*The Reference Reach - a Blueprint for Natural Channel Design

by
Dave Rosgen, P.H.1

Abstract

The reference reach is used to develop natural channel design criteria based upon
measured morphological relations associated with the bankfull stage for a specific stable
stream type. Specific data on stream channel dimension, pattern and profile are collected
and presented by dimensionless ratios by stream type. The reference reach is a portion of
a river segment that represents a stable channel within a particular valley morphology.
The morphological data collected is used for extrapolation to disturbed or unstable
reaches in similar valley types for the purposes of restoration, stream enhancement,
stabilization, and stream naturalization schemes. Bankfull discharge and dimensions
from streamgage stations for particular hydro-physiographic provinces are correlated with
drainage area to develop regional curves for extrapolation to non-gaged reaches.
Regime equations often used for river restoration design represent data developed
empirically from a range of stream types. If the streams where the regime equations are
being implemented are not similar to the streams from which the equations were
developed, resultant designs can be incompatible with natural channel morphology. This
problem can be offset if the source of the empirical equations can be identified and
published by specific stream type. Reference reach data can be used to validate and sort
appropriate regime equations by stream type prior to implementation. Examples of field
methods, analytical procedures, and applications in natural channel design are presented

Introduction

To understand, predict and describe each complex process of the mutual

integration of the independent and dependent variables that shape and maintain the stream
channel in the present climate has challenged the river engineer for centuries. The
requirements of modern civilizations have placed great stress on rivers and even greater
demands on river engineers. The activities of river control, floodplain encroachment,
channelization, levee construction, and stabilization with concrete, rip-rap, interlocking
blocks, and other “hard” control measures have made canals out of natural rivers. The
loss of physical and biological function in these altered river systems have induced large-
scale adverse response by the general public throughout North America and portions of
Europe…they want their rivers back. However, they also want their rivers to be stable
and their homes to be protected from flooding. These new questions have put the river
engineer in a real dilemma due to the often, conflicting and competing objectives.

____________
1
Wildland Hydrology, 1481 Stevens Lake Rd. Pagosa Springs, Colorado, 81147
*Permission to reproduce granted by ASCE, July 24, 2001. From proceedings of the
Wetlands and Restoration Conference, March, 1998, Denver, Co.
 A majority of river engineering designs are based on clear water discharge,
rigid boundary theory, uniform flow, smooth beds, and uniform channel materials (Chow,
1959). These underlying assumptions are often violated in natural rivers, thus their
validity for use in natural channel design needs to be tested. Empirically derived regime
equations, often used to establish channel dimensions and slope, can be very appropriate
if the stream being restored is similar to the stream from which the relations have been
developed. In design manuals, however, it is difficult to determine the source of the
relation and/or the stream type(s) that a model represents. It is desirable then, to stratify
regime equations associated with the stream types from which they were derived.
Morphologically described stream types based on field measurements are
described in Rosgen (1994). An assortment of stream types are presented that are
delineated by slope, channel materials, width/depth ratio, sinuosity and entrenchment
ratio. The stream types are described at the morphological description level (level II) of
the hierarchical system for classification. At this level of inventory, the existing
dimension, pattern, profile, and materials are described. The descriptions however, do
not necessarily represent a stable form or describe the potential of the stream. An
assessment of condition or state is determined in level III and verified in level IV
(Rosgen, 1994,1996).
The use of a reference reach data base, characteristic of the stable channel
morphology for a similar valley type can provide an integrative approach which has the
stable dimension, pattern and profile to keep the stream from aggrading or degrading.
Morphological measurements must be obtained by unique stream types in order to
extrapolate these data. The use of a classification system for this purpose is essential in
order to group variables by morphological similarity and to reduce statistical variance
between the groups.
If the condition of the river being restored is extremely unstable, a dilemma
often exists in the selection of the potential stream type and the associated morphological
characteristics for a given flow and sediment regime, valley slope, and channel materials.
Reference reach data using dimensionless ratios can be used to establish design values as
long as the reference reach is representative of the same valley type and sediment regime
(see valley type descriptions, Rosgen, 1996). The reference reach data is not to be
confused with stream type data from level II analysis, since this data summarized from
various streams of the same type, does not always represent the stable form.
Other approaches to restoration design involve modeling sediment transport
to then back-calculate the corresponding effective discharge, depth and slope (Thomas, et
al, 1994). Any significant errors in the sediment transport model, however, can be
directly transferred to the corresponding design discharge, directly influencing the
dimensions and corresponding meander pattern and slope of the restored channel.

Stream stability

A “stable” stream, as used in this paper, is defined as: the ability of a

stream, over time (in the present climate), to transport the flows and sediment produced
by its watershed in such a manner that the dimension, pattern and profile are maintained
without either aggrading, nor degrading (Rosgen, 1996). The reference reach
characterizes the stable morphology, but does not necessarily require “pristine” or “relic”
reaches that are very rare to find. Verification of the stability of the reference reach is
done by procedures included in Level IV of the hierarchical river inventory (Rosgen,
1996). Aerial photographs can be used to provide additional evidence of stability by
depicting time-trends in comparing the morphological state prior to and following floods.
For example, a major flood estimated at a 10,000 year return period, occurred on Fall
River, in Rocky Mountain National Park, resulting from the breach of Lawn Lake. The
C4 stream type in Horseshoe Park was stable prior to the flood, and the author observed
this river the following day of the flood. The stream channel maintained the same pre-
flood dimension, pattern and profile. The channel did not aggrade nor degrade which met
the criteria of a stable channel. The width of the stream did not reflect the rare, large
flood, but maintained the width associated with the bankfull discharge. Fall River, thus
makes a good candidate for the reference reach database for a stable C4 stream type.

The Bankfull discharge

The use of bankfull discharge is similar to effective, dominant, and channel

forming streamflows. It is imperative that the selection of a reference discharge be
consistent among rivers in order to: a) classify streams; b) extrapolate morphological
relations from similar stream types but of different size; and c) to develop dimensionless
ratios from these relations. A design that emulates natural, stable channels, associated
with self-formed and self-maintained reference reaches, allows a determination of
dimension, pattern and profile using dimensionless ratios. Measurements of width and
depth to obtain width/depth ratio are associated with the dimensions corresponding to the
bankfull stage. Without a consistent reference discharge, dimensionless ratios for a
variety of stream types, could provide confusion among river engineers.
The bankfull discharge or the discharge associated with the stage at the
incipient point of flooding is a frequently occurring flow of moderate magnitude. The
author for a period of the last 12 years has visited approximately 10 USGS gage sites/year
throughout North America. The average return period from field calibration using
bankfull field indicators is 1.1 to 1.8 years. This includes ephemeral streams in Arizona,
spring-fed streams in Eastern Oregon, Streams in Delaware, Alabama, Maryland,
Montana, California, Wisconsin, Minnesota, North and South Dakota and many other
vastly different hydro-physiographic provinces. The field determination of bankfull
discharge using morphological indicators needs to be calibrated with measured values at
gaged sites. Once this is done, regional curves are developed using drainage area to
predict bankfull discharge and bankfull dimensions for each hydro-physiographic region
where precipitation/runoff relations vary. These curves are used to assist in bankfull
discharge determination in highly unstable systems where field evidence of bankfull is
extremely difficult to detect. Regional curves for bankfull discharge and bankfull cross-
sectional area versus drainage area have been recently developed by a diverse group of
field observers in various agencies and universities (Table 1). The excellent correlation
coefficients in these studies show the apparent application of using drainage area for
estimating bankfull discharge on unstable reaches and un-gaged rivers.
Table 1. Correlation coefficients for regional curve development using morphologically
determined bankfull stage for various hydro-physiographic provinces throughout the
United States.

Location/source of data Variable correlated with drainage Correlation coefficient (r2)

area
Upper San Juan River, Colorado, Bankfull discharge vs drainage 0.988
(Rosgen, 1997, research in area
progress)
Bankfull cross-section area 0.977
Bankfull width 0.885
Bankfull depth 0.599
Maryland Piedmont,1998, Bankfull discharge 0.905
(USFWS, R.Everitt and T.
McCandless, in progress)
Cross-section area 0.914
Bankfull width 0.837
Bankfull depth 0.779
North Central Kansas, 1997, Bankfull discharge 0.911
(Phil Balch, in progress)
Delaware, 1997 (R. Smith, in Bankfull cross-section area 0.956
progress)
Bankfull width 0.823
Bankfull depth 0.698
Arizona, Northern Arizona Bankfull discharge 0.954
Univ.,1998, (T.Moody and W.
Odem, in progress)
Central Oklahoma, 1997 Bankfull discharge 0.99
lowlands,USDA,NRCS
(R.Riley,et al, in progress)
Bankfull cross-section area 0.989
Bankfull width 0.841
Bankfull depth 0.884
Wisconsin, 1997, USDA,NRCS Bankfull cross-section area 0.866
(L. Steffen)
Bankfull width 0.913
Bankfull depth 0.798
The regional curves assist in providing a closer "bracket" for consistency of the design
discharge determination using bankfull discharge in the absence of gaged data,
inexperienced observers, and/or lack of visible bankfull indicators.

The similarity of bankfull and effective discharge, as originally proposed by

Wolman and Miller (1960), has been verified by Andrews (1980), Andrews and
Nankervis (1995) and, Batalla and Sala, (1995). The calculation of effective discharge
involves flow duration and sediment rating curves, however, effective discharge, using
this calculation would have a greater magnitude than the bankfull discharge in incising
channels due to a shift upward of the intercept and slope values of the sediment rating
curve. The authors field observations associated with incised channels show an increase
in sediment supply reflected in the sediment rating curve shift, due to accelerated stream
bank and bed erosion. Calculating a larger magnitude discharge than bankfull for design
purposes in the incising channel will result in increased shear stress, stream power and
localized boundary stress. If the source of the increased sediment is from upstream
supply, then routing (sediment transport capacity) versus stream bank and bed erosion
needs to be distinguished. If an effective discharge calculation using a “steeper” sediment
rating curve in an incised river is much larger than the bankfull discharge, the design flow
can result in accelerated channel erosion. Incised rivers often make it difficult for field
observers to detect bankfull indicators, thus regional curves, calibrated for the appropriate
hydro-physiographic province, are recommended to obtain the bankfull discharge.

Development of design criteria

One group of regime equations are the hydraulic geometry relations,

(Leopold and Maddock, 1953). Hydraulic geometry relations have been very useful in
describing width, depth cross-sectional area and velocity as power functions of stream
discharge. Where the stream types associated with a bankfull width/depth ratio, slope,
and channel materials are similar to the streams from which the hydraulic geometry were
derived, extrapolation of these relations is appropriate for design purposes. If stream
types vary significantly from the empirical relations, the reference reach method that
integrates morphological variables into dimensionless ratios for unique sets of stream
types can help reduce errors. An example of this application compares bankfull width,
depth, and slope for the same discharge for an E4 stream type (an alluvial, gravel bed,
highly sinuous, meandering stream with a width/depth ratio of 3, with a well developed
floodplain) compared to a C4 stream type (an alluvial, gravel bed, meandering stream
with a width/depth ratio of 20 with a well developed floodplain). For a discharge of 3
cms (100 cfs), the corresponding dimensions for the two alluvial, gravel bed stream types
are shown in Table 2. The contrasting dimensions for the same discharge for these two,
stable, meandering gravel bed streams would not be correctly identified using hydraulic
geometry relations, unless initially stratified by stream type. Additional examples are
shown in Rosgen, (1994).
Some regime equations predict slope, width, and depth using discharge and
dominant channel materials (USACOE, 1994). In the example of the C4 and E4 stream
types, both of which have gravel beds, the dimensions for the C4 stream type would have
been correctly predicted using these regime equations. The E4 stream type, however,
would not have been correctly predicted due to its’ low width/depth ratio. Regime
equations can be stratified by individual stream types to avoid this problem.
Selection criteria for reference reaches involves characterization from a
representative segment of a valley type similar to the disturbed design reach. Valley types
are similar in basin relief, depositional materials and features of the stream to be restored.
Quantitative morphological data is collected for the reference reach, then converted to
dimensionless ratios by dividing the dimension, pattern and profile variables by the
bankfull values of the same feature. The purpose of the dimensionless ratios are to
calculate actual design values for width, depth, meander length, radius of curvature, pool
depth, pool slope, cross-sectional area of riffles and pools, riffle slope, maximum riffle
depth and many other channel properties.

Table 2. Relations of width, depth, velocity and cross-sectional area to discharge for two
different gravel bed streams for a bankfull discharge of 3.0 cubic meters/sec (100 cfs).

Variables E4 stream type C4 stream type

Bankfull surface width 2.2 meters (7.1 ft.) 7.3 meters (23.9 ft.)

Bankfull mean depth (meters) 0.73 (2.4 ft.) 0.37 meters (1.2 ft.)

Width/depth ratio 3 20

Bankfull mean velocity (meters/sec.) 1.8 meters/sec (6 ft./sec.) 1.1 meters/sec (3.5 ft./sec.)

Bankfull cross-sectional area (square 1.5 square meters (16.7 ft.2) 2.7 square meters (28.6 ft.2 )
meters)

Valley slope .010 .010

Channel sinuosity 1.4 2.5

Channel slope .007 .004

The dimensionless ratio values (Table 3) can be obtained directly and

applied to streams of various sizes and bankfull discharge, but of the same stream type.
For example, the bankfull width for a stable C4 stream type is calculated by the square
root of the product of width/depth ratio of the stable reference reach stream type times the
cross-sectional area at the bankfull stage. Mean bankfull depth is calculated by dividing
bankfull surface width by width/depth ratio. Additional examples of the data collected
and variables computed from reference reach data are shown in Table 3.

Procedural rules for data collection and documentation of the reference reach
• verify bankfull discharge with regional curves from gage station data
• stratify by representative valley type (width, valley slope, same channel
materials, landform/landtype association)
• stratify by morphological stream type
• be stable (in equilibrium or in regime) but not required to be pristine
• have at least two full meander wavelengths, or 20 widths of length of
consistency for measurements
• be free to adjust channel boundaries for the frequent high flows
• select cross-sections and long. Profile to represent typical bed features
• establish range of values as well as average for “natural variance”
• if located at gage station, work up hydraulic geometry by stream type
• complete a level III, (Rosgen, 1996) condition assessment of stream type
Table 3. Reference reach data and example computations for design.

Morphological Morphological Morphological Morphological Morphological

Variables Variables Variables Variables Variables
1}Bankfull width 10)Channel 19)Riffle slope 28) Glide slope 37) Ratio of run
materials:D15 depth/mean depth
D35,D50,D84,D95,
D100
2)Bankfull mean 11)Bar material: 20) Ratio riffle 29) Ratio of glide 38) Run w/d
depth D15, D35,D50, D84, slope/ave. slope slope/ave. slope
D95, D100
3)Width/depth ratio 12} Stream type 21) Riffle max. depth 30) Glide depth 39) Ratio of run
(1)/(2) ratio (4)/(2) w/d/ave. w/d
4)Bankfull max. 13) Bankfull cross- 22) Pool slope 31) Ratio glide 40)Meander length
depth section area (1)x(2) depth/mean depth (Lm )
5)Width of 14) Drainage area 23) Ratio of pool 32) Glide w/d ratio 41) Ratio of
floodprone area slope/ave. slope Lm/bankfull width
6)Entrenchment ratio 15) wetted perimeter 24) Max. pool depth 33) Ratio of glide 42) Radius of curvature
(5)/(1) (P) w/d/riffle w/d (Rc)
7) valley slope 16) Hydraulic radius 25) Ratio of max. 34) Run slope 43) Ratio of Rc/
(13)/(15) depth/mean depth bankfull width
8) Average water 17) Bankfull velocity 26) W/D ratio of 35) Ratio of run 44) Belt width
surface slope pool slope/ave. slope
9) Sinuosity (7)/(8) 18) Bankfull 27) Ratio pool 36) Run depth 45) Meander width ratio
chan.Length/valley L discharge w/d/riffle w/d (Wblt/Wbkf)

Restoration Concepts

Restoration of natural stable channels, as used in this paper, is defined as

the establishment of the dimension, pattern and profile of the appropriate stable stream
type in order to restore it’s physical and biological function. Observations are necessary
of the cause of the instability, and the current state of the existing stream type. Knowing
where the stream is in the evolutionary sequence of channel adjustment, often helps in
selecting the appropriate stable stream type. The evolution model proposed by Hupp and
Simon (1991) can be compared to the quantitative, morphological relations by stream
type shown in association with the various phases of channel evolution (Rosgen, 1996).
Restoration often helps “speed up” the process of evolution, toward a faster recovery to
the stable form. Channel form as described using dimensionless ratios of the reference
reach has evolved over time, resultant of an integration of a channel forming discharge
(bankfull) and sediment regime. The basis for this type of natural channel design utilizes
this assumption

Summary

The morphological variables used to establish design dimensions, patterns

and the profile for natural channel design can be developed from stable reference reaches.
Due to the large number of variables and their wide variability, stratification by stream
type assists in the extrapolation of dimensionless ratios to streams being restored.
Caution should be used in applying regime equations, unless the empirical relations have
been stratified by the various stream types that the restoration represents.
The advantage of using the stable reference reach for design criteria is the
integration of the dependent variables of the streams dimension, pattern and profile with
the independent variables of streamflow, sediment regime, channel materials and valley
slope. The stream has evolved, over time, to have just the slope, width, depth and other
morphological features to transport the flows and sediment produced by its’ watershed
without either aggrading or degrading. The design engineer must be able to “read the
river” and select design specifications compatible with the natural stable channel.

References

Andrews, E.D. (1980). Effective and bankfull discharges of streams in the Yampa river basin, Colorado and
Wyoming. Journal of Hydrology, 46, 311-110.

Andrews, E.D. and Nankervis, J.M. (1995). Effective discharge and the design of channel maintenance
flows for gravel-bed rivers. In: Natural and Anthropogenic Influences in Fluvial Geomorphology,
Geophysical Monograph 89, American Geophysical Union.

Balch, P. (1997). Regional curve development using bankfull discharge from morphological field evidence
at gage stations. Information developed by Kansas Department of Environment, Topeka, Ks. Personal
communication.

Batalla, R.J. and Sala, M. (1995). Effective discharge for bedload transport in a sub-humid Mediterranean
sandy gravel-bed river ( Arbucies, North-East Spain). River Geomorphology, Ed. E. Hickin, John
Wiley and Sons, Ltd.

Chow, V.T. (1959). Open channel hydraulics. London: McGraw-Hill.

Everitt, R. and McCandless, T. (1998). Regional curve development using bankfull discharge from
morphological field evidence at gage stations. Draft information developed by USFWS under contract
with Maryland Department of Highways, Baltimore, Maryland, Personal communication.

Hupp and Simon (1991). “Bank accretion and the development of vegetated depositional surfaces along
modified alluvial channels”, Geomorphology, Vol. 4, pp. 111-124.

Moody, T. and W. Odem (1998). Regional curve development using bankfull discharge from
morphological field evidence at gage stations.in Arizona . Draft information developed by Northern
Arizona University, Flagstaff, Az. Personal communication.

Rosgen, D.L. (1994). A classification of natural rivers. Catena, Vol. 22, 169-199. Elsevier Science, B.V.
Amsterdam.

Rosgen, D.L. (1996). Applied River Morphology. Wildland Hydrology Books, Pagosa Springs, Colo.

Riley, R. (1997). Regional curve development using bankfull discharge from morphological field evidence
at gage stations. Draft information developed by USDA Natural Resources Conservation Service,
Watershed Science Institute, North Carolina State University, Raleigh, N.C. Personal communicatio
Smith, R. (1997). Regional curve development using bankfull discharge from morphological field evidence
at gage stations in Delaware, State of Delaware and USDA,NRCS, personal communication
Steffen, L.J. (1997). Regional curve development using bankfull discharge from morphological field
evidence at gage stations. In Wisconsin. Draft information developed by USDA Natural Resources
Conservation Service, Lincoln, Nebraska, personal communication.

Thomas, W.A., Copeland, R.R.,Raphelt, N.K. and McComas, D.N. (1994). “User’s manual for the
hydraulic design for channels (SAM)” (in preparation), U.S Army Engineer Waterways
Experiment Station, Vicksburg, Ms.

Wolman, M.G. and Miller, J.P. (1960). Magnitude and frequency of forces in geomorphic processes.
Journal of Geology, 68, 54-74.

Leopold, L.B. and Maddock, T. (1953). The hydraulic geometry of stream channels and some
physiographic implications. U.S. Geol. Survey, Professional Paper, 252, 56 pp.

United States Army, Corps of Engineers, (1994). Channel stability assessment for flood control projects,
Manual No. 1110-2-1418, Wash. D.C.