Channel Design: Plan and Cross-SectionKareem Saint-Lot (

Stream restoration is commonly defined as the return of a degraded ecosystem to a close approximation of its remaining natural potential (U.S. Environmental Protection Agency (USEPA), 2000). This definition has developed over time with a lack of agreement on what constitutes a successful restoration project, and some would argue that restoration to a truly pre-disturbed state is rarely, if ever, accomplished (Kochel & Miller, 2010; Palmer et al., 2005). In an attempt to define an ecologically successful restoration project Palmer et al. (2005) proposed a standard method which can be used to measure the success of a restoration project. The effectiveness of a restoration project heavily depends on the design used during project implementation. Channel design within a stream aims to provide the river with a channel structure that can withstand relatively stable flow conditions without major aggradation or degradation. Within the restoration community exist two main design approaches: process based design and form based design. Both of the design approaches attempt to capture stream geomorphology and implement a design which is suitable for the project site. Although stream restoration has become a growing enterprise around the world, there remains controversy between the two common practices for restoration: Process Based Design and Natural Channel Design (NCD) (Lave, 2009). This page presents the basis for channel design and discusses both the form and process based methods in an unbiased and summarized manner, and then both design approaches are compared to one another.
Hydraulic engineering, in the past, focused primarily on the efficient conveyance of floodwaters. The environment and stream ecosystem, more often than not, were not of concern as the focus remained on the economic value of water (Lave, 2009). Thus, hydraulic engineering to control erosion, flooding, or drainage has led to the straightening, enlarging, relocation, and stabilization of stream channels by man (Shields, 1983). Degradation of streams and their ecosystems are often times attributed to engineering practices. Channel modifications, which produced more uniform velocities and depths and removed many natural structural elements (i.e. boulders and logs) also, unintentionally, decreased the habitat diversity. Lack of habitat diversity resulted in the decline of aquatic species populations (Shields, 1983). Out of the degradation of stream ecosystems evolved the concept of stream restoration, which was meant to return a degraded stream to a healthier and more dynamic state. Engineers must attempt to find a balance between efficiency in water conveyance and maintaining a dynamic equilibrium state for the stream that promotes a healthy and diverse ecosystem.
Habitat Design
Engineers have implemented the idea of habitat design in areas where a stream has experienced loss of species habitat diversity. Shields (1983) presents the design of habitat structures for open channels which is typically used to increase the biological recovery rate of relocated or enlarged streams. Although beneficial in improving the quality of aquatic environment for species, habitat structures do not restore the stream habitat to its natural state. Additionally, implementation of these structures is complex, as they are designed to withstand high velocities without compromising the ability of the engineered channel to convey flood waters or sediment transport (Shields, 1983).
Habitat structures are usually built using natural materials (i.e. logs, rocks), but are sometimes combined with concrete and gabions to increase stability. These structures can be categorized into four main categories: sills, deflectors, random rocks, or cover. Sills are defined as structures which stretch the full width of the channel (see Figure 1). These structures are typically intended to create a small pool upstream or a downstream scour hole, although sometimes sills are designed to achieve both outcomes. These structures are often designed with a notch or gap to concentrate low flows which create deeper pools and maintain scour holes. Sill structure design is usually accompanied by bank protection adjacent to the structure to prevent flanking and downstream to prevent bank erosion (Shields, 1983).
Structures which extend partially into the channel are placed into the category of deflectors (see Figure 2). Positioning of deflectors can differ from angling upstream, to perpendicular, to angling downstream dependent upon what conditions are desirable. These structures are usually used to create local increases or decreases in velocity and depth. Whereas, random rocks are used in the channel away from the banks to create lower velocities and small scour holes (see Figure 3).
Figure 1: Sill: Plan & Cross Section
Figure 2: Deflectors: Angled, Perpendicular, and Triangular

Finally, cover structures (see Figure 4), which do not influence flow patterns, are intended to create refuge for fish from predators, and sometimes provide stable substrate for invertebrates (Shields, 1983).
Site selections along with effective habitat structures are both essential for improving habitat diversity. As noted by Shields, the “structures which created diversity in substrates, depths, velocities, and illumination conditions that mimicked the natural stream’s pool riffle sequence” were most successful in creating the suitable aquatic habitat. Case studies show that site selection involves the placement of structures where water quality and flow quantity provide favorable conditions for biological recovery. Though it is important to mention modified streams tend to experience natural biological recovery, and the purpose of implementing habitat structures is to accelerate the recovery of the stream ecosystem (Shields, 1983). For more information on habitat structures see the Physical Habitat section by M. Sims and A. Martinez of the wiki page.
Figure 3: Random Rock Placements

Figure 4: Cover Structure

Channel Discharge Design Parameters
As stated above stream restoration projects often rely on relatively stable flow conditions. Doyle et al. (2007) present a framework for assessing and selecting channel forming discharge in river restoration. There are three main measures of channel discharge most commonly used in the design of stream restoration projects: effective discharge (Qeff), bankfull discharge (Qbf), and discharge of a certain recurrence interval (Qri). The effective discharge of a channel refers to “the discharge (or range of discharges) which, over time, transport the greatest amount of sediment” (Doyle et al., 2007). Qeff is calculated by multiplying the flow frequency curve by the sediment discharge curve and finding the maximum. An example of this is demonstrated in Figure 5. Effective discharge is more difficult to attain given the necessary data (it is typically recommended to use 10 or more years of data) required for evaluation, but is the most useful tool for channel design as it provides crucial information regarding channel processes (Doyle et al., 2007).
Figure 5: Effective Discharge Curve adapted from Wolman and Miller (1960)

Qri is usually determined to be the flow with a 1-2 year return interval and is usually, in stable systems, used as the first approximation of Qeff (Doyle et al., 2007). In urbanized watersheds Qri is considered to be inappropriate for estimating effective discharge because it relies on the assumption of stationary hydrologic conditions (see the Urban Streams section by J. Reale of the wiki page for more information on the effects of urbanization) (Shields et al., 2003). Qbf is defined as the flow at which a river’s banks are barely overtopped (Emmett & Wolman, 2001). Qbf can only be estimated for unstable channels and sometimes retains high levels of uncertainty due to the quality of field indicators and reference reaches, which are sometimes used for determining bankfull discharge (Johnson & Heil, 1996; Williams, 1978). Differences between Qbf and Qeff in unstable channels can be used as an indicator of channel stability. In stable, alluvial channels it is theoretically appropriate to assume Qbf and Qeff are equivalent (Shields et al., 2003).
The channel forming discharge (Qcf) is referred to as the maximum discharge an alluvial channel can accommodate in a state of equilibrium. “Hydraulic geometry relations, assessment of dynamic equilibrium, meander formulas, and floodplain design are all based on estimates of Qcf for a particular river system” (Goodwin, 2004). The three measures of channel discharge listed above (Qeff, Qbf, Qri) are used as different approaches for measuring or estimating Qcf (Doyle et al., 2007). Using these methods for describing channel forming discharge allows stream restoration practitioners to create a design which is stable within the specified discharge ranges. Hydrology in relation to stream restoration has been very briefly summarized for the purpose of this paper. For in depth coverage see the Hydrology section by A. Heermann of the wiki page.
Design Methods for Stream Restoration
There exists a suite of design approaches for stream restoration which includes designs based on: 1. Stream classification and regional curves for hydraulic geometry (Rosgen D. L., 1996; Riley, 1998) 2. Regime and tractive force equations (Ministry of Natural Resources 1994), 3. Reference reaches (Newbury and Gaboury 1993), 4. and combinations of the above listed approaches (Gillilan 2001). For the purposes of this paper, a focus will be made on summarizing and comparing Natural Channel Design and Process Based Design methods.
Process Based Design
In channel reconstruction there exists a tension between restoring natural fluvial processes and ensuring stability of the completed design (Shields et al., 2003). To determine which of the two scenarios is more suitable for a specific restoration project, all involved stakeholders must first set the project objectives. Process based design looks “to couple reach-level channel processes with the spatial arrangement of reach morphologies” to assess channel condition and response (Montgomery & Buffington, 1997). In this first step current habitat state and sources of degradation are often uncovered (Society for Ecological Restoration, 2002). Next, involves conducting a stability assessment which requires establishment of a spatial domain to uncover the instabilities that exist within the defined reach. Table 1 provides an overview of the stability assessment tools. This creates a basis for design and allows for predictable response of the river system (Shields et al., 2003). The stability assessment may be conducted qualitatively through the means of visual inspection (usually performed by experienced personnel) or quantitatively through the collection of data which describes channel geometry, bed sediments, hydrology, and land use in the past and present. It is important to note that a “stream may be highly dynamic but considered geomorphically stable if its long-term temporal average properties are stationary” (Shields et al., 2003). In other words, the channel reaches a state of dynamic equilibrium where its morphological characteristics remain relatively constant even though the channel position may change (Knighton, 1996). (See the Fluvial Geomorphology section by B. Levine and L. Begay for more information on geomorphology.)
Table 1: Overview of Channel Stability Assessment Tools

Determination of channel forming discharge, Qcf (described earlier), would follow the stability assessment. In process based design there exists a concept where the channel forming discharge, if given enough time, would create stream geomorphology similar to that which a natural hydrograph would impose on the river system (Shields et al., 2003). Therefore, using the stability assessment and the channel forming discharge, the project team can now create a design for the stream restoration project. Thorne recommends minimizing channel modifications in stable river systems, though it is recognized that channel modifications may sometimes be necessary to restore ecosystem functionality (Thorne et al. 1996) (see Ecosystem services by A. Griego for more information on ecosystem functions).
The design approach Shields presents is noted to be suited for perennial, moderate to low-energy meandering alluvial streams, where the stream is approaching or has approached stable conditions. In this case the primary role of stream restoration design is to push the stream towards a state of dynamic equilibrium, or provide a catalyst to aid in the stream’s ability to recover. This allows the restored stream to pass its sediment load during channel forming discharge without significant effects of aggradation or degradation (Shields et al. 2003). Thus, one can foresee the difficulty in stream restoration projects for unstable river systems as there is uncertainty in defining the Qcf. Different modeling techniques are used to predict channel response to restoration activities. For more information on modeling and hydraulics see the Hydraulics section by K. Steinhaus of the wiki page.
Threshold and Active-Bed Methods
At this point process based projects use one of two defined methods to complete design: the threshold channel method or the active-bed method. Threshold methods are described for cases where the stream bed remains relatively immobile at high flows and the incoming sediment load is negligible (Shields et al., 2003). This method may also be used if bed materials originated from events or processes that no longer exist. Design velocities within the channel are based on experience and various observations or can be analyzed using the tractive force method. “The basic derivation of equations used in the tractive force approach assumes that channel cross sections and slopes are uniform, beds are flat, and bed material transport is negligible. (Shields et al., 2003)” The threshold method, as defined by Shields, has been broken down to five basic steps:
  1. Determination of bed material and appropriate discharge design (usually use Qri or Qbf since boundary conditions assume an immobile channel)
  2. Computation of a preliminary average width (use hydraulic geometry or regime formulas)
  3. Estimation of critical bed shear stresses using bed material size (consider sediment gradation and local conditions)
  4. Estimation of a flow resistance coefficient using bed material size, channel sinuosity, bank vegetation and flow depth
  5. Computation of average depth and bed slope needed to pass design discharge using the continuity and uniform flow equation (see Hydraulics by K. Steinhaus)

It is recommended to use several iterations to refine initial estimates and obtain a proper design. This method assumes bed material transport is negligible and channel cross sections and slopes are uniform. These assumptions rarely occur realistically, which usually makes this threshold method approach inappropriate for channel design intended to encourage natural fluvial processes (Shields et al., 2003); although, this method has been previously used in a successful manner as shown by Newbury and Gaboury in 1993 to size stone used in the construction of a channelized stream. For another example using this threshold method, see Beck et al. 2000.
The active-bed method is used in cases where stream beds are mobile at high flows. In this method the relationship between channel geometry and sediment transport is recognized and evaluated using available modeling techniques (Shields et al., 2003). This method has been conveniently summarized by Shields in four basic steps:
  1. Determination of sediment load for the project reach (use hydraulics and sediment transport models)
  2. Development of multiple slope-width solutions that satisfy resistance and sediment transport equations (check width to depth ratios)
  3. Reduce the number of solutions to meet site constraints (ensure neither aggradation nor degradation is likely to occur see Figure 6)
  4. Computation of sediment transport capacity downstream of project reach

This method, as summarized by Shields, is useful only for single-thread channels because it was developed using one-dimensional models. For use within a braided channel Shields recommends using two or three-dimensional models (Shields et al., 2003). The key to success within the active-bed method is sediment continuity for the defined project reach, as well as, upstream and downstream boundaries. In order to achieve continuity the design variables of slope, width, and depth must be obtained and refined using discharge, sediment load, and channel composition. If this data is unavailable, a reference reach may be used (Shields et al., 2003).

Figure 6: Stable Design Chart

Final design of the channel and geometric detail, for both the threshold and active-bed approach, should reflect conditions present in nearby reference reaches that are in a current state of dynamic equilibrium. Reference reaches used in adjusting design should reflect similar characteristics, such as bed material, channel slope, Qcf, and hydrologic conditions, for the project reach. Additionally, it is often useful to perform stability checks on the final design to ensure project success (Shields et al., 2003).
Stability checks are encouraged to ensure the design aspects meet the requirements. The use of sediment transport models are used to simulate certain hydrologic events within the design reach to ensure aggradation or degradation does not occur along the project reach (See Sediment Transport Models section by A. Symonds for details on sediment transport modeling). Using bank stability models to analyze the stability of the design is also useful (Shields et al., 2003) (See Bank Stability section by G. Slape for in depth coverage on bank stability).
The final channel alignment can be determined from historic aerial photos or maps when appropriate, i.e. slope and channel discharge match historical records. Sinuosity should be kept similar to surrounding reference reaches when appropriate but bends should not keep constant dimensions (radius or length). River dimensions, such as slope, width, and depth, should not remain constant, but instead should be used as averages where the design “captures the spatial variability typical of lightly degraded systems” (Shields et al., 2003).
Natural Channel Design
Rosgen’s method of natural channel design (NCD) has three main components: the classification system, the 40-step design guideline, and stabilizing structures. The primary goal of Rosgen’s classification system is to determine what processes are affecting a given reach through examination of its physical form (Lave, 2009). As Rosgen stated its purpose is to “predict a river’s behavior from its appearance” (Rosgen D. L., 1996). This classification system works by dividing channels into alpha-numeric categories based on physical characteristics such as slope, planform, level of entrenchment, width to depth ratio, and bed material. The idea behind the classification system is the correlation between the morphological features and the processes which caused formation of those features. Rosgen’s use of the classification system attempts to indirectly assess the driving forces without direct measurement (see Stream Classification Techniques section by T. Gillihan for details on stream classification) (Lave, 2009).
The second component of NCD is the 40-step design process which has been summarized by Rosgen into eight phases in the USDA-NRCS Stream Restoration Design Handbook:
  1. Clearly and concisely define the project’s objectives, be they related to physical, chemical, biological, or human goals.
  2. Develop or verify regional curves for geomorphic, hydrologic, and hydraulic data. Determine valley types and stream classifications.
  3. Assess the stability and sediment supply of the restoration reach in relation to its watershed to determine the cause and direction of change or impairment. Obtain and analyze reference reach data.
  4. Seek a passive solution, such as change in land use management. If none is available, move on to Phase 5.
  5. Combine all of the data gathered in the previous steps. Based on this data, complete a design and test its compatibility with the hydraulic and sediment regimes in the watershed.
  6. Select and design appropriate enhancement and stabilization structures, such as cross-vanes, W-weirs, or j-hooks.
  7. Implement the design, including daily construction supervision.
  8. Develop and implement a plan for monitoring and maintenance (Rosgen D. L., The Rosgen Geomorphic Approach for Natural Channel Design, 2007).

From the eight phases listed above it can be noticed that the NCD approach depends on regional curves, reference reaches, and dimensionless ratios. “Regional curves describe empirically derived relations between drainage basin area within a given hydro-physiographic province and channel mean depth, width, cross-sectional area, and discharge” (Lave, 2009). The Natural Channel Design method relies on the regional curves in addition to physical indicators of bankfull discharge to determine the bankfull elevation and discharge. If the site has no evidence of physical indicators the regional curves are solely used (Lave, 2009).
In classifying the stream of interest the NCD approach requires characterization of the stream and bank sediments. Sediment samples are taken along both banks (below the bankfull levels) and across the channel bed to aid in characterization of the channel. These samples are then combined in order to create a particle-size distribution. This then allows the designer to classify the stream in relation to sediment composition (Simon et al., 2007).
Once the stream has been classified and its conditions characterized through physical inspection and data collection, the next step in the NCD approach is to predict channel adjustments. Using the channel form from the stream classification, NCD practitioners can then predict channel adjustments, sediment transport rates, and system disturbances (Rosgen D. L., 2001). It is important to note that most of the data collected for this method and used in predictions is focused on characterizing the stream at bankfull discharge (Simon et al., 2007; Kochel & Miller, 2010).
In order to evaluate bank stability the bank-erosion hazard index (BEHI) and near bank stress (NBS) are used in combination. The BEHI method is a qualitative technique in which bank properties such as height, angle, protection (vegetative cover), bankfull height, and root depth and density are used to evaluate bank stability. The bank properties are used in a worksheet to obtain a BEHI rating (U.S. Environmental Protection Agency (EPA)). The NBS method is another qualitative technique used to estimate the near bank stress. This method uses properties such as slope, depths, radii, and ratios in a worksheet to obtain a NBS rating (U.S. Environmental Protection Agency (EPA)). Worksheets and a more in depth description for bank erosion rates can be found on the EPA website on Bank Erosion Prediction.
The next step in Natural Channel Design is to find a reference reach on which to base the design. This reference reach should have the same classification, discharge, and sediment characteristics in order to be a reasonable match. Rosgen stresses the use of as many examples as possible for reference reaches, in order to establish a range of values on which to base the stream design (Lave, 2009). In some cases where data is unavailable for the project reach, data collected from reference reaches are scaled appropriately and applied to define the stream design parameters (Kochel & Miller, 2010).
The final step in the design process includes the use of in-stream structures such as cross-vanes, J-hooks, W-weirs, and single-arm vanes (Kochel & Miller, 2010; Lave, 2009). These structures are used to stabilize the channel while complimenting design aesthetics through the use of natural materials such as boulders, logs, and root wads (Lave, 2009). Using natural materials to stabilize the channel also increases habitat diversity, but in turn often has an unspecified design life (Kochel & Miller, 2010).
Comparison of Process-Based Design and Natural Channel Design
The biggest advantage of using the Natural Design Method can be realized from the nation-wide use of the stream classification system due to its simple application and repeatability (Lave, 2009). On the contrary the disadvantage of NCD lies in its simplistic empirical approach which fails to identify the fluvial processes inherent in channel formation. The biggest benefit of Process-Based Design lies in its ability to identify the links between channel formation and fluvial processes (Simon et al., 2007). The disadvantage associated with Process-Based Design can be attributed to the large datasets necessary and the amount of time required for analysis, which can drive up the costs associated with river restoration. In summary, NCD can be described as a empirical method to determine stream form, where Process-Based Design relies heavily on an analytical approach to determine stream form and define the level of stability within the stream.
Natural Channel Design’s classification system has been adopted by governmental agencies and is sometimes even required to obtain funding for river restoration projects (Lave, 2009; Simon et al., 2007). The method used by practitioners to create the particle-size distribution curves uses a soil sample containing both the bed and bank materials. This proves to be disadvantageous and, in fact, inconsistent for classification because two streams with the “C” classification could potentially end up being classified as the same stream type of “C5” (see Stream Classification Techniques by T. Gillihan for more information). This is made possible because both streams can have the same median diameters in the sand range, but differ in that one stream can have a gravel bed and silt-clay banks, whereas the other stream has a sand bed and sandbanks (Simon et al., 2007). The issue then compounds as the designer begins to compute sediment transport regimes for either of the streams, where it is clear one is transporting sand like sediments and the other may classify as a threshold channel depending on the gravel bed size. This emphasizes the importance of taking separate samples for both the bed and the banks in order to obtain proper classification and sediment transport regimes. This however does not prove the classification system to be inconsistent, because if used properly, each stream can be classified appropriately and consistently by different practitioners.
Process-Based Design has its own form of a classification system in the use of channel evolution models (CEM), which are used to determine the stage or state of energy within a given stream. However, practitioners only use these CEMs as estimations or early predictions because the stream’s state of energy will become clear through analysis. During analysis designers quantify the resistors and drivers of channel form through the stream power proportionality: Stream_Power_Proportionality.JPG. Where Q is discharge, Sb is bed slope, Qs is bed material discharge (sometimes known as effective discharge), and d50 is the median grain size of bed material (Simon et al., 2007). NCD bases its channel form on the classification system and uses this form to predict channel adjustments (Rosgen D. L., 2001). Use of the NCD approach can have drastic consequences for a stream, creating unstable morphological conditions. Figure 7 shows an example project which used the NCD approach and failed after a six year flood event (Kondolf et al., 2001). “Understanding alluvial channel behavior, channel response to disturbances, and stable channel forms can be accomplished by concentrating on those factors that directly control the balance or imbalance between applied forces and boundary resistance” (Simon et al., 2007). If the processes inherent to channel formation had been fully understood before design, this failure could have been avoided.
Figure 7: Uvas Creek, California (a) after restoration using the NDC approach and (b) result after a six year flood

The use of channel evolution models differs from the stream classification system in that, CEMs are based on the energy shift between driving and resisting forces and “are not tied to specific ranges of channel shape” (Simon et al., 2007). Additionally, CEMs consist of a sequence of stages but do not imply that each reach will go through each stage in order (see Channel Evolution Models by J. Collison of the wiki page). It is also advantageous to use CEMs as it allows the designer to understand whether the channel instabilities are just localized (within the defined reach) or if the instabilities represent channel adjustments of the entire stream system (Simon et al., 2007). NCD’s stream classification system is time independent (Rosgen D. L., 1994), thus it does not capture the dynamics (defined as the magnitude and rate of erosional and depositional processes through time) of alluvial streams (Kochel & Miller, 2010; Simon et al., 2007). Use of the NCD in an unstable reach of a stable system has the potential to be successful because of the localized issue which can be effectively uncovered using the stream classification system (Simon et al., 2007).
The Natural Channel Design method provides the user with the 40-step design process which enables designers to follow a set rule rules in the design process with specifics on how to accomplish individual steps. Critics argue though that this “cookbook” approach to channel restoration design fails to capture the regional specificity of stream channels (Lave, 2009). Process-Based Design provides basic steps but allows for different methods of analysis with the use of differing modeling techniques and equations which better define the stream system (Shields et al., 2003). This is where the argument begins to take a different direction in identifying the need for a common approach or set of guidelines for river restoration (Lave, 2009). In an attempt to develop national restoration handbooks the National Resource Conservation Service created a design handbook (National Resource Conservation Service (NRCS), 2007).


The common goal of both the Natural Channel Design process and Process-Based Design, to maintain or increase ecosystem good and services, can be used as a basis for which to create standards for river restoration (Palmer et al., 2005) (see Setting restoration goals and objectives by Rsazo-Hinds for details). The most effective river restoration should balance the overall stakeholder success, ecological success, and learning success of the project. Where stakeholder success is defined by the human satisfaction with the project outcome and learning success is defined as “advances in scientific knowledge and management practices that will benefit future restoration action” (Palmer et al., 2005). Thus, both channel design methods can provide successful projects if used appropriately, though the use of NCD is more appropriate for well-experienced practitioners. Process based design focuses on the link between channel formation and fluvial processes, but in doing so requires more time, data, and money for analysis. Although Natural Channel Design is defined as a "cookbook" approach, it requires less time, data, and money for implementation and if used by an experienced practitioner can produce a successful stream restoration project. In summary, process based design uses an analytical approach to define fluvial processes, whereas Natural Channel Design focuses on the physical form and uses an empirical approach.


Beck, S., Ramey, M., & Stilwell, C. (2000). Sedimentation Engineering Restoration of Silver Bow Creek near Butte, Montana. Proc. 1999 Int. Water Resources Engineering Conference. Reston, Va.: Environmental and Water Resources Institute (EWRI).
Doyle et al. (2007). Channel-Forming Discharge Selection in River Restoration. ASCE: Journal of Hydraulic Engineering, 831-837.
Emmett, W. W., & Wolman, M. G. (2001). Effective Dishcarge and Gravel-Bed Rivers. Earth Surfaces Processes Landforms, 1369-1380.
Federal Interagency Stream Restoration Working Group (FISWRG). (1998). Stream Corridor Restoration: Principles, processes and practices. Springfield, Va.
Goodwin, P. (2004). Analytical Solutions for Estimating Effective Discharge. Journal of Hydraulic Engineering, 729-738.
Johnson, P. A., & Heil, T. M. (1996). Uncertainty in Estimating Bankfull Conditions. Water Resources Bulletin, 1283-1291.
Knighton, D. (1996). Fluvial Forms and Processes: A New Perspective. Arnold, London.
Kochel, R. C., & Miller, J. R. (2010). Assessment of Channel Dynamics, In-Stream Structures and Post-Project Channel Adjustments in North Carolina and its Implications to Effective Stream Restoration. Environmental Earth Sciencs, 1681-1692.
Kondolf et al. (2001). Design and Performance of a Channel Reconstruction Project in a Coastal California Gravel-Bed Stream. Environmental Management, 761-776.
Kondolf, G. M. (1996). A Cross Section of Stream Channel Restoration. Journal of Soil and Water Conservation, 119.
Lave, R. (2009). The Controversy Over Natural Channel Design: Substantive Explanations and Potential Avenues for Resolution. Journal of the American Water Resources Association, 1519-1532.
Montgomery, D. R., & Buffington, J. M. (1997). Channel Reach Morphology in Mountain Drainage Basins. Geological Society of America Bulletin, 596-611.
National Resource Conservation Service (NRCS). (2007). Stream Restoration Design. In National Engineering Handbook (p. Part 654). Des Moines, Iowa: Department of Agriculture National Resources Conservation Service.
Newbury, R. W., & Gaboury, M. N. (1993). Stream Analysis and Fish Habitat Deisgn: A Field Manual. Gibsons, B.C.: Newbury Hydraulics, Ltd.
Palmer et al. (2005). Standards for Ecologically Successful River Restoration. Journal of Applied Ecology, 208-217.
Riley, A. L. (1998). Restoring Stream in Cities: A Guide for Planners, Policymakers, and Citizens. Washington: Island Press.
Rosgen , D. L. (2001). A Stream Channel Stability Methodology. Seventh Federal Interagency Sedimentation, (pp. II-18 to II-26). Reno, Nevada.
Rosgen, D. L. (1994). A Classification of Natural Rivers. Catena, 169-199.
Rosgen, D. L. (1996). Applied River Morphology. Pagosa Springs: Wildland Hydrology.
Rosgen, D. L. (2007). The Rosgen Geomorphic Approach for Natural Channel Design. In Stream Restoration Design (p. Part 654). Des Moines: U.S. Department of Agriculture National Resources Conservation Service.
Shields et al. (2003). Design for Stream Restoration. ASCE: Journal of Hydraulic Engineering, 575-584.
Shields, F. D. (1983). Design of Habitat Structures for Open Channels. ASCE: Journal of Water Resource Planning and Management, 331-344.
Simon et al. (2007). Critical Evaluation of How the Rosgen Classification and Associated "Natural Channel Design" Methods Fail to Integrate and Quantify Fluvial Processes and Chanel Response. Journal of the American Water Resources Association (JAWRA), 1117-1131.
Society for Ecological Restoration. (2002). The SER Primer on Ecological Restoration. Retrieved from Soceity for Ecological Restoration Science and Policy Working Group:
U.S. Environmental Protection Agency (EPA). (n.d.). Bank Erosion Prediction. Retrieved from Watershed Assessment of River Stability and Sediment Supply:
U.S. Environmental Protection Agency (USEPA). (2000). Principles for the Ecological Restoration of Aquatic Resources. Office of Water, Washington, D.C.
Williams, G. P. (1978). Bankfull Discharge of Rivers. Water Resources Res., 1141-1154.