The maturation of the Clean Water and the Endangered Species Acts-with their regulatory reach to stormwater, wetlands, natural resource management, instream habitats, and most recently nonpoint sources of water pollution-has fueled a revival of interest in rivers. The country is moving away from its midcentury mindset, wherein we reveled in our postwar technological ability to master the forces of nature. We've looked over our shoulders at a half century or more of harnessing rivers and have learned a few things. A great popular interest has sprung up in stream restoration. In some areas, stormwater utility groups are reinventing themselves as the watershed people and many state and federal agencies have come to regard streams as the ultimate expression of the health of ecosystems at the landscape scale.

A 1969 photorevision (in purple) of this 1949 topo clearly shows the dynamic changes of the alluvial Sacramento River.

Managers face thorny questions when it comes to "fixing" riverbanks. Public safety, facilities, and infrastructure might need to be protected, but protecting a streambank from erosion in one location might create new problems both upstream and downstream. A stream may be changing because of a change in discharge characteristics, sediment supply, or other reasons, and it can be difficult to determine what channel shape and runoff events to expect under current and future conditions. No matter what the issue, decisions about river treatments are difficult and complex. They are made more so by increasing scrutiny of projects with respect to their impacts on fish passage and instream habitats and increasing common knowledge about the dynamics and functions of river systems.

Well-intentioned decision-makers need theoretical and practical frameworks within which to consider whether riverbank restoration or protection projects might be successful. They need background to help them become more savvy consumers of professional services for river projects. And they need to know what kinds of projects are likely to be successful and when to walk away from potential disasters.

Do No Harm

The first rule of stream restoration or rehabilitation is: Do no harm. Is the project even necessary? Not all bank erosion is bad, and not all "problems" need to be "fixed." An interdisciplinary analysis of alternatives, costs, and risks can assist with decision-making at the "go" or "no-go" stage.

Don't Go It Alone

No single discipline has the whole package of technical or analytical skills to go it alone where river design is concerned. By its very nature, this field is interdisciplinary. Depending on the cultural and physical setting of a site and the natural resource issues involved, project teams may consist of people with expertise ranging from invertebrate taxonomy to transportation engineering. The insights of people with professional depth in the biological, physical, and cultural attributes of the site and its contributing watershed need to inform the project at the outset.

To Go Forward, First Take a Big Step Back

Before considering what might be done to fix a problem at the site, it is imperative to understand the hydrology; the dynamics of sediment production, transport, and deposition in the system; and the morphology of the stream before present-day influences. Historic mapping, original survey notes, records of gravel mining, gauge data, land-use mapping for the watershed, old aerial photos, and interviews with long-time residents can all contribute essential pieces to the puzzle of the stream's original condition. Government offices at all levels can be gold mines of historic information. Local universities might have master's theses that shine light on the system, and musty documents in local government repositories can hold the only surviving records of former river projects and, incidentally, goals and beliefs about river systems. An undisturbed reference stream-one with a similar watershed size, geology, aspect, elevation range, and vegetation-could provide important clues.

For many streams, a background search might disclose that runoff quantity and frequency have increased as a result of development or land-use activities in the watershed. Or perhaps a water-diversion or gravel-mining operation has upset the water/sediment balance in the stream. Maybe a few curves were taken out of the river a long time ago, and the channel responded to an increased gradient by downcutting. Perhaps vegetation or sediments have been removed from the channel to improve flood flow, increasing the stream's erosive energy. A headcut might be making its way upstream. Whatever the cause of streambank or channel change at the problem site, it is important to understand both the source of the change and the response of the channel.

The Present Is the Key to the Past

In 1785, a Scots doctor and farmer, James Hutton, published a paper explaining how streams can cut deep gorges. The principle of uniformitarianism grew from Hutton's observation and became codified nearly two centuries later as the first fundamental concept of geomorphology.

The top of the active channel bar indicates bankfull. Bankfull can readily be identified as the top of the active channel bar and the location trim line in the background.

The principle states that the same physical processes and laws that operate today operated throughout geologic time, although not necessarily always with the same intensity. Considering the late discovery of this phenomenon relative to the long history of Western civilization's engineering and scientific achievements, our knowledge of river behavior is still young. Notwithstanding, Hutton's principle is as good a place as any to begin to understand rivers.

One way to internalize this principle is to stand on the bank of a river where there is a problem and understand the problem as the expression of several variables, of which one or more might be changing while the others are responding. Ah, yes, of course, we might observe. Given the unique circumstances of gradient, discharge, sediment size, and load, this river is behaving exactly as all rivers throughout the history of time would behave under the same conditions.

In other words, river behavior is predictable. Modern fluvial morphologists explain that the following major variables work independently to define channel shape: stream discharge at bankfull stage, the amount and particle size of sediments, and the slope of the stream. When one factor changes, the others will adjust. Viewed from this perspective, we are more likely to look for the source of riverbank erosion when we are considering potential fixes for streambank erosion. This realization is revolutionizing the way we treat rivers.

The Principle of Least Work

Most people are fascinated by big runoff events. Big events remind us of the incredible energy of moving water. They refresh institutional resolves to rely on interdisciplinary decision-making about such actions as land-use planning and natural resource and range management-actions that can have far-reaching impacts on streams, streambanks, and floodplains. Big events make us humble. They reinforce our appreciation of the complex interface between nature and culture.

But despite the dramatic results of big events, current wisdom in fluvial morphology is that channel shape is not controlled by catastrophic events but by bankfull flow. This is a cornerstone concept for those involved with streambank restoration, for it changes long-held cultural expectations about the river stage to which streambank treatments should be designed.

Bankfull flow is the dominant high flow conveyed by the channel-the one that occurs about once each year. Bankfull discharge moves the most sediment and water for the least amount of energy. If we consider an alluvial stream on a topo map, it is easy to see how this is so. We see a sinuous channel with length greater than the length of the valley that contains it. We can take a piece of string and lay it out on the meanders of the river, then measure the length of the string. If we measure valley length over the same horizontal distance, it will be much shorter. Therefore, the gradient of the valley is steeper than the gradient of the river.

But during bankfull flow, the wetted perimeter of the channel increases in depth and width, and as a consequence, the channel at bankfull flow is less sinuous and therefore steeper. The differences in the slope of the water surface in pools and riffles at low water balance out during bankfull flow. Thus, the straighter channel and steeper gradient achieve an efficiency to convey both greater discharge and sediments during bankfull flow. This principle should not be construed to mean that low-flow channels should be straight. On the contrary: Low-flow channels are the most sinuous in nature; bankfull channels less so. Floodways, which typically include the entire meander belt of a river system, are often relatively straight.

This principle of least work, as it is sometimes called, can seem suspiciously anthropomorphic-as if a river is similar to a recalcitrant teenager figuring out how to vacuum without using any elbow grease. We wonder: How can a river decide to spend less energy? But those who have studied the physics and energy of rivers more or less agree that the meander is the most efficient way in which gradient and velocity are balanced under different conditions of discharge, sediment load, and channel roughness. William Morris Davis observed that the large volume of large rivers in their lower reaches enable them to run rapidly, even on a gentle slope. Most people would be surprised to learn that mountain streams cascading down their bouldery channels actually are flowing at about the same speed as alluvial streams on floodplains-about 3-5 ft./sec.

One hallmark of current approaches to river projects is the identification of bankfull elevation in a channel. The Rosgen method (www.wildlandhydrology.com) turns on this, as do several other methodologies. This is an important starting point, for it can help a project team understand whether and how a river channel might be changing. Bankfull is determined in the field by noting the elevations of indicators, including the tops of active depositional surfaces in the channel; the level of wash or scour of exposed roots; the elevation of persistent woody vegetation, such as alders; the upper limit of recent rock scour; and bedrock benches.

Although mature conifers have been retained at the river's edge, bank erosion has occured because of the removal of shrubs and ground cover.

It is important to note that bankfull is not always synonymous with top-of-bank. Everyone involved in the project needs to get on the same page with respect to the location of bankfull elevation, otherwise some serious mistakes can be made. After bankfull elevation is determined, the flood-prone area is qualitatively identified as two times maximum bankfull depth.

Floodplain and Channel Roughness

In the process of establishing bankfull, it may become apparent that the flood-prone area includes a portion of the upper bank as well as the terrace immediately adjacent to the river on one or both sides. For many managers, this zone tends to be a trouble spot. Large rock, living vegetation, and large organic debris in this zone may be viewed as creating an impediment to passage of flood flows. If these impediments are removed, however, in order to create a more efficient channel that can move both water and sediments through the reach more quickly, problems might merely be passed downstream where, ultimately, lower velocities, greater channel roughness, and increased deposition will create more flooding. This is the Catch-22 of simplification for flood control.

In this steep colluvial channel, energy is expended in turbulent flow.

Ironically, today's fluvial morphologists tell us that channel and floodplain roughness provide an essential way for stream energy to be used up before it can get to the damaging stage of bed and bank erosion. In fact, stream energy is used in this order: overcoming internal friction, overcoming the friction of bed and banks, transporting organic debris and sediment, then eroding streambed and banks. Therefore, the removal of channel and floodplain roughness elements actually facilitates the onset of bed and bank erosion.

What's more, if the flood prone area is a grassy park or golf course that is maintained in a smoother state than the channel, flooding is apt to result in greater erosion than if the area possessed its original roughness elements. Further, flood-prone areas that have only a fringe of vegetation parallel to the river are more apt to experience erosion behind this vegetation during flood events than if the vegetation extends perpendicular to the river. An analogy of this would be if you were to take a turn that puts you on the wrong side of a landscaped median. Once you're on the wrong side, it's hard to get back.

Finally, stream meandering is understood today to correlate to lack of channel roughness. But not too long ago, the opposite was believed to be true, and the result is that we've inherited a lot of rivers that were intentionally straightened and stripped of vegetation. In a study of 120 streams in the Pacific Northwest, fluvial morphologist Janine Castro of the US Department of Agriculture's Natural Resources Conservation Service found that relatively straight channel form correlated to channel roughness and the presence of large wood.

Not All Channels Are Created Equal

Geologist William Thornbury-whose 1954 text, Principles of Geomorphology, is perhaps the pinnacle of the descriptive school of geomorphology-stressed the importance of understanding the effects of climate zones and geologic materials and structure on local geomorphic processes and their role in the evolution of landforms.

Channel downcutting is accomplished by abrasion.

It's good to keep in mind when evaluating the dynamics of streams. Characteristics and processes to be expected in one system might not necessarily be expected in another. A case in point: In the arid and semiarid Great Basin, where rainfall is scant, sediments are swilled down mountain channels during episodic runoff events associated with infrequent convectional thunderstorms. Where the channels debouch onto the basin floor, the sediments are deposited in alluvial fans. Flooding events on alluvial-fan surfaces tend to involve multiple active channels and periodic avulsion or rapid establishment of new channels as older ones become filled with sediments.

Over time, these processes build coalesced fan systems, or bajadas, that begin to bury the mountain fronts in their own waste. In better-watered regions, rock waste delivered to the lowlands tends to be carried away by streams. The source mountains do not get buried but, instead, are worn down. Under certain circumstances, braiding may occur around bed load deposited at the footslopes of mountains.

With this in mind, it is helpful to be aware that many stream-assessment protocols were developed for a specific region and reflect characteristic fluvial processes of the region. They might not be applicable elsewhere. The Rosgen system of stream classification, however, classifies rivers according to relationships of channel variables independent of geomorphic setting and, in general terms, can be used to predict channel responses to changes in these variables.

Sediments Are in Motion

If we envision a stream in the Temperate Zone as rising in the mountains and flowing to low-gradient lowlands, we can think of it as huge conveyor belt moving earth materials from higher to lower ground. In fact, one current definition of a watershed states that it is merely an open system in which energy and earth materials are moving toward the lowest place in the landscape. In the uppermost-or sediment source-zone, materials are delivered to confined channels by gravitational processes: rockfall, debris avalanches, landslides, creep, and ravel. Once in the channel, this rock material grinds, bumps, and rubs against the streambed and banks, incising the bed and undercutting the slopes above. Sediment deposition in the channel may occur in this zone, but it is forced by such channel obstructions as large rock and woody debris. These may give way in large events, resulting in pulses or slugs of sediment being moved downstream all at once.

The channel materials may be so large in the sediment source zone that their presence in the channel makes water flow around them, further undercutting the steep hillslopes. These processes eventually result in backwasting of the slopes that bound the channel and widening of the channel. The intermediate zone between the upland's steep, turbulent, colluvial channel and the more sedate pool and riffle sequence of the valley floor's low-gradient, finer-grained alluvial channel is an interesting and dynamic zone of sediment transport and deposition. In an alluviated canyon, as this zone is sometimes called, increased discharge, increased channel width, and decreased slope have combined to scour and deposit discontinuous terraces at the channel margin. These often become veneered with boulders, cobbles, gravel, and sand and, while they may be colonized by riparian plants, are subject to inundation. High-flow channels cross these terraces, and the successional stage of riparian plants on them can deceive managers into thinking that these terraces are abandoned features no longer used by the river.

But in fact, the sediments stored at the river's margin in alluviated canyons should be considered to be in long-term transit down the river corridor. This zone is best understood as a balanced system in which the amount of sediment moving in is equal to the amount being transported out. It is safest to assume that nothing here is permanent. After a big event, the active channel bars appear to have remained in place, but in actuality, the original sediments have been carried downstream and have been replaced by newcomers transported from upstream.

Examples of classic eroded, alluviated, and terraced valleys.

The direction of flow during big events in alluviated canyons and alluvial valleys can take managers by surprise. When the discharge no longer fits within the sinuous low-flow channel and spreads out to occupy the width of the canyon, the channel is effectively straightened and steepened. This can result in flow direction during a big event being as much as 90º different in some areas (up to 180º in the case of very sinuous channels) than direction of flow during low flow. Knowing this and knowing how to interpret bars, terraces, and high-flow channels can help managers defend these channel margin areas from development and other incursions, including simplification of channel-margin vegetation.

An accurate understanding of the dynamics of cobble- and gravel-bedded streams is essential for design of riverbanks or in-channel projects in streams with deformable boundaries. Researchers at the Stream Systems Technology Center of the Rocky Mountain Research Station (stream/rmrs@fs.fed.us) have identified 10 general attributes of alluvial channels. Among these attributes are a channel-bed surface that is frequently mobilized (every one to two years), periodic channel-bed scour and fill, periodic channel migration, and infrequent channel resetting floods (10- to 20-year recurrence).

The larger materials of such channels tend to become armored by the deposition of smaller grains between them. It takes a threshold mobilizing flow to dislodge the packed materials, and when this occurs, the entire bed to a depth of several grain diameters may be moving. This is contrary to the popularly understood model, which is that sediments are changing places on point bars in a downstream direction.

Deformable bed streams are expected to cyclically reset their channels. The new channel will have the same radius of curvature, width, pool-riffle ratio, and gradient as the old channel. Because of this, current thinking is that setback levees can allow deformable bed streams some wiggle room. This way, a channel-resetting flow does not become a disaster. The dynamics involved in a channel-resetting flow can also set the timer over again for riparian plant associations, and this contributes to habitat diversity in the alluvial stream corridor.

Choose Stream Projects Carefully

It's no secret that we've built plenty of roads, houses, bridges, sewer treatment plants, refineries, and parks too close to dynamic channels. Yet the protection of structures and infrastructures at the river's edge can result in long-term expenditures as well as additional costs of fixing secondary downstream impacts. Perhaps because of this, we are beginning to see a new trend of floodplain reclamation, in which the costs of acquisition have become more favorable than the ongoing costs of protection.

But there always seems to be a need for streambank armoring to protect structures or facilities. Today it is recognized that to be successful in the long run, streambank armoring requires more than bigger rock. The design must take into account how the channel will continue to respond to ongoing watershed disturbances and how it will respond to the streambank treatment itself. And it is essential to have people on-board who can measure the variables and crank the equations that will prescribe the channel shape, roughness, curvature, and slope that will represent equilibrium for a reach.

At a time when fish passage is receiving much attention, it is important not to rush out and remove all culverts. It is essential to know the location of the natural grade control in the system before embarking on the removal project. Where a culvert outfall has scoured a plunge pool, or the culvert has been placed in a "cut" section of road, the pool and/or cut will behave similar to a headcut in an agricultural field. Unless the stream has a bedrock floor, the nickpoint, as it is called, will migrate upstream after the culvert is removed, resulting in grade adjustments throughout the watershed above. This means the stream will respond to the new, steeper gradient by eroding its bed. The sediments will be transported downstream.

Sediments being transported by a stream will be deposited according to their mass when the stream loses velocity. Velocity loss may occur as a result of a decrease in flow or slope, such as when a confined channel debouches onto a valley floor. Or deposition may be forced behind channel obstructions, such as rocks or large organic debris that locally impede flow. Typically, as flood levels drop, deposition occurs in areas of the channel where velocity is lowest. An engineered change in channel shape might bring about inadvertent deposition. If a channel is made wider and flatter to accommodate flood flows, this might in fact result in lower velocities, subsequent deposition, and increased flooding.

Most practitioners will say that it is easier to work on a small system than a large one. This is easy to understand in light of one of the most fundamental principles of hydrology-The Law of Transportation. This principle states that the transporting power of a current varies as the sixth power of velocity. In real terms, this means that if velocity doubles, it can transport a rock mass 64 times greater. Make no mistake: The size of river-rounded materials in a channel are an accurate indication of the competence of the stream to transport this material at some river stages. Ironically, as a rule, alluvial streams do not tend to have large rock. This reflects channel width, which increases downstream as the square root of bankfull discharge. The wider channel, with its flow core centered in the deepest part of the channel, conveys the flow at a low gradient with minimum friction and relatively low velocity.

People who work with and study rivers today have found new and compelling ways to explain the complex and delicate relations between energy and mass in river channels. Their discoveries echo Sir Isaac Newton's observation in Mathematical Principles of Natural Philosophy in 1687: "To every action there is always opposed an equal reaction: or, the mutual actions of two bodies upon each other are always equal and directed to contrary parts."

There are many benefits to working with, not against, the natural tendencies of streams, including a decrease in initial costs for projects and a decrease in costs for long-term maintenance. One aspect of a successful project is that the site does not need to be reentered and disturbed over and over again to fix the "fix." This is especially germane when there is a need to protect sensitive, threatened, and endangered species associated with river-corridor habitats. This new school of river design also extends to river-corridor benefits increasingly prized by managers: enhanced nutrient filtering, enhanced public greenspaces, improved groundwater levels, and better habitats for fish and wildlife.

In many places, human uses of the landscape are imposing changes on streams and watershed hydrology at a more rapid rate than nature can respond. In many cases, the fix for this disturbed stream morphology will not be to restore it to a previous condition but to create a new channel that is adjusted to the current basinwide conditions. River work has entered an interdisciplinary era in which practitioners must concede to the complexity of the riverine environment. This means design solutions must be informed by fluvial morphologists; hydrologists; civil, structural, and geotechnical engineers; geologists; land-use and recreation planners; landscape architects; economists; and politicians. Ironically, such an intertwined approach reflects the complex nature of rivers.