"Use of Riprap in Soil
Bioengineering Streambank Protection"

Soil bioengineering techniques offer a natural way to repair streambank erosion while benefiting the environment

Robbin B. Sotir & Associates Inc., Marietta, GA USA


Vegetation is, perhaps, the most important component of a riverine system. The role of vegetation in streambank stability is well known and widely accepted. Woody vegetation slows velocities in the vicinity of the bank, and the root systems help support the bank and reduce scour. On small streams, trees and shrubs provide shade that helps prevent solar radiation from increasing water temperatures, and overhanging vegetation provides much needed cover for fish and organic debris that is used for cover and food by aquatic organisms.

Clearly, streambank protection systems that incorporate woody vegetation provide additional benefits over those that do not. Soil bioengineering, a technology developed and refined largely in Europe, and more recently in North America, employs woody vegetation as the major structural component in streambank protection designs. This approach to streambank protection is being accepted increasingly in the United States and Canada, especially in areas where environmental quality is a major concern. Examples include streams in urban areas, parks, scenic locations, and streams with important salmonid fisheries.


TIn some applications adequate protection against erosion can be provided by vegetative systems alone. Most applications, however, require the use of some rock in conjunction with vegetation to prevent damage to the system that would impair its effectiveness or reduce its environmental benefits. Several examples are provided in this section to illustrate how stone is employed in soil bioengineering designs.

1.1 Vegetated dikes

Vegetated dikes, sometimes referred to as live booms in soil bioengineering, are dikes (or groins) constructed from live fascines, live stakes, soil, and rock. They are built upon a rock foundation and vegetated bundle that extends from the bed to a depth sufficient to prevent failure by undercutting. Ordinarily, we wrap this rock in Tensar or some other geogrid material. This allows the structure to settle as a unit if any undercutting does occur, and it also permits the use of excavated stream gravels larger than the grid openings, thereby helping to reduce costs.

Beginning at the bed elevation a gridwork of live fascine bundles is constructed, consisting of long, live fascines placed lengthwise about 30 to 60 cm apart and short, live fascine bundles placed on top and perpendicular to the long, live fascines with the same spacing. After securing the live fascines by driving wooden stakes through the point of intersection, the grid is backfilled with rock. This process continues until the structure attains a height equivalent to the elevation of the normal high water level (defined as the discharge that is exceeded 5% of the time), except that soil is substituted for rock at the mean low water elevation (which is defined as the elevation of the discharge that is exceeded 50% of the time).

Once the desired height is reached, the entire structure is covered with hand-placed riprap and live stakes are tamped in between the rocks. The 80 to 100 cm long stakes are cut from dormant native plants 2 to 4 cm in diameter and are installed on 30 to 60 cm centers. Typically, willow (Salix) species make the best cuttings for this purpose. Large stone is then dumped around the upstream side and the nose of the structure. See Figures 35.1 and 35.2.

1.2 Toe protection

A number of systems constructed of live woody vegetation are placed directly on, or parallel to, the bank to trap sediment and to protect against scour. Examples include live siltation construction structures, brushmattress, live fascines, and vegetated geogrids. See Figures 35.3, 35.4, 35.5 and 35.6. Although these can be used without toe protection on some streams (such as streams with well-armoured beds or other non-scouring situations), most applications require some toe protection to prevent undermining and subsequent failure. Riprap is the most common form of toe protection used when stone of sufficient size and quality is available. Small stone that otherwise would not be stable can be wrapped to provide toe protection.

Figure 35.1 Spring, prior to vegetative growth. Note deposition between the live blooms.

Figure 35.2 Same location as Figure 35.1, in fall, with vegetative growth that developed within the same year.

Figure 35.3 Live siltation contruction and live boom in the early spring.

Figure 35.4 Brushmattress in the second year of growth.

Figure 35.5 Live fascine growth two years after construction.

Figure 35.6 Vegetated geogrid in its first year of growth.

Figure 35.7 A joint planted steambank one year after installation.

1.3 Vegetated riprap

Sometimes riprap is the form of bank protection that is preferred by a client for cost or other reasons. Where not prohibited by institutional constraints, environmental benefits of riprap can be enhanced considerably by live staking, as described previously. When live stakes are used with riprap, the process is called joint planting to distinguish it from live staking without rock. The live stakes should be tamped into the soil below the riprap and any filter layer to a depth of a least 50 cm at an angle perpendicular to the slope and angled slightly downstream. Joint planting can provide a considerable amount of shade and cover, as well as trapping sediment, immediately from the first growing season. See Figure 35.7.


The major design considerations encountered when using riprap in the situations just described involve issues of stability--bed degradation, scour depth, and appropriate rock size and blanket thickness. Each of these is discussed separately.

2.1 Bed degradation or bed scour

Most streambank protection work is performed on streams with inherent instability. This instability can result from channel modifications such as straightening or enlargement; altered rainfall-run-off relationships due to activities such as urbanization, agriculture, or forestry; or from base level changes, just to mention a few examples. When installing foundations or toe protection for streambank protection it is essential to have a clear idea whether bed degradation might be occurring and, if so, how rapidly it is occurring and what the ultimate depth is likely to be.

Predicting the amount of bed degradation is a difficult and risky business, and one that is likely to tax the skills of even the most innovative and experienced practitioner. Although we are unable to suggest a single approach that works in a majority of situations, we can propose an approach that may have potential when degradation has been going on for several years and reliable measurements of bed elevation have been taken at several time periods. Plotting the elevation data against time and fitting a regression line to the plot as shown in Figure 35.8 may suggest a limiting depth. The data in Figure 35.8 apply to a Mississippi stream that has been undergoing bed scour for a number of years.

Figure 35.8 Estimated scour depth for Osborne Creek

2.2 Predicting scour depth in bends and around structures

Even in streams with stable beds, localized scour caused by secondary currents can undermine streambank protection structures. The two most common situations where this is encountered are along outside meander bends and around the tips of dikes.

Outside meander bends are zones of scour, as indicated by the pools that normally occupy these locations. Streams with beds composed of sand and gravel may be scoured to substantial depths during the passage of flood events. Some of the greatest scour occurs in meander bends. On one Mississippi stream where an experimental streambank protection design was tested that consisted of closely spaced posts with overlapping automobile tires, tires came off the bottoms of the posts when the bed was scoured below the 10 foot dept to which the posts were installed. Although the pool in a meander bed may refill as the falling limb of the hydrograph passes, any rock placed on the bed would be buried or swept downstream during the scouring event.

The authors are presently unaware of any reliable method of predicting scour depth, given the lack of data and the time and cost constraints that are usually encountered in stream-bank protection work. Accordingly, we have adopted two rules of thumb that we have used. First, we use a minimum key-in depth of 1-m on small streams (bankfull discharge less than about 15-20-m3/s) and 2-m on larger streams. Second, we add additional stone, either as a surface blanket extending away from the buried toe or buried in the toe trench as additional width.

Dikes are used on large rivers to contract flow through a narrower width and scour channels for navigation purposes. These dikes are normally placed in series and are long enough to cause significant channel contraction. Dikes used for streambank protection typically contract the channel very little and cause local scour around the tip of the dike only.

Numerous studies about dikes have been published. Some of them propose empirical formulas for predicting scour depth around the tips of dikes. These relationships are usually derived from studies done with flumes or with movable bed models of large streams. We have had little success with these formulas for various reasons. Some employ empirical constants that do not seem to be applicable to most streams; some require data, such as discharge and bed material size, that are unavailable and too time consuming or expensive to collect; and, some apply only to dikes with specific orientations, heights, relative lengths, or other specific design criteria.

There is significant lack of agreement on how to design dikes to maximize bank protection. When unsubmerged, all flow is forced around the tip of the structure, causing acceleration of flow and scour. Thus, the higher the dike, the deeper the scour hole tends to be. There is evidence from model studies that dikes angled upstream develop deeper scour holes than those perpendicular to the bank or angled downstream. For these reasons we tend to avoid alignments angled upstream and build our structures with sloping crests that intersect the bank at about the same elevation as the natural sedimentary berms present in many streams (about the elevation of the flow that is exceeded 5% of the time). So far, we have experienced few problems with undercutting, except on streams undergoing active degradation.

2.3 Rock size and blanket thickness

In sizing rock and determining blanket thickness we prefer the guidance developed by the US Army Engineer Waterways Experiment Station (WES). Not only do we find this approach to be easy to use and cost effective, but it lends itself readily to hand placement of stones and joint planting in most cases. Sometimes governmental clients have their own guidance for sizing stone that they prefer us to use.


Aside from availability of stone suitable for riprap, which is sometimes a problem, the major threats to successful project performance are improper site assessment, design and installation, and lack of monitoring and maintenance. Improper site assessment and design occurs due to failure to consider or to fully understand all pertinent hydraulic, geotechnical, and hydrologic information. Improper installation occurs for a variety of reasons. Contractors employed to do the work are usually general contractors who lack experience in soil bioengineering projects and requirements for handling dormant, living vegetation. Clients often try to save money by shortening the reach after design or by modifying designs through elimination or substitution of elements.

Monitoring and maintenance are critical to any streambank protection project, but especially so to those employing vegetation. During the first year after installation, projects should be inspected following all high flow events. Thereafter, they should be inspected annually and following all major flood events. Shortly after the growing season commences, all vegetative systems should be inspected to identify and correct any problems of vegetative growth.


Vegetation is an important element in streambank stability and a critical element in aquatic and riparian habitat. Streambank protection that incorporates vegetation offers far more environmental benefits than structural designs without vegetation. Stresses imposed on vegetative systems by the flow of water often necessitate the use of stone in conjunction with vegetation to protect against undercutting and scour. When used together they can provide effective protection against erosion in addition to environmental benefits. See Figure 35.9.

Figure 35.9 Rock used in conjunction with vegetated soil bioengineering treatments offering both protection against erosion as well as broad environmental benefits


5.1 Discussion by S.T. Maynord

It is certain that bioengineering will soon be incorporated into bank protection projects on most small streams. Guidance is needed to define what is the minimum amount of structural protection (such as riprap) required on the lower portion of the bank and in the toe. To assist in that determination, the Corps is developing guidance in nomographic form on predicting the velocity distribution in bendways from toe of bank to top of slope.

5.2 Discussion by C.R. Thorne

What criteria are used to design the vegetation structure in terms of: flexibility; flow resistance; root strength; permissible velocities; and near-toe scour?

How are seasonal aspects of construction/contracting handled with regard to timing; seasons; weather; flood events; availability of plants; and dormancy of plants?

At present, the implementation of schemes using vegetation as an integral part of the structural strength and erosion resistance of the protection works is relatively rare. If this situation is to change, as it should and must in my opinion, we will need to gear up for very widespread use of vegetative solutions. This means massively increasing the availability of suitable plants, providing construction crews capable of installing the schemes, and, perhaps most difficult of all, finding people with the experience and expertise to produce efficient, economical, and environmentally sound designs. The analogy with chemistry may be illustrative; it is no small step to go from a bench-top experiment to large-scale production at an industrial plant in the synthesis of a new compound. Widespread adoption of vegetative solutions to bank erosion problems will be much more difficult. My question is, what do you see as the key steps that should be taken now to facilitate this transformation and what is a realistic time scale for expansion of vegetative solutions to be used in say, 25%, 50%, and 75% of all bank protections schemes?

5.3 Discussion by S. Kurnzer

Would you comment on efficacy of vegetative solutions in desert environments since all of your examples seem to be in moist to wet climates?

5.4 Closure by R.B. Sotir and N.R. Nunnally

5.4.1 Closure to S.T. Maynord

The guidelines for setting the top of the riprap in mixed schemes vary. In the Pacific Northwest on streams with anadromous fish runs, the Ordinary High Water (OWH) is used as a reference elevation. This elevation typically is locally defined as mean high water or the elevation where permanent wood vegetation begins. In other regions, such as the Midwest and South, we often use the elevation of the 50% duration discharge. If no flow records are available, we use the elevations where permanent woody vegetation begins. In most cases, the top of the riprap is set 1 or 2 feet above the OWH or line of woody vegetation.

In general, we encourage the use of woody vegetation rather than grass for two reasons. First, woody vegetation produces roots that are stronger and penetrate more deeply than grass. Second, woody stems have more resistance and this reduces flow velocities in the vicinity of the bank. The main exceptions would be in settings where the height of woody vegetation might interfere with visibility (such as some residential and park settings) and on small flood control channels where woody vegetation might reduce flow capacity too much.

5.4.2 Closure to C.R. Thorne

There are no generally accepted criteria for designing the vegetative components of streambank protection structures. We rely heavily on our own experience and that of European soil bioengineers. The issues you have raised are valid ones, and each is addressed separately.

Flexibility and flow resistance: The vegetative component is intended as a foundation into which other native species invade. Over time, natural succession occurs, and the character of the vegetation changes. Flexibility is an issue only on small flood control channels, usually in urban areas. In these circumstances, we often recommend species of willow that are very flexible and often suggest a regular maintenance program of pruning or mowing to maintain flexibility. On new flood control channels we sometimes use roughness values that would be typical of well vegetated natural channels (Manning's n values in the range of 0.07 to 0.10) and oversize the channel to accommodate woody bank vegetation. Root Strength: Root strength is a factor that needs to be considered when dealing with soils subject to shallow slope failures. Unfortunately, not much information is available regarding tensile strength and density of the roots of various plant species. Fortunately, plants such as willow and dogwood which are widely used in soil bioengineering systems have good root strength. Permissible velocities: Virtually no research has been done on permissible velocities of woody plant species. However, much anecdotal evidence exists that indicates that woody vegetation can withstand scour at much higher velocities than non-woody species. After all, nearly all high gradient mountain streams have well-vegetated banks. In addition, European soil bioengineers have successfully established woody vegetation on many high gradient flood channels. Near-toe scour: Unfortunately, we are not exactly sure what this means. If you are referring to scour in the bed near the toe, we can only refer you to our comments in this paper. If you are referring to scour low on the bank, that generally is not a problem if toe protection is extended high enough and woody vegetation is used above (see answer to Steve Maynord's question.)

The woody plants used in soil bioengineering must be installed during the dormant season. On some streams dormancy coincides with critical periods such as fish spawning runs. In these situations, we often employ staged construction. The instream works (rock toes and foundations) are installed during the summer, and the upper bank work is done during dormancy. Although weather is not a limiting factor, it is a complicating one. Adverse weather may prolong construction and especially harsh conditions may require special equipment or special construction techniques as with any system. On small and medium size streams, flooding during construction is normally not a problem, although work may be shut down for a few days at a time. Flood delays and flood damage likely are less of a problem on soil bioengineering projects than on conventional projects. In our experience, working across the USA and Canada, plant availability is not a problem. Plants can be harvested and hauled economically as far as 50 miles from the construction site. Regrowth at previously harvested sites makes excellent source material, and plant material can be harvested from established soil bioengineering projects.

Gearing up for widespread use of vegetative solutions will be difficult if the technology gains widespread acceptance. We consider the lack of acceptance, especially by civil engineers, as being the major factor limiting use of the technology. Aside from that, the lack of properly trained individuals to evaluate problems and develop mechanically and environmentally sound, economical designs is the biggest constraint. Although the technology appears simple and straightforward, there are many detailed aspects in design and construction that are critical to project success.

From a construction standpoint, the biggest problem is lack of familiarity with vegetative construction. General contractors are likely to overbid to protect themselves because they are unfamiliar with the technology, and they often fail to understand the need to follow specifications closely. Availability of plants should not pose a problem. As pointed out previously, harvesting sites can be reused frequently (every year or two, depending on the location and growing season) and old construction projects can be harvested. In addition, in the USA, the Soil Conservation Service is already experimenting with nursery grown stock for soil bioengineering, with the ultimate goal of promoting commercial productions. As far as the timetable for extending woody vegetative schemes to 25%, 50%, and 75% of all bank protection projects, we cannot see the first level (25%) occurring before the year 2005. Even this modest level will not be achievable unless training becomes more widely available through workshops, short courses, and, we hope, academic curricula.

5.4.3 Closure to S. Kurnzer

Soil bioengineering is more difficult in arid environments, but can be carried out successfully by using native plants that are adapted to those areas. Additionally, soil nutrient testing and amelioration can be effective in promoting quick establishment and rapid plant growth. Plants that normally grow along watercourses are good candidates, such as willow and cottonwood in much of the western USA. The major criteria used in plant selection are rooting ability and habitat value.

Robbin B. Sotir & Associates, Inc.
3221 Stoney Acres Drive
Kennesaw, Georgia 30152
Phone: 770-424-0719
©2020 Robbin B. Sotir & Asociates, Inc.