CATENA

Catena 22 (1994) 1699199

A classification of natural rivers

David L. Rosgen
Wildland Hydrology, I Steven’s Lake Road, Pagosa Springs, CO 81147, USA

Abstract

A classification system for natural rivers is presented in which a morphological arrangement

of stream characteristics is organized into relatively homogeneous stream types. This paper
describes morphologically similar stream reaches that are divided into 7 major stream type
categories that differ in entrenchment, gradient, width/depth ratio, and sinuosity in various
landforms. Within each major category are six additional types delineated by dominate channel
materials from bedrock to silt/clay along a continuum of gradient ranges. Recent stream type
data used to further define classification interrelationships were derived from 450 rivers
throughout the U.S, Canada, and New Zealand. Data used in the development of this classi-
fication involved a great diversity of hydro-physiographic/geomorphic provinces from small to
large rivers and in catchments from headwater streams in the mountains to the coastal plains. A
stream hierarchical inventory system is presented which utilizes the stream classification system.
Examples for use of this stream classification system for engineering, fish habitat enhancement,
restoration and water resource management applications are presented. Specific examples of
these applications include hydraulic geometry relations, sediment supply/availability, fish
habitat structure evaluation, flow resistance, critical shear stress estimates, shear stress/velocity
relations, streambank erodibility potential, management interpretations, sequences of morpho-
logical evolution, and river restoration principles.

1. General statement

It has long been a goal of individuals working with rivers to define and understand
the processes that influence the pattern and character of river systems. The differences
in river systems, as well as their similarities under diverse settings, pose a real
6 challenge for study. One axiom associated with rivers is that what initially appears
complex is even more so upon further investigation. Underlying these complexities is
an assortment of interrelated variables that determines the dimension, pattern, and
_- profile of the present-day river. The resulting physical appearance and character of
the river is a product of adjustment of its boundaries to the current streamflow and
sediment regime.
0341-8162/94/$07.00 0 1994 Elsevier Science B.V. All rights reserved
SSDZO341-8162(94)EOO46-3
170 D.L. Rosgen 1 Catena 22 (1994) 169-199

River form and fluvial process evolved simultaneously and operate through mutual
adjustments toward self-stabilization. Obviously, a classification scheme risks over-
simplification of a very complex system. While this may appear presumptuous, the
effort to categorize river systems by channel morphology is justified in order to
achieve, to some extent, the following objectives:
1. Predict a river’s behavior from its appearance;
2. Develop specific hydraulic and sediment relations for a given morphological
channel type and state;
3. Provide ‘a mechanism to extrapolate site-specific data collected on a given
stream reach to those of similar character;
4. Provide a consistent and reproducible frame of reference of communication
for those working with river systems in a variety of professional disciplines. r

2. Stream classification review

A definition of classification was offered by Platts (1980) where “classification in

the strictest sense means ordering or arranging objects into groups or sets on the basis
of their similarities or relationships.” The effort to classify streams is not new. Davis
(1899) first divided streams into three classes based on relative stage of adjustment:
youthful, mature, and old age. Additional river classification systems based on qua-
litative and descriptive delineations were subsequently developed by Melton (1936)
and Matthes (1956).
Straight, meandering, and braided patterns were described by Leopold and
Wolman (1957). Lane (1957) developed quantitative slope-discharge relationships
for braided, intermediate, and meandering streams. A classification based on
descriptive and interpretive characteristics was developed by Schumm (1963) where
delineation was partly based on channel stability (stable, eroding, or depositing) and
mode of sediment transport (mixed load, suspended load, and bedload).
A descriptive classification was also developed by Culbertson et al. (1967) that
utilized depositional features, vegetation, braiding patterns, sinuosity, meander
scrolls, bank heights, levee formations, and floodplain types. Thornbury (1969)
developed a system based on valley types. Patterns were described as antecedent,
superposed, consequent, and subsequent. The delineative criteria of these early
classification systems required qualitative geomorphic interpretations creating
delineative inconsistencies. Khan (197 1) developed a quantitative classification for
sand-bed streams based on sinuosity, slope, and channel pattern.
To cover a wider range of stream morphologies, a descriptive classification scheme
was developed for and applied on Canadian Rivers by Kellerhals et al. (1972, 1976),
Galay et al. (1973), and Mollard (1973). The work of these Canadian researchers
provides excellent description and interpretation of fluvial features. This scheme has
utility both for aerial photo delineation and for describing gradual transitions
between classical river types. and to date offers the most detailed and complete list
of channel and valley features. The large number of possible interpretative
 D.L. Rosgen / Catena 22 (1994) 169-199 171

delineations, however, makes this scheme quite complex for general planning
objectives.
An attempt to classify rivers in the great plains region using sediment transport,
channel stability, and measured channel dimensions was developed by Schumm
(1977). Classifying stream systems on the basis of stability is often difficult because
of the qualitative criteria can vary widely among observers leading to inconsistencies
in the classification. Similarly, data on ratio of bedload to total sediment load as
needed in this classification, while useful, often is not readily available to those
who need to classify streams.
Brice and Blodgett (1978) described four channel types of: braided, braided point-
bar, wide-bend point-bar, and equi-width point-bar. A descriptive inventory of
alluvial river channels is well documented by Church and Rood (1983). This data
set can be very useful for many purposes including the grouping of rivers based on
similar morphological characteristics. Nanson and Croke (1992) presented a classifi-
cation of flood plains that involved particle size, morphology of channels, and bank
materials. This classification has some of the same criteria of channel type as pre-
sented in this paper, but is restricted to flood plains. Pickup (1984) describes the
relation of sediment source and relative amounts of sediment to various aspects of
river type, but is not a classification of channels. Recent documentation by Selby
(1985) showed a relationship between the form and gradient of alluvial channels and
the type, supply and dominant textures (particle sizes) of sediments. This relationship
utilizes the Schumm (1977) classification in that an increase in the ratio of bed
material load to total sediment load with a corresponding increase in channel
gradient leads to a decrease in stability causing channel patterns to shift from a
meandering to braided channel form. In his classification, Selby (1985) treats anasto-
mosed and braided channel patterns similarly. However, the anastomosed rivers are
not similar to braided rivers in slope, adjustment processes, stability, ratio of bed
material to total load or width/depth ratios as shown by (Smith and Smith, 1980).
Typically, theoretically derived schemes, often do not match observations. To be
useful for extrapolation purposes, restoration designs, and prediction, classification
schemes should generally represent the physical characteristics of the river. With
certain limitations, most of these classification and/or inventory systems met the
objectives of their design. However, the requirement for more detailed, repro-
ducible, quantitative applications at various levels of inventory over wide hydro-
physiographic provinces has led to further development of classification schemes.

2. Stream classification concepts

The morphology of the present day channel is governed by the laws of physics
through observable stream channel features and related fluvial processes. Stream
pattern morphology is directly influenced by eight major variables including channel
width, depth, velocity, discharge, channel slope, roughness of channel materials,
sediment load, and sediment size (Leopold et al., 1964). A change in any one of
these variables sets up a series of channel adjustments which lead to a change in
Table 1
Hierarchy of river inventories

Level of detail Inventory description Information required Objectives

I Broad morphological Landform, lithology, soils, climate, depositional To describe generalized fluvial features using
characterization history, basin relief, valley morphology, river remote sensing and existing inventories of
profile morphology, general river pattern geology, landform evolution, valley morphology.
depositional history and associated river slopes,
relief and patterns utilized for generalized
categories of major stream types and associated
interpretations.

II Morphological Channel patterns, entrenchment ratio, width/ This level delineates homogeneous stream types
description (stream depth ratio, sinuosity, channel material, slope that describe specific slopes, channel materials,
types) dimensions and patterns from “reference reach”
measurements. Provides a more detailed level of
interpretation and extrapolation than Level I.

Ill Stream “state” or Riparian vegetation, depositional patterns, The “state” of streams further describes existing
condition meander patterns, confinement features, fish conditions that influence the response of channels
habitat indices, flow regime, river size category, to imposed change and provide specific
debris occurrence, channel stability index, bank information for prediction methodologies (such
erodibility as stream bank erosion calculations, etc.).
Provides for very detailed descriptions and
associated prediction/interpretation.

IV Verification Involves direct measurements/observations of Provides reach-specific information on channel

sediment transport, bank erosion rates, processes. Used to evaluate prediction
aggradation/degradation processes, hydraulic methodologies; to provide sediment, hydraulic
geometry, biological data such as fish biomass, and biological information related to specific
aquatic insects, riparian vegetation evaluations, stream types; and to evaluate effectiveness of
etc. mitigation and impact assessments for activities
by stream type.
 D.L. Rosgen / Catena 22 (1994) 169-199 173

the others, resulting in channel pattern alteration. Because stream morphology is the
product of this integrative process, the variables that are measurable should be used
as stream classification criteria.
The directly measurable variables that appear from both theory and experience to
govern channel morphology have been included in the present classification proce-
dure. These “delineative criteria” interact with one another to produce a stream’s
dominant features.
The present classification system has evolved from field observation of hundreds of
rivers of various sizes in all the climatic regions of North America, experience in
stream restoration, extensive teaching, and practical applications of the classification
system by many hydrologists, geomorphologists, fisheries experts, and plant
ecologists. Initial efforts to develop the classification procedure began in 1973, and
a preliminary version was presented to the scientific community (Rosgen, 1985). The
present paper includes notational changes from the earlier publication.

3. Stream classification system

The classification of rivers is an organization of data on stream features into

discreet combinations. The level of classification should be commensurate with the
initial planning level objective. Because these objectives vary, a hierarchy of stream
classification and inventories is desirable because it allows an organization of stream
inventory data into levels of resolution from very broad morphological characteri-
zations to discreet, measured, reach-specific descriptions. Each level should include
appropriate interpretations that match the inventory specificity. Further, general
descriptions and characteristics of stream types should be able to be divided into
even more specific levels. The more specific levels should provide indications of
stream potential, stability, existing “states”, etc., to respond to higher resolution
data and interpretations when planning needs change. A proposed stream inventory
system, including an integrated stream classification, is shown in Table 1.
Current river “state” and influences on the modern channel by vegetation, flow
regime, debris, depositional features, meander patterns, valley and channel confine-
ment, streambank erodibility, channel stability, etc., comprise additional parameters
that are considered critical to evaluate by stream type at a more detailed inventory
level (Level III). However, for the sake of brevity and clarity, this paper will focus on
the first two levels, the broad geomorphic characterization (Level I) and the morpho-
logical description (Level II) which incorporates the general character of channel
form and related interpretations. Portions of the data used for detailed assessment
levels are contained in the sub-type section of the earlier classification paper (Rosgen,
1985).

4.1. Geomorphic characterization (level I)

The purpose of delineation at this level is to provide a broad characterization that

integrates the landform and fluvial features of valley morphology with channel relief,
 D.L. Rosgen / Catena 22 (1994) 169-199

. I

Fig. 1. Longitudinal, cross-sectional and plan views of major stream types.

pattern, shape, and dimension. Level I combines the influences of climate,

depositional history, and life zones (desert shrub, alpine, etc.) on channel
morphology.
The presence, description, and dimensions of floodplains, terraces, fans, deltas and
outwash plains are a few examples of valley features identified. Depositional and
erosional history overlay channel patterns at this level. Generalized categories of
“stream types” initially can be delineated using broad descriptions of longitudinal
profiles, valley and channel cross-sections, and plan-view patterns (see Fig. 1 and
Table 2).

Longitudinal profiles
The longitudinal profile, which can be inferred from topographic maps, serves as
the basis for breaking the stream reaches into slope categories that reflect profile
morphology. For example, the stream types of Aa + (Fig. 1) are very steep, (greater
than 10%), with frequently spaced, vertical drop/scour-pool bed features. They tend
to be high debris transport streams, waterfalls, etc. Type A streams are steep (4-10%
slope), with steep, cascading, step/pool bed features. Type B streams are riffle-
dominated types with “rapids” and infrequently spaced scour-pools at bends or areas
of constriction. The C, DA, E and F stream types are gentle-gradient riffle/pool types.
Type G streams are “gullies” that typically are step/pool channels. Finally, the D type
streams are braided channels of convergence/divergence process that lead to
localized, frequently spaced scour/depositional bed forms.
Bed features are consistently found to be related to channel slope. Grant et al.
(1990) described bed features of pools, riffles, rapids, cascades, and steps as a function
 D.L. Rosgen / Catena 22 (1994) 169-199 175

Fig. 2. Relationship of bed slope to bed forms for various stream types (from Grant et al., 1990)

of bed-slope gradient. Using their bed form descriptions, the above described stream
types were plotted against the corresponding slope ranges reported by Grant et al.
(1990). “Groupings”, (Fig. 2), were apparent for riffle/pool stream types (C, E, and F)
at less than 2%) rapids at 2-4% in “B” and “G”, cascades in slopes 4- 10% in type A
streams, and steps for slopes 4-40% in types A and Aa + streams. Because gradient
and bed-feature relationships are integral to the delineation of stream type categories,
“stream types” are more than just “arbitrary units”. Bed morphology can be pre-
dicted from stream type by using bed-slope indices.

Cross-section morphology
The shape of the cross-section that would indicate a narrow and deep stream as
opposed to a wide and shallow one can be inferred at this broad level. The manner in
which the channel is incised in its valley can also be deduced at this level as well as
information concerning floodplains, terraces, colluvial slopes, structural control
features, confinement (lateral containment), entrenchment (vertical containment),
and valley vs. channel dimension. For example, the type A streams are narrow,
deep, confined, and, entrenched. The width of the channel and valley are similar.
This contrasts with type C streams, where the channel is wider and shallower with a
well-developed floodplain and a very broad valley. Type E streams have a narrow and
deep channel (low width/depth ratio) but have a very wide and well developed flood-
plain. Type F streams have wide and shallow channels, but are an entrenched
meandering channel type with little to no developed floodplain. Type G channels
have low width/depth ratio channels similar to type E streams except they are well
entrenched (no floodplain), are steeper, and less sinuous than type E streams (see
Fig. 1).

Plan view morphology

The pattern of the river is classed as relatively straight (A stream types), low
sinuosity (B stream types), meandering (C stream types), and tortuously meandering
Table 2 b
Summary of delineative criteria for broad-level classification b
k
Stream General description Entrenchment W/D Sinuosity Slope Landform/soils/features z
type ratio ratio 39

Aa+ Very steep, deeply entrenched, debris < 1.4 < 12 1.0 to 1.1 > 0.10 Very high relief. Erosional, bedrock or depositional F
transport streams. features; debris flow potential. Deeply entrenched g
streams. Vertical steps with/deep scour pools; 2
waterfalls. 2
E
A Steep, entrenched, cascading, step/pool < 1.4 < 12 1 .O to 1.2 0.04 to 0.10 High relief. Erosional or depositional and bedrock 5
streams. High energy/debris transport forms. Entrenched and confined streams with
5
associated with depositional soils. Very cascading reaches. Frequently spaced, deep pools in I
stable if bedrock or boulder dominated associated step-pool bed morphology. s
channel.

B Moderately entrenched, moderate 1.4 to 2.2 > 12 > 1.2 0.02 to 0.039 Moderate relief, colluvial deposition and/or residual
gradient, riffle dominated channel, with soils. Moderate entrenchment and W/D ratio.
infrequently spaced pools. Very stable Narrow, gently sloping valleys. Rapids predominate
plan and profile. Stable banks. with occasional pools.

C Low gradient, meandering, point-bar, > 2.2 > 12 > 1.4 < 0.02 Broad valleys with terraces, in association with
riffle/pool, alluvial channels with broad, floodplains, alluvial soils. Slightly entrenched with
well defined floodplains well-defined meandering channel. Riffle-pool bed
morphology.
D Braided channel with longitudinal and n/a > 40 n/a < 0.04 Broad valleys with alluvial and colluvial fans.
transverse bars. Very wide channel with Glacial debris and depositional features. Active
eroding banks. lateral adjustment, with abundance of sediment
supply.

DA Anastomosing (multiple channels) > 4.0 < 40 variable < 0.005 Broad, low-gradient valleys with fine alluvium and/
narrow and deep with expansive well or lacustrine soils. Anastomosed (multiple channel)
vegetated floodplain and associated geologic control creating fine deposition with well-
wetlands. Very gentle relief with highly vegetated bars that are laterally stable with broad
variable sinuosities. Stable streambanks. wetland floodplains.

Low gradient, meandering riffle/pool > 2.2 < 12 > 1.5 < 0.02 Broad valley/meadows. Alluvial materials with
stream with low width/depth ratio and floodplain. Highly sinuous with stable, well
little deposition. Very efficient and stable. vegetated banks. Riffle-pool morphology with very
High meander width ratio. low width/depth ratio.

Entrenched meandering riffle/pool < 1.4 > 12 > 1.4 < 0.02 Entrenched in highly weathered material. Gentle
channel on low gradients with high gradients, with a high W/D ratio. Meandering,
width/depth ratio. laterally unstable with high bank-erosion rates.
Riffle-pool morphology.

Entrenched “gulley” step/pool and low < 1.4 < 12 > 1.2 0.02 to 0.039 Gulley, step-pool morphology with moderate slopes
width/depth ratio on moderate and low W/D ratio. Narrow valleys, or deeply
gradients. incised in alluvial or colluvial materials; i.e., fans or
deltas. Unstable, with grade control problems and
high bank erosion rates.
178 D.L. RosEen I Catena 22 (1994) 169-199

STREAM TYPE

1.1 / 3.7

L 1-2
I
1 2-8

Fig. 3. Meander width ratio (belt width/bankfull

2-10 /4-20 / 20-40

width) by stream type categories.

1

(E stream types). Complex stream patterns are found in the multiple channel, braided
(D) and anastomosed (DA) stream types. Sinuosity can be calculated from aerial
photographs and often, like slope, serves as a good initial delineation of major
stream types. These river patterns have integrated many processes in deriving their
present form and thus, provide interpretations of their associated morphology.
Even at this broad level of delineation, consistency of dimension and associated
pattern can be observed by broad stream types. Meander width ratio (belt width/
bankfull surface width) was calculated by general categories of stream types for a
wide variety of rivers. Measured mean values and ranges by stream type are shown in
Fig. 3. Early work by Inglis (1942) and Lane (1957) discussed meander width ratio but
the values were so divergent among rivers that the ratio appeared to have little value.
When stratified by general stream types, however, the variability appears to be
explained by the similarities of the morphological character of the various stream
types. This has value not only for classification and broad-level delineations, but also
for describing the most probable state of channel pattern in stream restoration work.

Discussion
Interpretations of mode of adjustment - either vertical, lateral, or both - and
energy distribution can often be inferred in these broad types. Many variables that are
not discrete delineative variables integrate at this level to produce an observable
morphology. A good example of this is the influence of a deep sod-root mass on
type E streams that produces a low width/depth ratio, low meander length, low radius
of curvature, and a high meander width ratio. Vegetation is not singled out for
mapping at this level, but is implicit in the resulting morphology. If this vegetation
is changed, the width/depth ratio and other features will result in adjustments to the
 D.L. Rosgen / Catena 22 (1994) 169-199

SIN. a.2 x.2 >l.4 <l.l l.l-I.6 >1.5 H.4 >1.2

W/D (12 >12 >12 >40 <40 <12 >12 <I2
SLOPE .04-,099 .02-.039 co2 <.02 <.005 <.02 <.02 .02-.039

Fig. 4. Illustrative guide showing cross-sectional configuration, composition and delineative criteria of
major stream types.

type C stream morphology. Detailed vegetative information, however, is obtained at

the channel state level (Level III, Table 1).
Delineating broad stream types provides an initial sorting within large basins and
allows a general level of interpretation. This leads to organization and prioritization
for the next more detailed level of stream classification.

4.2. The morphological description (level II)

General description
This classification scheme is delineated initially into the major, broad, stream
categories of A-G as shown in Fig. 1 and Table 2. The stream types are then broken
into discreet slope ranges and dominant channel-material particle sizes. The stream
types are given numbers related to the median particle size diameter of channel
materials such that 1 is bedrock, 2 is boulder, 3 is cobble, 4 is gravel, 5 is sand, and
6 is silt/clay. This initially produces 42 major stream types as shown in (Fig. 4).
A range of values for each criterion is given in the key to classification for 42 major
stream types (Fig. 5). The range of values chosen to represent each delineative
criterion is based on data from a large assortment of streams throughout the United
States, Canada and New Zealand. A recent data set of 450 rivers was statistically used
to refine and test previous ranges of delineative criteria as described in the author’s
earlier publication (Rosgen, 1985).
Histograms were drawn of the distribution of values of each delineative criterion
for each channel type. From the histograms of 5 criteria for 42 major stream types, the
mean and “frequent range” of values were recorded. The most frequently observed
values seemed to group into a recognizable “river form” or morphology. When values
180 D.L. Rosgen / Catena 22 (1994) 169-199
- -
I SlNGLE lwE*Ll CHINNELS

‘Entrenchment SLIGHTLI EllTRErnHED ,.I P

*‘WidthlDepth2

‘Sinuosity’

cl c3

1 values can valy by * 0.2 units as a function of the continuum of physical variables within stream reaches.
2 values can vary by f: 2.0 units as a function of the continuum of physical variables within Stream reaches.

Fig. 5. Key to classification of natural rivers.

were outside of the range of the “most frequently observed” condition, a distinctly
different morphology was identified. As a result, the delineation of unique stream
types representing a range of values amongst several variables were established. These
variables and their ranges make up the current morphological description of stream
types as shown in Figs. 4 and 5.
The classification can be applied to ephemeral as well as perennial channels with
little modification. Bankfull stage can be identified in most perennial channels
through observable field indicators. Although, these bankfull stage indicators, are
often more elusive in ephemeral channels.
The morphological variables can and do change even in short distances along a
river channel, due to such influences of change as geology and tributaries. Therefore,
the morphological description level incorporates field measurements from selected
reaches, so that the stream channel types used here apply only to individual reaches of
channel. Data from individual reaches are not averaged over entire basins to describe
stream systems. A category may apply to a reach only a few tens of meters or may be
applicable to a reach of several kilometers.
Data is obtained from field measurements of representative or “reference reaches.”
The resultant stream type as delineated can then be extrapolated to other reaches
where detailed data is not readily available. In similar valley and lithological types,
stream types can often be delineated using these reference reaches through the use of
aerial photos, topographic maps, etc.

Continuum concept
When the variables which make up the range of values within a stream type change,
 D.L. Rosgen / Catena 22 (1994) 169-199 181

there is more often than not, a change in stream type. The ranges in slope, width/
depth ratio, entrenchment ratio and sinuosity shown in Fig. 4 span the most fre-
quently observed values. Exceptions occur infrequently, where values of one variable
may be outside of the range for a given stream type.
This level recognizes and describes a continuum of river morphology within and
between stream types. The continuum is applied where values outside the normal
range are encountered but do not warrant a unique stream type. Often the general
appearance of the stream and the associated dimensions and patterns of the stream do
not change with a minor value change in one of the delineative criteria. For example,
slope values as shown in Fig. 5, using the continuum concept, are not “lumped”, but
rather are sorted by sub-categories of: a+ (steeper than O.lO), a (0.04-O.lO), b (0.02-
0.039), c (less than 0.02) and c- (less than 0.001).
The application of this concept allows an initial classification of a C4 stream type (a :
gravel bed, sinuous, high width/depth ratio channel with a well-developed floodplain.
If the slope of this stream was less than 0.00 1, then the stream type would be a C4c-.
Rivers do not always change instantaneously, under a geomorphic exceedance or
“threshold”. Rather, they undergo a series of channel adjustments over time to
accommodate change in the “driving” variables. Their dimensions, profile and
pattern reflects on these adjustment processes which are presently responsible for
the form of the river. The rate and direction of channel adjustment is a function of
the nature and magnitude of the change and the stream type involved. Some streams
change very rapidly, while others are very slow in their response.

Delineative criteria
At this level of inventory each reach is characterized by field measurements and
validation of the classification. The delineation criteria and ranges for various stream
types are shown in Fig. 5. This classification key also represents the sequential process
for classification. The classification process starts at the top of the chart (single or
multiple thread channels), and proceeds downward through channel materials and
slope ranges.

Entrenchment
An important element of the delineation is the interrelationship of the river to its
valley and/or landform features. This interrelationship determines whether the river is
deeply incised or entrenched in the valley floor or in the deposit feature. Entrench-
ment is defined as the vertical containment of river and the degree to which it is incised
in the valley floor (Kellerhals et al., 1972). This makes an important distinction’of
whether the flat adjacent to the channel is a frequent floodplain, a terrace (abandoned
floodplain) or is outside of a flood-prone area. A quantitative expression of this
feature, “entrenchment ratio” was developed by the author so that various mappers
could obtain consistent values. The entrenchment ratio is the ratio of the width of the
flood-prone area to the bankfull surface width of the channel. The flood-prone area is
defined as the width measured at an elevation which is determined at twice the
maximum bankfull depth. Field observation shows this elevation to be a frequent

---A
182 D.L. Rosgen 1 Catena 22 (1994) 169-199

MODERATELY SLIGHTLY
ENTRENCHED ENTRENCHED ENTRENCHED

STREAM TYPE STREAM TYPE

ENTRENCHMENT RATIO E
------- --------
FLOdD-PRONE WIDTH

ENTRENCHMENT RATIO = FLOOD-PRONF w[DTY FLOOD-PRONE WIDTH = WATER LEVEL

BANKFULL WIDTH ‘3% 2 X MAX. DEPTH

Fig. 6. Examples and calculations of channel entrenchment.

flood (50 year return period) or less, rather than a rare flood elevation. The categories
are illustrated in Figs. 4, 5 and 6.
Entrenchment ratios of l- 1.4 represent entrenched streams, 1.41-2.2 represent
moderately entrenched streams and ratios greater than 2.2 are slightly entrenched
(well-developed floodplain). These categories were empirically derived based on
hundreds of streams. As with other criteria, the measured entrenchment ratio value
may lie somewhat outside of the classification range. When this occurs, the author
applies the continuum concept which allows for a category description where the
entrenchment is either greater or less than the most frequently observed value for a
given morphology. The continuum allows for a change of \pm 0.2 units where the
corresponding delineative criteria still match the range of variables consistent for that
type. In this case, all of the other attributes must be considered before assigning a
stream type.

Width/depth ratio
The width/depth ratio describes the dimension and shape factor as the ratio of . .
bankfull channel width to bankfull mean depth. Bankfull discharge is defined as the
momentary maximum peak flow; one which occurs several days in a year and is often
 D.L. Rosgen / Catena 22 (1994) 169-199 183

related to the 1.5 year recurrence interval discharge. Specific discussions on the
delineation and significance of bankfull discharge are found in Leopold et al.
(1964), Dunne and Leopold (1978), and Andrews (1980). Hydraulic geometry and
sediment transport relations rely heavily on the frequency and magnitude of bankfull
discharge.
Osborn and Stypula (1987) utilized width/depth ratio to characterize stream
channels for hydraulic relations using channel boundary shear as a function of
channel shape.
For this classification, values of low width/depth ratio are those less than 12. Values
greater than 12 are moderate or high. Average values and ranges are shown in the
stream type summaries. As in the continuum concept, applied to entrenchment ratio,
there is an occasion where width/depth ratio values can vary by f2 units without
showing a different morphology. This does not occur very frequently, but the
continuum allows for some flexibility to fit the stream type into a “dominant”
morphology.

Sinuosity
Sinuosity is the ratio of stream length to valley length. It can also be described as
the ratio of valley slope to channel slope. Mapping sinuosity from aerial photos is
often possible, and interpretations can often be made of slope, channel materials, and
entrenchment once sinuosity is determined. Values of sinuosity appear to be modified
by bedrock control, roads, channel confinement, specific vegetative types, etc.
Generally speaking, as gradient and particle size decreases, there is a corresponding
increase in sinuosity. The continuum as mentioned earlier also applies and adjust-
ments of + or -0.2 can be applied to this delineative criteria. Meander geometry
characteristics are directly related to sinuosity following minimum expenditure of
energy concepts. Initial studies by Langbein and Leopold (1966) suggested that a
sine generated curve describes symmetrical meander paths. From this observation
they predicted the radius of curvature of meander bends from meander wavelength
and channel sinuosity. In comparing observed versus predicted values of radius of
curvature for 79 streams, Williams (1986) found this relation to be highly correlated
when applied to an expanded data set. This demonstrates the interrelationship of
sinuosity to meander geometry. Based on such relations and the relative ease of
determination, sinuosity was selected as one of the delineative criteria for stream
classification.

Channel materials
The bed and bank materials of the river is not only critical for sediment transport
and hydraulic influences but also modifies the form, plan and profile of the river.
Interpretations of biological function and stability also require this information.
Often a good working knowledge of the soils associated with various landforms
can predict the channel materials at the broad delineation level. Reliable estimates
of the soil characteristics for glacial till, glacial outwash, alluvial fans, river
terraces, lacustrine and eolian deposits, and residual soils can be derived from
mapped lithology.
184 D.L. Rosgen / Catena 22 (1994) 169-199

r-----SANDS

3”

.I 10 100 1000

PARTICLE SIZE (MILLIMETERS)

Fig. 7. Channel material sizes showing cumulative and percent distributions.

Field determination of channel materials for this classification system utilizes the
“pebble count” method developed by Wolman (1954), with a few modifications to
account for bank material and for sand and smaller sizes. This is a determination the
frequency distribution of particle sizes that make up the channel. The pebble count
data is plotted as cumulative percent and percent of total distribution (Fig. 7). The
dominant particle size is identified in the cumulative percent curve as the median
size of channel materials or size that 50% of the population is of the same size
or finer (D&. The percent distribution shown in Fig. 7 is often used to detect
bimodal distributions that may be hidden in cumulative plots. This data is used
in biological evaluation, sediment supply assessment, and other interpretative
applications.

Slope
Water surface slope is of major importance to the morphological character of the
channel and its sediment, hydraulic, and biological function. It is determined by
measuring the difference in water surface elevation per unit stream length. Typi-
cally, slope is often measured through at least 20 channel widths or two meander
wavelengths. As observed with the other delineative variables, slope values less or
greater than the most frequently observed ranges can occur. These can occur without
a significant change in the other delineative criteria for that stream type. The most
frequently observed slope categories and applications of the continuum concept for
slope is shown in Fig. 5.
In broad-level delineations, slopes can often be estimated by measuring sinuosity
from aerial photos and measuring valley slope from topographic maps (valley
slope/sinuosity = channel slope). The basin and associated landform relief can also be
used to estimate stream slope ranges, as for example terraces and slopes of alluvial
fans.
 D.L. Rosgen / Cafena 22 (1994) 169-199 185

Fig. 8. Progressive stages of channel adjustment due to imposed stream bank instability.

5. Application

Past observations of adjustments of stream systems often provide insight into

sensitivity and consequence of change. Stream system changes can be due to flow,
sediment, or many of the interrelated variables that have produced the modern
channel. If changes produces disequilibrium, similar streams types receiving similar
impacts may be expected to respond the same. If the observer knows the stream type
of the disturbed reach, and has cross-section, bank erosion, sediment data, riparian
vegetation and fisheries data, this information can be used predictively to evaluate the
risk and sensitivity to disturbance.

5.1. Evolution of stream types

In reviewing historical aerial photos, observations can be made of progressive

stages in channel adjustment. These adjustments occur partially as a result of change
in stream-flow magnitude and/or timing, sediment supply and/or size, direct distur-
bance, and vegetation changes. These observed changes in channel morphology over
time can be communicated in terms of stream type changes. For example, due to
streambank instability, and a resultant increase in bank erosion rate, the stream
increased it’s width/depth ratio; decreased sinuosity; increased slope; established a
. =
bimodal particle size distribution; increased bar deposition; accelerated bank erosion;
and decreased the meander width ratio. These changes can be described more simply
 D.L. Rosgen / Catena 22 (1994) 169-199

G4 F4
I E4
SLOPE
.008
I .015 1 .012 .008

CROSS-

‘SECTION
WID RP;TIO 28
SINUOSITY 1.8

PLAN VIEW

2 5

Fig. 9. Evolutionary stages of channel adjustment.

as a series of progressive changes of channel adjustment in stream type from an E4 to

C4 to C4 (bar-braided) to D4 (Fig. 8).
Another example of channel adjustment where morphological patterns are changed
sufficient to indicate a shift in stream type is shown in Fig. 9. In this scenario, a change
in streambank stability led to an increase in width/depth ratio and slope, and a
decrease in sinuosity and meander width ratio. As the slope steepened along with a
high width/depth ratio, chute cutoffs occurred across large point bars creating a
gulley. The stream abandoned its floodplain, decreased the width/depth ratio,
steepened the slope and decreased sinuosity. This resulted in a change in base level
as all of the tributaries draining into this stream were over-steepened. Sediment from
both channel degradation and bank erosion was increased. As the banks continued to
erode, the width/depth ratio and sinuosity both increased with a corresponding
decrease in slope. The channel was still deeply entrenched, but eventually started to
develop a floodplain at a new elevation. This stream eventually evolved under a
changed sediment and flow regime into a sinuous, low gradient, low width/depth
ratio channel with a well developed floodplain which matched the original mor-
phology, except now exists at a lower elevation in the valley. This case is shown
more simply in Fig. 9 as a shift from an E4 stream type to C4 to G4 to F4 and
back to an E4 type.
These changes have been well documented throughout western North America due
to various reasons including climate change and adverse watershed impacts. The
knowledge provided by observing these historical adjustments and the understand-
 D.L. Rosgen / Catena 22 (1994) 169-199 187

ing of the tendency of rivers to regain their own stability can assist those restoring
disturbed river systems. Often the works of man try to “restore” streams back to a
state that does not match the dimension, pattern and slope of the natural, stable form.
As stream types change, there are a large number of interpretations associated with
these “morphological shifts”. Stream types can imply much more than what is
initially described in it’s alphanumeric title.

5.2. Fish habitat

When physical structures are installed in channels to improve the fish habitat, the
adjustment processes that occur sometimes create more damage than habitat. For
example, Trail Creek in southeast Colorado, a C4 stream type, had a gabion check
dam installed at 80% of the bankfull stage to create a plunge pool for fish. The results :
were; decreased upstream gradient; width/depth ratio increase; decreased mean bed
particle diameter; and decreased competence of the stream to move its own sediment.
The longitudinal profile of the river changed creating headward aggradation. With a
decrease in slope, there was a corresponding increase in sinuosity that resulted in
accelerated lateral channel migration and increased bank erosion. Subsequently, the
stream abandoned the original channel and created a “headcut gulley” with a gra-
dient that was twice the valley slope. This converted the C4 stream type to a G4 type
in a period of approximately two years. The “new” stream type has abandoned its
floodplain, is rejuvenating tributaries headward and creating excess sediment from
stream degradation and bank erosion. This disequilibrium caused by the check dam is
long-term and has deteriorated the habitat that the structure was initially designed to
improve. Unfortunately, structures like this continue to be installed by well-meaning
individuals without a clear understanding of channel adjustment processes.
To prevent similar problems and to assist biologists in the selection and evaluation
of commonly used in-channel structures, guidelines by stream type were developed
(Rosgen and Fittante, 1986). In the development of these guidelines hundreds of fish
habitat improvement structures were evaluated for effectiveness and channel
response. A stream classification was made for each reach containing a structure.
From this data, the authors rated various structures from “excellent” to “poor” for
an extensive range of stream types. These guidelines provide “warning flags” of
potential adverse adjustments to the river so that technical assistance may be
obtained. In this manner, structures may be better designed to not only meet their
objectives, but help maintain the stability and function of the river. Fisheries habitat
surveys presently integrate this stream classification system (USDA, 1989). The
objective for this integration is to determine the potential of the stream reach, current
state, and a variety of hydraulic and sediment relations that can be utilized for habitat
and biological interpretations.

5.3. Flow resistance

Application of the Manning’s equation and the selection of a roughness coefficient

N value to predict mean velocity is a common methodology used by engineers and
 D.L. Rosgen / Catena 22 (1994) 169-199

LEGEND

COMBlNED AVE.
HICKS 8. MASON

i3 A2 F2 G6 82 83
““e6 83~ 65 I=6 F6 64 84 F3 C5 65
~6 81 F4
E3
~4 G3 C1 C3 C4

Fig. 10. Bankfull stage roughness coefficients (“W’ values) by stream type for 140 streams from the United
States and New Zealand.

hydrologists. The lack of consistent criteria for selection of the correct N values,
however, creates great variability in the subsequent estimate of flow velocity. Barnes
(1967) and Hicks and Mason (1991) produced photographs and a variety of stream
data which was primarily a visual comparison approach for the selection of roughness
coefficients. However, using these books for a visual estimate of roughness, actually
involves looking at various stream types. The author classified each of the 128 streams
described in both publications, noted the occurrence of vegetation influence, and
plotted the bankfull stage N values by stream type (Fig. 10). The remarkable simi-
larity of N values by stream type for two data bases from two countries revealed
another application for estimating a bankfull stage roughness coefficient using stream
classification. This may help in developing more consistent roughness estimates and
provide an approach for improving stream discharge estimates by using the
manning’s equation. Roughness values increase as stage decreases, thus, the N values
shown in Fig. 10 are for bankfull conditions only. The Hicks and Mason (1991) work
is exemplary in terms of evaluating and displaying variations in N with changes in
stream discharge. These variations can potentially be developed as a rate of change
index for changes in stage by stream type. The influence of vegetation is shown to
cause a marked adjustment in values by stream type. As would be expected, this
relationship suggests the vegetation influence on roughness is diminished as channel
 D.L. Rosgen / Catena 22 (1994) 169-199 189

gradient and bed material particle size increase. Stream types essentially integrate
those variables affecting roughness, such as; gradient, shape and form resistance,
particle size, and relative depth of bankfull discharge to the diameter of the larger
particles in the channel. Rather than looking at discrete predictors, stream types
integrate the many variables that influence resistance. Another recommended
application to roughness estimation is to develop specific relations of roughness
and associated velocity as recently developed for “mountain streams” by Jarrett
(1984, 1990). In this method, equations were stratified for steeper slopes and
cobble/boulder channel materials, using hydraulic radius and slope in the
equations. Jarrett’s results were valuable in that they produced values much different
from most published equations. This work could be even more effective if the stream
data were further stratified into stream types and size of stream. In this manner, much :
like the Manning’s N values, equations could be developed using the integrating
effects of stream types and thereby advance the state of the art of applications.

5.4. Hydraulic geometry relations

The original work of Leopold and Maddock (1953) made a significant contribution
to the applied science in the development of hydraulic geometry relations. The
variables of; depth, velocity, and cross-sectional area were quantitatively related to
discharge as simple power functions for a given river cross-section. Their findings
prompted numerous research efforts over the years. To refine average values of
exponents, and to demonstrate the potential for applications of hydraulic geometry
relations by stream types, this author assembled stream dimensions, slopes, and
hydraulic data for six different stream types having the same discharge and channel
materials. The objective was to demonstrate how the shape (width/depth ratio),
profile (gradient), plan view (sinuosity), and meander geometry affect the hydraulic
geometry relations. For example channel width increases faster than mean depth,
with increasing discharge in high width/depth ratio channels. The opposite is true
in low width/depth ratio channels. Streamflow values from baseflow of approximately
4 cfs up to bankfull values of 40 cfs were compared for each cross-section, and the
corresponding widths, depths, velocities, and cross-sectional area for each stream
type were computed. The A3, B3, C3, D3, E3 and F3 stream types selected for
comparisons all had a cobble dominated bed-material size. The resultant hydraulic
geometry relations for the selected array of stream types at the described flow ranges
are shown in Fig. 11. Except for the E3 stream type for the plot of width/discharge,
the slope of the plotted relations did not significantly change nearly as much as the
intercept values.

6. Shear stress/velocity relations

. _ Using the same data from the six stream types described previously, a “lumped”
data base for all stream types from low to high flow was made for the corresponding
shear stress (T = yRS) (Shields, 1936) vs. mean velocity, where; r = shear stress,
 D.L. Rosgen / Catena 22 (1994) 169-199

Discharge vs. Cross Sectional Area

5 10 50
Stream Discharge (cfs)

Discharge vs. Width

g 100
f
P
s
E
B
2 10
5
5 10 50
Stream Discharge (cfs)

Discharge vs. Mean Velocity

10
Stream Discharge (cfs)

Discharge vs. Mean Depth

I I I
.I 1 1
2 5 10 50
Stream Discharge (cfs)
Fig. Il. Hydraulic geometry relations for selected stream types of uniform size.
 D.L. Rosgen / Catena 22 (1994) 169-199 191

2.5

.1 5

I - I I I I I Illll
.l .15 .2 .25 .3 .4 .5 .6 .7 .8.91D 1 1.5 2 2.5 3 4 5 6 78910

SHEAR STRESS (LESiFT/SEC.)

Fig. 12. Relationship of mean velocity vs. shear stress for six stream types from base flow (3-4 cfs) to
bankfull discharge (40-41 cfs).

y = density of water, R = hydraulic radius, and S = channel slope. As expected, a

meaningful relation was not found. However, plotting shear stress and velocity
stratification by stream type provided a trend that did shows promise (Fig. 12).
While more data are needed to establish mathematical and statistical relationships,
the comparisons arranged by stream type may have potential for future applications.

6.1. Critical shear stress estimates

Previous investigations of the magnitude of shear stress required to entrain various

particle diameters from the stream-bed material have produced a wide range of
values. A number of investigators have assumed the critical dimensionless shear
192 D.L. Rosgen / Catena 22 (1994) 169-199

.20

Field Data for: From:

STREAM TYPE
Fig. 13. Relationship of field verification of critical dimensionless shear stress values for various stream
types.

stress values of 0.06 for computations of bedload transport using Shield’s (1936)
criteria (Baker, 1974; Baker and Ritter, 1975; Church, 1978; Bradley and Mears,
1980; Simons and Senturk, 1977; Simons and Li, 1982). In addition, critical
dimensionless shear stress, values computed from data compiled by Fahnestock
(1963), Ritter (1967), and Church (1978) for the entrainment of gravels and cobbles
from a natural river-bed, as reported by Andrews (1983) showed a range of approxi-
mately 0.02 to 0.25. The mean of the computed values was 0.06, which is the value
suggested by Shields (1936).
Andrews (1983) described a relationship where to the ratio of surface (pavement)
bed particles to sub-surface (sub-pavement) particles that yielded an estimate of
critical dimensionless shear stress vaIues (r,i*) from 0.02 to 0.28. Additional work
using the same equation was applied to several Colorado gravel-bed streams with
similar results (Andrews, 1984).
It is sometimes difficult for many engineers to obtain pavement and sub-pavement
data along with the required channel hydraulics information to refine critical
dimensionless shear stress estimates using the Andrews (1983, 1984) equation. The
use of stream types to help bridge this gap of estimating the critical dimensionless
 D.L. Rosgen / Catena 22 (1994) 169-199 193

shear stress value (TC) has potential where these study streams have been analyzed and
classified. The study streams by Andrews (1984) were classified, data compiled and
the values of rc* (critical dimensionless shear stress) were plotted (Fig. 13). A2 and D4
stream types were obtained from field measurements of bedload sediment and bed-
material size distribution for those types (Williams and Rosgen, 1989). Stream types
and their morphologic/hydraulic characteristics do not substitute for detailed on-site
investigations as described by Andrews (1983, 1984); however, calculations of rci* are
often made without the benefit of site-specific investigation. Based on the great
variability in the estimate of Tci*, sediment transport prediction errors can be from
one to several orders of magnitude. A closer approximation of Tci* for stream reaches
that cannot be investigated in detail, is possible using the extrapolation approach
shown in Fig. 13.
A similar analysis has been made but not included here using unit stream power ’
rather than critical shear stress. This analysis again demonstrated that stratification
by stream type improved sediment transport/stream power relations as an integrative
function of the supply/energy distribution/resistance factors for specific stream types.

6.2. Sediment relations

Stream types have been used to characterize sediment rating curves that reflect
sediment supply in relation to stream discharge. For example, a sediment rating
curve regression relation for an A2 stream type would have a characteristic low
slope and intercept. The sediment rating curve for the C4 stream type, however,
has a higher intercept and steeper slope. The author has used this procedure for
both suspended and bedload rating curves. These relationships were initially plotted
as a function of channel stability ratings as developed by Pfankuch (1975).
Applications for cumulative effects analysis for non-point sediment sources utilized
this approach (USEPA, 1980). Subsequent comparisons of data with stream type
delineations indicated similar relations.
The ratio of bedload to total sediment load can also be stratified by stream type
where measured data is available. Ranges of less than 5% bedload to total sediment
load for C3 stream types have been reported, but values greater than 75% bedload to
total load for G4 stream types have also been measured (Williams and Rosgen, 1989).
The “high ratio” bedload streams are the A3, A4, A5, D3, D4, D5, F4, F5, G3, G4,
and G5 stream types.

6.3. Management interpretations

The ability to predict a river’s behavior from its appearance and to extrapolate
information from similar stream types helps in applying the interpretive information
in Table 3. These interpretations evaluate various stream types in terms of; sensitivity
to disturbance, recovery potential, sediment supply, vegetation controlling influence,
and streambank erosion potential. Application of these interpretations can be used
for; potential impact assessment, risk analysis, and management direction by stream
type. For example, livestock grazing effects were related to stream stability and
194 D.L. Rosgen / Catena 22 (1994) 169-199

Table 3
Management interpretations of various stream types

Stream Sensitivity Recovery Sediment Streambank Vegetation

type to potential b supply c erosion controlling
disturbance a potential influence d . I

Al very low excellent very low very low negligible

A2 very low excellent very low very low negligible
A3 very high very poor very high high negligible
A4 extreme very poor very high very high negligible
A5 extreme very poor very high very high negligible
A6 high poor high high negligible
Bl very low excellent very low very low negligible
B2 very low excellent very low very low negligible :
B3 low excellent low low moderate
B4 moderate excellent moderate low moderate
B5 moderate excellent moderate moderate moderate
B6 moderate excellent moderate low moderate
Cl low very good very low low moderate
c2 low very good low low moderate
c3 moderate good moderate moderate very high
c4 very high good high very high very high
c5 very high fair very high very high very high
C6 very high good high high very high
D3 very high poor very high very high moderate
D4 very high poor very high very high moderate
D5 very high poor very high very high moderate
D6 high poor high high moderate
DA4 moderate good very low low very high
DA5 moderate good low low very high
DA6 moderate good very low very low very high
E3 high good low moderate very high
E4 very high good moderate high very high
E5 very high good moderate high very high
E6 very high good low moderate very high
Fl low fair low moderate low
F2 low fair moderate moderate low
F3 moderate poor very high very high moderate
F4 extreme poor very high very high moderate
F5 very high poor very high very high moderate
F6 very high fair high very high moderate
Gl low good low low low
G2 moderate fair moderate moderate low
G3 very high poor very high very high high
G4 extreme very poor very high very high high
G5 extreme very poor very high very high high
G6 very high poor high high high

a Includes increases in streamflow magnitude and timing and/or sediment increases.

b Assumes natural recovery once cause of instability is corrected. . f
’ Includes suspended and bedload from channel derived sources and/or from stream adjacent slopes.
d Vegetation that influences width/depth ratio-stability.
 D.L. Rosgen 1 Catena 22 (1994) 169-199 195

sensitivity using stream types (Meyers and Swanson, 1992). They summarized their
study results on streams in northern Nevada that “ . . . range managers should con-
sider the stream type when setting local standards, writing management objectives, or
determining riparian grazing management strategies.”
This interpretive information by stream type can also apply to establishment of
watershed and streamside management guidelines dealing with; silvicultural
standards, surface disturbance activities, surface disturbance activities, gravel and
surface mining activities, riparian management guidelines, debris management, flood-
plain management, cumulative effects analysis, flow regulation from reservoirs/
diversions, etc. An example of the implementation of these guidelines by stream type
are shown in the Land and Resource Management Plan (USDA, 1984).
Applications for riparian areas (USDA, 1992), have utilized the stream classifi-
cation system into their recently developed “Integrated Riparian Evaluation Guide”:
- Intermountain Region. The classification system was used to help stratify and
classify riparian areas based on natural characteristics and existing conditions. It is
also used to evaluate the potential risks and sensitivities of riparian areas.

6.4. Restoration

The morphologic variables that interact to form the dimensions, profile and
patterns of modern rivers are often the same variables that have been adversely
impacted by development and land use activities. To restore the “disturbed” river,
the natural stable tendencies must be understood to predict the most probable form.
Those who undertake to restore the “disturbed” river must have knowledge of fluvial
process, morphology, channel and meander geometry, and the natural tendencies of
adjustment toward stability in order to predict the most effective design for long-term
stability and function. If one works against these tendencies, restoration is generally
not successful. Restoration applications using stream classification and the previously
discussed principles are documented in the “Blanco River” case study (National
Research Council, 1992).

7. Summary

Rivers are complex natural systems. A necessary and critical task towards the
understanding of these complex systems is to continue the river systems research.
In the interim, water resource managers must often make decisions and timely pre-
dictions without the luxury of a complex and thorough data base. Therefore, a goal
for researchers and managers is to properly integrate what has been learned about
rivers into a management decision process that can effectively utilize such knowledge.
There is often more data collected and available on rivers than is ever applied. Part of
the problem is the large number of “pieces” that this data comprises and the difficulty
I
of putting these pieces into meaningful form.
The objective of this stream classification system presented here is to assist in
bringing together these “pieces” and the many disciplines working with rivers
196 D.L. Rosgen 1 Catena 22 (1994) 169-199

under a common format - a central theme for comparison, a basis for extrapolation,
prediction, and communication. The stream classification system can assist in
organizing the observations of river data and of molding the many pieces together
into a logical, useable, and reproducible system.
With the recent emphasis on “natural” river restoration or “naturalization” .
throughout Europe and North America, understanding the potential versus the
existing stream type is always a challenge. The dimension of rivers related to the
flow, and the patterns, which in turn are related to the dimensions, have to be further
stratified by discrete stream types. In this way, the arrangement of the variables that
make up the plan, profile and section views of stable stream types that are integrated
within their valley’s can be emulated. This also involves re-creation of the correspond-
ing appropriate bed morphology associated with individual stream types with the
observed sequence of step/pool and/or riffle pool bed features as a function of
the bankfull width. The use of meander width ratios by stream type helps to establish
the minimum, average and ranges of lateral containment of rivers. This often helps
the design engineer/hydrologist determine appropriate widths that need to be accom-
modated when natural, stable rivers are re-constructed within their valleys. River and
floodplain elevations, which need to be constructed, can be often determined by the
used of the entrenchment ratio, which depicts the vertical containment of rivers in the
landform. Using these integrative, morphological relations by stream type, can avoid
the problematic “works” done on streams which create changes in the dimensions,
pattern and profile of rivers which are not compatible with the tendencies of the
natural stable form.
A classification system is particularly needed to stratify river reaches into’groups
that may be logically compared. Such stratification reduces scatter that mightappear
to come from random variation, whereas the scatter often results from attempting to
compare items generically different. For example, data developed from empirical
relations associated with process oriented research in natural channels such’ as trac-
tive force relations, resistance and sediment transport equations, etc., can be sratified
by stream type. This can help reduce the scatter when applied to stream types different
than those from which the relations were developed. ,! c
Utilizing quantitative channel morphological indices for a classification procedure
insures for consistency in defining stream types among observers for a great diversity
of potential applications. The classification presented here may be the
first
approximation of a system that undoubtedly will be refined over the years with
continued experience and knowledge. This stream classification system hopefully
can be a vehicle to provide better communication among those studying river
systems and promote a better understanding of river processes, helping put principles
into practice.

Acknowledgements

I would like to thank the many individuals who have contributed in data collection,
 D.L. Rosgen / Catena 22 (1994) 169-199 197

analysis, thought, and spirit towards this classification. The individuals are too
numerous to mention, but in particular I would like to mention my river compa-
nions, Dr. Luna Leopold, Hilton “Lee” Silvey, Dale Pfankuch, Owen Williams, Alice
Johns, Jim Nankervis and Steve Belz. Their contributions to consultation, analysis,
and review are much appreciated. The majority of the illustrations were drawn by
H. Lee Silvey. Appreciation of the fine work of word processing is acknowledged to
Kay McElwain.

References

Andrews, E.D., 1980. Effective and bankfull discharges of streams in the Yampa River Basin, Colorado and
Wyoming. J. Hydrol., 46: 311-330.
Andrews, E.D., 1983. Entrainment of gravel from natural sorted riverbed material. Geol. Soc. Am. Bull.,
94: 1225-1231.
Andrews, E.D., 1984. Bed-material entrainment and hydraulic geometry of gravel-bed rivers in Colorado.
Geol. Soc. Am. Bull., 95: 371-378.
Baker, V.R., 1974. Paleohydraulic interpretation of Quaternary alluvium near Golden, Colorado. Quat.
Res., 4: 944112.
Baker, V.R. and Ritter, D.F., 1975. Competence of rivers to transport coarse bedload material. Geol. Soc.
Am. Bull., 86: 975-978.
Barnes, H.H., 1967. Roughness characteristics of natural channels. U.S. Geol. Survey, Water Supply Paper
1849, 213 pp.
Bradley, W.C. and Mears, A.I., 1980. Calculations of flows needed to transport coarse fraction of Boulder
Creek alluvium at Boulder, Colorado. Geol. Soc. Am. Bull., Part II, 91: 105771090.
Brice, J.C. and Blodgett, J.C., 1978. Counter Measures for Hydraulic Problems at Bridges. Vol. 1,
Analysis and Assessment. Report No. FHWA-RD-78-162, Fed. Highway Admin., Washington, DC,
169pp.
Church, M., 1978. Paleohydrological reconstructions from a Holocene valley. In: A.D. Miall (Editor.),
Fluvial Sedimentology. Can. Soc. Pet. Geol. Mem., 5: 743-772.
Church, M. and Rood, K., 1983. Catalogue of alluvial river channel regime data. Natural Sciences and
Engineering Research Council of Canada. Dept. Geology Univ. British Columbia, Vancouver, B.C.,
Edition 1 .O.
Culbertson, D.M., Young, L.E. and Brice, J.C., 1967. Scour and fill in alluvial channels. U.S. Geological
Survey, Open File Report, 58 pp.
Davis, W.M., 1899. The geographical cycle. Geogr. J., 14: 481-504.
Dunne, T. and Leopold, L.B., 1978. Water in Environmental Planning. Freeman, San Francisco, CA,
818 pp.
Fahnestock, R.K., 1963. Morphology and hydrology of a glacial stream - White River. Mount Rainier,
Washington. U.S. Geol. Survey Prof. Paper 422-A, 70 pp.
Galay, V.J., Kellerhals, R. and Bray, D.I., 1973. Diversity of river types in Canada. In: Fluvial Process and
Sedimentation. Proceedings of Hydrology Symposium. National Research Council of Canada, pp. 217-
250.
Grant, G.E., Swanson, F.J. and Wolman, M.G., 1990. Pattern and origin of stepped-bed morphology in
high-gradient streams, Western Cascades, Oregon. Geol. Soc. Am. Bull., 102: 340-352.
Hicks, D.M. and P.D. Mason, 1991. Roughness characteristics of New Zealand rivers. New Zealand Dept.
of Scientific and Industrial Research, Marine and Freshwater, Natural Resources Survey, Wellington,
329 pp.
Inglis, C.C., 1942. Rapid westerly movement of the Kosi River. Central Board of Irrigation, India
(Technical), Poona, Bombay, pp. 7 and 8. F
Jarrett, R.D., 1984. Hydraulics of high-gradient streams. J. Hydraul. Eng., llO(11): 1519-1539.

I
198 D.L. Rosgen / Catena 22 (1994) 169-199

Jarrett, R.D., 1990. Hydrologic and hydraulic research in mountain rivers. Water Res. Bull. WARBAQ,
26(3): 419-429.
Kellerhals, R., Neill, C.R. and Bray, D.I., 1972. Hydraulic and geomorphic characteristics of rivers in
Alberta. Research Council of Alberta, River Engineering and Surface Hydrology Report 72-1, p. 52.
Kellerhals, R., Church, M. and Bray, D.I., 1976. Classification and analysis of river processes. J. Hydraul.
Div., ASCE, 102(HY7): 813-829. .;
Khan, H.R., 1971. Laboratory studies of alluvial river channel patterns. Ph.D. Dissertation, Dept. of Civil
Engineering Department, Colorado State University, Fort Collins, CO.
Lane, E.W., 1957. A study of the shape of channels formed by natural streams flowing in erodible material.
Missouri River Division Sediment Series No. 9, U.S. Army Engineer Division, Missouri River, Corps of
Engineers, Omaha, NE.
Langbein, W.B. and Leopold, L.B., 1966. River meanders - Theory of minimum variance. U.S. Geol.
Survey Prof. Paper No. 422-H, 15 pp.
Leopold, L.B. and Maddock, T., 1953. The hydraulic geometry of stream channels and some physiographic
implications. U.S. Geol. Survey Prof. Paper No. 252. Supt. of Docs, U.S. Government Printing Office, :
Washington, DC, 57 pp.
Leopold, L.B. and Wolman, M.G., 1957. River channel patterns: braided, meandering, and straight. U.S.
Geological Survey Prof. Paper 282-B.
Leopold, L.B., Wolman, M.G. and Miller, J.P., 1964. Fluvial processes in geomorphology. Freeman, San
Francisco, CA, 522 pp.
Matthes, G., 1956. River engineering. In: P.O. Abbott (Editor), American Civil Engineering Practice.
Wiley, New York, Vol. II, pp. 15-56.
Melton, F.A., 1936. An empirical classification of flood-plain streams. Geogr. Rev., 26.
Mollard, J.D., 1973. Air photo interpretation of fluvial features. In: Fluvial Processes and Sedimentation.
Research Council of Canada, pp. 341-380.
Meyers, T.J. and Swanson, S., 1992. Variation of stream stability with stream type and livestock bank
damage in Northern Nevada. Water Resources Bull. AWRA, 28(4): 743-754.
Nanson, G.C. and Croke, J.C., 1992. A genetic classification of floodplains. In: G.R. Brakenridge and J.
Hagedorn (Editors), Floodplain Evolution. Geomorphology, 4: 459-486.
National Research Council, 1992. Restoration of aquatic ecosystems. Restoration case studies. National
Academy Press, Washington, DC, pp. 470-477.
Osborn, J.F. and Stypula, J.M., 1987. New models of hydrological and stream channel relationships. In:
Erosion and sedimentation in the Pacific Rim. IAHS Publ. No. 165, Oregon State Univ., Corvallis, OR.
Pfankuch, D.J., 1975. Stream reach inventory and channel stability evaluation. USDA Forest Service, Rl-
75-002. Government Printing Office #696-260/200, Washington DC, 26 pp.
Pickup, G., 1984. Geomorphology of tropical rivers. 1. Landforms, hydrology and sedimentation in the fly
and lower Purari, Papua New Guinea. In: A.P. Schick (Editor), Channel Processes, Water, Sediment,
Catchment Controls. Catena Suppl., 5: 1-17.
Platts, W.S., 1980. A plea for fishery habitat classification. Fisheries, 5(l): l-6.
Ritter, J.R., 1967. Bed-material movement. Middle Fork Eel River, California. U.S. Geol. Survey Prof.
Paper 575-C, pp. C219-C221.
Rosgen, D.L., 1985. A stream classification system. In: Riparian Ecosystems and Their Management. First
North American Riparian Conference. Rocky Mountain Forest and Range Experiment Station, RM-
120, pp. 91-95.
Rosgen, D.L. and Fittante, B.L., 1986. Fish habitat structures - A selection guide using stream classitica-
tion. In: J.G. Miller, J.A. Arway and R.F. Carline (Editors), 5th Trout Stream Habitat Improvement
Workshop, Lock Haven Univ., Lock Haven, PA. Penn. Fish Comm. Publics, Hamsburg, PA, pp. 163-
179.
Schumm, S.A., 1963. A tentative classification of alluvial river channels. U.S. Geological Survey Circular
477. Washington, DC.
Schumm, S.A., 1977. The Fluvial System. Wiley, New York, 338 pp.
Selby, M.J., 1985. Earth’s Changing Surface: an Introduction to Geomorphology. Oxford University Press,
Oxford.
 D.L. Rosgen 1 Catena 22 (1994) 169-199 199

Shields, A., 1936. Application of similarity principles and turbulence research to bed-load movement. In:
W.P. Ott and J.C. Uchelen (translators), Mitt. Preuss. Verschsanst., Berlin. Wasserbau Schiffbau.
California Institute of Technology, Pasadena, CA, Report No. 167, 43 pp.
Simons, D.B. and Senturk, F., 1977. Sediment Transport Technology. Water Resource Publications, Fort
Collins, CO, 807 pp.
_- Simons, D.B. and Li, R.M., 1982. Engineering Analysis of Fluvial Systems. SLA, P.O. Box 1816, Fort
Collins, CO.
Smith, D.G. and Smith, N.D., 1980. Sedimentation in anastomosing river systems: examples from alluvial
valleys near Banff, Alberta. J. Sediment. Petrol., 50: 157-164.
Thornbury, W.D., 1969. Principles of Geomorphology. 2nd Edition. Wiley, New York.
USDA (U.S. Department of Agriculture), Forest Service, 1984. Land and Resource Management Plan -
Arapaho and Roosevelt National Forests and Pawnee Grassland. Fort Collins, CO.
USDA (U.S. Department of Agriculture), Forest Service, 1989. Fisheries Habitat Surveys Handbook.
Region 4 - FSH 2609.23. Ogden, UT.
USDA (U.S. Department of Agriculture), Forest Service, 1992. Integrated Riparian Evaluation Guide. j
Intermountain Region - Ogden, UT.
USEPA (U.S. Environmental Protection Agency), 1980. An approach to water resources evaluation of non-
point silvicultural sources. Chap. VI, Total Potential Sediment, D. Rosgen. EPA-600/8-80-012, Athens,
GA, pp. 39-41.
Williams, G.W., 1986. River meanders and channel size. J. Hydrol., 88: 147-164.
Williams, G.P. and Rosgen, D.L., 1989. Measured total sediment loads (suspended loads and bedloads) for
93 United States streams. Open File Report 89-67. U.S. Geol. Survey, Denver, CO, 128 pp.
Wolman, M.G., 1954. A method of sampling coarse river-bed material. Trans. Am. Geophys. Union, 35:
951-956.