. Typically, these
gages have short records or have been discontinued.
Therefore, measured streamflows at locations without
gages are determined from hydrologic models or from
regional discharge curves. One source of regional
discharge information is the National Flood Frequency
(NFF) Program. The NFF Program includes 2,065
regression equations for 289 flood regions nation-
wide. These equations are contained in a Windows®
program for estimating the magnitude and frequency
of peak discharges for unregulated rural and urban
watersheds. This program can be obtained at the fol-
lowing Web site:
http://water.usgs.gov/software/

Since most two-stage channels are of fairly uniform

section and constant slope, a resistance equation, such
as Manning’s equation, can be used to calculate the
depth corresponding to the design flow recurrence
intervals.

(210–VI–NEH, August 2007) 10–7

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

interval data, or estimates based on the USGS Urban

654.1007 Spreadsheet tools for Method for Ohio, are used together with Manning’s
equation to calculate the flow stage associated with
data analysis and design each recurrence interval. The ODNR channel design
spreadsheet is programmed to obtain coefficients for
Many of the calculations for two-stage channel as- recurrence intervals of 0.25, 0.5, 1.0, and 1.5 years.
sessment and design can be performed with the help
of computer spreadsheets. One set has been devel- In the spreadsheet, bed-load transport in the bankfull
oped by the Ohio Department of Natural Resources channel is calculated based on the probable discharg-
(ODNR). This suite of spreadsheet tools can be ob- es that will occur during a 100-year time period, the D50
tained from the following Web site: of the bed material, and based on estimates obtained
from the Meyer-Peter and Müller bed-load transport
http://www.ohiodnr.com/soilandwater/ equation (Ward and Trimble 2004). Estimates are
streammorphology.htm obtained for the bed load, recurrence interval of the
bankfull discharge, and probable stage of the bankfull
These spreadsheets aid in designing a new bankfull discharge.
channel together with various size benches, based on
the following: While these spreadsheet tools have been developed to
• a regional curve for the area aid in the analysis of stream form and processes, they
are best applied to alluvial stream systems that are a
• cross-sectional data for a ditch or channel function of the bankfull discharge.
reach
• profile data for the reach that can include bed, (a) Site selection and reconnaissance
water elevation, bench, and top of ditch data
• the D50 fraction of the bed material Potential sites for regional curve measurements
were marked on Michigan, Indiana, and Ohio State
• user-defined adjustments to the channel, bench, Gazetteers published by DeLorme. Sites were selected
and ditch geometry to provide data for several sized drainage areas within
each log cycle. Due to the remoteness of the area
The channel width, depth, and cross-sectional area from Columbus, Ohio, no preliminary reconnaissance
associated with the bankfull discharge at each location was conducted. At each site, the widths and depths
surveyed are entered into the spreadsheet to develop a associated with grade breaks were measured using a
regional curve. The calculated drainage area and bank- 100-foot tape and a telescoping surveying rod. A more
full discharge at each location are also entered into the detailed and accurate survey was conducted at the
same spreadsheet. A log-log plot is then made of each Hillsdale ditch using a laser level, 100-foot tape, and
bankfull discharge dimension versus drainage area. a telescoping rod with a laser receiver. At that site,
A least-squares analysis is then used to fit a power cross-sectional information was obtained every 100
regression line (a trend line) through each set of data feet, and the location of the thalweg was noted. On
and calculate the coefficients of the regional curve. November 16, just prior to construction of the two-
stage geometry, a pebble count was conducted for
In the spreadsheet, stage-discharge relationships for reaches 600 to 800 feet, 800 to 1,000 feet, and 1,000 to
each site are obtained based on Manning’s equation. 1,200 feet.
A separate Manning’s n value is used for the bankfull
channel and the vegetated benches and banks of the
ditch. The roughness of the bed, banks, and benches (b) Regional curve
vary seasonally based on winter conditions, vegeta-
tion growth, maintenance, and scour or deposition Data were obtained at 14 locations within a 600-
on these features. Therefore, the approach used only square-mile drainage area. A regression analysis of the
provides a general representation of roughness con- data (fig. 10–7) indicates that the bankfull discharge
ditions. User-defined discharge versus recurrence dimensions are highly correlated with drainage area.

10–8 (210–VI–NEH, August 2007)

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

The poorest correlation is with channel depth, perhaps At this site, a laser level was used to determine the
because some of the channels were associated with bankfull discharge and water surface elevation. Real-
streams that were highly connected to the flood plain, time gage data were available at the time the survey
while others were associated with grade breaks and was performed and downloaded from the Internet.
small bench formation in ditches. The Hillsdale ditch Due to deep flow conditions, it was only possible to
has a 4.5-square-mile drainage area, and its measured estimate the width of the river by making a measure-
channel dimensions are located almost exactly on the ment on the road across the bridge.
regression lines.

(d) Discharge versus recurrence interval

(c) Discharge data
An annual series of peak flow data for the period
The regional curve analysis was extended to the USGS 1947 through 2002 were available for the gage on the
gage on the St. Joseph River near Newville, Indiana. St. Joseph River near Newville, Indiana. The Weibull
Streamflow data for this gage was obtained from the method (Ward and Trimble 2004) was used to develop
following Web site: a plot of discharge versus recurrence interval data
resulting in a high correlation (r2=0.96) between peak
http://water.usgs.gov/oh/nwis/rt
discharge and recurrence interval.

Example 10–1: Hillsdale County case study

The Hillsdale County, Michigan, case study was conducted as part of a demonstration project for The Nature
Conservancy, Upper St. Joseph River Project Office, and is funded with a grant from the Great Lakes Commission.
The survey was conducted in July 2003, and an existing ditch was modified to a two-stage geometry in November
2003 (fig. 10–6). The project site is located in Hillsdale County, within the St. Joseph Watershed (MI). In 1997, the
ditch was cleaned out as part of a maintenance action and in July 2003, had 0.5 to 2.0 feet of sediment deposits on
the bed and had formed small intermittent benches.

Figure 10–6 Ditch before construction at location 1,000–1,200 ft, and after widening of small benches at location 1,600–
2,100 ft (Hillsdale County, MI)

(a) Before (b) After

(210–VI–NEH, August 2007) 10–9

 Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

On the day of the survey, the water depth at the gage surprising. In flat, poorly drained areas in the Midwest,
was about 3.7 feet (based on the real-time measure- where subsurface drainage is widely used, the bank-
ment from the USGS NWIS Web site), and the bankfull full discharge occurs frequently and primarily due to
discharge dimensions were measured on both banks subsurface drainage discharges. Since this bankfull
at about 3.7 to 5.4 feet above the water elevation. The discharge is associated with high subsurface flows, it
most dominant feature was a continuous approximate- is usually associated with a recurrence interval that is
ly 20-foot high bench that was located on the left bank. much more frequent that one year (Ward 2005; Ward et
A shorter bench at a similar elevation was located on al. 2003).
the right bank. Therefore, it was estimated that the
bankfull stage was 7.4 to 9.1 feet. An analysis of the daily flow records shows that dis-
charges within this range or larger occur on average
An approximate rating curve for the gage was cre- 40 to 80 days annually. This range of flow seems to be
ated using data from the USGS Web site. The bankfull too frequent to correspond to the bankfull or channel-
discharge was estimated from the survey data, and the forming discharge. However, an analysis of the daily
rating curve to be between 740 and 1,330 cubic feet flows exceeding 1,330 cubic feet per second revealed
per second. While this is a wide range, an analysis of that, on average, they are associated with 1 to 13 dis-
the recurrence interval curve indicates that this cor- charge events annually, and the duration of these flows
responds to a recurrence interval that is much more ranged from 1 to 49 days. High flows lasting many days
frequent than 1 year. This frequent occurrence is not typically occurred between November and April. On

Figure 10–7 Bankfull discharge channel dimensions for the St. Joseph River (OH) upstream of the gage near Newville, IN

Regional curve
cross-sectional geometry
1,000.0
Channel cross-sectional area
Channel width A=6.84DA0.75
Channel depth r 2=0.93
Width, depth, and bank ht (ft)

100.0
Cross-sectional area (ft2)

10.0

1.0
W=5.34DA0.45 D=1.28DA0.29
r 2=0.96 r 2=0.82

0.1
0.1 1.0 10.0 100.0 1,000.0
Drainage area (mi2)

10–10 (210–VI–NEH, August 2007)

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

average, annually, there are slightly more than five The results of discharge versus recurrence interval
events, with an average duration of 8 days that exceed- estimation analysis are presented in table 10–1 for this
ed 1,330 cubic feet per second. example. The gage data results were obtained from
the regression equation. At the gage, the USGS Rural
A bankfull channel associated with very frequent flows method gave similar results to the gage data. However,
is consistent with observation by Ward, Mecklenburg, the USGS Urban method greatly overestimated the
and Brown (2002) in Wood County, Ohio. However, in discharges, even though an annual precipitation of
a recent study, they noted that typically, only about 10 only 32 inches was used, rather than the 34 to 35
percent of the sediment is transported by flows that inches suggested by the annual precipitation map for
are less than double the mean discharge. For most of Ohio (Ward and Trimble 2004). For the rural equation,
the gages, less than 25 percent of the sediment load a slope of 0.1 percent (5.2 ft/mi) and a storage value of
is transported by flows that are 3 to 5 times the mean 3 were used. At the Hillsdale ditch, the urban method
discharge. (Ward et al. 2003). For example: the mean also gave much higher estimates than the rural equa-
annual discharge for the St. Joseph gage is about 540 tion. However, if the urban method were calibrated
cubic feet per second, so it is probable that a discharge based on the ratio of the urban to gage data results,
of 1,330 cubic feet per second or higher corresponds the urban and rural methods gave similar results,
to the bankfull discharge at the gage. except for a recurrence interval of 2 years. It was
decided to base the analysis on the rural equation re-
At the gage, the river was very entrenched the top sults. For this ditch, knowledge of the actual discharge
of the bank corresponded to a stage of at least 16 to versus recurrence interval has little influence on the
18 feet (not measured). From further analysis, it was design. The ditch is extremely large, and regardless of
estimated that the out-of-bank discharge is 6,000 to what estimates are used, the out-of-bank discharge is
9,000+ cubic feet per second and corresponds to a 4- associated with a recurrence interval greater than 100
to 20-year recurrence interval flow. Therefore, at this years.
location, the behavior of the river is similar to that of
a ditch.

Table 10–1 Discharge vs. recurrence interval results at the gage and the Hillsdale ditch

St. Joseph gage Hillsdale ditch

Recurrence Gage USGS Urban USGS Rural USGS Urban Urban USGS Rural
interval (years) data method method method calibrated method
     2   4160   4902   5261    176    149    110
     5   6337   9735   7365     302    197    171
    10   7984 13920   8635    373    214    205
    25 10162 22495 10148    546    247    246
    50 11809 30121 11141    663    260    274
   100 13456 37488 12148    786    282    301

(210–VI–NEH, August 2007) 10–11

 Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

(e) Ditch geometry data are relative to an arbitrary datum. For most of the
ditch, the bed slope varied from 0.05 to 0.2 percent.
A survey of a 2,100-foot length of ditch was performed
on July 17, 2003. Station 1+00 is located 100 feet south A typical cross section is shown in figure 10–9, with
of the upstream bridge, and station 21+00 is located a possible new design with a 4:1 overtopped width to
near the southeast corner of the field on the right channel-width ratio. For a 4:1 ratio, the total width of
bank. This location is close to the Michigan-Ohio state the benches is three times the width of the bankfull
line and the point where the ditch enters a wooded channel. The top width of the bankfull channel is 10
area. feet, the mean depth is 1.8 feet, and the maximum
depth is 2.3 feet. Based on an analysis of the data, the
Working conditions in the ditch were difficult because maximum stable size of this channel might be 12.3 feet
of steep slopes, dense vegetation, and deep deposits wide, with a mean depth of 2.2 feet, and a maximum
of fine sediment in the bottom of the ditch. Therefore, depth of 2.8 feet. The channel has a 0.1 percent slope,
cross-sectional data were obtained by locating a per- the channel dimensions estimated from the regional
son with a rod and receiver on each side of the ditch curve (top width of 10 ft and maximum depth of 2.3
and stretching a tape between the two people. Each ft), and an over bench flow width to channel width
person took elevation and position data on their side ratio of 4:1. The 0.2-year recurrence interval discharge
of the ditch and part of the bankfull channel. Notes almost fills the small channel, the 1.6-year recurrence
were made indicating the location of the thalweg in interval discharge fills the channel to a depth of about
the bankfull channel, and the water depth. Thalweg 4.5 feet, and the stage for the 100-year discharge is just
data were then used in place of conducting a separate over 5.5 feet. For these conditions, fine sediment will
profile survey. Also, grade-break data and top-of-bank be flushed from the bankfull channel, and substrate
data were extracted from the cross-sectional data to with a mean size of about 3 to 4 millimeters will be
obtain profiles of these features. Profiles of the vari- established.
ous features are shown in figure 10–8. All elevation

Figure 10–8 Profiles of the bed, bench, and top of ditch

Drainage ditch in Hillsdale County

714

Top of ditch
712

710
Elevation (ft)

708
Bench or bankfull grade break
706

Channel bed
704

702

700
0 100 200 300 400 500 600 700 800 900 1,000 1,100 1,200 1,300 1,400 1,500 1,600 1,700 1,800 1,900 2,000 2,100
Station or channel distance (ft)

10–12 (210–VI–NEH, August 2007)

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

This ditch was cleaned out about 6 years prior to this counts for the last 200 feet (1,000–1,200 ft, fig 10–6a)
survey and only exhibited intermittent small bench had the coarsest substrate, widest benches, and nar-
formations along much of its length. There was up to rowest bankfull channel.
2 feet of sediment deposited on the bed in the first few
hundred feet, perhaps because of the culvert configu- The mean bed-material size is associated with the trac-
ration and a rapid change in bed elevation. Further tive force (mean shear stress) on the bed and can be
downstream from the bridge, the bench formations estimated as (Ward and Trimble 2004):
improved, the depth of the sediment deposits de-
D50 = 1000ds
creased, and in places (1,000 to 1,400 ft), clean, coarse
where:
substrate was observed in the bottom of the bankfull
D50 = particle size (mm)
channel.
d = flow depth (m)
s = bed slope (ft/ft)
(f) Bed material and bed-load transport
Therefore, a 0.6-meter (2.0 ft) bankfull discharge depth
The measured D50 and D84 for reaches 600 to 800 feet, in a channel with a bed slope of 0.1 percent might
800 to 1,000 feet, and 1,000 to 1,200 feet were <1 mil- result in a D50 of 6 millimeters. The bed slope varies
limeter, <1 millimeter and 12 millimeters, and 3 mil- from 0.05 to 0.2 percent, and the bankfull discharge
limeters and 10 millimeters, respectively. More than depth is 1.8 to 2.5 feet, so the probable D50 is about 3
80 percent of the bed material was clay and silt where to 13 millimeters. This is in good agreement with the
there were only small intermittent benches (600–800 measured D50 and D84 of 3 millimeters and 10 millime-
ft). In the next two reaches, the bench development ters in the channel at reach 1,000 to 1,200 feet, where
was more pronounced, and the main channel was nar- fluvial benches have formed.
rower, resulting in the coarse substrate sizes. Pebble

Figure 10–9 Pre- (dashed line) and post- (solid line) maintenance geometries at a location with a grade break, but weak
bench formation at the elevation where it was determined that a bench would naturally form. The existing
main channel has a similar geometry to the projected geometry.

712

711

710

709
Elevation (ft)

708

707

706

705

704

703

702
-40 -30 -20 -10 0 10 20 30
Width from left to right (ft)

(210–VI–NEH, August 2007) 10–13

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

While this equation is readily applied, it should be Vegetation is left along the fringe of the existing chan-
noted that it contains some inherent assumptions. Its nel, and no work is done to reshape or narrow the cur-
use assumes that sufficient coarse material is available rent channel. The benches will vegetate quickly, and it
to form the armor layer. If sufficient coarse material is is anticipated that the channel will adjust its shape as a
not available, then this approach may not be advisable. function of fluvial processes.

Actual bed-load transport is difficult to quantify be- A much debated and often controversial issue is the
cause of the complexity of the system and the lack of type of vegetation that should be established on the
any sediment transport data. Relative bed-load esti- benches and at the top of the ditches. Trees provide
mates were obtained by relating bed-load transport many benefits in natural stream systems and are par-
to the current channel conditions: a bed slope of 0.1 ticularly important for the aquatic biota. However, in
percent and a D50 of 2 millimeters. The geometry for straightened, channelized systems, grass might pro-
current conditions was approximated as a cross-sec- vide better overall benefits. Often trees will affect the
tional area that is three times the area predicted by ability of nature to establish stable benches, as much
the regional curve and an over-bench flow width to of the stability of these systems depends on the dense
channel width ratio of 1.5. The results of the bed-load grass cover that quickly establishes, in the absence of
transport analysis results are summarized in table trees. A way of viewing these systems is to think of the
10–2. It is anticipated that fluvial processes will estab- small bankfull channels as meadow streams (Rosgen
lish a coarse substrate with a mean particle size of 4 to type E channels) that lack the sinuosity that occurs in
8 millimeters, and bed-load transport will be less than natural systems. Therefore, trees will often provide the
half current rates. Following the flushing of deposited most benefit if they are set back from a grass buffer at
fines, and bench building by fluvial processes, the total the top of the two-stage system or at locations where
sediment export will primarily be a function of conser- there is a wide, well-attached flood plain—something
vation practices on the landscape and ditch instability that is rarely found in watersheds with extensive
problems upstream of this reach. networks of agricultural ditches. Constructing wide
benches with a 10:1 or larger flood-width ratio might
be considered, but that approach will be very expen-
(g) Discussion sive in locations where the main function of the ditch-
es is primarily to convey discharges from subsurface
In establishing two-stage geometry, the ditch is drainage systems. In those situations, the ditches must
widened at the elevation that corresponds to exist- be more than 5 feet deep and sometimes are more than
ing bench features or the elevation at which these 10 feet deep.
benches are predicted to form from fluvial processes.

Table 10–2 Relative bed-load transport for various channel conditions

Geometry Bed slope 0.05% Bed slope 0.1% Bed slope 0.2%
Relative D50 Relative D50 Relative D50
bed load (mm) bed load (mm) bed load (mm)
Current 0.36 2 1 2 2.45 2
3:1 bench ratio 0.06–0.18 4–6 0.28–0.49 6–8 1.03–1.38 8–10
4:1 bench ratio 0.04–0.16 4–6 0.25–0.45 6–8 0.96–1.30 8–10

10–14 (210–VI–NEH, August 2007)

Chapter 10 Two-Stage Channel Design Part 654
National Engineering Handbook

The primary costs of two-stage ditches are associ-

ated with the increased ditch width required. This 654.1008 Conclusion
increased width requires additional initial earthwork.
Costs for construction increase with both watershed
size and ditch depth and might range from $5 to $20 In channelized ditches and streams that are en-
per linear foot. trenched and have overwidened bed widths, alluvial
channel processes try to develop a flood plain that
Creating a low bench typically requires the top width consists of low benches. Often, these ditches show
of the ditch to be greater than what would be required improved stability. The techniques presented in this
for a traditional trapezoidal channel. It is important to chapter are based on observations and analysis of the
note that the wider ditch top width results in the sur- behavior and evolution of traditional trapezoidal earth
rendering of surrounding agricultural production land. channels.
To offset landowner costs, the potential for includ-
ing the bench width in buffer conservation programs The elevation of the benches and size of the bankfull
should be considered. Buffers have typically been channel can be determined from regional curves that
measured from the top of the ditch. Alternatively mea- relate channel dimensions to drainage area. The design
suring from the top of the small channel to include the approach considers the magnitude and design frequen-
bench and the main side slope of the ditch is prefer- cy of discharges for both stages of the channel. It is
able from a water-quality perspective and profitability anticipated that total bench widths that are two to four
perspective. times the bankfull channel width will result in a stable
geometry and a channel with low sinuosity. The overall
In many locations, a do nothing approach might be conveyance capacities of these two-stage systems can
considered. Removing benches will only provide an be sized based on the probability of out-of-ditch flood-
increase in the conveyance capacity of the ditch. ing into adjacent areas.
However, this improvement might only be temporary.
If subsurface drains are free flowing and not blocked Construction of a two-stage channel system requires a
by bench formations, constructing wider benches may significant capital investment to create a wider design
provide limited benefit and could disrupt a functional top width. However, it is anticipated that two-stage
system that currently provides aquatic and terrestrial systems will have improved conveyance capacity, be
habitat and water quality benefits. more self-sustaining, and create and maintain im-
proved aquatic habitat.

(210–VI–NEH, August 2007) 10–15