The geomorphology of the watershed and channel play key roles when locating bridges. Basic geomorphic principles help designers understand the geomorphic processes and difficulties that can arise when bridges cross streams at various positions in the watershed. These processes change with the crossing's location in the watershed and along the reach that will be crossed. Channels are extremely dynamic, responding to changes in the watershed by propagating changes downstream to upstream and vice versa, depending on the channel's elevation in the watershed, the type of disturbance, and the channel types along the stream. To choose the best location for a bridge, the designer should address the following questions:

• Where is the crossing location in the watershed and how does the stream transport water, sediment, and wood?
  
  
• How is the channel configured?
  
  
  • What is the degree of channel containment/entrenchment?
    
    
  • During high flows is water conveyed in the flood plain (flood plain conveyance)? If so, how much? Are there side channels or flood swales?
    
    
  • Can the stream move laterally and affect the crossing during the structure's design life? Are the stream's banks erodible or not?
    
    
  • What is the range of vertical fluctuation of the streambed during the structure's design life?
    
    
• How well does the trail and bridge alignment mate with the stream alignment?
  
  
• Is the channel stable?
  
  
• Is the channel adjusting to recent large-scale disturbances (such as landslides)?

The location of a stream reach in its watershed determines the reach's channel morphology and responsiveness to natural or manmade disturbances (Gubernick and others 2003). Slope, discharge, sediment, and vegetation are the main controlling factors. The way a channel is configured provides information that can help you decide whether a crossing is in a good, safe location or whether the location will require extensive analysis and design and where the crossing may be costly. Channel classification has been an excellent tool for describing stream configurations and for interdisciplinary communication. The two main channel classification schemes are Montgomery and Buffington (Montgomery and Buffington 1993) and Rosgen (Rosgen 1994).

The Montgomery and Buffington system is based principally on watershed position, slope, and the geomorphic description of bed characteristics. Additional information may be found at http://www.fgmorph.com/.

The Rosgen system (figures 23 and 24) is based on slope, entrenchment ratio (figure 25), bankfull width to bankfull depth ratio, sinuosity, and bed material. Both the Rosgen and the Montgomery and Buffington systems have utility, but this report relies on the Rosgen system (Rosgen 1994, 1996). The U.S. Environmental Protection Agency has developed a watershed management training Web site. The analysis and planning modules include the "Fundamentals of the Rosgen Stream Classification System" http://www.epa.gov/watertrain/.

At an ideal bridge crossing, all floodwater and debris would stay in the confines of the existing channel. Such crossings would have high banks with a narrow flood plain or no flood plain at all. Rosgen's channel classification system illustrates that certain channel types are more vertically contained than others.

The entrenchment ratio is the flood stage width (see figure 13) divided by the bankfull width.

In channels with low entrenchment ratios (channel types A, B, F, and G, see figure 24), the majority of the discharge remains in the confines of the bankfull or active channel area even during floods (the flooded area does not get wider and wider as water rises, figures 26, 27, 28, and 29). When a bridge crosses such channels, it is relatively easy to provide good vertical clearance between the stream and the bottom of the bridge's girder. Channels with high entrenchment ratios (channel types C, D, DA, and E, see figure 24) tend to have active flood plains with low banks (figures 30, 31, and 32). Bridges built at such sites will require deep fills to provide enough vertical clearance. Streams with high entrenchment ratios often require additional drainage structures on the flood plain and wider crossings. Bridges built on such streams may pose problems for animals that need to cross the area.

Figure 23—The Rosgen system of stream classification showing examples of a broad-level
delineation of stream types. (Rosgen 1998, courtesy of Wildland Hydrology).

Figure 24—Broad-level delineation of major stream types showing longitudinal, cross-sectional,
and plan views (Rosgen 1998, courtesy of Wildland Hydrology).

Figure 25—Illustration of different entrenchment ratios (ER). Wfp is the
width of the flood plain. Wbf is the bankfull width.

Figure 26—Type A channel.

<
Figure 27—Type B channel.

Figure 28—Type F channel.

Figure 29—Type G channel.

Figure 30—Type C channel.

Figure 31—Type D channel.

Figure 32—Type E channel.

Identifying how much water flows over a flood plain and the width of the floodflow is a major consideration when channels are only slightly to moderately entrenched (entrenchment ratio of 1.4 or greater). Bridge designs should consider flood plain conveyance, the width of the bridge's opening, its vertical clearance, and the site's scour potential. If the flood plain has high conveyance, constrictions increase the chance that erosion may scour the streambed and the banks.

At a minimum, the bridge should span the stream's bankfull width with no piers in the stream. Additional culverts, slab structures (such as low water crossings), or fords can help reduce the constriction caused by trail approaches and help maintain a functioning flood plain. If crossings must be located in these areas, riprap or other materials are recommended to prevent excessive scour of bridge abutments.

All stream channels migrate laterally over time. Confined channels usually migrate more slowly than unconfined channels because of their underlying geology and bank composition. If the banks are composed of highly erodible materials (noncohesive finer grained sands, gravels, and cobbles), the banks adjust more easily than if they are composed of nonerodible materials (boulders, bedrock, and cohesive materials). Vegetation can also be a major factor influencing a bank's susceptibility to erosion. When deeply rooted vegetation is present, banks are less erodible. Material alone is not always the sole indicator of the likelihood of lateral movement. Streams with low entrenchment ratios and lower width-to-depth ratios (channel types A, B, F, and G, see figure 24) tend to have lower migration potential (less lateral movement) than those with high entrenchment and high width-to-depth ratios (channel types C, D, and E, see figure 24). Type E channels and channels with dense, deep-rooted woody vegetation can be very stable.

A transport reach (typically a length of river with moderate slopes) has a heavily armored streambed and tends to be stable (channel types A, B, and G, see figure 24). A response reach (typically a length of river with gentler slopes) usually has a fine-grained, noncohesive streambed (channel types C, D, DA, E, and F, see figure 24) and is more susceptible to scour and erosion. Streambeds in transport reaches tend to be less susceptible to downcutting (when the streambed is cut away and material washes downstream). Streambeds in the response reaches tend to aggrade (when the streambed is built up by materials that washed downstream) or degrade more readily with changes in sediment supply and discharge.

In streams where the response reaches are composed of cohesive materials (clay), the channel tends to be very stable and may have good sites for crossings. Establishing solid foundations at such crossings can be very expensive. Flood plain issues and stream sinuosity will need to be addressed.

Depending on their depth, headcuts (when a channel causes localized erosion upstream) can undermine bridge foundations or materials intended to prevent scour, such as riprap or gabion baskets filled with stones. Characterize the bed materials, using a longitudinal profile to determine potential headcut locations. A method to evaluate headcuts and vertical changes in the streambed is provided in "Stream Simulation: An Ecological Approach Providing Aquatic Organism Passage at Road-Stream Crossings" (Stream Simulation Working Group 2008).

When a wide stream flows into a narrow bridge opening or when the structure is not hydraulically aligned with the stream, back eddies can form, constricting the portion of the channel with unrestricted flow. Sediment transport and localized scour will increase. Field evidence of this condition includes aggradation above the structure, usually seen in the longitudinal profile as a flat wedge of sediment or as gravel bars. Bank scour can occur above or below the constriction because the changes in the channel's cross section create back eddies, increasing the boundary shear stresses and directing flow into the banks instead of parallel to them. Bed scour commonly occurs downstream, caused by increased outlet velocities and increased slope of the water's surface. Avoid locating structures in reaches with poor hydraulic alignment, such as a curve. If a structure is not aligned hydraulically, the flow could be restricted, raising the stream's surface. Bank armor may be needed farther up and downstream from the structure than if the structure had been hydraulically aligned.

Understanding how dynamic landforms behave over time can help when planning for maintenance and when considering alternative bridge designs or locations. For example, active alluvial fans are sediment deposition zones. Their channels change location frequently, sometimes rapidly, when sediment and debris deposits cause the channel to seek a lower level along the path of least resistance. If a crossing is on an active fan, streams can abandon their historic channels after a flood event, depositing trees or excessive sediment in the channel upstream. The crossing may fail catastrophically because of sediment or debris deposition, (which can reduce the area where the stream can flow, its cross section) at the structure.

It's best to avoid active fans. If you must locate a crossing in such areas, the best crossings would be below the alluvial fan or near its apex. These locations are beyond the active areas and may be better suited for siting a structure that will survive with the least maintenance.

In addition, low-cost structures such as a simple ford or low-water crossing may be most appropriate in flood plains or on alluvial fans.

If the crossing must be on an alluvial fan, large channel changes should be anticipated and the design should minimize the downstream consequences of the structure's failure by reducing the possibility that the stream could cut a new channel around the structure (Grant 1988).

All channels need to be assessed for stability at both the watershed (broadest) and reach (more narrowly focused) scales. It is particularly important to identify systemwide instability such as head cutting, because the structure's design needs to account for predicted changes in the channel. It is best to avoid crossings in unstable channels because predicting changes in width and depth can be difficult. System-wide instability usually can be seen in a series of aerial photos as noticeable changes in channel width, rapid growth and movement of depositional bars, alluvial fans at tributary mouths, and so forth. Frequently, large-scale channel changes are associated with land-use changes such as mining, agriculture, subdivision and road development, or logging.

As a rule of thumb, the heavily armored transport reaches (channel types A, B, and G with cobble and larger substrates) tend to be more stable and less affected by watershed changes than the response reaches (channel types C, D, E and F, see figure 24).