Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures (2016)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

90 4.1 Overview This chapter contains both general and site-specific guidance for the design, installation, monitoring, and maintenance of environmentally sensitive stream bank protection measures. From a macro-scale perspective, selection and design of an environmentally sensitive stream bank treatment must consider the watershed and river system processes that drive river channel response, including climatic, hydrologic, hydraulic, and geomorphic considerations. Ultimately, however, the success or failure of a selected treatment will depend on its ability to adapt to site-specific micro-scale processes in a given stream reach, including hydraulic, geomorphic, and geotechnical considerations. Conformance to design and construction guidance as well as a rigorous monitoring and maintenance effort will help ensure the success of the treatment. Monitoring and maintenance of, in particular, the vegetative component of an environmentally sensitive treatment is often the key to a successful installation (see Section 4.2.5). Equally important are issues involving potential impacts to the aquatic habitat (both positive and negative) at the treatment site (Section 4.2.6). In addition, for the engineer involved in the multidisciplinary design of a treatment, guidance from FHWA on bioengineered treatments should be considered (Section 4.2.7). The engineer on the design team must also consider potential professional liability issues related to “stamping” an environmentally sensitive design. Some thoughts on these issues are provided in Section 4.2.8. The focus of the chapter is on detailed standalone design guidelines for the two environmen- tally sensitive treatments subjected to hydraulic laboratory testing under Task 7 of this project (as described in Chapter 3). Detailed design guidelines updated by the results of hydraulic labo- ratory testing are presented in Section 4.3.1 (Live Siltation and Live Staking with Rock Toe) and Section 4.3.2 (VMSE Without Hard Toe). An overview of design guidelines for a related treat- ment (vegetated riprap) provides value-added guidance on a third treatment in Section 4.3.3. Section 4.4 features two detailed examples of the application of environmentally sensitive stream bank protection measures employed in conjunction with stream channel restoration projects. Recognizing that dryland landscapes are quite different from those of more humid regions, one example involves a detailed engineering analysis underpinning the application of environmentally sensitive techniques on an arid region perennial stream—the Rio Grande near Bernalillo, NM (Section 4.4.2). The second example deals with a smaller stream in a humid region with significant infrastructure issues—Malletts Creek near Ann Arbor, MI (Section 4.4.3). The chapter concludes with an appraisal of results from this research project (Section 4.5). Key observations derived from various research activities (survey and site visits) are provided together with a summary of advances in the state of practice. C H A P T E R 4 Design Guidelines and Appraisal of Research Results

Design Guidelines and Appraisal of Research Results 91 4.2 General Considerations 4.2.1 Introduction Factors that affect stream stability at highway infrastructure can be classified as hydrologic, hydraulic, and geomorphic. Rapid and unexpected changes can occur in streams in response to human activities in the watershed and/or natural disturbances of the fluvial system, making it important to anticipate changes in channel geomorphology, location and behavior. Geomorphic characteristics of particular interest are the alignment, geometry, and form of the stream channel. The behavior of a stream depends not only on the apparent stability of the stream at a specific reach of interest, but also on the behavior of the stream system of which it is a part. Upstream and downstream changes may affect future stability at the treatment site. Natural disturbances such as floods, drought, earthquakes, landslides, forest fires, etc., may result in large changes in sedi- ment load in a stream and major changes in the stream channel. These changes can be reflected in aggradation, degradation, or lateral migration of the stream channel. The following section provides an overview of general considerations relevant to implementing the guidelines for spe- cific treatments that follow. For more detailed coverage of these important topics, reference to FHWA’s Hydraulic Design Series (HDS) 6 (Richardson et al. 2001) or Hydraulic Engineering Circular- (HEC)-20 (Lagasse et al. 2012), and the American Society of Civil Engineers’ (ASCE’s) sedimentation engineering handbook (ASCE 2008) is suggested. 4.2.2 Hydrologic, Hydraulic, and Geomorphic Considerations Hydrologic Factors Magnitude and Frequency of Floods. The hydrologic analysis for a stream reach consists of establishing peak-flow frequency relationships and such flow-duration hydrographs as may be necessary. Flood-frequency relationships are generally defined on the basis of a regional analysis of flood records, a gaging station analysis, or both. Regional analyses have been completed for all states by the USGS. Flood-frequency relationships at gaged sites can be established from station records that are of sufficient length to be representative of the total population of flood events on that particular stream. The Pearson Type III distribution with log transformation of flood data is recommended by the Water Resources Council (1981) for station flood data analysis. Where flood estimates by regional analysis vary from estimates by station analysis, factors such as gaging station record length and the applicability of the regional analysis to that specific site should be considered, as well as high-water information, flood data, and information on flood levels at existing structures on the stream. FHWA’s HDS 2 should be referred to for more detailed information and guidelines on hydrologic analysis (McCuen et al. 2002). Flood History and Rainfall-Runoff Relations. Consideration of flood history is an integral step in attempting to characterize watershed response and morphologic evolution. Although the occurrence of single large storms can often be directly related to system change in any region of the country, this is not always the case. In particular, the succession of morphologic change may be linked to the concept of geomorphic thresholds. Under this concept, although a single major storm may trigger an erosional event in a system, the occurrence of such an event may be the result of a cumulative process leading to an unstable geomorphic condition. Where available, the study of flood records and corresponding system responses, as indicated by time-sequenced aerial photography or other physical information, may help determine the relationship between morphological change and flood magnitude and frequency. Evaluation of wet-dry cycles can also be beneficial to an understanding of historical system response. Observable historic change may be found to be better correlated with the occurrence of a sequence of events

92 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures during a period of above average rainfall and runoff than with a single large event. The study of historical wet-dry trends may explain certain complex aspects of system response. For example, a large storm preceded by a period of above-average precipitation may result in less erosion, due to better vegetative cover, than a comparable storm occurring under dry antecedent conditions; however, runoff volumes might be greater due to saturated soil conditions. A good method to evaluate wet-dry cycles is to plot annual rainfall amounts, runoff volumes, and maximum annual mean daily discharge for the period of record. A comparison of these graphs will provide insight into wet-dry cycles and flood occurrences. Additionally, a plot of the ratio of rainfall to runoff is a good indicator of watershed characteristics and historical changes in watershed condition (Lagasse et al. 2012). Special Considerations in Arid Regions. Since dryland landscapes are quite different from those of more humid regions, analysis of flood history is of particular importance to under- standing arid region stream characteristics. In an arid region, the topography and landforms are more abrupt, the soils are thinner, the bedrock exposures are usually more pronounced, and the streams are smaller and are likely to be dry for at least part of the year. Overall, the physical environment reflects the lack of water and mechanical weathering and erosion predominate over chemical weathering and solution, as compared to a humid environment. In a humid environment, high precipitation produces vegetation and soils that are well developed and stabilized. Under these conditions, natural streams generally carry small suspended sediment loads, reflecting this stability in the upland watersheds. Additionally, high precipitation produces a dilution effect on the sediments that are eroded (Simons, Li & Associates 1982). In contrast, dryland streams normally carry large sediment loads from erosion by both wind and water. The precipitation generating the erosion in a dryland environment usually results from small storm cells that may be limited in areal extent, but can produce high-intensity rain and rainfall energy. This type of storm produces “flashy” runoff with pronounced capacity for sedi- ment removal and transportation. Only rarely does a single storm produce runoff in all parts of a dryland stream basin and extended periods may pass with no streamflow at all. Many dryland streams flow only during the spring runoff and immediately after major storms. For example, Leopold et al. (1966) found that arroyos near Santa Fe, New Mexico, flow only about three times a year. As a consequence, dryland stream response can be considered to be more hydrologically dependent than streams located in a humid environment. Whereas the simple passage of time may be sufficient to cause change in a stream located in a humid environment, time alone, at least in the short term, may not necessarily cause change in a dryland system due to the infrequency of hydrologically significant events. Thus, the absence of significant morphological changes in a dryland stream or river, even over a period of years, should not necessarily be construed as an indication of system stability. The unique hydrologic and morphologic characteristics of a dryland stream can have a significant impact on the utility or survivability of the vegetative component(s) of environmentally sensitive treatments (see Section 4.4. for an example of an arid region application). Hydraulic and Geomorphic Factors Mechanics of Channel Response. Because of the number of interrelated variables that can react simultaneously to natural or imposed changes in a river system, river response to both natural and human-induced forces is complex and varied in nature, but trends are generally predictable. However, Richardson et al. (2001) details the variables affecting alluvial channel geometry and bed roughness and concludes that the nature of these variables is such that, unlike rigid boundary hydraulics problems, it is difficult to isolate and study the role of an individual variable. For example, evaluation of the effects of increasing channel depth on average velocity is

Design Guidelines and Appraisal of Research Results 93 hampered because related variables such as flow resistance, the form of bed roughness, channel cross-section shape, and sediment discharge also respond to the changing depth. Position and shape of alternate, middle, and point bars can also be expected to change. Geomorphic and geotechnical factors that can influence stream stability include stream size, flow habit (i.e., ephemeral or perennial), and the characteristics of channel boundaries. The bed material of a stream can be a cohesive material, sand, gravel, cobbles, boulders, or bedrock. Bank material is also composed of these materials but may be dissimilar in composition from the bed material. The stability and rate of change in a stream are dependent on material in the bed and banks. Other natural factors such as the stream’s relationship to its valley, floodplain and planform characteristics, and features such as natural levees, incision, and riparian vegetation are important indicators of stream stability (or instability). Hydraulic factors that affect stream channel stability are numerous and include bed forms and their effects on sediment transport, resistance to flow, flow velocities, and flow depths. They also include the magnitude and frequency of floods; characteristics of floods (i.e., duration, time to peak, and time of recession); flow classification (e.g., unsteady, nonuniform, turbulent, super- critical, or subcritical); ice and other floating debris in the flow; and flow constrictions. Other factors include the effects of natural and human-induced changes that affect the hydrology and hydraulic flow conditions of the stream. Human-induced changes in the drainage basin and the stream channel, such as alteration of vegetative cover and changes in a pervious (or impervious) area can alter the hydrology of a stream, sediment yield, and channel geometry. Channelization, stream channel straightening, stream-side levees and dikes, bridges and culverts, reservoirs, gravel mining, and changes in land use can have major effects on streamflow, sediment transport, and channel geometry and location. Water discharge is the key factor that affects channel shape and sediment transport. Related hydraulic variables, such as velocity, depth, and flow area, are important in the analysis of channel response. Coinciding with these are channel shape, channel slope, and flow resistance from grain roughness and bed forms. Geology and soils of the channel bottom and banks help determine the relative erodibility of the system, and therefore its response to other changes or alterations. Local variations in geology, soils, vegetation, and flow rate play important roles in determining bank stability in channels. Consequently, an alluvial river will generally change its position and shape as a result of hydraulic forces acting on its bed and banks. These changes may be slow or rapid and may result from natural environmental changes or from changes by human activities in the watershed or river channel itself. When a river channel is modified locally, this local change frequently causes modification of channel characteristics both up and down the stream. The response of a river to human-induced changes often occurs in spite of attempts to keep the anticipated response under control. Such changes can have a significant effect on the success or failure of an environmentally sensitive stream bank treatment intended to correct a localized problem. Sediment Transport. The movement of sediment is described by the sediment continuity concept. This concept, like water continuity, describes the input, output, and storage change of sediment in the watershed or stream channel. Sediment transport is usually considered in three parts: 1. Bed load—sediment movement primarily along the stream bed. 2. Suspended load—sediment transport primarily in the flow with some contact and inter- change with the stream bed, and 3. Wash load—fine particles suspended in the flow that are not found in appreciable quantities in the bed material.

94 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Bed load and suspended load can be described by transport equations as they are usually related to transport capacity. Wash load, however, is dependent primarily on supply conditions. In upland (low order) watersheds, the concept of supply and capacity is crucial, as it determines the sediment yield. Typically, transport capacity for small sizes (e.g., wash load) is in excess of the supply. Therefore, the yield is related to the supply. Conversely, for large sizes, the capacity is much less than the supply (e.g., large cobbles in a mountain stream), and therefore capacity controls. The relationship between sediment supply and sediment transport capacity influences long-term processes in the stream system such as aggradation and degradation which, in turn, influence the selection, design, and performance of environmentally sensitive stream bank protection measures (see Section 4.2.3). 4.2.3 Site-Specific Physical Processes Affecting Environmentally Sensitive Treatments Introduction In spite of their complexity, all rivers are governed by the same basic factors discussed in Section 4.2.2. In the design of environmentally sensitive treatments, one must understand and work with these natural factors: • Geologic factors, including soil conditions; • Hydrologic factors, including possible changes in flows, runoff, and the hydrologic effects of changes in land use; • Hydraulic characteristics such as depths, slopes, and velocity of streams and what changes may be expected in these characteristics in space and time; and • Geomorphic characteristics of the stream, including the probable geometric alterations that will be activated by the changes a treatment project and future projects will impose on the channel. The following subsections present several site-specific concepts important to the design of environmentally sensitive stream bank protection measures. Bankfull Discharge. NCHRP Report 544 (McCullah and Gray 2005) contains detailed cov- erage on a series of special topics relevant to the design of environmentally sensitive stream bank treatments (see also Section 2.1). McCullah and Gray (2005) note that dynamically stable stream channels formed in fully alluvial materials (sediments that can be eroded and deposited by the stream) tend to have widths, depths, and slopes that reflect a balance among the geologic and hydrologic variables that interact to create the fluvial system. Many engineers and resource man- agers have found the concept of channel-forming discharge to be a useful tool in understanding and managing streams. This concept is based on the idea that even though channel width, depth, and slope vary along a stream and through time, average values of width, depth, and slope tend to be constant for a reach with a given drainage area if: 1. The stream bed and banks are alluvial; 2. There have not been any extreme floods, droughts, earthquakes, forest fires, or other catastrophic events in the recent past; 3. The watershed is largely free of human-caused disturbances, such as land-use changes, grazing, mining, road building, dams, or channelization; and furthermore 4. The channel geometry is such that the greatest discharge the channel will carry without overflowing is not a rare flood (which moves tremendous amounts of sediment, but occurs only rarely) or a low flow (which occurs frequently, but has relatively little erosive power), but is an intermediate magnitude, such as the one- or two-year flood. This characteristic discharge is referred to as the “channel-forming” or “dominant” discharge, Qcf . This discharge therefore dominates channel form and process, at least for streams in humid

Design Guidelines and Appraisal of Research Results 95 regions and for perennial streams in semi-arid environments (Soar and Thorne 2001 and Biedenharn et al. 2001). Clearly, very few stream reaches meet the four criteria outlined above, and few can be described as “dynamically stable” or “fully alluvial” without qualification. Therefore, channel geometries often vary considerably from sizes needed to convey Qcf. Many workers ignore departures from ideal conditions, and determine bankfull stage based on field indicators like permanent vegetation or terraces, but this approach can lead to errors. One of the principal reasons for estimating “channel-forming discharge” (or dominant dis- charge) is to ensure that any planned modification to channel cross section (as a result of pro- posed bank or channel protection measures along a reach of stream) will be compatible with this discharge. There are several approaches that can be employed to estimate this channel-forming discharge. Under the right conditions, other types of flows, including “bankfull discharge,” can be used as surrogates. At least three approaches are available for determining Qcf : effective discharge (Qeff), bankfull discharge (Qbf), or the discharge that corresponds to a given return interval, Qri (Table 4.1). The expression, “bankfull discharge,” Qbf, should be used to refer to the maximum discharge that the channel can convey without overflow onto the floodplain. Although this definition, proposed by Copeland et al. (2001) differs from that used by others (e.g., Rosgen 1996), it eliminates confusion. As noted above, theoretically Qbf and Qeff are generally equivalent in channels that have remained stable for a period of time, thus allowing the channel morphology to adjust to the current hydrologic and sediment regime of the watershed allowing Qbf to be used as a surrogate for Qeff. In such a chan- nel, the bankfull discharge generally corresponds to a sharp change in the slope of the rating curve. It must be noted, however, that in an unstable channel that is adjusting its morphology to changes in the hydrologic or sediment regime, Qbf can vary markedly from Qeff. Finally, the quantities Qeff, Qbf, and Qri are estimates of Qcf, and thus more than one of these should be considered (Biedenharn et al. 2001). Computed effective and bankfull discharges outside the range between the 1- and 3-year recurrence intervals should be questioned. The computed effective and recurrence interval discharges should be compared with field evidence to ascertain if these discharges have geomorphic significance. Channel performance should be Quantitative Estimate of Qcf Data Requirements Recommended For Limitations Effective Discharge (Qeff) Historical hydrology for flow- duration curve (10 years or more recommended) or synthetic flow-duration curve; channel survey; hydraulic analysis; sediment gradation; sediment transport analysis and model calibration (if possible) Channel design Requires large dataset and training in hydraulic engineering or fluvial geomorphology Bankfull Discharge (Qbf) Channel survey; hydraulic analysis and model calibration using observed stage-discharge relation (if possible). Identification of field indicators in a stable, alluvial reach Stability assessment; estimation of Qeff in stable channels Can be very dynamic in unstable channels/watersheds; field indicators can be misleading Return Interval Discharge (Qri) Historical hydrology for flood frequency analysis, regional regression equations, or hydrologic model First approximation of Qeff and/or Qbf in stable channels No physical basis; relations to Qeff and Qbf inconsistent in literature Table 4.1. Comparison of approaches for finding the channel-forming discharge (Qcf) (McCullah and Gray 2005).

96 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures examined for a range of discharges that represent key levels for aquatic habitat, riparian vegetation, channel stability, or flow conveyance (Copeland et al. 2001). The bankfull discharge special topic in NCHRP Report 544 includes detailed guidance for determining effective discharge, bankfull discharge, and discharge for a specific return period for use in design of environmentally sensitive stream bank treatments at both gaged and ungaged sites. An ongoing project (NCHRP Project 24-40) at CSU will address issues related to “Design Hydrology for Stream Restoration and Channel Stability at Stream Crossings.” This project (Bledsoe et al. in process) is scheduled for completion in early 2016. The objectives of NCHRP Project 24-40 are to develop a scientifically supported method for defining the design hydrol- ogy for DOT stream restoration projects along with an understanding of how that design hydrology might change with land-use changes. Steps to produce that methodology include the following: • Investigate flow metrics other than peak annual flood frequency curves for more consistent correlation with channel-forming processes (such as distribution of daily mean discharge, flow duration, key points on a flow-duration curve, etc.). • Develop quantitative methods for estimating the impact of land-use change on the design metric that is appropriate for design. • Investigate the connection between these changes and changes in channel-forming discharge, and consequently bankfull channel hydraulic geometry. These methods should include changes due to urbanization, surface mining, agriculture, and forestry practices. When available, the results of NCHRP Project 24-40 could change, update, or expand bankfull discharge concepts related to design of environmentally sensitive stream bank protection measures. In relation to the vegetative component of environmentally sensitive treatments, McCullah and Gray (2005) also note that establishing the annual high water (AHW) serves as a guide to identify where vegetative techniques should be positioned on a given stream bank (in some stable streams systems, AHW may be equivalent to bankfull discharge). The AHW may or may not represent the average annual peak discharge or the elevation for the 1-year event, but it does represent the lower limit suitable for establishing permanent woody vegetation. This is impor- tant because vegetation usually should not be planted below the AHW elevation. Generally the channel boundary below the AHW elevation is subjected to higher velocities and boundary shear stresses than regions above, therefore AHW roughly delineates the upper boundary of the zone in which structural support should be applied to the toe of the bank. The annual low water (ALW) indicates the general elevation the roots must be able to penetrate down to in order to have access to water during the dry season. ALW approximates the depth of the vadose zone in a stream bank soil profile. The elevation of the vadose zone is increasingly dictated by soil type as distance from the stream lengthens (McCullah and Gray 2005). Design high water (DHW), as calculated and specified by the designer, defines the upper elevation extent of a structure or technique, not including the required level of freeboard. The DHW elevation simply depicts the extent to which a technique may be inundated during a rare (design) hydrologic event. The designer must select techniques and materials suitable for hydraulic forces (i.e., waves, currents, seepage) or gravitational loading anticipated under design conditions. Design hydraulic loading may or may not coincide with the highest water levels. The relationship between river stage and hydraulic loading is site-specific due to differences in energy slope, channel roughness, and channel geometry. Figure 4.1 provides a visual representation of DHW, AHW, ALW, and examples of techniques suitable for particular elevations (McCullah and Gray 2005). For more details, reference to NCHRP Report 544 is suggested.

Design Guidelines and Appraisal of Research Results 97 Conveyance. NCHRP Report 544 also contains a special topic on management of conveyance in relation to design of environmentally sensitive stream bank treatments. Here the term convey- ance refers to the amount of flow a channel can carry with a given energy slope (i.e., it represents the frictional controls imposed on discharge rate by channel cross-sectional shape and roughness). Bed and bank stabilization treatments can modify cross-sectional shape, influence roughness, and other- wise influence channel conveyance. The incorporation of vegetation and large wood into channel and bank stabilization measures frequently results in rougher channel boundaries than more tra- ditional measures. For example, plantings of woody vegetation may provide more flow resistance than riprap revetment. Changes in conveyance properties of a channel can have both engineering and ecological implications. If flooding or upstream drainage is an issue at the site in question, the designer may have to estimate the impact of proposed measures on flood stages. During low-flow periods conveyance properties affect both depth and velocity, which have important implications for fish and other stream dwelling organisms (see Section 4.2.6). In either case, an environmentally sensitive approach to channel protection implies a thoughtful review of the consequences of any large alteration in existing channel conveyance properties (McCullah and Gray 2005). This special topic in NCHRP Report 544 includes guidance on such stream-related hydraulic concepts as: • Uniform flow equations, • Estimating Manning n values for simple and complex boundaries, and • Selecting Manning n values for vegetated boundaries. For uniform flow conditions NCHRP Report 544 notes that the oldest and perhaps simplest approach for computing channel conveyance for steady flow conditions involves the use of a uniform flow equation like the Manning or Chezy equations that contain a coefficient that rep- resents the combined effect of all of the channel characteristics that contribute to flow resistance. For example, the coefficient n in the Manning equation: Q 1.486 n AR S (4.1)2 3 1 2= reflects channel bed material, bank conditions, planform, and cross-section shape for an entire reach, while Q is the discharge in ft3/s, A is the cross-sectional area of the flow in ft2, R is the Figure 4.1. Elevation diagram of DHW, AHW, and ALW (McCullah and Gray 2005).

98 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures hydraulic radius in ft, and S is the energy slope. R is equal to A/P, where P is the wetted perimeter in ft, and S is equal to the bed slope when flow is uniform. If SI units are used, the conversion factor in the numerator is simply 1.0 instead of 1.486. Experienced designers often select n values based on experience or calibrate their n values based on stage-discharge curves from nearby gaging stations. Typically the Manning equation as expressed above is applied to an entire reach only for preliminary estimates or as a first step in design. More detailed analyses are typically based on computer models that represent the channel as a series of cross sections, for example, the USACE 1-dimensional HEC-RAS model (USACE 2010). When trying to simulate flow effects of vegetation, the user must decide if seasonal factors should be considered. Case studies show that flow resistance due to deciduous vegetation in full leaf is much greater than for dormant (winter) conditions. HEC-RAS allows input of constant factors for adjusting n values by month. Finally, over the long term, vegetation may impact conveyance by inducing sediment deposition. Inclusion of sedimentation in the hydraulic analysis for a project will increase the cost and complexity of the analysis several times (McCullah and Gray 2005). Again, referring to NCHRP Report 544 for more detailed guidance is suggested. HEC-20 (Lagasse et al. 2012) and HDS 6 (Richardson et al. 2001) also provide guidance on applications of the Manning and Chezy equations and detailed guidance for estimating the roughness coefficient. References cited in Section 2.2.5 under “Fluid Mechanics” provide additional detail on recent findings regarding the effect of woody vegetation on flow resistance. Analyzing Aggradation and Degradation. For the successful performance of any stream bank protection it is essential to determine, and, if necessary, control the vertical stability of the stream bed in the reach of concern. If progressive long-term degradation problems in particular exist, most hard engineering and environmentally sensitive treatments will ultimately fail unless this process is controlled by a countermeasure to correct and hold the vertical position of the stream bed. The typical effects associated with bed elevation changes at highway infrastructure are the exposure and undermining of foundations from degradation or a reduction in flow area from aggradation (resulting in more frequent overbank flow). Bank caving associated with degradation poses the same problems at structures as lateral erosion from bend migration, but the problems may be more severe because of the lower elevation of the stream bed. Aggrading stream chan- nels also tend to become wider as aggradation progresses, eroding floodplain areas and highway embankments on the floodplain (Lagasse et al. 2012). It has been reported that there are serious problems at about three degradation sites for every aggradation site (Brown et al. 1980). This is a reflection of the fact that degradation is more common than aggradation, and also the fact that aggradation does not directly endanger the infrastructure foundation. It is not, however, an indication that aggradation is not a serious problem in some channel reaches. Problems commonly associated with degrading channels include the undermining of cutoff walls, other flow-control structures, and bank protection. Bank sloughing because of degradation often greatly increases the amount of debris carried by the stream and increases the potential for blocked waterway openings, reduced conveyance, and increased scour at bridges (Lagasse et al. 2010). The hazard of local scour becomes greater in a degrading stream because of the lower stream bed elevation. Aggradation in a stream channel increases the frequency of higher stages that can cause damage. In the case of highway infrastructure, bridge decks and approach roadways become inundated more frequently, disrupting traffic, subjecting the superstructure of the bridge to hydraulic forces that can cause failure, and subjecting approach roadways to overflow that can erode and cause

Design Guidelines and Appraisal of Research Results 99 failure of the embankment. Where lateral erosion or increased flood stages accompanying aggra- dation increase the debris load in a stream, the hazards of clogged bridge waterway, reduced conveyance, and hydraulic forces on bridge superstructures are increased. Data records for at least several years are usually needed to detect bed elevation problems. This is due to the fact that the channel bottom often is not visible and changes in flow depth may indicate changes in channel width, flow rate, local scour, or obstructions rather than bed eleva- tion changes. Reach-scale bed elevation changes typically develop over long periods of time even though rapid change can occur during an extreme flood event. The data needed to assess bed elevation changes include historic stream bed profiles and long-term trends in stage-discharge relationships. Occasionally, information on bed elevation changes can be gained from a series of maps prepared at different times. Bed elevations at railroad, highway, and pipeline crossings monitored over time may also be useful. On many large streams, the long-term trends have been analyzed and documented by agencies such as the USGS and the USACE. As noted in Section 4.2.2, long-term bed elevation changes may be the natural trend of the stream or may be the result of some modification to the stream or watershed. The stream bed may be aggrading, degrading, or in relative equilibrium in the vicinity of a planned bank-protection project. Long-term aggradation and degradation do not include the cutting and filling of the stream bed at a stream crossing that might occur during a runoff event (contraction and local scour). A stream may cut and fill at specific locations during a runoff event and also have a long-term trend of an increase or decrease in bed elevation over a longer reach of a stream. The problem for design of stream bank protection measures is to estimate the long-term bed elevation changes that will occur during the life of the planned treatment. A long-term trend may change during the life of a project as a result of modifications to the stream or watershed (see Section 4.2.2). Such changes may be the result of natural processes or human activities. The designer must assess the present state of the stream and watershed and then evaluate potential future changes in the river system. From this assessment, the long-term stream bed changes must be estimated. Factors that affect long-term bed elevation changes are dams and reservoirs (up- or downstream of a study reach), changes in watershed land-use (urbanization, deforestation, etc.), channelization, cutoffs of meander bends (natural or of human origin), changes in the downstream channel base level (control), gravel mining from the stream bed, diversion of water into or out of the stream, natural lowering of the fluvial system, and movement of a bend with respect to stream planform. Tidal ebb and flood may degrade a coastal stream; whereas, littoral drift may result in aggradation. The elevation of the bed under stream crossings on streams tributary to a larger stream will follow the trend of the larger stream unless there are controls. Controls could be bed rock, dams, culverts, check dams, or other structures. Data from the USACE, USGS, and other federal and state agencies should be considered when evaluating long-term stream bed variations. If no data exist or if such data require further evaluation, an assessment of long-term stream bed elevation changes for riverine streams should be made using the principles of river mechanics [see HDS 6 (Richardson et al. 2001)]. Such an assessment requires the consideration of all influences upon the study reach, i.e., runoff from the watershed to a stream (hydrology), sediment delivery to the channel (watershed erosion), sediment transport capacity of a stream (hydraulics), and response of a stream to these factors (geomorphology and river mechanics). To organize an assessment of long-term aggradation and degradation, a three-level fluvial system approach can be used. The three-level approach consists of (1) a qualitative determination based on general geomorphic and river mechanics relationships, (2) an engineering geomorphic analysis using established qualitative and quantitative relationships to estimate the probable behavior of the

100 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures stream system to various scenarios or future conditions, and (3) physical models or physical pro- cess computer modeling using mathematical models such as the USACE HEC-RAS (USACE 2010) to make predictions of quantitative changes in stream bed elevation due to changes in the stream and watershed. Methods to be used in Levels 1 and 2 are presented in HEC-20 (Lagasse et al. 2012) and HDS 6 (Richardson et al. 2001). Another source for guidance on qualitative and engineer- ing geomorphic analyses is the USACE manual Channel Stability Assessment for Flood Control Projects (1994). The biennial bridge inspection reports for bridges on the stream reach under study are an excellent source of data on long-term aggradation or degradation trends. Also, inspection reports for bridges crossing streams in the same area or region should be studied. In most states the biennial inspection includes taking the elevation and/or cross section of the stream bed under the bridge. These elevations are usually referenced to the bridge, but these relative bed elevations will show trends and can be referenced to sea level elevations. Successive cross sections from a series of bridges in a stream reach can be used to construct longitudinal stream bed profiles through the reach. The USGS and many state water resource and environmental agencies maintain gaging sta- tions to measure stream flow. In the process, they maintain records from which the aggradation or degradation of the stream bed can be determined. Where an extended historical record is available, one approach to using gaging station records to determine long-term bed elevation change is to plot the change in stage through time for a selected discharge. This approach is often referred to as establishing a “specific gage” record. Figure 4.2 shows a plot of specific gage data for a discharge of 500 cfs (14 m3/sec) from about 1910 to 1980 for Cache Creek in California. During that period, Cache Creek experienced significant gravel mining with records of gravel extraction quantities available since about 1940. When the historical record of cumulative gravel mining is compared to the specific gage plot, the potential impacts are apparent. The specific gage record shows more than 10 ft (3 m) of long-term degrada- tion in a 70-year period. In addition, the geology and geomorphology of the site need to be studied to determine the potential for long-term bed elevation changes at the project site. Quantitative techniques for stream bed aggradation and degradation analyses are covered in detail in HEC-20 (Lagasse et al. 2012). These techniques include: • Incipient motion analysis, • Analysis of armoring potential, • Equilibrium slope analysis, and • Sediment continuity analysis. Sediment transport concepts and equations are discussed in detail in HDS 6 (Richardson et al. 2001), HDS 7 (Zevenbergen et al. 2012), and HEC-20 (Lagasse et al. 2012). Sediment transport computer models can be used to determine long-term aggradation or degradation trends. These computer models route sediment down a channel and adjust the channel geometry to reflect imbalances in sediment supply and transport capacity. The USACE HEC-RAS (USACE 2010) model is an example of a sediment transport model that can be used for single-event or long-term estimates of changes in bed elevation. Sediment routing is essentially an application of the sediment continuity concept. Here, sediment inflow and outflow for a specific reach are analyzed and the sediment transport capacity is used to update cross-section geometry, which is then used to update the hydraulic calculations. The geometry is updated for individual cross sections, though the hydraulic vari- ables can be weighted with up- and downstream cross sections. A flood hydrograph or long-term

Design Guidelines and Appraisal of Research Results 101 flow hydrograph is entered as a series of constant flows. Within each flow time step, many sedi- ment transport and cross-section updating time steps are often required. The model does not assume that transport capacity is reached at every cross section, but limits erosion based on potential entrainment rates and limits deposition based on fall velocity, flow velocity, and water depth. Sediment layer depths, as well as lateral limits for erosion and deposition are also input. Sediment transport modeling generally requires greater model extent upstream and down- stream than a hydraulic flow model, as well as careful consideration of all boundary conditions (hydraulic and sediment). The Channel Evolution Model. Earth scientists (geomorphologists) have historically con- cerned themselves with documenting and explaining the changing morphology of the landscape through time and have documented the changing character of a landscape during long periods of geologic time. Initially, this type of evolution of landforms would appear to be of no inter- est to the highway engineer, but it serves as an alert that change can be expected at the scale of individual landforms (hillslopes, channels), and that the change can be sufficiently rapid to cause problems with the design and maintenance of environmentally sensitive stream bank protection measures (Lagasse et al. 2012). In the case of incised channels (gullies, arroyos) rapid incision can be followed by channel adjust- ment (deepening, widening) to a new condition of relative stability (on an engineering time scale) as erosion decreases, sediment storage increases, and a floodplain develops (Figure 4.3). Simon (1989) obtained data on the sediment loads transported through incised channels in Tennessee (Figure 4.4). The stages of channel evolution shown in Figure 4.3 are reflected in the changing Figure 4.2. Specific gage data for Cache Creek, California.

102 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Figure 4.3. Evolution of incised channel from initial incision (A, B) and widening (C, D) to aggradation (D, E) and eventual relative stability; h is bank height (Schumm et al. 1984). Figure 4.4. Sediment loads following channel incision: (A) bed-material load transported by incised Tennessee streams for each stage of incised-channel evolution (Figure 4.3), (B) hypothetical (dashed line) and measured (solid line) sediment volumes transported through Grand Canyon (Gellis et al. 1991).

Design Guidelines and Appraisal of Research Results 103 sediment loads of Figure 4.4. Note that there is an apparent increase of sediment load at stage E (Figure 4.4A) as some stored sediment is remobilized. Although applicable only to incised channels, this and similar evolution models (Harvey and Watson 1986) have been of value in developing an understanding of watershed channel dynamics and in characterizing whether or not a reach is stable (ASCE 2008). This model (Figure 4.3) was originally based on observations of Oaklimiter Creek, a channelized (straightened) stream with sandy bed and cohesive banks in northern Mississippi (Schumm et al. 1984). The sequence described a systematic response of a channel to base level lowering and encompasses conditions that range from disequilibrium (Figure 4.3E) to a new state of dynamic equilibrium (Figure 4.3B and 4.3E). Stages C and D in Figure 4.3 illustrate the widening that can accompany incision. These stages are only conceptual and variations may be encountered in the field; however, the sequence enables the evolutionary state of the channel to be determined from field observations that record the characteristic channel forms associated with each stage of evolution. The mor- phometric characteristics of the channel reach types can also be correlated with hydraulic, geo- technical, and sediment transport characteristics (ASCE 2008). Field investigations in the upper Colorado River basin have also revealed that the large arroyos formed by incision of valley-floor alluvium in the latter part of the nineteenth century are at present storing sediment in newly developed floodplains (Gellis et al. 1991). Daily sediment-load data downstream from these arroyos for 1930–1963 were examined, with later data excluded to avoid confounding effects of reservoirs. These incised channels are also behaving as illustrated in Figure 4.3. At the later stages of adjustment they are eroding less sediment and storing larger amounts of sediment. As a result, sediment loads at the Grand Canyon gaging station have decreased during the period of record, prior to closure of Glen Canyon Dam and other upstream dams in 1963 (Figure 4.4B). In addition, sediment deposition in Lake Powell between 1963 and 1986 is only 43 percent of that estimated prior to dam construction (Ferrari 1988), which indicates that the channel adjustment process is occurring throughout the upper Colorado River basin in a manner similar to that in the incised channels of Tennessee (Figure 4.4A). Because of climatic differences, the evolutionary changes involved in the complex response of Figures 4.3 and 4.4 require about 100 years in the Southwest but only about 40 years in the Southeast. As noted in Section 4.2.4, mature trees on a graded bank slope are convincing evidence of bank stability. A detailed study of bank erosion on streams in southern British Columbia pro- vides a quantitative assessment of the role of riparian vegetation in limiting channel width adjustments in bends during major flood events (Beeson and Doyle 1995). A total of 748 bends in four stream reaches were assessed by comparing pre- and post-flood aerial photography. Bends without riparian vegetation were found to be nearly five times as likely as vegetated bends to have undergone detectable erosion during the flood events. The likelihood of erosion on semi-vegetated bends was between that of the vegetated and non-vegetated category of bends. Most of the non-vegetated bends experienced major erosion (widening) in excess of 150 ft (45 m). Similar findings have been reported for streams in other regions, but the effect of riparian vegetation diminishes as bank heights and angles increase so that root reinforcement does not extend deeply enough to intersect failure planes. Also, large rivers deeply inundate the lower bank so that vegetation does not grow in that critical zone. Although the landscape as a whole may appear unchanging except over long periods of time, components of the landscape can evolve or adjust to human activities and hydrologic variations (Figures 4.3 and 4.4) during relative short periods of time and can pose serious problems for the design and successful implementation of environmentally sensitive stream bank protection treatments as successive waves of aggradation and degradation sweep through the system on an engineering time scale.

104 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Consequently, one of the key factors recommended for evaluation during a field site visit of an existing environmentally sensitive treatment or during the reconnaissance phase for the selection and design of a new treatment is to establish whether or not the channel is vertically stable and, if it is not, what the channel evolution stage for the reach of interest is (see Appendix C, Field Data Forms, Part 2). The Key to Stability. Both NCHRP Report 544 and FHWA’s HEC-23 (Lagasse et al. 2009) note that a common cause of failure for both “hard” and “soft” stream bank protection measures is erosion at the upstream and downstream “flanks” on the bank and/or erosion at the toe on the stream bed. McCullah and Gray (2005) address the issue of bank or flank keys in a special topic “The Key to Stability is the Key.” They observe that many stream bank protection projects fail at the upstream or downstream end. One of the important lessons derived from the Section 32 Program (see Table 2.1) is that the most common mistake in designing protection for an eroding bank is to extend bank protection too far upstream and not far enough downstream. Local flow acceleration or “expansion” at the downstream end of a protected bank often leads to local scour that progressively undercuts the protection and eats its way upstream. At the upstream end, erosion associated with impinging flow sometimes results in flow above or behind the protective measures and eventual failure. Keys consist of rock riprap or other heavy granular material buried in deep trenches dug at 30 degree angles (not perpendicular) to the flow (see Fig- ure 4.5) (McCullah and Gray 2005). Design considerations include: • Soil types. • Flow velocities. • Flood crests, durations, and recurrence intervals. • Active failure areas. Figure 4.5. Plan and cross-section views of keys (McCullah and Gray 2005).

Design Guidelines and Appraisal of Research Results 105 • Vegetation. • Height of bank. • Exposed key rocks in banks that can be angled to help redirect flow. • Crests of keys should be flush (even) with the bank, as this will decrease turbulence and chance of scour (see Figure 4.5). If key rock is exposed on the surface of the bank, it can be angled to help redirect flow on the bank back toward the stream. • Keys at the upstream and downstream ends of continuous longitudinal features should not be at a 90 degree angle to the feature, but at a 30 degree angle. HEC-23 “Bridge Scour and Stream Instability Countermeasures: Experience, Selection and Design Guidance” (Lagasse et al. 2009) provides guidance for the toe scour issue for stream bank protection measures. The recommendations for riprap revetment regarding toe scour are typi- cal (see HEC-23 Design Guideline 4) where the depth of maximum scour for countermeasure “toe down” into the stream bed includes the sum of long-term degradation, contraction scour, and toe scour. As noted, FHWA’s HEC-20 (Lagasse et al. 2012) provides guidance for estimating long-term degradation in a stream reach and HEC-18 (Arneson et al. 2012) provides equations for calculating contraction scour. HEC-23 includes design guidance for estimating scour in a protected bendway (toe scour) based on work by Maynord (1996). While these techniques were developed for armoring countermeasures such as riprap or articulating concrete block systems, they are applicable to estimating “toe down” requirements for any environmentally sensitive treatment that includes a “hard” toe (e.g., longitudinal stone toe). HEC-23 notes that deep regions along the toe of the outer bank of a bendway are the result of scour. High velocity along the outer bank is caused by secondary currents and greater outer-bank depths, and together with the resultant shear stress, produce scour and cause a difference between the sediment load entering and exiting the outer-bank zone (see Figure 4.9). Since secondary currents transport sediment supplied, in large part, from outer-bank erosion toward the inner bank of a bend, hardening of the outer bank by longitudinal bank-protection structures may cause the channel cross section to narrow and deepen by preventing the recruitment of eroded outer-bank sediments. Experience is usually the most reliable means of estimating scour depth when designing a bank-protection project for a particular stream. Lacking experience on a particular stream, scour depths may be estimated using physically based analytical models or empirical methods. Although scour depth can be estimated empirically or analytically, empirical methods are generally found to provide better agreement with observed data. Maynord (1996) provides an empirical method for determining scour depths on a typical bendway bank-protection project. Although his studies are restricted to sand-bed streams, the Maynord method agrees reasonably well with the limited number of gravel-bed data points obtained by Thorne and Abt (1993). Nonetheless, the techniques presented by Maynord are restricted to meandering channels having naturally developed widths and depths, and cannot be applied to channels that have been confined to widths significantly less than would occur naturally. Maynord’s method of estimating scour depth is based on a regression analysis of 215 data points. The scour data used in developing his equation were measured at high discharges that were within the channel banks and had return intervals of 1–5 years. Maximum depth as defined in his best-fit equation for scour depth estimation is a function of Rc/W, width-to-depth ratio, and mean depth as follows: D D 1.8 0.051 R W 0.0084 W D (4.2) mxb mnc c mnc = −     +  

106 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures where: Rc = Centerline radius of the bend, ft (m) W = Width of the bend, ft (m) Dmxb = Maximum water depth in the bend, ft (m) Dmnc = Average water depth in the crossing upstream of the bend, ft (m) The terms Dmxb and Dmnc are defined in Figure 4.6. The applicability of Maynord’s equation is limited to streams with Rc/W from 1.5 to 10 and W/Dmnc from 20 to 125 because of the lack of data outside these ranges. He recommends that for channels with Rc/W < 1.5 or width-to-depth ratios less than 20, the scour depth for Rc/W = 1.5 and W/Dmnc = 20, respectively, be used. In addition, Thorne and Abt (1993) suggest these methods are valid unless there is significant interaction between the main channel flow and overbank flow. Therefore, Maynord (1996) recommends that application of these empirical methods to overbank flow conditions should be limited to overbank depths less than 20% of the main channel depth. Lateral Channel Stability. From a bank-protection standpoint, ideally a stable channel is one that does not change in size, form, or position over time. However, all alluvial channels change to some degree and, therefore, have some degree of inherent instability. For purposes of this document, an unstable channel is defined as one with a rate or magnitude of change that is sufficiently large to be a significant factor in the design and maintenance of engineered structures and countermeasures within the river environment (Lagasse et al. 2012). Although a stream or river may appear unstable, this does not necessarily indicate that it is not an equilibrium or regime channel. Based on the relationship of channel width, depth, and slope to discharge, most natural alluvial channels have probably attained or approached a state of equilibrium at one time or another. Yet, these channels migrate laterally at rates ranging from imperceptible to very rapidly. Thus, equilibrium or regime channels may not necessarily be stable in the practical engineering sense. An actively migrating channel may maintain its equilibrium Figure 4.6. Definition sketch of width (W) and mean water depth at the crossing upstream (Dmnc ) of the bend and maximum water depth in the bend (Dmxb ).

Design Guidelines and Appraisal of Research Results 107 slope and cross section while posing a threat or hazard to engineered structures and both “hard” and “soft” countermeasures. Some types of lateral instability are shown in Figure 4.7. Meandering streams (Sketches 1, 2, and 3 in Figure 4.7) are classified as either actively or passively meandering. An actively meandering stream has sufficient stream power to deform its channel boundaries through active bed scour, bank erosion, and point bar growth. Conversely, while a passively meandering stream is sinuous, it does not migrate or erode its banks. Such channels are sometimes termed, “ideally stable,” or “moribund.” Although there is no completely satisfactory explanation of how or why meanders develop (Knighton 1998), it is known that meanders are initiated in straight channels by localized bank retreat which alternates from one side of the channel to the other in a more or less regular pattern. In addition, deformation of the channel bed may be an important prerequisite that modifies the pattern of flow prior to meandering. It is believed that secondary helicoidal flow develops spon- taneously in straight channels as a result of vortices generated at the boundary walls (Figure 4.8) (Einstein and Shen 1964, Shen and Komura 1968). A pair of surface-convergent helical cells will form if vortices develop along both banks. Inequalities in bank roughness may induce asymmetry in these cells and periodic reversal of the dominant cell. This periodically reversing helicoidal flow has an important influence on the pattern of erosion and deposition through meanders, and more specifically by forming a meandering thalweg and alternating bars (Einstein and Shen 1964). In addition, macroturbulent flow and the bursting process (i.e., streamwise fluctuations in the velocity field) are also important components in bank deformation (Yalin 1971 and 1992). The primary features of the flow pattern through meander bends are: • Superelevation of the water surface against the outside (convex) bank (Figure 4.9) • Transverse current directed toward the outer bank at the surface and toward the inner bank at the bed producing a secondary circulation additional to the main downstream flow Figure 4.7. Types of lateral activity and typical associated floodplain features (Thorne 1998).

108 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Figure 4.8. Models of flow structure and associated bed forms in straight alluvial channels: (A) Einstein and Shen’s (1964) model of twin periodically reversing, surface-convergent helical cells (the dark stippled line shows the trace of the thalweg) and (B) Thompson’s (1986) model of surface-convergent flow produced by interactions between the flow and a mobile bed, creating riffle-pool units of alternate asymmetry. Black lines indicate surface currents, and white lines represent near-bed currents (Knighton 1998). • Maximum-velocity current which moves from near the inner bank at the bend entrance to near the outer bank at the bend exit, crossing the channel at the zone of maximum bend curvature The interaction between centrifugal force acting outwardly on the water as it flows around the bend and an inward-acting pressure gradient force driven by the cross-stream tilting of the water surface is reflected in the above characteristics. The transverse current and the primary downstream flow component combine to produce the helicoidal motion to the flow. The superelevation of the water surface against the outer bank of a bend produces a locally steep downstream energy gradient and, in turn, a zone of maximum boundary shear stress in close proximity to the outer bank just downstream of the bend apex (Figure 4.9). The maximum shear stress zone shifts outward further upstream as a result of the bar-pool topography and cross- sectional asymmetry characteristic of meander bends.

Design Guidelines and Appraisal of Research Results 109 Secondary currents, which are usually weaker than primary ones, influence the distribution of velocity and boundary shear stress. Markham and Thorne (1992) divided the bend cross section into three regions relative to the pattern of secondary flow (Figure 4.9): • Mid-channel region, helicoidal flow is well established passing nearly 90 percent of the flow; • Cell of opposite circulation develops in the outer-bank region: the strength of this cell increases with discharge, the steepness of the bar, and the acuteness of the bend; and • Inner bank region where shoaling over the point bar induces a net outward flow, forcing the core of maximum velocity more rapidly toward the outer bank (Dietrich and Smith 1983, Dietrich 1987); increasing stage tends to reduce the shoaling, allowing an inward component of near-bed flow over the bar top. Understanding the role of secondary currents is particularly important in the design of “redirective techniques” that disrupt the secondary circulation, particularly in larger, deeper channels. For more detail on the utility of redirective techniques in stream bank protection see McCullah and Gray (2005). In summary, the pattern of primary and secondary currents influences the distribution of erosion and deposition in meanders. In general, erosion in the bend is concentrated along the outer bank downstream of the bend apex where the currents are strongest, while point bar building predominates in a parallel position along the opposite bank, with material supplied Figure 4.9. Flow patterns in meanders: (A) (i) location of maximum boundary shear stress (to ), and (ii) flow field in a bend with a well-developed point bar (after Dietrich 1987); (B) secondary flow at a bend apex showing the outer-bank cell and the shoaling-induced outward flow over the point bar (after Markham and Thorne 1992); and (C) model of the flow structure in meandering channels (after Thompson 1986). Black lines indicate surface currents and white lines represent near-bed currents (Knighton 1998).

110 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures by longitudinal and transverse currents. This produces a largely down-valley component to meander migration. The next section introduces and references a relatively simple technique for predicting the rate and direction of meander bend movement that could be applied in the design, installation, and monitoring of environmentally sensitive stream bank treatments. Predicting Meander Migration. In general, most streams are sinuous to some degree and the majority of bank retreat and lateral migration occurs along meander bends. As such, the following discussion on evaluating and predicting lateral migration will focus on meander bends. A relatively accurate method of determining migration rates and direction is through the com- parison of sequential historical aerial photography (photos), maps, and surveys. In the 1970s Brice introduced a methodology for conducting a stream stability and meander migration assessment using a comparative analysis of aerial photos, maps, and channel surveys for both the Army Research Office (ARO) and FHWA (Brice 1975, 1982). Today, historical aerial photos and maps can be obtained from a number of federal, state, and local agencies (see HEC-20 Section 6.3 for a detailed listing of sources). Developing a practical methodology to predict the rate and extent of channel migration (i.e., lateral channel shift and down-valley migration) in proximity to transportation facilities was the objective of NCHRP Project 24-16 (Lagasse et al. 2004). This research produced NCHRP Report 533: Handbook for Predicting Stream Meander Migration using aerial photographs and maps. The handbook deals specifically with the problem of incremental channel shift and pro- vides a relatively simple methodology for predicting the rate and extent of lateral channel shifting and down-valley migration of meanders. The methodology is based, primarily, on the analysis of bend movement using map and aerial photo comparison (overlay) techniques. For additional details and a demonstration of the procedure refer to NCHRP Report 533 (Lagasse et al. 2004) and/or HEC-20 (Lagasse et al. 2012). As with any analytical technique, aerial photograph comparison technologies have limitations. The accuracy of photo comparison is greatly dependent on the period over which migration is evaluated, the magnitude of internal and external perturbations forced on the system over time, and the number and quality of sequential aerial photos and maps. The analysis will be much more accurate for a channel that has coverage consisting of multiple datasets (aerial photos, maps, and surveys) covering a long period of time (several decades to more than 100 years) versus an analysis consisting of only two or three datasets covering a short time period (several years to a decade). Predictions of migration for channels that have been extensively modified or have undergone major adjustments attributable to extensive land-use changes will be much less reliable than those made for channels in relatively stable watersheds. Since the scale of aerial photography is often approximate, contemporary maps are usually needed to accurately determine the scale of air photos without the use of sophisticated photogrammetric instruments. In addition to scale adjustment and distortion problems that are inherent in the use of aerial photography for comparative purposes, there are a number of physical characteristics of the river environment that can complicate the prediction of meander migration impacts on transportation facilities. Countermeasures to halt bank erosion or protect a physical feature within the floodplain can have an impact on the usefulness of the overlays and these features should be identified prior to developing the overlays. Anomalous changes in the bend or bankline configuration or a major reduction in migration rates may suggest that bank protection is present, especially in areas where the bankline is not completely visible or on images with poor resolution. Geologic features such as clay plugs or rock outcrops in the floodplain can also limit the usefulness of the overlays because they can have a significant influence on migration patterns. Bends can become distorted as they impinge on these features and localized bankline erosion rates may decrease significantly as these erosion resistant features become exposed in the bank.

Design Guidelines and Appraisal of Research Results 111 In reaches where geologic controls are exposed predominantly in the bed of the channel, migration rates may dramatically increase because the channel bed is not adjustable. A fundamental assumption of overlay techniques based on aerial photo or map comparison is that a time period sufficient to “average out” such anomalies will be available, making the historic meander rates a reasonable key to the future. Even with these limitations, however, determining the rate and direction of movement of a specific bend can provide insight and specificity to the design, installation, and monitoring of environmentally sensitive stream bank protection measures. Hydraulic Stress on a Bendway. As an indicator of the potential for success with a specific treatment for erosion of the stream bank in a meander bend, the ratio of bend radius of curvature to flow width can provide insight into the force on the meander bend margin. This parameter does not include discharge. A quantitative technique that considers a single-event discharge and an estimate of the radial stress on a meander bend margin was developed to evaluate the perfor- mance of alternative stream bank erosion protection techniques for the USACE, Vicksburg District [Water, Engineering & Technology (WET) 1990]. This technique could also be used to evaluate alternative channel instability countermeasures for an erosion site located in a meander bend (Lagasse et al. 2009). For this technique, Begin (1981) defines radial stress as the centripetal force divided by the outer-bank area. The centripetal force is responsible for deflecting the flow around the bend and is equal to the apparent reactive force of the flow on the bend. Thus, the radial stress is defined as a force per unit area (lbs/ft2 or N/m2). Although it is not suggested that the radial stress is directly responsible for meander bend migration or failure of bank-protection countermeasures, Begin did show that the radial stress is related to meander migration. It is assumed that shear stress is related to radial stress because of water surface superelevation and increased near-bank velocity gradients (see Figure 4.9). Field investigations and computation of radial stress on banklines for channels in the Yazoo River basin in Mississippi clearly showed that rudimentary countermeasures (such as used-tire revetment) were generally unsuccessful even in bends with low to moderate radial stress (WET 1990). The study also showed that stone structures including longitudinal stone dikes and stone spurs performed well in reaches of high radial stress. Isolated failures of stone structures did occur at locations with the highest radial stress. The 2-year storm discharge was used in the computations for radial stress at these sites. As an alternative, the increased shear force on the outside of bends can be calculated by multiplying the bed shear stress, t0, by a dimensionless bend coefficient, Kb. The sharper the bend, the greater the shear stress imposed on the outer bank. The bend coefficient, Kb, is related to the ratio of the bend radius of curvature, Rc, divided by the top width of the channel, T, as shown in Figure 4.10. While the simplistic techniques presented in this section can provide a “quick estimate” of relative stress on a bendway, design practice is rapidly moving toward the use of 2-D hydro- dynamic models that provide shear stress and velocity values for each grid cell at each time step. 4.2.4 Geotechnical Considerations Overview As discussed in Sections 4.2.2 and 4.2.3, geomorphic factors that influence the success or failure of an environmentally sensitive stream bank protection measure include aggradation or degradation of the stream bed, the evolution of incised channels, meander migration, channel width adjustments, and radial stress on meander bendways. There is an obvious interplay between

112 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures these geomorphic factors and the geotechnical considerations that affect the stability of the channel banks on which the erosion protection measures are to be placed. The vertical stability of the channel directly affects the stability of the channel banks, and the lateral stability of the channel is a function of the hydraulic stresses to which the channel banks are subjected. In this section the general processes affecting channel bank geotechnical stability are discussed, followed by consideration of several site-specific factors that influence the success or failure of stream bank erosion control measures. The appearance of the stream bank is generally a good indication of relative stability. A field inspection of a channel will help to identify characteristics that are associated with erosion rates: • Unstable banks with moderate to high erosion rates usually have slopes that exceed 30 percent, and a cover of woody vegetation is rarely present. At a bend, the point bar opposite an unstable cut bank is likely to be bare at normal stage, but it may be covered with annual vegetation and low woody vegetation, especially willows. Where very rapid erosion is occurring, the bank may have irregular indentations. Fissures, which represent the boundaries of actual or potential slump blocks along the bankline, indicate the potential for very rapid bank erosion. • Unstable banks with slow to moderate erosion rates may be partly reshaped to a stable slope. The degree of instability is difficult to assess, and reliance is placed mainly on vegetation. The reshaping of a bank typically begins with the accumulation of slumped material at the base such that a slope is formed and progresses by smoothing of the slope and the establishment of vegetation. • Eroding banks are a source of debris when trees fall as they are undermined. Therefore, debris can be a strong indicator of unstable bank conditions. • Stable banks with very slow erosion rates tend to be graded to a smooth slope of less than about 30 percent. Mature trees on a graded bank slope are convincing evidence of bank stability. In most regions of the United States, the upper parts of stable banks are vegetated, but the lower part may be bare at normal stage, depending on bank height and flow regime of the stream. Where banks are low, dense vegetation may extend to the water’s edge at normal stage. Where banks are high, occasional slumps may occur on even the most stable graded banks. 1 1.5 2 2 3 4 5 6 7 8 9 10 11 12 Rc/T B en d Co ef fic ie nt K b = τ b/ τ 0 Kb = 2.0 for Rc/T < 2 Kb = 2.38 - 0.206(Rc/T) + 0.0073(Rc/T)2 for 2 < Rc/T < 10 Kb = 1.05 for Rc/T >10 Figure 4.10. Shear stress multiplier, Kb, for bends (Kilgore and Cotton 2005).

Design Guidelines and Appraisal of Research Results 113 Active bank erosion can be recognized by falling or fallen vegetation along the bankline, cracks along the bank surface, slump blocks, deflected flow patterns adjacent to the bank- line, live vegetation in the flow, increased turbidity, fresh vertical faces, newly formed bars immediately downstream of the eroding area, and, in some locations, a deep scour pool adja- cent to the toe of the bank. These indications of active bank erosion can be noted in the field and on stereoscopic pairs of aerial photographs. Color infrared photography is particularly useful in detecting most of the indicators listed above, especially differences in turbidity (Shen et al. 1981). Figure 4.11 illustrates some of the features that indicate that a bankline is actively eroding. Physical Processes Controlling Bank Failure Bank Materials. Resistance of a stream bank to erosion is closely related to several charac- teristics of the bank material. Bank material deposited in the stream can be broadly classified as cohesive, noncohesive, and composite. Typical bank failure surfaces of various materials are shown in Figure 4.12 and are described as follows (Brown 1985): • Noncohesive bank material tends to be removed grain by grain from the bank. The rate of particle removal and, hence, the rate of bank erosion is affected by factors such as particle size, bank slope, the direction and magnitude of the velocity adjacent to the bank, turbu- lent velocity fluctuations, the magnitude of and fluctuations in the shear stress exerted on the banks, seepage force, piping, and wave forces. Figure 4.12(a) illustrates failure of banks of noncohesive material from flow slides resulting from a loss of shear strength because of saturation and failure from sloughing resulting from the removal of materials in the lower portion of the bank. • Cohesive material is more resistant to surface erosion and has low permeability, which reduces the effects of seepage, piping, frost heaving, and subsurface flow on the stability of the banks. However, when undercut and/or saturated, such banks are more likely to fail due to mass- wasting processes. Failure mechanisms for cohesive banks are illustrated in Figure 4.12(b). • Composite or stratified banks consist of layers of materials of various sizes, permeability, and cohesion. The layers of noncohesive material are subject to surface erosion, but may be partly protected by adjacent layers of cohesive material. This type of bank is also vulnerable to ero- sion and sliding as a consequence of subsurface flows and piping. Typical failure modes are illustrated in Figure 4.12(c). Figure 4.11. Active bank erosion illustrated by vertical cut banks, slump blocks, and falling vegetation.

114 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Piping. Piping is a phenomenon common to alluvial stream banks. With stratified banks, flow is induced in more permeable layers by changes in stream stage and by waves. If flow through the permeable lenses is capable of dislodging and transporting particles, the material is slowly removed, forming “pipes” that undermine portions of the bank. Without this foundation material to support the overlying layers, a block of bank material drops down and results in the development of tension cracks as sketched in Figure 4.12(c). These cracks allow surface flows to enter, further reducing the stability of the affected block of bank material. Bank erosion may continue on a grain-by-grain basis or the block of bank material may ultimately slide downward and outward into the channel, with bank failure resulting from a combination of seepage forces, piping, and mass wasting. Mass Wasting. Local mass wasting is another form of bank failure. If a bank becomes satu- rated and possibly undercut by flowing water, blocks of the bank may slump or slide into the channel. Mass wasting may be caused or aggravated by the construction of homes on river banks, operation of equipment adjacent to the banks, added gravitational force resulting from tree Figure 4.12. Typical bank failure surfaces: (a) noncohesive, (b) cohesive, and (c) composite (after Brown 1985).

Design Guidelines and Appraisal of Research Results 115 growth, location of roads that cause unfavorable drainage conditions, agricultural uses on adjacent floodplain, saturation of banks by leach fields from septic tanks, and increased infiltration of water into the floodplain as a result of changing land-use practices. Various forces are involved in mass wasting. Landslides, the downslope movement of earth and organic materials, result from an imbalance of forces. These forces are associated with the downslope gravity component of the slope mass. Resisting these downslope forces are the shear strength of the materials and any contribution from vegetation via root strength or engineered slope reinforcement. When the toe of a slope is removed, as by a stream, the slope materials may move downward into the void in order to establish a new equilibrium. Often, this equilibrium is a slope configuration with less than original surface gradient. The toe of the failed mass then provides a new buttress against further movements. Erosion of the toe of the slope then begins the process over again (see discussion of basal endpoint control, below). General mass wasting often accompanies channel incision as bank heights increase. Bank Erosion and Failure. The erosion, instability, and/or retreat of a stream bank are dependent on the processes responsible for the erosion of material from the bank and the mech- anisms of failure resulting from the instability created by those processes. Bank retreat is often a combination of these processes and mechanisms operating at various timescales. While the detailed analysis of bank stability is, primarily, a geotechnical problem, insight on the relation- ship between stream channel degradation and bank failure, for example, can be important to the designer concerned with the influence of bank instability on environmentally sensitive bank- protection measures. For a detailed discussion of the processes responsible for bank erosion and bank failure mechanisms, refer to HEC-20, Appendix B (Lagasse et al. 2012). Site-Specific Considerations Basal Endpoint Control. Material is delivered to the basal area of a bank by mechanical bank failures and erosion. The removal of this material from the basal area depends almost entirely on fluvial entrainment and downstream transport (Figure 4.13). The amount of basal accumulation of bank material depends on the relative rates of supply by bank failures and erosion and removal by fluvial entrainment. Where the flow is able to remove all the sediment supplied to the basal area and scour of the basal area continues, bank erosion will also continue. In contrast, where the rate of supply exceeds the rate of removal, bank stability will be increased with respect to gravity failures because loading and buttressing the base of the slope effectively reduces the bank angle and height. Neill (1984) has argued that the bedload transport rate must set an upper limit to local erosion rates over a period of time, and Nanson and Hickin (1986) support this view. However, if the failed material is cohesive, the failure blocks will gradually break apart and most of it will be transported as wash load. Carson and Kirkby (1972) characterize the balance between basal supply and removal in terms of three states of basal endpoint control, as follows: 1. Impeded Removal. If bank failures supply material to the base at a higher rate than it is removed, then basal accumulation results, thus decreasing the bank angle and vertical height and increas- ing bank stability. 2. Unimpeded Removal. Bank failures and erosion supply material to the base at the same rate that it is removed resulting in bank recession by parallel retreat, the rate being controlled by the degree of fluvial activity at the base of the bank. Slope angle and basal elevation remain relatively unchanged. 3. Excess Basal Capacity. Basal scour is greater than the rate of supply of material. This causes bed scour and basal lowering, which increases the bank height and angle and promotes bank failure.

116 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Estimating Critical Bank Height. The stability of the bank with respect to mass failure is dependent on soil properties and bank geometry. Bed lowering and lateral erosion are the two most common processes that act to steepen the bank and cause bank instability. For estimating critical bank height for steep, cohesive banks, a simple slope stability analysis can be developed. Refer to the analysis approach derived by Osman and Thorne (1988) to predict bank stability response to lateral erosion and bed degradation. Thorne and Osman (1988) also developed a modeling technique to study the effects of channel widening and bank-sediment contribution on flow energy, stream power, and the rate and extent of bed lowering during degradation and outer-bank stability using a critical shear stress concept to account for lateral erosion and a slope stability criterion for mass failure. Again, a review by the designer is recommended prior to evaluating lateral erosion and bank instability problems in detail for a given site where environmentally sensitive treatments are being considered. Deterministic Modeling. Currently there are a number of deterministic tools available to the designer of environmentally sensitive stream bank treatments to evaluate geotechnical factors that could influence the choice of treatment or support developing design of treatment components. An example of a deterministic approach is the bank stability and toe erosion model (BSTEM) developed at the USDA-ARS National Sedimentation Laboratory. BSTEM is a spread- sheet tool used to simulate stream bank stability. The user supplies bank geometry, soil proper- ties, vegetation cover and bank water-table levels, and the model outputs factors of safety. If banks fail or erode, new bank geometry is generated. Fluvial erosion of material at the bank toe is simulated using a simple excess-shear stress approach (Simon et al. 2011). Figure 4.13. Schematic representation of sediment fluxes to and from river bank basal zones (Thorne and Osman 1988).

Design Guidelines and Appraisal of Research Results 117 Many river management situations require information on the stability of the channel banks. These may include assessing the stability of existing channel banks, predicting the effect that changes in riparian land use will have, or designing new channels. The BSTEM is a spreadsheet model that calculates bank Factor of Safety (Fs) for new or existing banks. BSTEM features include: • Limit equilibrium analysis of planar shear failures with and without tension cracks; • User-defined and automatically-generated bank geometries: compound and undercut banks allowed; • Up to five distinct bank material layers; • Simulated saturated and unsaturated soil strength; • User-definable positive and negative pore-water pressures or pressures calculated from water table position; • Optional fiber-bundle root-reinforcement model with data from 22 vegetation species, including willows, grasses and large trees, or the users may enter their own data; • Hybrid random-walk and random-leap search algorithm for the minimum Fs; • Clear-water scour hydraulic erosion model; • Simulated potential options to protect the bank and/or bank toe against hydraulic erosion. As noted in the BSTEM Users Manual, the model is a physically based model. It represents two distinct processes: the failure by shearing of a soil block of variable geometry and the erosion by flow of bank and bank-toe material. The effect of toe erosion, vegetative treatments, or other bank and bank-toe protection measures can be illustrated by calculating the actual Fs of the bank (Bankhead et al. 2013). The BSTEM combines three limit equilibrium-method models that calculate Fs for multi-layer stream banks. The methods simulated are horizontal layers, vertical slices with tension crack, and cantilever failures. The model can be adapted to incorporate the effects of geotextiles or other bank stabilization measures that affect soil strength. The model accounts for the strength of up to five soil layers, the effect of pore-water pressure [both positive and negative (matric suction)], confining pressure due to streamflow, and soil reinforcement and surcharge due to vegetation. BSTEM can be used as a tool for making reasonably informed estimates of hydraulic erosion of the bank and bank toe by hydraulic shear stress. The model is primarily intended for use in studies where bank-toe erosion threatens bank stability. The effects of erosion protection on the bank and toe can be incorporated to show the effects of erosion control measures. The model estimates boundary shear stress from channel geometry and considers the critical shear stress and erodibility of two separate zones with potentially different materials: the bank and bank toe. The bed elevation is assumed to be fixed. To evaluate stream bank stability, the shear strength of saturated soil can be described by the Mohr-Coulomb criterion (Simon et al. 2011). Driving forces for stream bank instability are controlled by bank height and slope, the unit weight of the soil and the mass of water within it, and the surcharge imposed by any objects on the bank top. The ratio of resisting to driving forces is commonly expressed as the Fs, where values greater than one indicate stability and those less than one, instability. Figure 4.14 provides a flow chart illustrating the bank stability model inner workings to output an Fs and erosion from both hydraulic and mass failure processes. Both beneficial and detrimental effects of vegetation on bank stability have been observed: plant roots reinforce soils and remove moisture by evapotranspiration, increasing matric suction, but very large trees impose weight loading on banks and woody vegetation may also enhance infiltration of pre- cipitation (Simon and Collison 2002, Gray and Barker 2004, Pollen et al. 2004). Earlier models of root reinforcement of soils ignored the effects of soil type and moisture on root-soil bonds; root tensile strength was simply added to soil strength and roots were assumed to be oriented

118 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures perpendicular to failure planes. Work by Pollen (2007) has advanced understanding of tempo- ral and spatial variability in root reinforcement due to variations in soil type and moisture, and work by her team (Pollen et al. 2004, Pollen and Simon 2005) has shown that soil-root matrices are better simulated by fiber-bundle models that allow progressive failure of roots (weakest first) rather than the “all-roots-break-instantaneously” assumption of the older perpendicular mod- els. The fiber-bundle model algorithm has been incorporated into the BSTEM. A relevant example of the application of the BSTEM model can be found in Simon et al. (2008) with the application of the model to design a reach-scale restoration project. Specifically, this project deals with restoration design using environmentally sensitive stream bank treatments on Goodwin Creek in Mississippi—NCHRP Project 24-39 field site MS3 (see Section 2.4.2 and Figure 2.3). As described by Simon et al. (2008), because of continued land loss in adjacent agriculture fields by mass failure of the stream banks on Goodwin Creek, a restoration project was designed to stabilize the banks and to protect a road running parallel to the bendway. To provide a stable alternative, the analysis of the restored configuration needed to address both hydraulic erosion and geotechnical stability. The proposed design was limited to 1:1 bank slopes due to the proximity of the road and included longitudinal stone-toe protection and bendway weirs to counter basal erosion by hydraulic shear. Worst-case geomorphic conditions under the proposed design were simulated by modeling (1) typical, annual high flows to evaluate the amount of bank-toe erosion that would occur and (2) geotechnical stability where groundwater levels were high and flow had receded to low-flow conditions in the channel (drawdown case). Results showed that the bank would still be unstable at 1:1 under the drawdown case but that the addition of specific riparian Figure 4.14. Flow chart illustrating the BSETM inner workings and output (Bankhead et al. 2013).

Design Guidelines and Appraisal of Research Results 119 vegetation on the slope would stabilize the bank even under worst-case conditions. The design was, therefore, implemented and constructed. Thus, for the Goodwin Creek project the use of riparian vegetation to increase shear strength by root reinforcement was central to the design. Post-project monitoring for 10 months after implementation revealed that the bank slope has remained stable and that bed-material size has returned to pre-project conditions. However, some bed scour has occurred. This was expected given that flows were re-directed away from the bank toe and into the center of the channel. Further scour is not expected as bed material has coarsened. This successful project shows how deterministic approaches to design and implementation of a reach-scale restoration project can provide reasonable confidence in developing and testing designs (Simon et al. 2008). 4.2.5 Monitoring Success of the Vegetative Component Monitoring During Initial Establishment Since vegetative components of biotechnical stabilization countermeasures grow, develop, experience dormancy and sometimes die, periodic inspection is necessary, particularly during the critical period of early establishment just after installation. If the installation is determined to have been improper, any warranty or dispute resolution clauses in the plant installation contract may be invoked. In particular, irrigation programs require attention to see that they are effective. Inspection frequency should be high during the first growing season, particularly after high-flow events and during droughts. After the first year, semiannual to annual evaluations should be suf- ficient in most cases. Sites should be visually checked for drought stress, herbivory, trampling, competition from undesirable species, and vandalism (FISRWG 2001). McCullah and Gray (2005) also highlighted excessive soil moisture, insufficient soil nutrients, toxic soil conditions (high alkalinity or acidity), and inadequate light as additional issues of concern. Monitoring the effects of riparian revegetation on ecological values is discussed by Guilfoyle and Fischer (2006), and a broader protocol that includes water quality, biotic factors, and habitat for rapid bioassessment of wadeable streams is provided by Barbour et al. (1999). Visual Assessment Jones and Johnson (2015) adapted parts of the Barbour et al. (1999) protocol to create a damage assessment framework for wadeable, 1st to 4th order, modified streams in “urban or otherwise constrained” settings. Specifically, the types of stream projects considered were ones that aimed to improve natural channel functioning and protect infrastructure, property, or other physical assets in or near the channel. Furthermore, these streams were located in such close proximity to infrastructure that a state of static equilibrium was required. Static equilibrium was defined (after Rhoads et al. 2008) as lateral and vertical rates of change that were relatively slow on an engineering time scale, were not accelerating, and were occurring within the context of a balanced sediment regime. An adaptation of the Jones and Johnson (2015) approach is suggested here for qualitative, visual monitoring of environmentally sensitive stream bank protection measures. More quantitative assessment methods are presented below. The visual assessment monitoring protocol for vegetative measures described here consists of scoring the treated bank in five categories, the first two of which are directly or indirectly related to vegetation (Table 4.2). The damage assessment framework developed by Jones and Johnson (2015) includes up to seven categories, but only five are used here. The omitted categories (flood hazard and thalweg degradation) deal with properties of the channel as a whole and not of a protected bank. General health and vigor of plants may be visually assessed. Leaf color, evidence of insects, herbivory, disease, and growth vigor are readily obvious and should be considered when assigning

120 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Category Excellent (1-3) Good (4-6) Poor (7-9) Failed (10-12) Stream bank vegetation Stream bank vegetation is in good condition1 and showing progress along all stream bank surfaces. 50%–70% of the bank is covered by vegetation in good condition or showing progress. 50%–70% of the bank is covered by stressed or dying plants. Less than 50% of the bank is covered by vegetation or disruption due to grazing and mowing is evident. Bank stability and migration Banks are stable and vegetated. Isolated instances of bank failures (mass wasting, undercut, etc.) or raw banks, affecting 5%–30% of treated bank. Bank failures or raw banks frequent, describing 30%– 60% of treated bank segment. Channel migration is evident anywhere in reach, but thalweg is within design channel limits. Bank failures or raw banks prevalent, describing >60% of treated bank segment. Thalweg has migrated outside design channel limits anywhere in reach. Infrastructure protection Infrastructure is not in immediate danger. Erosion has left infrastructure (1) nearer stream flow or (2) with more surface exposed to stream flow than as-built condition. Infrastructure shows unexpected signs of vulnerability that has the potential to impact the integrity or functioning of the infrastructure. Infrastructure is exposed to stream flow. The structural integrity of infrastructure is compromised or infrastructure has failed as a result of stream flow. Structural integrity2 Structure and structural components have not been displaced and there is no visible erosion. At least 10% of the structure is displaced from the as-built location and/or structure is attached to bank but erosion is visible everywhere structure is in contact with bank. 25%–75% of the structure is displaced from as- built location and/or structure is partially detached from bank. More than 75% of structure is displaced from as- built location and/or structure is detached from bank. Flow obstruction and sedimentation Less than half of the bottom is affected by sediment deposition. Pools are not filling in and there are few to no unintended obstructions. Occasional unintended obstructions are present; minor local scour at these obstructions. Sediment deposition is affecting 50%– 80% of the channel bottom or pool depths have measurably decreased. Moderately frequent unintended obstructions. Unintended obstructions are frequent or have significantly altered the design capacity of the channel. Aggradation is evident or sediment deposition affects >80% of the channel bottom. 1Key indicators of condition include leaf color and evidence of insects, herbivory, disease, trampling, competition from undesirable species, and vandalism. 2Score each structure in the reach and use the median score for the overall site score. Table 4.2. Visual assessment monitoring protocol for environmentally sensitive bank protection measures (adapted from Jones and Johnson 2015). scores for the stream bank vegetation category (Table 4.2). Assessment of the structural integ- rity category should be informed using failure modes tabulated by McCullah and Gray (2005), who frequently highlighted toe scour and flanking due to inadequate end treatments (keys) at upstream and downstream ends of protection structures (see Section 4.2.3). Scores from the five categories are combined by adopting the lowest score observed in any category as the overall site score. Table 4.2 has been incorporated into the proposed field data form for evaluating environ- mentally sensitive bank-protection treatments (see Appendix C).

Design Guidelines and Appraisal of Research Results 121 Quantitative Assessment Inspection of sites stabilized with biotechnical measures should include assessment of three properties of the vegetation: a. Overall health and vitality of the plants, b. Ability of plants to shield bank sediments from flow shear stresses by suppressing flow velocity in the near-bank region, and c. Ability of plant roots to reinforce soils and thus make them more erosion resistant. Each of these factors is discussed below. Overall Health and Vitality of Vegetation. Vitality of dormant woody vegetation may be checked by testing the elasticity of stems (whether they bend rather than break when loaded) and by visually examining the cambium layer under the bark to see if it is green. Counts of living and dead individuals may be used to compute survival percentages, but survival percentages may be misleading for two reasons. First, high survival percentages may occur in cases where survival patterns are patchy and “weak spots” in coverage create vulnerabilities and potential failure zones and second, low survival percentages may occur in cases where stand development is quite strong and natural succession is causing a decline in stem density but an overall increase in the size of individual plants. In the latter case, a low survival percentage would not be a negative finding with regard to the vegetative component. Satisfactory survival rates can be established in advance based on common-sense decisions regarding the adequacy of establishment relative to the objec- tives. Johnson and Stypula (1993) in FISRWG (2001) suggest using a rule of thumb that open spaces should not be larger than “2 ft in dimension.” Beyond the methods described above, health and vitality of vegetation may be examined using ground-based multispectral imagery to detect chlorophyll levels (Kancheva et al. 2014). Plant vitality and vigor may be assessed using a large number of indices computed from plant reflectance, transmittance, and absorbance for a variety of wavelengths. Efficacy of Plant Cover for Counteracting Surface Erosion. Plants protect against surficial erosion by deflecting flow and shielding bank surfaces. The effective shear forces acting on the bank are therefore reduced. In addition, shallow roots bind soils and increase effective cohesion, increasing the resistance of soils to erosion. For example, the universal soil loss equation (USLE) and its descendants are intended to predict erosion of soil due to raindrop impact and overland flow but not stream bank erosion; however, a vegetative component is considered: A R K LS C P (4.3)p=    where: A = Computed soil loss for a given storm period or time interval in tons per acre R = Rainfall factor LS = Slope length and steepness factor C = Vegetation factor P = Erosion control practice factor The variables that control the vegetation factor, C, in the USLE for woody vegetation are indicative of the plant cover characteristics that are important in controlling erosion. Tables of C-factors for brush, bushes, and trees show variation of raised canopy height, type and cover percentage as well as ground canopy type and cover as controlling variables. The importance of these characteristics has been confirmed by more recent research (Zuazo and Pleguezuelo 2008). Selected C-factor values for “permanent pasture, grazed forest land, range, and idle land,” tabulated by NRCS (2008) are presented in Figure 4.15. Erosion is inversely proportional to C,

122 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures so Figure 4.15 shows that predicted soil erosion is more strongly affected by the amount of cover that contacts the soil surface (reducing C-factor from 0.45 to near 0) than the type of vegetation (trees versus grass/weeds). Beasley (2011) and Flikweert et al. (2013) examined resistance of turfed levee slopes to erosion and considered several measures of turf status as erosion protection (root length, root surface area, root volume, biomass, and ground cover). The volume of roots per unit volume of soil was selected as the best index of a continuous cover of intact turf for erosion protection because it captures both root length and thickness. Longer roots provide more interaction with the soil, and thicker roots are less likely to break. Observation and measurement of the above-ground portion (“canopy coverage”) of the turf was not an adequate measure of “vegetative rooting and engineering performance.” However, measures of root properties for dormant and actively growing turf were not significantly different. Most biotechnical bank-protection measures employ woody species rather than turf. The value of this type of vegetation in countering surface erosion is directly related to its stem density and flexibility. Density. Density may be assessed by counting and measuring stems per unit bed area within selected quadrants or by using an approach that includes the frontal area of plants as described by Petryk and Bosmajian (1975) with digital imaging. Although shrubby species can persist in dense, low, supple stands for long periods, many woody stands become much less dense as they mature due to succession or self-thinning as stronger individuals outcompete and shade out their neighbors. Larger trees, although they may provide excellent habitat features, may provide relatively little protection against surface erosion and actually accelerate or concentrate flows between trunks or underneath canopy. This hazard may be counteracted by development of a Figure 4.15. Effect of vegetation characteristics on vegetative C-factor in USLE.

Design Guidelines and Appraisal of Research Results 123 healthy understory of shade-tolerant species. López and García (2001) showed that the hydraulic effects (e.g., flow depth, Manning n) of submerged, rigid vegetation were negligible until the density of stems (basal area of stems per unit area of stream bed) exceeded about 3 × 10-4, but increased as a linear function of density beyond that threshold. White et al. (2014) surveyed riparian communities in the vicinity of Louisville, Kentucky, and found mean values of stem basal area per unit ground surface area of 11 × 10-4 to 29 × 10-4. Flexibility. As long as flexible plants are not broken off and destroyed by the flow, they may be assumed to provide superior protection to that provided by rigid stems since they physi- cally shield the bank as they lay over, increasing the local stem density and reducing near-bank stresses. Vegetative flexibility has been assessed using the “board drop” test described by Kouwen (1988) and by measuring tree stem deflection associated with measured forces applied by a winch (Stone et al. 2011). These tests, although interesting, are primarily research tools. Quan- titative assessment of flexibility must therefore rely on an estimate of the likely behavior of the plants growing at a given site when subjected to flow depths and velocities typical of higher flows at the site. Observation of similar stands of vegetation under typical flow conditions is most reliable. In the absence of such observations, application of the theory for bending a cylindrical cantilever beam (as shown by Stone et al. 2011) yields the following relationship between deflec- tion angle and applied drag force: Tan FL 2IE (4.4) 2 θ = where: q = Angle between plant stem and vertical F = Applied force assumed to be concentrated at a height, L, above ground I = Second moment of area E = Tree’s modulus of elasticity The quantity I may be computed by assuming the plant behaves as a cylinder: I D 64 (4.5) 4 = pi where: D = Stem diameter and reasonable assumptions may be made for the value of E. Stone et al. (2011), measured values of E for three tree species and found a range from near zero to about 1.5 × 105 lb/ft2 or 7 × 106 N/m2. Assuming the individual plants have a frontal area when subjected to flow of 1 ft2, an effective height of 5 ft, a stem diameter of 0.25 ft and a drag coefficient of 1.0, then an approach velocity of only about 1 ft/s would produce a deflection of 30 degrees if E = 1.5 × 105 lb/ft2. This is a low velocity, but considering the shielding effects of adjacent plants in the near-bank region, it might correspond to a much higher mean channel velocity. Efficacy of Vegetative Component for Soil Reinforcement. Roots can increase slope stability when they intersect failure planes. Norris and Greenwood (2006) provide guidance on conducting geotechnical investigations of vegetation effects on slope stability at proposed construction sites (Table 4.3). Some of this material is relevant to assessment of existing biotechnical stream bank stabilization projects. Investigations may be staged, with preliminary work done in the office and more detailed investigation on site or in the laboratory with samples from the site. Desk study in the office can be done using maps, construction plans, aerial photos, and soil surveys.

124 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures 4.2.6 Aquatic Habitat Issues Introduction As noted, CRP-CD-58 that accompanies NCHRP Report 544 (McCullah and Gray 2005) contains detailed coverage on a series of special topics relevant to the design of environmentally sensitive stream bank treatments (see also Section 2.1e). McCullah and Gray (2005) address issues related to “physical aquatic habitat” in some detail as a special topic on this CD. This material is presented with minor editorial changes made by this projects research team in the paragraphs that follow. This special topic has been reviewed and supplemented, as appropriate, by the a fisheries biologist on the NCHRP Project 21-39 research team. Definitions and Diversity Habitat is the place or environment where a plant or animal naturally or normally lives and grows. Environment here implies the sum total of all influences within the living space of a plant or an animal. For fish, habitat includes the stream, its boundaries (bed and banks), existing vegeta- tion, and other animals. Physical factors such as water depth, velocity, cover, and bed material are referred to as physical habitat. In fact, these four factors are most often used to describe physical aquatic habitat. Streams tend to provide complex, dynamic physical habitat. For example, water depth and velocity vary continuously in time and space. Deep, slow pools lie adjacent to swift, shallower runs and riffles. Bed material, although slightly less dynamic than depth or velocity, also varies to produce a high level of spatial and temporal heterogeneity. The high level of physical diversity typical of natural (lightly impacted by humans) streams provides niches for many types of plants and animals, and thus supports relatively high levels of biological diversity. Human influences often result in simplification of stream habitats, making them more uni- form and adversely impacting biological communities (see Figure 4.16). Meandering streams with deep pools on the outside of bends and gravelly riffles at thalweg crossings (inflection points) between bends are often straightened and channelized to improve alignments for bridges or Study Phase Topic Subtopic Information Derived Desk Study Soils Topsoil Suitability as plant-growth media Subsoil Likely penetration by roots. Fill Soil classification and moisture regime. Vegetation Presence and distribution of vegetation. Analysis Slope stability Preliminary analysis of slope stability based on assumed properties for soil, hydrology and vegetation. Software tools such as BSTEM1 (http://www.ars.usda.gov/research/docs.htm?docid=5044) are helpful, as they allow use of assumed default values. Field Soils Shallow pits, perhaps hand dug Classify soils using texture. Root size, depth, density, spatial distribution. Consider seasonal effects of soil moisture regime on plants. Borehole, direct shear tests In situ measurement of soil shear strength (Lutenegger 1987) Vegetation Verify desk study findings regarding vegetation types, sizes, density. Apparent effects of planted and naturally occurring vegetation on stability of adjacent sites. Roots Root strength In situ pullout resistance. (Pollen-Bankhead et al. 2009) In situ shear tests on root reinforced soils (larger projects). Seasonal monitoring of moisture content profiles. (Shields et al. 2009, Pollen-Bankhead and Simon 2010) In-depth Assessment Analysis Slope stability Analysis of slope stability based on measured properties for soil, hydrology, and vegetation. BSTEM1 or more sophisticated tools may be used. 1See Section 4.2.4. Table 4.3. Site investigations for contribution of vegetation to slope stability (adapted from Norris and Greenwood 2006).

Design Guidelines and Appraisal of Research Results 125 highway embankments. Sinuous channels with a nonuniform cross section are sometimes altered to become a prismatic trapezoid, and pool-riffle sequences are commonly replaced with uniform runs. Woody debris, an essential component of many habitats, is usually removed or displaced as riparian vegetation is removed and banks are either cleared or stabilized. However, it is important to note that the spatial heterogeneity typical of natural streams is not random. Specific patterns occur that are essential for various populations. For example, swift waters immediately adjacent to eddies and regions of depressed velocity allow some organisms to obtain food with minimal expenditure of energy. An engineer or geo- morphologist could design a stream with a checkerboard pattern that would be physically diverse but ecologically barren since natural patterns would not occur. Lotic (moving water) ecosystems also depend on temporal patterns. Human activities tend to perturb natural hydrographs, either exaggerating extremes or making them more uniform by removing flood peaks and elevating base flows, as is the case with reservoirs. Urbanization and other types of watershed development often increase the fraction of precipitation that reaches the stream channel as surface runoff. This makes flood peaks higher and sharper, but depresses base flows and effective precipitation since less of the total precipitation infiltrates into the soil profile or aquifer, which supplies the base flow. Occasionally, human impacts are extreme enough to cause a stream to regress from perennial to seasonal in duration. Habitat Scale in Time and Space The design of erosion protection countermeasures typically focuses on the hydraulic and structural properties of a relatively short reach during higher flow periods when boundary shear stresses are at a maximum. Consideration of aquatic habitat, however, requires a much larger scale of reference both in time and space. Fish (and most other organisms specialized for life in river systems) are highly mobile creatures that live out their lives in a series of places (habitats) which can be separated by up to 100 km (60 miles) of river channel. Fisheries ecologists typically partition these habitats into basic types such as feeding habitats, resting habitats, spawning habitats, and nursery habitats. Each type has relatively distinct hydraulic and structural properties. Many fish populations are limited by the quantity or quality of one or more of these habitats, or by the lack of connectivity between key habitats. While it is unlikely that a fish population would ever be entirely dependent on the relatively small areas typically affected by the installation of a Figure 4.16. Stone toe provides stable benthic habitat and cover for smaller fish, but little diversity or pool habitat (McCullah and Gray 2005).

126 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures single erosion countermeasure, opportunities for increasing either the amounts or quality of fish habitat could potentially pay high dividends both for the environment and in terms of public appreciation and resource agency approvals. Large-scale channel restoration or stabilization work (directly affecting a reach longer than 20 channel widths) could have major effects on an aquatic community. The design of erosion protection countermeasures typically requires a focus on high-flow hydraulic conditions. This is entirely reasonable because erosion processes are usually limited to (or maximal at) high-energy events. Habitat design, however, must again take a wider view because fish live in their habitats at all times of the year and are affected by both high- and low-flow hydraulic environments. The habitat implications of a particular erosion control or habitat improvement technique can be quite different during high-flow and low-flow periods (see Figure 4.17). Therefore, environmentally sensitive design requires consideration of hydrau- lics and habitat requirements across the range of flows and seasons. Water Temperature An important first step toward environmentally sensitive design is to recognize that water temperature and site spatial context (with regard to both landscape and river network position) do more to shape the basic structure and productivity of fish communities than local physi- cal features. Because fish and their food organisms are cold-blooded, growth and productivity rates always vary strongly with temperature. Typically, both increase with rising temperature until a physiological maximum is exceeded. Thermal optima vary between species, but fisheries managers often make a basic distinction between sites and their fish communities based upon the thermal tolerances of the dominant species present. “Coldwater streams,” for example, are dominated by a few species such as trout and salmon that have rather low thermal optima. “Warmwater streams” are dominated by a larger number of species such as sunfish and suckers with higher thermal optima. In between these extremes, scientists recognize coolwater systems where members of both groups (and other fish with intermediate thermal preferences) may coexist. Water temperature in these transitional sites is an extremely important variable, since small changes in one direction or the other can cause tremendous alterations in the composition and productivity of the fish community. This variability in thermal setting directly affects the ecological value of more local habitat features like shade and groundwater inflows. Figure 4.17. Weirs can create pool-riffle habitats and help stabilize beds, but should be designed carefully in streams with low bed slope to prevent eliminating current during low flow (McCullah and Gray 2005).

Design Guidelines and Appraisal of Research Results 127 Geomorphic Context The landforms and hydrology of the upstream watershed directly affect the water temperature and the water and sediment flow regimes that govern habitat quality of a specific river reach. The details of linkages between climate, landscape position, fluvial geomorphology, and hydrology are very complex (see Section 4.2.2). Simply put, however, the flow regime interacts with the physical structure of the channel to produce physical aquatic habitat are characterized in terms of depth, velocity, substrate composition, etc. For example, coldwater streams tend to be steeper, found at higher elevations, and have coarser bed material (gravel, cobble and boulders) than warmwater streams. Habitat Issues and Opportunities High-Flow Issues. Both fish and invertebrates have normal ranges of velocity in which they are typically found, and threshold velocities above which they cannot survive. Habitat suitability models, for example, give standardized curves and rules for velocity and depth requirements based on published data. Local or state fisheries agencies may have developed regional hydraulic criteria for species of interest. Flow refugia are places fish and other organisms can go to escape excessive velocities. Natural current refugia can be either structural (e.g., boulders, undercuts, the downstream edge of point bars or abutments) or spatial (e.g., floodplains, high-water cross channels). The availability of refugia is obviously more or less critical, depending upon the gen- eral hydraulics in a reach during high-flow events. In high-energy channels, velocity refugia can be critical. From an environmental design perspective, the inclusion of elements providing velocity refuge can be beneficial to both fish and invertebrate populations. Boulder clusters, vegetated floodways, large woody debris, live cribwalls and live brush layering, spurs, and stone weirs etc. all may have potential application in this context (see Figure 4.18). However, these elements may not provide significant benefits in reaches where hydraulics are not an important constraint on the biological community, or where normal floodplain and geomorphic activity provides adequate refuge opportunities. Inundation of floodplains during high flows provides large areas of refuge, but also facilitates a host of other ecological functions, particularly along larger rivers where flooding durations are measured in days or weeks. For example, fish feed on terrestrial plants and animals (insects, earthworms, etc.) trapped by rising floodwaters. Receding floodwaters carry organic matter back into the riverine ecosystem for additional cycling and spiraling. In order to spawn, some species Figure 4.18. Spurs, barbs, and bendway weirs provide local zones of low velocity that are effective refuges (McCullah and Gray 2005).

128 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures of fish are dependent on the low-energy habitats found on flooded, forested floodplains. Isolation of floodplains from the stream by channel incision, channelization, levee installation, or other forms of structural flood control can be detrimental. Restoration of connections between the stream and temporarily flooded areas is often beneficial. Localized bank erosion is often identified as detrimental to fish habitat, usually because it is thought that the eroded sediments may fill pools or cover gravelly riffles downstream. Actual effects of local bank erosion vary based on the overall sediment balance within the watershed and reach. Bank erosion is not necessarily problematic from a habitat perspective; likewise, every bank stabilization project is not necessarily a habitat improvement measure. In some rivers, gravel and cobble riffles require eroded material from banks to replace downstream transport. In other reaches, flow energy is sufficient to transport sediments from eroding banks through the reach with little or no accumulation. Obviously, there are many river systems in which bank erosion can lead to habitat deteriora- tion. For example, when low-flow deposition (see below) is particularly problematic, spawning substrate (coarse bed material) is in critically short supply, or erosion and transport of sand banks contribute to a lack of deep pools and hydraulic diversity in the reach. In these settings, almost any form of bank stabilization might have habitat benefits, provided the design does not adversely affect other aspects of habitat quality. Channel Adjustment. Systemic erosion associated with lateral migration or other forms of channel adjustment occurs when there is a large-scale disequilibrium between water and sediment loads, channel shapes, and slope. Under these conditions, extensive bank erosion, bottom scour, sediment transport, and downstream bed aggradation usually occur, more or less simultaneously, across large sections of the channel network. Causes for large-scale/systemic erosion vary, but fre- quently involve natural climatic variations, human alterations of watershed hydrology, or changes in the channel base level or other downstream hydraulic controls. In such situations, high-flow erosion is evident, and biological communities are often heavily impacted. Streams in urbanizing watersheds frequently suffer from these effects, compounded by deteriorations in water quality and low-flow dewatering. Opportunities for habitat enhancement in these settings are minimal without addressing the underlying issues. From a habitat perspective, erosion control countermeasures are difficult to design, implement and justify in such a setting. Watershed management is typically the most effective scale for corrective measures. Low-Flow Issues. Design considerations for protecting habitat quality during low flows and mitigating high-flow erosion can be quite divergent. Environmentally sensitive erosion protection should include a careful review of habitat impacts during summer and fall when habitat quality is typically limiting the reproduction and growth of both adults and juvenile fish. Low-flow magnitudes vary widely from site to site, adding further complexity. During periods of base flow, stream water velocities usually do not produce biologically detrimental levels of shear and drag forces. However, many riverine organisms depend upon the transport of food to their relatively stationary feeding habitats. For example, in both filter- feeding insects and drift-feeding fish, rate of food capture increases (up to a limiting maximum) with increasing flow velocity. Furthermore, the relatively slow rate of molecular diffusion in water frequently limits physiological uptake of essential inputs like oxygen, and dissolved nutrients (for plants). As a result of these mechanisms, excessively low velocities (e.g., behind a weir) can be as detrimental to aquatic habitat quality as excessively high velocities. Erosion protection countermeasure design for low-flow habitat quality includes emphasizing techniques that do not increase low-flow roughness elements or decrease low-flow hydraulic radius. Two-stage channel designs can be a useful way to mimic natural channel configurations

Design Guidelines and Appraisal of Research Results 129 by providing a relatively high radius compact channel for low flow, and a larger but lower radius cross section during high flows (as observed naturally occurring in valley floodplains). Fish populations generally suffer high rates of mortality throughout their life span; visual pre- dators, typically birds, other fish, and humans, inflict the bulk of this mortality. It is not surprising then that most species of fish show a strong attraction to structural elements that obscure the vision of potential predators and provide a complex hunting stage that might favor escape by the victim. Fisheries biologists use the term “cover” to very broadly refer to any physical structure that might provide such refuge, including pools, undercut banks, submersed living vegetation, and woody debris. During low-flow periods, much of the cover near the margins of the streambed may be exposed and unusable. Likewise, pools may become too shallow to provide cover at low flow. Design features that address needs for increased low-flow cover include using instream structures that promote pool formation (e.g., boulder clusters, spurs, stone weirs, etc.) or constructing chan- nels with low-flow channels that contain some deep pools. Installing measures that increase pool cover in slowly moving streams must be weighed against potential reductions in flow velocity. Bed Composition. Stream bed material is an important component of physical aquatic habitat. Key aspects of bed material include its size distribution, how frequently the particles move, and how open the interstitial spaces within the particle matrix are. In general, the diversity and abun- dance of aquatic insects are lowest for frequently shifting sand beds and highest for cobble and gravel beds that have a wide gradation (see Figure 4.19). When finer sediments deposit within gravel or cobble matrices, water circulation and oxygen supply to the areas beneath the surface of the bed are impeded or eliminated. These habitats are extremely important for incubating eggs of fish, like salmon, that spawn in gravel and for many species of insects (benthic macro- invertebrates). Sandy beds typically support lower densities of all but the smallest invertebrates, and stable objects such as woody debris, clay outcrops, or stone may be heavily colonized by invertebrates in sand-bed streams. Summary Designing erosion protection countermeasures in an environmentally sensitive way means that design decisions are made in light of the larger ecological context of the site. Here, “ecological Figure 4.19. Stable gravelly riffles that are free of fine material typically support rich communities of aquatic invertebrates.

130 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures context” means the sum of the physical constraints and the relevant biological constraints. When proposing hydraulic, structural, or environmentally sensitive modifications either to abate erosion or improve biological habitat, solutions should be chosen in light of the specific hydraulic and ecological processes relevant at the site of interest. In other words, designs should be ecologically tailored to the site. Some basic habitat improvement issues and opportunities are described above and identified by the Greenbank Decision Support Tool in NCHRP Report 544 (McCullah and Gray 2005) (see also Section 2.1, above); however, most designers will find consultation with regional fisheries management agencies a useful step in gathering information about ecological context, and in identifying which habitat management goals might be consistent with the erosion protection goals of a particular project and at the same time, be ecologically appropriate for the site. 4.2.7 FHWA Perspective and Guidance Overview The FHWA’s current guidance and perspective on the use of stream bank erosion protection treatments in the vicinity of highway infrastructure is contained in HEC-23, Volume 1, Chapter 6 (Lagasse et al. 2009). The following paragraphs and subsections provide extracts of this guidance as they relate to application of environmentally sensitive treatments to protect bridge waterway crossings, in particular, and highway infrastructure, in general. FHWA notes that there are several synonymous terms that describe the field of vegetative stream bank stabilization and countermeasures. Terms for the use of “soft” revetments (consisting solely of living plant materials or plant products) include bioengineering, soil bioengineering, ground bioengineering, and ecological bioengineering. Terms describing the techniques that combine the use of vegetation with structural (hard) elements include biotechnical engineering, biotechnical slope protection, bioengineered slope stabilization, and biotechnical revetment. The terms soil bioengineering and biotechnical engineering are most commonly used to describe stream bank erosion countermeasures and bank stabilization methods that incorporate vegetation. Where riprap constitutes the “hard” component of biotechnical slope protection, the term vegetated riprap is also used. The FHWA Perspective Based on a 1998 scanning review of European practice for bridge scour and stream instability problems, it was observed that most hydraulic engineers in Europe would not recommend the reliance on bioengineering countermeasures as the only countermeasure technique when there is a risk of damage to property or a structure, or where there is potential for loss of life if the counter- measure fails (TRB 1999). Soil bioengineering is not suitable where flow velocities exceed the strength of the bank material or where pore-water pressure causes failures in the lower bank. In contrast, biotechnical engineering is particularly suitable where some sort of engineered struc- tural solution is required because the risk associated with using just vegetation is considered too high. Continuous and resistive bank-protection measures, such as riprap and longitudinal rock toes are primarily used to armor outer bends or areas with impinging flows. Since stream bank protection designs that consist of riprap, concrete, or other inert structures alone may be unacceptable for lack of environmental and aesthetic benefits, there is increasing interest in designs that combine vegetation with inert materials into living systems that can reduce erosion while providing environmental and aesthetic benefits (Sotir and Nunnally 1995). For example, the concerns over the poor aquatic habitat value of riprap, both locally and cumu- latively, have made the use of riprap alone controversial in some jurisdictions (Washington

Design Guidelines and Appraisal of Research Results 131 Department of Fish and Wildlife 2003). In general, any negative environmental consequences of riprap can be reduced by minimizing the height of the rock revetment up the bank and/or including biotechnical methods, such as vegetated riprap with brush layering and pole planting, vegetated riprap with soil, grass, and ground cover, vegetated riprap with willow (Salix spp.) bundles, and vegetated riprap with bent poles (see Section 4.3.3). Combining riprap with deep vegetative planting (e.g., brush layering and pole planting) is also appropriate for banks with geotechnical problems, because additional tensile strength is often contributed by roots, stems, and branches. In contrast, trees and riparian vegetation planted only on top of the bank can sometimes have a negative impact (Simon and Collison 2002). Correctly designed and installed, vegetated riprap offers an opportunity for the designer to attain the immediate and long-term protection afforded by riprap with the habitat benefits inherent with the establishment of a healthy riparian buffer. The riprap will resist the hydraulic forces, while roots and branches increase geotechnical stability, prevent soil loss (or piping) from behind the structures, and increase pullout resistance. Above-ground components of the plants will create habitat for both aquatic and terrestrial wildlife, provide shade (reducing thermal pollution), and improve aesthetic and recreational opportunities (see Section 4.2.6). The roots, stems, and shoots will help anchor the rocks and resist “plucking” and gouging by ice and debris (McCullah and Gray 2005). Advantages and Limitations of Biotechnical Engineering Specific ways vegetation can protect stream banks as part of a biotechnical engineering approach include: • The root system binds soil particles together and increases the overall stability and shear strength of the bank. • The exposed vegetation increases surface roughness and reduces local flow velocities close to the bank, which reduces the transport capacity and shear stress near the bank, thereby inducing sediment deposition. • Vegetation dissipates the kinetic energy of falling raindrops, and depletes soil water by uptake and transpiration. • Vegetation reduces surface runoff through increased retention of water on the surface and increases groundwater recharge. • Vegetation deflects high-velocity flow away from the bank and acts as a buffer against the abrasive effect of transported material. • Vegetation improves the conditions for fisheries and wildlife and helps improve water quality. In addition, biotechnical engineering is often less expensive than most methods that are entirely structural and it is often less expensive to construct and maintain when considered over the long term. The critical threats to the successful performance of biotechnical engineering projects are improper site assessment, design or installation, and lack of monitoring and maintenance (especially following floods and during droughts). Some of the specific limitations to the use of vegetation for stream bank erosion control in the vicinity of highway infrastructure include: • Lack of design criteria and knowledge about properties of vegetative materials, • Lack of long-term quantitative monitoring and performance assessment, • Difficulty in obtaining consistent performance from countermeasures relying on live materials, • Possible failure to grow and susceptibility to drought conditions, • Depredation by wildlife or livestock, and • Significant maintenance requirements.

132 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures More importantly, the type of plants that can survive at various submersions during the normal cycle of low-, medium-, and high-stream flows is critical to the design, implementation, and success of biotechnical engineering techniques. In addition, the combination of riprap and vegetation may be inappropriate if flow capacity is an issue, since bank vegetation can reduce flow capacity, especially when in full leaf along a narrow channel (see management of conveyance discussion in Section 4.2.3). Design Considerations for Biotechnical Countermeasures In an unstable watershed, careful study should be made of the causes of instability before biotechnical countermeasures are contemplated (see HEC-20) (Lagasse et al. 2012). Since bank erosion is tied to channel stability, a stable channel bed must be achieved before the banks are addressed. Scour and erosion of the bank toe produce the dominant failure modes and con- sequently, most biotechnical engineering projects documented in the literature contain some form of structural (hard) toe stabilization, such as rock riprap (Figure 4.20), rock gabions, cribs, cable-anchored logs, or logs with rootwads anchored by boulders (Figure 4.21). Note the use of a fascine bundle in Figure 4.20 as part of the rock toe protection. Toe protection should be keyed into the channel bed sufficiently deep to withstand significant scour, and the biotechnically engineered revetment should be keyed into the bank at both the upstream and downstream ends (called refusals) to prevent flanking (see discussion of “The Key” in Section 4.2.3). Deflectors such as fences, dikes, and pilings may also be utilized to deflect flow away from the bankline. Figure 4.20. Details of brush mattress technique with stone-toe protection (FISRWG 2001).

Design Guidelines and Appraisal of Research Results 133 Other factors that need to be considered when selecting a design option include climate and hydrology, soils, cross-sectional dimensions (is there sufficient room for the countermeasure?), flow depth, flow velocity (both magnitude and direction), and slope of the bankline being pro- tected. Most methods of biotechnical engineering will require some amount of bank regrading. Because structure design is based on flood velocities and depths, one or more design flows will need to be analyzed. Of particular interest is the bankfull or overtopping event, since this event generates the greatest velocities and tractive forces. Local (at or near the project site) flow velocities should be used for the design, especially along the outside of bends. The erosion protection should extend far enough downstream, particularly on the outer banks of bends. The highest velocities generally occur at the downstream arc of a bend and on the outer bank of the exit reach imme- diately downstream. As noted, the countermeasures should be tied into the bank at both ends to prevent flanking. Summary Biotechnical engineering can be a useful and cost-effective tool in controlling bank or channel erosion, while increasing the aesthetics and habitat diversity of the site. However, vegetation alone should not be seriously considered as a countermeasure against severe bank erosion where a highway facility is at risk. At such locations, vegetation can best serve to supplement other countermeasures. Where failure of the countermeasure could lead to failure of a bridge or high- way structure, the only acceptable solution in the immediate vicinity of a structure is a traditional, Figure 4.21. Details of root wad and boulder revetment technique (FISRWG 2001).

134 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures “hard” engineering approach. Biotechnical countermeasures need to be applied in a prudent manner, in conjunction with channel planform and bed stability analysis and rigorous engineering design. Designs must account for a multitude of factors associated with the geotechnical charac- teristics of the site, the local and watershed geomorphology, local soils, plant biology, hydrology, and site hydraulics. Finally, programs for monitoring and maintenance, which are essential to the success and effectiveness of any biotechnical engineering project, must be included in the project and adhered to strictly. 4.2.8 Engineering Liability Issues The FHWA perspective on the application of biotechnical engineering treatments summa- rized in the preceding section brings out, by inference, a number of issues relevant to potential professional liability concerns when the treatment must be certified as having been “engineered.” As noted by FHWA (Lagasse et al. 2009) certifying design and performance of biotechnical treat- ments becomes problematic for the PE on the design team primarily in relation to the vegetative component of the treatment. While design and performance standards are clearly established for the “hard” component of a treatment (e.g., riprap), the same cannot be said for the vegetative compo- nent. Here, unknowns and conditions generally out of the control of the engineer exist, including (1) lack of design criteria and knowledge of the “engineering” properties of vegetative materials, (2) lack of long-term performance criteria, (3) difficulty in obtaining consistent performance from the “live” material component of a treatment, and (4) introduction of factors beyond the control of the engineer (e.g., drought and/or depredation by wildlife or livestock). Extracts from a Journal of Hydraulic Engineering Forum paper prepared by members of the ASCE Environmental and Water Resources Institute (EWRI) River Restoration Committee (Slate et al. 2005) highlight the issues that confront the engineer when engaged in river restoration design and the design and implementation of environmentally sensitive stream bank protection measures. The fundamental issue is that by sealing river restoration or environmentally sensitive designs, engineers assume the burden of professional liability for those designs. In many cases, stream restoration or biotechnical engineering projects require the seal of a licensed PE. Upon affixing his or her seal to design documents and drawings, the engineer assumes the responsibility for the accuracy of the design and affirms that the work was directly conducted or overseen by him or her. Professional ethics dictate that the work be within the engineer’s area of expertise, that the engineer has kept abreast of the state of the practice through continuing education, and that a reasonable standard of care has been exercised in developing the project design. In many areas of civil engineering, the design standards are such that following those standards will ensure that these criteria are met. However, as enumerated above, for many environmentally sensitive treatments the factors necessary to ensure successful projects are less clear. It should be noted that engineers employed by government agencies typically are shielded from professional liability of this type; however, engineers serving as consultants to an agency are not. Overall, the lack of rigorous engineering standards for critical elements of a restoration project produces difficulty for designers in (Slate et al. 2005): • Identifying an appropriate design procedure and choosing which techniques are most suitable for given conditions. • Effectively communicating with stakeholders on the suitability of a particular design procedure. • Ascertaining the level of documentation necessary to convey design analysis into plans to ensure successful project implementation. • Identifying measurable performance standards that can be monitored and assessed, thereby supporting an adaptive management approach to advance design methodologies. • Managing risk and liability.

Design Guidelines and Appraisal of Research Results 135 Slate et al. (2005) note that if engineering is required to meet project objectives, ideally the designer or project engineer works with other professionals throughout the design process in crafting a solution. Engineers may receive input from fluvial geomorphologists, geologists, fish- eries biologists, ecologists, or other professionals. The engineer then converts that input into a solution (reports, drawings, specifications) affixed with seal and signature, which is presented with supporting information that clearly defines design criteria, risks, and measurable performance standards. While affixing of an engineering seal to a design does not guarantee “success” of a project, the seal does indicate that the engineer has exercised his or her best professional judgment upholding the industry “standard of care” in the design process. Consequently, the EWRI River Restoration Committee points out that there is a need for members of the design team to appreciate and understand the roles and responsibilities of each discipline. While some disciplines fall squarely into the realm of traditional engineer- ing practice, such as hydraulics and geotechnical engineering, others are based in physical, biological, or social sciences. Still others could be considered an indistinguishable blend of engineering and science, such as river mechanics and bioengineering. Clearly, there is a need for a suite of design approaches that combine engineering, geomorphology, hydraulics, and biology. The Committee strongly advocates the development of performance-based design guidance that incorporates a broad base of disciplines (hydrology, hydraulics, geomorphology, river mechanics, sediment transport, biology and ecology), and thus open up more options for channel design. The Committee concludes that there is a need for objective, performance-based guidelines or a manual of practice for river and stream restoration design as well as improved channel design standards. The scope and objectives of NCHRP Project 24-39 did not include developing a broadly-based manual of engineering practice for environmentally sensitive stream bank protection treatments. However, through a compilation of design practices observed during field site visits to a wide range of restoration projects (Task 6) and by subjecting selected environmentally sensitive treat- ments to rigorous hydraulic engineering testing under prototype-scale laboratory conditions (Task 7), the design guidelines developed under this project together with the guidance devel- oped under NCHRP Project 24-19 (McCullah and Gray 2005) represent advances in the state of practice in biotechnical engineering. The application examples of Section 4.4 demonstrate the successful integration of hydraulic engineering practice with a multidisciplinary design team in developing and implementing environmentally sensitive bank-protection designs responsive to the goals of stakeholders under a range of climatic and geomorphic conditions. 4.3 Guidelines for Specific Treatments 4.3.1 Live Siltation and Live Staking with Rock Toe Introduction As described in Chapter 3, prototype-scale laboratory testing of specific environmentally sensitive treatment configurations was a major component of this research. Tests focused on two representative biotechnical measures that were constructed with real plants in large planter boxes (6 ft wide by 20 ft long by either 12 or 18 in. deep). Plant materials were given time to establish prior to installation on the banks of an experimental trapezoidal channel. The various treatments presented in NCHRP Report 544 (McCullah and Gray 2005) (see Appendix A) were carefully evaluated from the perspective of having wide applicability across the nation, as well as practical issues (for testing) of: • Constructability, • Physical testing requirements/constraints,

136 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures • Quantitative measurements of key hydraulic variables, and • Condition monitoring of each component before testing and after each test flow event. The two biotechnical bank-protection treatments tested were (see Figures 3.3 and 3.4): 1. Live siltation with live staking and rock toe at a 3H:1V slope, and 2. VMSE (sometimes referred to as FES lifts) at a 2H:1V slope. The first treatment (live siltation with live staking and a rock toe) is referred to as “Tray 1” in Chapter 3. For the vegetative components of this treatment NCHRP Report 544 lists the following research opportunities (see Appendix A): Live Siltation—Research into velocities that this technique can withstand would be helpful. Live Staking—Studies regarding the effect live staking has on increasing the ability of other measures to withstand higher velocities and shear stresses would be valuable. Suggested design guidelines for this treatment are consolidated in this section. First, the design guidelines suggested for live siltation and live staking are summarized. Next, current guidance for the riprap toe of this treatment is referenced and updated. Finally, the results of laboratory testing of this treatment (see Chapter 3) as they address the research opportunities referenced above are summarized as an update to commonly used guidelines, specifically in relation to hydraulic issues of permissible velocities and shear stresses for this biotechnical configuration. Live Siltation Purpose and Advantages. Live siltation is a revegetation technique used to secure the toe of a stream bank, trap sediments, and create fish habitat. The system is normally constructed at the water’s edge. Its primary purposes are to help secure the toe of a stream bank and trap sediments. Live siltation is an appropriate practice along an outer bend with significant scour if toe protection is provided. Live siltation is a very effective and simple conservation method using local plant materials. This technique is particularly valuable for providing immediate cover and fish habitat while other revegetation plantings become established. The protruding branches provide roughness, slow velocities, and encourage deposition of sediment. The depositional areas are then available for natural recruitment of native riparian vegetation. Design. A typical design drawing for live siltation is shown in Figure 4.22. Cuttings should be placed adjacent to the water’s edge to ensure effective sediment trapping and velocity reduction at the toe of the slope. At least 12 branches per foot (40 branches per m) should be installed. This technique may be used for velocities up to 6.6 ft/sec (2 m/sec), but velocities should be at least 0.8 ft/sec (0.25 m/sec) for the system to function properly. For additional data on permissible velocities and shear stresses see Chapter 3 and the laboratory testing summary below. Materials and Equipment. Stone (generally riprap), willow wattles, logs or rootwad revetments are needed for toe and scour protection. The live siltation will require live branches of shrub willows 3.5 to 5 ft (1 to 1.5 m) in length. The branches should be dormant, and need to have the side branches still attached. Any woody plant material, such as alder, can be installed for a nonliving system. Construction and Installation. Construct a V-shaped trench at the AHW level, with hand tools or a backhoe. Excavate a trench so that it parallels the toe of the stream bank and is approxi- mately 2 ft (0.6 m) deep. Lay a thick layer of willow branches in the trench so that one-third of the length of the branches is above the trench and the branches angle out toward the stream.

Design Guidelines and Appraisal of Research Results 137 Figure 4.22. Live siltation typical drawing (McCullah and Gray 2005). Place a minimum of 12 branches per foot (40 willow branches per m) in the trench. Backfill over the branches with a gravel/soil mix. Both the upstream and downstream ends of the live siltation construction need to transition smoothly into a stable stream bank to reduce the potential for the system to be flanked. More than one row of live siltation can be installed running parallel to the channel. A living and growing siltation system typically is installed at AHW. If it is impossible to dig a trench, the branches can be secured in place with logs, armor rock, bundles made from wattles, or coir logs. Figure 4.23 illustrates the construction and installation process. This project included live siltation used in conjunction with TRM on Alamitos Creek in Santa Clara County, CA. The live siltation was installed in October 2003. This site was included as a Task 6 site visit under NCHRP Project 24-39 (see Figure 2.5). For additional information see Site CA4 in Tables 2.10 and 2.11 and the Compendium available with this report.

138 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Cost. An estimated level of effort for installing live siltation ranges from 0.2 to 0.6 work hours per linear foot (0.7 to 2 work hours per linear m), plus willow stock if not readily available on site. For the Alamitos Creek site illustrated in Figure 4.23, the coir netting was readily available. The Alamitos Creek project had value added because it was constructed as part of a training workshop for Santa Clara Valley Water construction crews. Maintenance and Monitoring. During the first year, a live siltation installation should be checked for failures after all overtopping flows, and repaired as necessary. During dry periods of the growing season of the first year, ensure that cuttings are not becoming dehydrated. Additional infor- mation on performance of a live siltation installation can be found in this project’s Compendium for the Guadalupe River near San Jose, CA (Site CA5), and the Russian River near Geyserville, CA (Site CA6) (see Table 2.10 and Section 4.5.3). Section 4.2.5 provides additional guidance on moni- toring the success of the vegetative component(s) of environmentally sensitive treatments. Common Reasons for Failure. Cuttings will not promote siltation if not located at the water’s edge. If located further up the bank, cuttings may dry out, and will only trap sediments Figure 4.23. Construction and installation of live siltation on Alamitos Creek, CA (McCullah and Gray 2005).

Design Guidelines and Appraisal of Research Results 139 and slow velocities during high flows. Cuttings may not grow well if not handled properly prior to installation. Live Staking Purpose and Advantages. Live stakes are pieces of freshly cut woody plant stem planted in the ground or into erosion control or stream bank stabilization structures. The branches vary from about 20 to 39 in. (50 to 100 cm) long, and typically ¾ to 3 in. (20 to 75 mm) in diameter. Live stakes are planted with the terminal buds or leaf nodes pointing up and the basal ends down into the soil. The buried portion of the cuttings develop roots, while the exposed por- tion produces branches and leaves. Depending on the species, the cuttings can grow into shrubs and/or trees. Because of its ability to root from cuttings, the preferred plant species for live staking is willow (salix ssp.), but cottonwood (Poplar ssp.), dogwood (Cornus ssp.), elderberry (Sambucus ssp.), coyote brush (Baccharis ssp.), and others have been used successfully. The concept behind live stake planting is that the live, vegetative cuttings are placed into the ground to allow the stakes to root and grow (see Figure 4.30). Even if the branches do not grow the stakes can provide, at least temporarily, reinforcement much like a wooden stake or steel rebar stake. Live stakes generally accomplish several purposes concurrently: 1. The stakes grow vegetatively thereby providing cover and erosion control. 2. The vegetative cover can provide improved aesthetics. 3. The cover provides shade and canopy cover where thermal pollution may be a concern. 4. The leaves, branches, and insects living on them can provide carbon and nutrient cycling and are an important food source for aquatic organisms. 5. The roots and branches provide for and improve geotechnical and soil stability. Using a sys- tem of live stakes creates a root mat that stabilizes the soil by reinforcing and binding soil particles together. Roots can also aid stabilization by extracting excess soil moisture and by binding fill soils to existing native soils. 6. Live stakes used as slope nails can stabilize slumps and slides through the mechanisms of “buttressing and arching.” 7. Leafy and brushy top growth benefits the stream bank by increasing roughness, thereby reducing boundary shear stress underneath the canopy. 8. Live staking can, especially when used in conjunction with biodegradable erosion control materials, enhance conditions for colonization of native species. Live staking has been successfully used in many different climatic, soil moisture regimes and elevations. There is a wide range of possible uses for and benefits of live staking, but the primary uses generally involve revegetation, anchoring, enhancing geotechnical strength (shear strength), or reducing erosion through increased cover (raindrop impact) and hydraulic roughness (reduced boundary shear). The practice is commonly used in combination with other treatments to pro- vide more stable site conditions and a more environmentally sensitive design. Among the advantages of live staking are: • Stake sources are plentiful and inexpensive. • Installation is rapid and inexpensive. • Stakes can be planted with minimal surface preparation or disturbance and can be placed into irregular (but stable) slope surfaces. • Stakes can be planted into already existing structures. • Stakes root rapidly and help to reduce slope soil moisture soon after installation. • Provides both environmental and aesthetic benefits. The direct environmental benefits are cover and shade, carbon and insects for basic food web and nutrient cycling, and the ability to provide stable areas for native plant re-establishment

140 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures (successional reclamation). Live stakes establish a root mat that stabilizes the soil. Stake estab- lishment can improve aesthetics and provide wildlife habitat. The stems, branches, and leaves slow high flows and provide shade, habitat, food, and shelter for stream corridor biota. When live stakes are placed on upper banks, they provide habitat, food, and shelter for terrestrial fauna as well. As a temporary, immediate measure, live staking performs an important function of stabilizing and modifying the soil, serving as a pioneer species until other plants become established. Stakes can play an important geotechnical function by providing apparent cohesion and soil reinforcement. Design. One of the most important design considerations for vegetative success is deter- mining whether the chosen species (willow, cottonwood, dogwood, etc.) is naturally occurring in the region and/or will allow successional reclamation—natural succession from willow shrubs to tall riparian overstory. Live stakes are useful for the following situations: 1. Live staking is useful as a revegetation technique and for establishing riparian plants in high- flow or droughty situations. 2. Live staking can be used in irrigated or non-irrigated conditions with the latter being more prevalent. Irrigation can greatly increase vegetative success. Most often live staking is installed during the dormant season or when climatic or soil moisture conditions are favorable for establishment in non-irrigated conditions. 3. Live staking provides an environmentally sensitive anchoring technique for geotextiles and erosion control materials. The anchoring can be temporary or permanent depending on whether the stakes take root. 4. Live staking adds immediate failure resistance to the soil mass. While providing geotechnical benefits by “buttressing and arching,” deep-seated failure planes underneath the bottom end of the cuttings will not usually be affected by live staking. These plants can remove excess soil moisture via evapotranspiration during the growing cycle; however, these benefits will not be realized during dormancy. The stakes should be harvested from relatively straight, disease- and insect-free branches. Stakes shall be ¾ to 3 in. (20 to 75 mm) in diameter and a minimum of 18 in. (0.5 m) long. The upper end of the stake should be cut square and the basal end of the stake should be cut at an angle. Preparing the stakes in this manner will aide insertion into the soil and also ensure the stakes are oriented correctly when installed (see Figures 4.20 and 4.30). Generally the deeper the branch is inserted into the soil, the better the chance of vegetative success and the greater the soil stabilization benefits, therefore the “80% Rule” should be strictly followed—at a minimum 80% of the branch should be placed in the soil and with 20% protruding above. For instance, an 18 in. (46 cm) stake will be installed at a minimum of 14.5 in. (37 cm) into the soil, and a 30 in. (91 cm) stake should be installed 24 in. (61 cm) in the soil. Deeper planting also reduces the chance that stakes can be pulled out by beavers, deer, or other wildlife. As shown in Figure 4.30 stakes should be inserted on the bank slope at a density of 1 to 3 ft (30 to 90 cm) apart. Ideally, the stakes should not be planted in rows or at regular intervals, but at random in the most suitable places at a rate of 2 to 5 cuttings/10 ft2 (2 to 5 cuttings/m2). Allowable shear stress for this technique is approximately 2.5 lb/ft2 (120 N/m2) (Schiechtl and Stern 1996), and allowable velocity is about 3 ft/s (0.9 m/s) (Gray and Sotir 1996). For additional data on permissible velocities and shear stresses see Chapter 3 and the laboratory testing sum- mary, below. Materials and Equipment. Live stakes are typically made of woody riparian plant stems, although fleshy plant stems can have some success as well. Willow, cottonwood, and dogwood are the most used woody plants; however, willow cuttings make the best material for live stakes.

Design Guidelines and Appraisal of Research Results 141 Willow species choice is highly dependent on locale; the best species for a given site are those found growing near the site. Stakes are typically harvested and planted when the willows, or other chosen species, are dormant, although the cuttings can do well other times of year when soil moisture is available. When harvesting cuttings, select healthy, live wood that is reasonably straight, and at least 2 years old. Make clean cuts without splitting ends. Trim branches from cutting as closely as possible. Cuttings should generally be ¾ in. (19 mm) in diameter and 18 in. (46 cm) long, or larger depending on the species. The butt end of the cutting should be pointed or angled and the top end should be cut square to help identify the top and bottom when planting. The top, square end can be painted and sealed by dipping the top 1 to 2 in. (2.5 to 5 cm) into a 50:50 mix of light colored latex paint and water. Sealing the top of the stake will reduce desiccation, ensure the stakes are planted with the top up, and make the stakes more visible for subsequent planting evaluations. Stakes must not be allowed to dry out. All cuttings should be soaked in water for 5 to 7 days (a minimum of 24 hours) and planted the same day they are removed from water. Construction and Installation. It is important not to damage the stakes during installation. Damaged and split stakes have increased incidence of dehydration, decay, and introduction of disease. Most compacted soils or soils with rocks and gravel will require the use of a “pilot bar” to make a hole prior to driving the stake. The use of a polyurethane hammer or rubber mallet will reduce splitting damage to the stake. Using a high-powered water jet to pilot the holes is also favorable as the holes are left well hydrated. The successful implementation of this technique requires care and consideration of the stakes during harvest, storage, transport, and installation. The stakes should never be allowed to dry out and should be moist and covered at all times. Soaking the stakes will increase success. The basal ends should be planted into the ground, with the leaf bud scars or emerging buds always pointing up. Care should be taken not to damage the buds, strip the bark, or split the stake during installation. As noted above the stake should be set as deep as possible into the soil, with 80% of its length into the soil (see Figure 4.30). Deep planting will increase the chances of survival. The stake should never protrude more than one-quarter of its length above the ground level to prevent it from drying. The excess stake or any damaged or split ends can be cut off after installation. At least two buds and/or bud scars should remain above the ground after planting. Add soil to the planting hole if necessary to ensure solid contact with the stem. It is important to tamp the soil around the cutting to insure good soil-stem contact. The best installations, espe- cially on droughty sites, will include “watering in” and slightly compacting the backfill or hole. Watering in, much like transplanting a container plant, can successfully be accomplished by pouring one to two gallons of water into the soil around the stake and planting hole, then slightly tamping or otherwise jarring the soil. This procedure will ensure intimate soil to stem contact. Cost. Costs range from $1.50 to $3.00 per stake (circa 2005), including harvesting, trans- portation, storage, and installation. Costs may be higher if labor costs are especially high or the harvesting location is a long way from the project site. Estimated labor allocations are 20 to 50 ft2 (2 to 5 m2) of live staking per work hour or approximately 10 to 25 stakes per hour. These estimates include all preparatory work. Maintenance and Monitoring. Without temporary irrigation, stakes have the highest sur- vival rate when installed during the dormant season, which may not coincide with the best time for construction of the rest of the project. Stakes do not become fully effective until one grow- ing season after installation, and thus provide limited immediate and areal stabilization unless combined with other practices. Stakes should be inspected every few weeks until well established, and irrigation, browse control (from livestock, deer, beavers, etc.), pruning, weed control, and fertilization should be implemented as needed.

142 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures For additional information on performance of live staking installations, see Table 2.10 which lists six sites inspected under Task 6 where live staking was part of the environmentally sensitive treatment implemented. Detailed information on these sites can be found in the Compendium. For example, live staking was included at the Malletts Creek site near Ann Arbor, MI (site M15, see Figure 2.4) and performance of the live staking component is shown in Figure 4.48g, h, and j. Section 4.2.5 provides additional guidance on monitoring the success of the vegetative component(s) of environmentally sensitive treatments. Common Reasons for Failure. Live staking can fail if vegetation is not handled properly prior to installation, is installed incorrectly (less than 80% of the cutting in the ground, bud scars facing down, poor soil contact, etc.), or not irrigated or watered in when installed in arid areas. Rock Toe Purpose and Advantages. The “engineered” component of the Tray 1 treatment (see Chapter 3) consisted of a rock riprap toe. This component of a biotechnical treatment is also known as lon- gitudinal peaked stone toe protection (LPSTP), stone toe, rock toe, stone-toe buttress, weighted riprap toe, and longitudinal fill stone-toe protection (LFSTP). When implanted with vegetation the rock toe is also referred to as “vegetated riprap” (see overview discussion in Section 4.3.3 and Figures 4.27 through 4.30). Longitudinal stone toe has proven cost effective in protecting lower banks and creating con- ditions leading to stabilization and revegetation of steep, caving banks (for example, see Shields et al. 1995). Stone toe is continuous bank protection consisting of riprap placed longitudinally at, or slightly streamward of, the toe of an eroding bank. The cross section of the stone toe is triangular in shape. The success of this method depends, in part, upon the ability of stone to self-adjust or “launch” into any scour holes formed on the stream side of the revetment. The stone toe does not need to follow the bank toe exactly, but should be designed and placed to form an improved or “smoothed” alignment (e.g., through the stream bend). The “smoothed” longitudinal alignment results in improved flow (less turbulence) near the toe of the eroding bank. This continuous bank-protection technique protects the toe from erosion. It is especially effective in streams where most erosion is due to relatively small but frequent events. It protects the toe so that slope failure of a steep bank landward of the stone toe will produce a stable angle. Such a bank is often rapidly colonized by natural vegetation (Figure 4.24). Figure 4.24. Revegetation of eroding bank landward of stone toe (photo by J. McCullah).

Design Guidelines and Appraisal of Research Results 143 Longitudinal stone toes are well suited for many situations where relatively low-cost, continuous bank protection is needed, and is particularly applicable for ephemeral, narrow, and small- to medium-sized streams. A stone toe is also well suited for areas where the toe is experiencing erosion but the mid and upper banks are fairly stable due to vegetation, cohesive soils, infrequent short-duration inundation, or relatively slow velocities. A longitudinal stone toe can be applied in some situations where the bankline needs to be built back out into the stream, where the existing stream channel needs to be realigned, where the outer-bank alignment makes abrupt changes (scallops, coves, or elbows), or where the stream is not otherwise smoothly aligned. Bank grading, reshaping, or sloping is usually not needed (existing bank and overbank vegetation need not be disturbed or cleared). Longitudinal stone toe is very cost effective and is relatively easy to construct. It is simple to design and specify and is a thoroughly tested method that has been used in a variety of situations and has been extensively monitored. Another advantage is that it is easily combined with other bank stability techniques that provide superior habitat compared to pure riprap. A longitudinal stone toe has documented environmental benefits, especially for aquatic habitat (see Lister et al. 1995, Dardeau et al. 1995, and Shields et al. 1995). Stone inter- stices provide cover and habitat for smaller fish and other organisms, and rocky surfaces provide stable substrate for benthic invertebrates (see Section 4.2.6 and Figure 4.16). Vegetative cover can become established, even growing through the rock, and can provide canopy and a source of woody debris. Design. Longitudinal stone toe can be specified by weight per unit length or to a specific crest elevation. A specific crest elevation may be specified when the bed of the stream is uneven or deep scour holes are evident. Maynord (1994) presents a design procedure for “launchable riprap” that may apply to stone toe design in situations where scour on the water side of the toe is possible. Longitudinal stone toe must be keyed deeply into the bank at both the upstream and downstream ends and at regular intervals along its entire length. On small streams, 75 to 100 ft (25 to 30 m) spacing between keys (tie-backs) is typical, while on larger streams and smaller rivers, one or two multiples of the channel width can be used as a spacing guide. Excavation of trenches for keys provides a good opportunity for deep planting willow (Salix ssp) posts or poles (Figure 4.25). Figure 4.25. Construction of keys provides an opportunity for deep planting willow poles (McCullah and Gray 2005).

144 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures The key trenches at the upstream and downstream ends should be excavated into the bank at an angle of approximately 30 degrees with the primary flow direction and of sufficient length that flows will not be able to get around them during the design storm. A gentle angle is important for the end keyways, often referred to as “refusals,” because it allows for smooth flow transitions coming into and flowing out of the treated reach. Tie-backs or “refusals” oriented at 90° to the bank have resulted in many failures at the downstream end of the structure, due to flow expansion at that point (see discussion of “The Key” in Section 4.2.3 and Figure 4.5). While “launchable riprap” as described by Maynord (1994) may offer ease of installation and reduce construction costs, it does not permit the use of a granular or geotextile filter under the stone toe. As noted in NCHRP Report 568 (Lagasse et al. 2006) and FHWA’s HEC-23 (Lagasse et al. 2009), the importance of the filter component of a riprap installation should not be under- estimated. Filters (either granular or geotextile) contribute to the long-term success of riprap (including a stone toe), particularly if significant toe scour is anticipated (see “The Key” discus- sion in Section 4.2.3). For further guidance on riprap design, reference to the publications cited above is suggested. NCHRP Report 568: Riprap Design Criteria, Recommended Specifications, and Quality Control (Lagasse et al. 2006) provides design guidance for sizing the rock for dumped riprap used for bank protection. That NCHRP study evaluated numerous procedures for sizing revetment rip- rap and suggests using the method developed by Maynord et al. (1989) and Maynord (1990) and published by the USACE as Engineering Manual No. 1110-2-1601 EM-1601 (USACE 1991). The procedure uses both velocity and depth as its primary design parameters. Design guidance is provided for riprap sizing, thickness, shape, and gradation, as well as for the design of granular and geotextile filters. The results of the NCHRP riprap study have been incorporated into FHWA’s HEC-23 (Lagasse et al. 2009) to include a general discussion of riprap design, filter requirements, and failure modes in Volume 1 (Chapter 5), and detailed design guidelines in Volume 2, Design Guideline 4 “Riprap Revetment” and Design Guideline 16 “Filter Design.” Reference to this guidance is suggested for the design of the rock toe component of biotechnical countermeasures. The issue of appropriate engineering design of a rock toe is also addressed in Section 4.3.3. Note that Figures 4.27 through 4.29 and associated discussion in Section 4.3.3 recognize the need for a filter and include the notes: • Filter layer of graded aggregate and/or filter fabric; • Graded, granular filter is preferable to filter fabric to improve root penetration; and • Inserting live poles/stakes through slits in a geotextile filter is discouraged, but can be used if no other alternative is available. A longitudinal stone toe only provides toe protection and does not protect mid- and upper-bank areas. Some erosion of these areas should be anticipated during long-duration, high-energy flows, or until the areas become otherwise protected (e.g., by vegetation). Stone toe is not suitable for reaches where rapid bed degradation (lowering) is likely, or where scour depths adjacent to the toe will be greater than the depth of the toe. Section 4.2.3 provides specific guidance for analyzing long-term degradation trends and design procedures for toe and bendway scour. In regard to hydraulic loading, permissible shear and velocity for a longitudinal stone toe are related to the size of rock used in construction. Other factors, such as the angularity of the stone, the thickness of the layers of stone, and the angle at which the faces of the stone structure are constructed also come into play. Materials and Equipment. Stone for the structure should be well graded and properly sized. Detailed guidance for sizing stone for bed and bank stabilization structures is beyond the scope of this guideline, and many approaches are available (see references cited in the design discussion above).

Design Guidelines and Appraisal of Research Results 145 Construction and Installation. All longitudinal stone toes should be constructed in an upstream to downstream sequence. This technique usually requires heavy equipment for exca- vation of keys (tie-backs) and efficient hauling and placement of stone. Longitudinal stone toes can be constructed from within the stream, from roadways constructed along the lower section of the stream bank itself, or from the top. The preferred method is from the point bar side of the stream (especially possible with ephemeral or intermittent streams), as this causes the least dis- turbance of existing bank vegetation. The least preferred is from the top of the bank, as it disturbs or destroys more bank vegetation and the machine operator’s vision is limited. Usually, the toe trenches are excavated first and a filter and rock are placed into the key. The rock is then formed into tie-backs (if needed) and finally the stone toe is constructed along a “smoothed” alignment, preferably with a uniform radius of curvature throughout the bend. In a multi-radius bend, smooth transitions between dissimilar radii are preferred. Installation of a stone toe is illustrated in Section 4.4.3, the “Humid Region” application example (see Figure 4.48a, b, i, and j). For additional information on this installation on Malletts Creek near Ann Arbor, MI, see Table 2.11 (Site MI5) and the Compendium. Cost. Costs of a riprap toe depend on the cost of stone, hauling, and amount of stone used. Including stone for keys and tie-backs, typically 120 to 140 tons (110 to 130 metric tons) of stone will be used for each 100 ft (30 m) of protected bank when toe is placed at a rate of 1 ton/ft (3 metric tons per lineal m) of protected bank. Based on typical unit costs for stone (including delivery and placement), cost for this type of toe ranges from $16 to $35 per foot ($50 to $115 per m) of pro- tected bank, although costs are highly dependent on regional considerations. Maintenance and Monitoring. Maintenance and monitoring requirements should be linked to consequences of failure. Detailed maintenance and monitoring requirements for riprap can be found in NCHRP Report 568 (Lagasse et al. 2006) and FHWA’s HEC-23 (Lagasse et al. 2009). Inspection of riprap placement typically consists of visual inspection of the installation procedures and the finished surface. Inspection must ensure (1) that a dense, rough surface of well-keyed graded rock of the specified quality and sizes is obtained, (2) that the layers are placed such that voids are minimized, and (3) that the layers are the specified thickness. The following general guidance for inspecting riprap is presented in HEC-23 (Lagasse et al. 2009): 1. Riprap should be angular and interlocking (old bowling balls would not make good riprap). Flat sections of broken concrete paving do not make good riprap. 2. Riprap should have a granular or synthetic geotextile filter between the riprap and the subgrade material. 3. Riprap should be well graded (a wide range of rock sizes). The maximum rock size should be no greater than about twice the median (d50) size. 4. For bridge piers, riprap should generally extend up to the bed elevation so that the top of the riprap is visible to the inspector during and after floods. 5. When inspecting riprap, the following are strong indicators of problems: – Has riprap been displaced downstream? – Has angular riprap blanket slumped down slope? – Has angular riprap material been replaced over time by smoother river run material? – Has riprap material physically deteriorated, disintegrated, or been abraded over time? – Are there holes in the riprap blanket where the filter has been exposed or breached? Common Reasons for Failure. Features that should be monitored are similar to those for all stone structures: loss of stone due to subsidence, leaching of underlying sediments, raveling, or excessive launching. Extreme scour or bed lowering on the stream side of the toe can cause the entire mass of stone to launch, creating an opening or gap in the longitudinal structure. If this

146 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures situation is anticipated or encountered, the problem can be remedied by adding more rock to restore design conditions. Longitudinal stone toes may be flanked during extremely high flows if the key trenches are incorrectly built or if the tie-backs are spaced too widely or are constructed with inadequate amounts of stone. Terminal keyways or “refusals” oriented at 90 degrees to the bank have resulted in many failures at the downstream end of the structure, due to flow expansion at that point. These terminal key trenches at the upstream and downstream ends should be excavated into the bank at an angle of approximately 30 degrees with the primary flow direction and of sufficient length that flows will not be able to get around them during the design storm (see Figure 4.5). Observations relevant to the performance of this multi-component treatment from the field site visits are summarized in Section 4.5.3. Laboratory Testing Results—Live Siltation and Live Staking with Rock Toe Full-scale laboratory testing conducted under this research project confirmed the suitability of existing design and installation guidelines for live siltation and live staking as stream bank protection treatments as described above in this section. Observations from laboratory testing include the following: 1. At the planting densities of live siltation (toe) and live staking (upper bank) examined in the testing program, the Mannings n resistance coefficient was found to range from 0.030 to 0.035. The lower Manning n values were associated with fully pronated willows. 2. Pronation of willows occurred at a unit discharge (velocity times depth) of 5.3 cfs/ft. 3. Permissible shear stress for this treatment was found to be 2.5 lb/ft2, which confirms existing guidance. 4. Permissible depth-average velocity V60 was found to be 7 to 9 ft/s, which is considerably higher than previous recommendations (3 ft/s). 5. Measurement of the vertical velocity distribution demonstrates the effectiveness of the live siltation component in moving high-velocity flow away from the bank toe and into the main channel. 6. The excellent performance of this treatment underscores the need for a stone toe component in combination with vegetative treatments. 4.3.2 Vegetated Mechanically Stabilized Earth Without Hard Toe Introduction As noted in Section 4.3.1 VMSE without a hard toe was the second of two environmentally sensitive treatments tested at prototype scale in an experimental trapezoidal channel (see Chap- ter 3 and Figure 3.4). This treatment is referred to as “Tray 2” in Chapter 3. For this treatment NCHRP Report 544 (McCullah and Gray 2005) lists the following research opportunities (see Appendix A): VMSE—Some uncertainty exists at present as to the exact permissible shear stresses and velocities for VMSE interfaces. Additional research would also be helpful on the nature of the interaction between roots and fabric and between root architecture and distribution in VMSE structures. Proposed design guidelines for VMSE treatment are consolidated in this section. Then the results of laboratory testing of this treatment (see Chapter 3) as they address the research oppor- tunities referenced above are summarized as an update to commonly used guidelines, specifically in relation to hydraulic issues of permissible velocities and shear stresses for this biotechnical configuration.

Design Guidelines and Appraisal of Research Results 147 Purpose and Advantages. Components and variations of this treatment are also known as FES, brush layering with soil wraps, and vegetated geofabric wrapped soil. This technique consists of live cut branches (brush layers) interspersed between lifts of soil wrapped in natural fabric, e.g., coir, or synthetic geotextiles or geogrids (which are not considered in this design guidance). The live brush is placed in a criss-cross or overlapping pattern atop each wrapped soil lift in a manner similar to conventional brush layering (see live brush layering in McCullah and Gray 2005 and Bischetti et al. 2009). The fabric wrapping provides the primary reinforcement in a manner similar to that of conventional mechanically stabilized earth. The live, cut branches eventually root and leaf out, providing vegetative cover and secondary reinforcement as well. Since the inert fabric wraps provide reinforcement and mechanical stabilization, they may permit steeper slopes to be constructed than would be possible with live brush layers alone. Brush layering treatment by itself is normally restricted to slopes no steeper than 1V:2H. The “Tray 2” VMSE as tested with a biodegradable coir fabric was installed at a slope of 1V:2H (see Chapter 3 and Figure 3.4). Design. VMSE as defined here can be used to stabilize slopes as steep as 1V:2H. This tech- nique provides an alternative to steep retaining structures, and to techniques that require slope flattening or bank lay back, which results in excessive right-of-way encroachment at the top of the bank. The use of geogrids permits even steeper slopes and provides greater long-term dura- bility and security (see McCullah and Gray 2005). The fabric or geotextile wrap also provides additional protection to upper portions of stream banks that are subject to periodic scour or tractive stresses. If either steady, long-term seepage, or temporary bank return flows after flood events are a problem, the brush layers act as a drainage layer or as conduits that relieve internal pore-water pressure, and favorably modify the groundwater flow regime within this slope to minimize slope stability problems. Design of VMSE is relatively complex, because it entails designing, melding together, and constructing two similar yet distinct methods, conventional MSE and live brush layering. Both techniques are widely used and well understood; however, simultaneous use introduces com- plexity. While many different types of inclusions with various shapes and properties can be used to reinforce and buttress earthen slopes, Tray 2 as tested should more properly be referred to as FES. The installation steps and design criteria for FES with brush layering are illustrated in Figure 4.26. Live cuttings inserted as shown in Figure 4.26 also act as tensile inclusions and help to stabilize a slope, embankment, or structural fill. Live brush layers behave exactly in this fashion. Gray and Sotir (1992) discuss how brush layers can be analyzed and their contribution to slope stability deter- mined in a rational, quantitative manner. In this combined approach, however, the contribution to mechanical reinforcement from the live cuttings is simply treated as a bonus, and the design analysis is focused on the fabric (or geogrid) reinforcements themselves. There appears to be little or no published test data for permissible hydraulic loading of VMSE structures. There does exist, however, published data on vegetated coir mats and live brush layers, respectively, as shown in Table 4.4. These data can be used to approximate permissible shear stresses and velocities for VMSE. However, one of the goals of testing the Tray 2 configuration under NCHRP Project 24-39 was to develop better hydraulic loading data for this particular environmentally sensitive bank-protection treatment. For additional data on permissible velocities and shear stresses see Chapter 3 and the laboratory testing summary below. Materials and Equipment. Select long branches of native tree species that are capable of vegetative propagation. Willows (Salix spp) are the most commonly used plant material, because

148 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Figure 4.26. VMSE typical drawing (McCullah and Gray 2005). Treatment Type Limiting Velocity ft/sec (m/sec) Limiting Shear Stress lbs/ft2 (Kg/m2) Vegetated coir mat 4 - 8 (1.2 - 2.4) 9.5 (46.4) Live brush mattress (initial) 0.4 - 4.1 (0.1 - 1.2) 4.0 (19.5) Live brush mattress (established) 3.9 - 8.2 (1.2 - 2.5) 12.0 (58.6) Brush layering (initial/established) 0.4 - 6.5 (0.1 - 2.0) 12.0 (58.6) Table 4.4. Limiting shear stress and velocity levels for selected soil bioengineering treatments (adapted from Fischenich 2001).

Design Guidelines and Appraisal of Research Results 149 they generally root well from cuttings. Alder, cottonwood (Populus deltoides), and dogwood (Cornus spp) can also be used effectively, particularly when mixed in with willow. The length of the branches will vary depending upon the desired depth of reinforcement, but they should be long enough to reach the back of an earthen buttress placed against a stream bank while protruding slightly beyond the face (see Figure 4.26). The diameter of the live cuttings will also vary depending on their length, but typically should range from ¾ to 2 in. (19 to 51 mm) at their basal ends. The inert construction materials in the configuration tested consisted of a natural geofabric. The coir netting is visually less obtrusive than other manufactured materials (e.g., geogrids) and can retain moisture helpful to vegetative establishment. As with most biotechnical structures, the vegetation is intended to provide long-term stability. Material properties such as longevity, durability, and resistance to abrasion/corrosion should be considered. Coir fabric or netting comes in various grades that have different size openings and unit tensile resistances. Construction and Installation. A VMSE structure must be constructed during the dormancy period to ensure good vegetative propagation and establishment. Alternatively, the live cuttings may be harvested during dormancy, and placed in temporary cold storage until they are ready for use during an out-of-dormancy period (e.g., during the summer months). The latter recourse increases the unit cost of the technique. For this treatment, materials procurement is more demanding, and installation more complex, because of the blending of two distinct methods, i.e., conventional MSE/FES and live brush layering, into a single approach. Costs will also be more than brush layering used alone, because of the added expense of the geotextile and the additional labor required to handle and construct the wraps. A VMSE installation begins at the base of the slope and proceeds upwards. A series of schematic drawings illustrating the installation procedures step by step are shown in Figure 4.26. The struc- ture should generally be supported on a rock toe or base (not shown in Figure 4.26 schematic) and be battered or inclined at an angle of at least 10 to 20 degrees to minimize lateral earth forces (as shown in Figure 4.26). It is critical that factors such as toe scour depth be determined for each particular project and be incorporated into project design. Note—the VMSE treatment as tested for NCHRP Project 24-39 had no hard toe in response to a tendency of some resource agencies to prefer a treatment with no rock component (see Figure 3.4 and summary of testing results, below). The following general guidelines and procedures apply: 1. Excavate a trench below the likely depth of toe scour and backfill it with rock to provide a base for the VMSE structure. The top surface of the rock should be inclined with the horizontal to establish the desired minimum batter angle for the overlying structure. 2. Construct an earthen structure reinforced with coir fabric and live brush on top of the rock base. For this purpose, select fabric rolls with a minimum roll width of 13 ft (4 m) and unit tensile strength predetermined from a stability analysis that takes into account the height and slope angle of the reinforced, earthen buttress fill (see McCullah and Gray 2005 for additional guidance). 3. Place select fill material on the fabric and compact it in 3 in. (7.5 cm) lifts to a nominal thickness ranging from 12 to 30 in. (30 to 76 cm). Thinner lifts are used at the base of the structure, where shear stresses are higher. Temporary batter boards may be required at the front face to confine the select fill during the installation process and to form an even face. 4. The fabric sheet should be allowed to drape down or protrude beyond the front edge of each underlying lift of earthen fill to create at least a 3 ft (0.9 m) overlap when it is pulled up and over the next lift. The exposed sections of fabric layers are pulled up and over the faces of the fill layers (see to Figure 4.26) and staked in place. The fabric should be pulled as uniformly as possible before staking to develop initial tension in the fabric. A tractor or winch pulling on a long bar with hooks or nails along its length works well for this purpose. The tensioned fabric

150 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures overlap sections should be secured in place using wood construction stakes spaced every 3 ft (0.9 m). 5. Layers of live cut branches are then placed criss-crossed atop the underlying wrapped soil lift (see Figure 4.26). In addition, 1 to 2 in. (22 to 50 mm) of topsoil should be mixed in with the cut branches. The top soil can be placed beforehand or spread over the top of a brush layer. Up to three (3) layers of live, cut branches interspersed with 1 to 2 in. (25 to 50 mm) of topsoil can be placed in this manner. 6. The process is repeated with succeeding layers of earth fill, live brush and fabric until the specified height or elevation is reached. Cost. Costs for VMSE structures are likely to be on the high end of environmentally sensi- tive bank-protection measures because of both design/construction complexity and material acquisition costs. In addition, site-specific considerations, such as access, also have a significant influence. VMSE treatments will obviously cost more than live brush layering used alone because of the presence of geotextile material reinforcements. Maintenance and Monitoring. For VMSE, monitoring should consist of inspecting the fabric for signs of breakage or tearing from scour damage or possibly from excessive tensile stresses due to higher than expected lateral earth pressures. Signs of uncontrolled seepage, such as weeping or wet spots in the structure, should also be noted. Finally, the site should be examined for pos- sible signs of flanking erosion, which must be addressed with ancillary protective measures as the flanking will threaten the integrity and effectiveness of the VMSE structure itself. For additional guidance on monitoring of the brush layering component see Section 4.2.5. Common Reasons for Failure. As of 2005 no known instances of failure had been published. The most likely causes of a hypothetical failure, however, would be the following (McCullah and Gray 2005): 1. Inadequate primary reinforcement from the inert tensile inclusions (fabric or geotextile), i.e., improper vertical spacing or lift thickness, insufficient allowable unit tensile resistance in the selected fabric or geotextile, and too short an embedment length for the given soil and site conditions, e.g., slope height, slope angle, and soil shear strength properties. 2. Failure to properly consider seepage conditions and install adequate drainage measures, e.g., chimney drain behind VMSE structure. 3. Inadequate attention to construction procedures and details. 4. As with all resistive, bank line protective structures, flanking, toe scour, and undermining are always potential problems. Observations relevant to the performance of this treatment from the field site visits are summarized in Section 4.5.3. Laboratory Testing Results—VMSE Without Hard Toe As modifications to the NCHRP Report 544 guidance extracted above, the following additional guidance based on the Task 3 laboratory testing is recommended. 1. The vegetative treatment of Tray 1 (live siltation willow and live staking with stone toe) exhibited significantly lower Mannings n resistance coefficients, bed shear stresses, and erosion compared to Tray 2 (VMSE soil lifts with soft toe) at the same discharge. As tested, the Tray 2 VMSE treat- ment resulted in Mannings n resistance coefficients ranging from 0.040 to 0.045. The lower Mannings n values were associated with fully pronated willows. 2. Pronation of willows occurred at a unit discharge (velocity times depth) of 5.3 cfs/ft. 3. A higher density of willows per square yard of surface area does not necessarily result in better performance. Rather, the overall geometry of the bank slope and planting configuration appears

Design Guidelines and Appraisal of Research Results 151 to be the more important factor in the overall performance of the vegetative component. For example, the stair-step configuration of the VMSE creates preferential pathways for high-velocity flow that results in the potential for damage to the coir/jute fabric of the soil lifts. 4. Vulnerable areas of the VMSE treatment include the faces of the individual soil lifts. In par- ticular, the lowermost lift at the toe of the bank slope is subject to damage if left unprotected by other means, such as a stone (i.e., riprap) toe armor or other “hard” engineered treatment. Qualitative observations suggest that preferential pathways for high-velocity flow along the faces of the soil lifts appear to be a shortcoming of this bank-protection treatment. 5. The uppermost soil lift of VMSE is also vulnerable to soil loss if it becomes fully submerged and is not vegetated on its top surface. In contrast, the top surfaces of soil lifts that are lower down on the slope were seen to be protected by the pronated willows of the lift above, which provided a “shielding” effect. 6. The point velocity, V60, corresponding to the onset of tears in the coir/jute fabric, as well as areas of significant soil erosion, was found to be in the range of 5 to 7 ft/s. The corresponding shear stress at these conditions was 2.5 lb/ft2. In the stair-step 2H:1V configuration of VMSE, the limiting values for velocity and shear stress are much lower than those for similar materials on a graded 3H:1V slope (see Table 4.4). At even higher velocities (7 to 9 ft/s) and shear stresses (3 to 4.5 lb/ft2), excessive damage and soil loss were observed. 7. The laboratory testing of this treatment configuration underscores the need for a hard armor toe component for vegetative stream bank protection measures. 4.3.3 Vegetated Riprap—An Overview Introduction Vegetated riprap is a layer of stone and/or boulder armoring that is vegetated, optimally during construction, using pole planting, brush layering, and live-staking techniques. The goal of this method is to increase the stability of the bank while simultaneously establishing riparian growth within the rock and overhanging the water to provide shade, water quality benefits, and fish and wildlife habitat. Vegetative riprap combines the widely accepted, resistive, and continuous rock revetment techniques with deeply planted biotechnical techniques. The analysis of performance and design guidelines presented for the rock toe with live siltation and live staking in Section 4.3.1 has direct application to this environmentally sensitive treatment for both the design of the stone and the need for a filter. One of the main conclusions drawn from extensive research on stream bank erosion and pro- tection in the 1970s was that simply grading the bank to a stable slope and planting vegetation without toe protection is ineffective (USACE 1981); similar conclusions have been reached by others (Shields et al. 1995). In most stream channels, shear stress reaches a maximum at the toe of concave banks, and this region is unsuited for terrestrial plants because it is either permanently or frequently inundated. Combining treatments allows maximum flexibility in meeting the objectives of bank stabilization and habitat development. Selection of an appropriate toe protection structure, e.g., longitudinal stone toe with spurs, can create ideal water/shore interface conditions and scour holes that may provide stable pool habitats. A well vegetated (or revegetated) bank can improve aquatic and riparian habitat in addition to providing important functional benefits (Coppin and Richards 1990). Stream bank vegetation provides cover, shade, and insect food sources for fish and other aquatic organisms near the water’s edge. Upper and mid-bank vegetation yields cover and habitat opportunities for small mammals and other riparian wildlife (McCullah and Gray 2005). Based on the results of the Task 2 survey of practitioners, vegetated riprap is a fairly common approach to designing environmentally sensitive stream bank treatments (17 out of 35 responses)

152 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures (see Table 2.9). Of the 16 field sites visited under Task 6, 11 used willow posts and poles and five sites included, specifically, vegetated riprap as a stream bank treatment. The following sections provide general design guidance for several vegetated riprap alternatives. Commonly Used Vegetated Riprap Techniques Five methods for constructing vegetated riprap have proven effectiveness. Typical design concept sketches of four of these five methods are provided as Figures 4.27 through 4.30. These sketches are reproduced from NCHRP Report 544 (McCullah and Gray 2005). While the key hydraulic design variable “design high water” is not defined in these sketches, AHW and ALW criteria are discussed in Section 4.2.3 under the Bankfull Discharge section. 1. Vegetated riprap with willow bundles (Figure 4.27): Vegetated riprap with willow bundles is the simplest to install, but it has a few drawbacks. This technique typically requires very long 10 to 23 ft (3 to 7 m) poles and branches, as the cuttings should reach from 6 in. (15 cm) below the low water table to 1 ft (30 cm) above the top of the rocks. In addition, only those cuttings that are in contact with the soil will take root, and therefore, the geotechnical benefits of the roots from those cuttings on the top of the bundle may not be realized. 2. Vegetated riprap with bent poles (Figure 4.28): Vegetated riprap with bent poles is slightly more complex to install. A variety of different lengths of willow cuttings can be used because they will protrude from the rock at different elevations. 3. Vegetated riprap with brush layering and pole planting (Figure 4.29): Vegetated riprap with brush layering and pole planting is the most complex type of riprap to install, but also provides the most immediate habitat benefits. The installation of this technique is separated into two methods; one method describes installation when building a bank back up, while the other is for Figure 4.27. Vegetated riprap—willow bundle method (McCullah and Gray 2005).

Design Guidelines and Appraisal of Research Results 153 Figure 4.28. Vegetated riprap—bent pole method (McCullah and Gray 2005). Figure 4.29. Vegetated riprap—brush layering with pole planting (McCullah and Gray 2005).

154 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Figure 4.30. Vegetated riprap with joint planting (McCullah and Gray 2005). a well-established bank. If immediate aquatic habitat benefits are desired, this technique should be used. However, vegetated riprap with brush layering and pole planting may not provide the greatest amount of root reinforcement, as the stem contact with soil does not extend up the entire slope. Combination of this technique with pole- or bundle-planted riprap will perform well, as the latter techniques typically have higher rooting success. 4. Vegetated riprap with soil cover, grass and ground cover: This technique is also known as “buried riprap,” and consists of infilling and covering a standard rock riprap installation with soil and subsequently establishing grass vegetation. Some stripping of the soil and grass may be expected during severe events. 5. Joint or live stake planted riprap (Figure 4.30): Joint or live stake planted riprap is reveg- etated riprap, as opposed to the other techniques, which are true vegetated riprap methods. This technique should be used only when attempting to get vegetative growth on previously installed riprap.

Design Guidelines and Appraisal of Research Results 155 Environmental Considerations and Benefits There are many environmental benefits offered by vegetated riprap, most of which are derived from the planting of willows or other woody species in the installation. Willow provides canopy cover to the stream, which gives fish and other aquatic fauna cool places to hide. The vegetation also supplies the river with carbon-based debris, which is integral to many aquatic food webs; birds that catch fish or aquatic insects will be attracted by the increased perching space next to the stream (Gray and Sotir 1996). The small spaces between the rocks also provide benthic habitat and hiding places for small fish and fry (also see Section 4.2.6). Construction/Installation Guidelines Vegetated Riprap with Willow Bundles (Figure 4.27): • Grade the bank to the desired slope where the riprap will be placed, such that there is a smooth base. • Dig a toe trench for the keyway below where the riprap will be placed on a granular or geo- textile filter. • Place 5 to 6 in. (10 to 15 cm or 5 to 8 stem) bundles on the slope, with the butt ends placed at least 1 ft (30 cm) in the low water table. This will probably involve placing the poles in the toe trench before the rock is placed, if standard riprap rock is being used. Digging shallow trenches for the willows prior to placing them on the slope will decrease damage to the cuttings from the rocks, and may increase rooting success because more of the cuttings will be in contact with soil. • The bundles should be placed every 6 ft (1.8 m) along the bank and point straight up the slope. Once the bundles are in position, place the rock on top of it at the top of the slope. The bundles should extend 1 ft (.3 m) above the top of the rock. If the bundles are not sufficiently long, they will probably show decreased sprouting success, and therefore, a different technique should be chosen. Vegetated Riprap with Bent Poles (Figure 4.28): • Grade back the slope where the riprap will be placed, such that there is a smooth base. • Dig a toe trench for the keyway below where the riprap will be placed on a granular or geo- textile filter. • If filter fabric is being used, lay the fabric down on the slope, all the way into the toe trench, and cut holes in the fabric about 2 to 3 ft (0.6 to 0.9 m) above the mean low water level. Slip the butt ends of the willow poles through the fabric and slide them down until the bases are at least 6 in. (15 cm) into the perennial water table, or at the bottom of the toe trench, whichever is deepest (Hoag and Fripp 2002). • If using filter gravel, lay it down on the slope, and place a layer of willow poles on top of the gravel, with the bases of the cuttings at least 6 in. (15 cm) into the perennial water table, or at the bottom of the toe trench, whichever is deepest. • Ensure that rocks in the toe trench lock together tightly, as they are the foundation for the structure. • Place the next layer of boulders such that it tapers back slightly toward the stream bank. • Bend several willow poles up, such that they are perpendicular to the slope, and tight against the first layer of rocks. Now place the next layer of rocks behind these poles. Placement will require an excavator with a thumb, as someone will have to hold the poles while the rocks are placed. As the poles are released, they should be trimmed to 1 ft (30 cm) above the riprap. • This last step should be repeated until all the poles have been pulled up and the entire slope has been covered.

156 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Vegetated Riprap with Brush Layering and Pole Planting (Figure 4.29): There are two methods of constructing brush layered riprap; one involves building up a slope, and the other works with a pre-graded slope. Method 1: • Lay the bank slope back to somewhat less than the desired finished slope. • Dig a toe trench and lay the key rocks into the trench on a granular or geotextile filter. Pack soil behind these rocks, with filter gravel in between the soil and rocks. Continue installing riprap 3 to 4 ft (0.9 to 1.2 m) up the bank. • Slope the soil back into the bank at a 45 degree angle, such that the bottom of the soil slope is in the vadose zone (see Figure 4.1 and associated discussion). Place a layer of willow cut- tings on top of the soil, with the butt ends extending into the vadose zone, and the tips of the branches sticking out 1 to 2 ft (30 to 60 cm). • Place the next layer of stones on top of the initial rocks, but graded slightly back, and repeat the soil and brush layering process. When finished, trim the ends of the willow branches back to 1 ft (30 cm). Do not cut shorter than 1 ft (30 cm) as the plant will have difficulty sprouting. Method 2: • Lay the bank slope back to the desired finished grade, and dig a toe trench if self-launching stone is not being used. • Place the rocks in the keyway on a granular or geotextile filter, and fill in behind with filter gravel and soil. Continue installing riprap 3 to 4 ft (0.9 to 1.2 m) up the bank. • Place the bucket of an excavator just above the layer of rocks at a 45 degree angle. Pull the bucket down, still at a 45 degree angle, until the water table is reached, or if the stream is dry, to the elevation at the bottom of the key trench. Pull up and back on the bucket: this will provide a slot in the bank into which willow poles can be placed. • Throw in some willow poles [6 poles per linear foot (about 18 poles per linear meter)], ensuring that the butt ends are at the bottom of the trench. • Release the scoop of earth, and allow it to fall back in place on the slope. Then place the next layer of rock on top of the branches, flush with the slope. If self-filtering stone is not being used, filter gravel should be placed behind the rocks. Repeat the process, beginning again with pulling back a scoop of soil. Continue this process to the top of the slope, or, if preferred, use joint-planted riprap on the upper slope, where it is difficult to reach the perennial water table with the excavator bucket. • When finished, trim the ends of the branches back such that only 1 ft (30 cm) extends beyond the revetment. Cost Installation of vegetated riprap will require about 2.5 to 6 work hours/m2 (2 to 5 work hours/yd2). The cost of rock will vary depending on availability in the local area, but typically ran between $20 to $60 per ton ($22 to $67 per metric ton), delivered (in 2005). Maintenance/Monitoring Riprap should be visually inspected following any one-year return interval or greater flow, with focus on potential weak points, such as transitions between undisturbed and treated areas. Soil above and behind riprap may show collapse or sinking, or loss of rock may be observed. Inspect riprap during low flows annually, to ensure continued stability of the toe of the structure. Treat bank or replace rock as necessary. See Section 4.3.1 for additional guidance on monitoring and inspection of the riprap component of any environmentally sensitive treatment.

Design Guidelines and Appraisal of Research Results 157 Common Reasons/Circumstances for Failure Flanking, overtopping, or undermining of the revetment due to improperly installed or insufficient keyways is one of the biggest reasons for failure of riprap. Improperly designed or installed filter material can also cause undermining and failure of the installation. Undersized stones can be carried away by strong currents, and sections of the revetment may settle due to poorly consolidated substrate. Vegetation may require irrigation if planted in a nondormant state or in extremely droughty soils. Also, vegetation may be limited by excess soil moisture (Pezeshki et al. 1998). Observations relevant to the performance of this treatment from the field site visits are summarized in Section 4.5.3. 4.4 Applications 4.4.1 Overview This section presents two detailed examples of the application of environmentally sensitive stream bank protection measures employed in conjunction with stream channel restoration projects. Recognizing that dryland landscapes are quite different from those of more humid regions, one example involves the application of environmentally sensitive techniques on an arid region perennial stream—the Rio Grande near Bernalillo, NM. The second example deals with a smaller stream in a humid region with significant infrastructure issues—Malletts Creek near Ann Arbor, MI. Compared to a humid region, the topography and landforms of an arid or semi-arid region are more abrupt, the soils are thinner, the bedrock exposures are usually more pronounced and the streams are smaller, carry larger sediment loads, and are likely to be dry for at least part of the year. Overall, the physical environment reflects the lack of water, and mechanical weathering and erosion predominates over chemical weathering and solution, as compared to a humid environment. As a consequence, dryland stream morphology significantly differs from that of humid-zone streams. In a humid environment, high precipitation produces vegetation and soils that are well developed and stabilized. Under these conditions, natural streams generally carry small suspended sediment loads, reflecting this stability in the upland watersheds. Additionally, high precipitation produces a dilu- tion effect on the sediments that are eroded. While these projects present certain commonalities in the design, implementation, and monitoring phases, the differences in the climatic, hydrologic, and geomorphic conditions, as well as the characteristics of the cultural environment bring out interest- ing contrasts in the approaches necessary to achieve project success. The examples have been chosen to illustrate the integration of hydraulic engineering analysis and design with the multidisciplinary approach necessary to achieving success with a river restoration project. The examples involve the cooperative efforts of multiple agencies and the challenges of meeting stakeholder goals and expectations. In both cases monitoring and maintenance issues are addressed and lessons learned are summarized. 4.4.2 Arid Region Example Problem Identification Water resource management activities (diversions, dams, levees, drains, channelization, jetty- jacks) by federal agencies and other entities have altered the hydrologic, ecologic, and sediment transport characteristics of the Rio Grande within New Mexico. Jemez Canyon, Cochiti, Abiqui, and Gallisteo Dams, operated for flood and sediment control, have contributed, in part, to the degradation of ecosystem functions and values of the Rio Grande in the restoration area.

158 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures The restoration area is located approximately 25 miles downstream of Cochiti Dam on the Rio Grande in what is designated as the Santa Ana Reach (Figure 4.31) of the Pueblo of Santa Ana (Pueblo—a sovereign tribal nation). This reach encompasses the length of the Rio Grande beginning immediately downstream from the confluence with the Jemez River (also known as Rio Jemez) to the Highway 550 bridge in Bernalillo, approximately 4.4 river miles. Along the approximately 4.4 miles of the Rio Grande within the Santa Ana Reach, several hydrologic and ecological problems have been identified: • The historically broad channel has incised up to 10 feet during the past 30 years, resulting in a narrow, entrenched channel. • The extent and quality of aquatic habitat for native fish have deteriorated due to increased water depth and velocity. • Channel incision has resulted in lowering the local water table in certain locations. • The lack of inundation, scouring, and sediment deposition within the “bosque” (riparian woodland) has curtailed native cottonwood and willow seedling recruitment. • Widespread invasion of non-native saltcedar and Russian olive trees has decreased the value of wildlife habitat and increased the threat of damaging fire. In cooperation with the U.S. Bureau of Reclamation (Reclamation) and the USACE, the Pueblo of Santa Ana has implemented restoration activities to restore the river channel, active flood- plain, and the historic floodplain. Channel Instability and Geomorphic Response A number of publications provide a detailed discussion of the history and response of the Rio Grande within the project reach to the construction of flood control measures, bank revetment, Figure 4.31. Santa Ana Reach of Santa Ana Pueblo.

Design Guidelines and Appraisal of Research Results 159 channelization works, and the construction of upstream dams (for example, see Lagasse 1980, Salazar 1998, Richard 2001, USACE 2002a, Grassel 2002, Sixta 2004, Ortiz 2004, Ayres Associates 2006a). Historically, the fluvial characteristics of the Middle Rio Grande were those of a wide and shallow river prior to the influence of flood control activities. The channel was described as a sand-bed stream, (Nordin and Beverage 1965), with a braided pattern (Lane and Borland 1953), probably due to sediment overload (Woodson 1961). The referenced studies indicate that the river followed a pattern of scour and fill during floods and was in an aggrading regime. Flood hazards associated with the aggrading riverbed prompted the Middle Rio Grande Con- servancy District to build levees along the floodway during the 1930s. However, the levee system confined the sediment and increased the aggradation in the floodway. By 1960 the river channel near Albuquerque was 6 to 8 ft above the elevation of lands outside the levees (Lagasse 1980). Additional channel rectification works included the Kellner jack system (Figure 4.32) for bank stabilization, which was installed during the 1950s and 1960s. (Note: In the Middle Rio Grande Valley, the Kellner jacks are referred to as “jetty-jacks.”) By 1962, a total of 115,000 jacks were in place along the river (Lagasse 1980). Jetty-jack fields placed along the river were highly effective. By trapping sediment, they filled in and trees grew on the new banks created by the jetties. These jetties and the new banks they created protected the newly constructed levees. The jetty fields and bosque vegetation have been considered as largely responsible for the stable channel posi- tion observed in the current river planform (Grassel 2002). In fact, Lagasse (1980) notes that the stabilization and rectification of the floodway channel between 1954 and 1962 had, by the early 1970s, begun to accomplish the desired effect of reversing the long-term aggradational trend and lowering bed elevations. Regardless of the effectiveness of the jetty-jack fields, under the Comprehensive Plan of Improve- ment for the Rio Grande in New Mexico, construction of dams at Cochiti (1973), Abiqui (1963), Jemez Canyon (1953), and Galisteo Creek (1970) were expected to slow aggradation or reverse the trend to degradation in the Middle Rio Grande Valley. As a result of the flood and sediment control measures, the Middle Rio Grande has experienced significant channel degradation and streambed coarsening. Construction of these dams has also cut off the historical floodplain from the river (Woodson and Martin 1963). By the early 1980s the resultant post-dam degradation and the development of streambed armoring (Lagasse 1980, 1981) within the project reach created unfavorable conditions for further jetty-jack stabilization. By the early 1990s the loss of frequent inundation of flood plain areas and the control of major sediment source areas such as the Jemez River resulted in significant channel degradation (Figure 4.33) and, as a result, significantly reduced the effectiveness of the jetty-jack fields. This ongoing degradational environment and the significantly reduced sediment supply Figure 4.32. Jetty-jacks on the Rio Grande.

160 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures have induced scour of the remaining jetty-jack fields exposed along the eroding river banks result- ing in undercutting and burial of the jacks. Hydromodification Hydrology in the Middle Rio Grande Valley (i.e., Cochiti Lake to Elephant Butte Lake) follows a pattern of high flows during spring snowmelt runoff and low flows during the fall and winter months (USACE 2008). Additional, short-duration, high flows result from thunderstorms that occur in late summer and fall. Middle Rio Grande hydrology has been altered due, in part, to the influence of flood control dams. Cochiti Dam primarily acts to decrease peak flows and has a much smaller impact on low flows; therefore, average annual flows have been less affected, while peak flows have been reduced. Average yearly hydrographs for pre- and post-Cochiti Dam periods through 1999 are shown in Figure 4.34. The annual hydrographs illustrate that the closure of Cochiti Dam has reduced the peak flows and extended the duration of the high-flow period. Average winter base flows are somewhat larger during the post-dam period. Review of annual peak discharge data also exhibits the influence of flood control. Historical annual peak discharges recorded at the San Felipe gage (approximately 15 river miles down- stream from Cochiti Dam) illustrate the effects of regulation on the Rio Grande (Figure 4.35). From 1927 to 1945 flows in excess of 20,000 cfs were experienced approximately every five years. From 1945 to the construction of Cochiti Dam in 1973, floods in excess of 10,000 cfs were fairly common with the exception of drought years. Following construction of Cochiti Dam, regulation has prevented downstream flows from exceeding 10,000 cfs. This has reduced the average annual instantaneous peak discharge from 9,800 cfs to 5,700 cfs for the pre- and post-dam periods, respectively. CO-27 Sta. 153+73.60 (US of GRF2) Rio Grande, NM Distance from left bank reference point (ft) 0 50 5,060 5,061 5,062 5,063 5,064 5,065 5,066 5,067 5,068 5,069 5,070 5,071 5,072 5,073 100 150 200 250 300 350 400 May-74 Nov-83 Jul-92 Aug-01 Jul-08 450 500 550 600 650 700 El ev at io n (ft ) Figure 4.33. Comparison of historic surveys at Cochiti transect CO-27 in the Santa Ana reach.

Design Guidelines and Appraisal of Research Results 161 Annual Hydrographs at Albuquerque Gage Date D is ch ar ge (c fs) Figure 4.34. Average annual mean daily hydrograph at Albuquerque gaging station for pre- and post-Cochiti Dam periods. Water Year San Felipe Gage pre-Cochiti Qavg = 9,800 cfs 0 5,000 19 27 19 30 19 33 19 36 19 39 19 42 19 45 19 48 19 51 19 54 19 57 19 60 19 63 19 66 19 69 19 72 19 75 19 78 19 81 19 84 19 87 19 90 19 96 19 93 10,000 15,000 20,000 25,000 30,000 post-Cochiti Qavg = 5,700 cfs 1973 Construction of Cochiti Annual Peak Discharges D is ch ar ge (c fs) Figure 4.35. Annual peak discharges at the San Felipe gage. Ecological Issues Historically, the Middle Rio Grande Valley had one of the highest-value riparian ecosystems in the Southwest (Crawford et al. 1993). However, the existing riparian community in much of the Middle Rio Grande Valley and in the project area specifically, is a result of alteration of the flow regime, drainage for agriculture and development, flood control, channelization and jetty-jack fields, livestock grazing, beaver activity, and the spread of non-native tamarisk (salt cedar) and Russian olive. For example, natural wetlands, which were common, no longer occur within the Santa Ana Reach of the Rio Grande. As the quality and quantity of the fish and wildlife habitat within the Middle Rio Grande Valley has decreased over time, so has its ability to sustain native flora and fauna. Several spe- cies endemic to the valley have been placed on the federal threatened and endangered species list under the Endangered Species Act. Listed species that could potentially occur within the project area include the Rio Grande silvery minnow and Southwestern willow flycatcher. No

162 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures federally-listed plant species are likely to occur within the project area, and none have been detected by USACE and Pueblo biologists. Stakeholder Activities The Pueblo has implemented an ecosystem-based restoration program, designed to reverse the impacts of 60 years of flood and sediment control and channelization projects on the Middle Rio Grande and restore a healthy, functioning Rio Grande ecosystem. The Pueblo has been assisted in implementing their overall restoration plan by several agencies. The Bureau of Indian Affairs and U.S. Environmental Protection Agency provided financial assistance in clearing non-native vegetation for the purpose of fire management and habitat improvement. The USFWS provided funding toward soil, wildlife, and vegetation surveys, and native riparian vegetation plantings. The Albuquerque area Reclamation office provided analysis, design, and construction support under their River Maintenance, Priority Site Program and the USACE Albuquerque District provided similar support under the USACE’s Section 1135 program. In 1996, in response to the hydrologic and ecological problems within the reach, the Pueblo initiated a restoration plan encompassing approximately 1,400 acres of riparian communities adja- cent to the Rio Grande. The Pueblo has discontinued livestock grazing in the area and manages it as a nature preserve. Baseline vegetation, soil, and hydrologic data have been compiled. A mature cottonwood overstory is present throughout approximately one-third of this area. Saltcedar and Russian olive are common understory plants, replacing native vegetation such as cottonwood and coyote willow in many areas. In accordance with their overall restoration plan, the Pueblo has cleared non-native vegetation from nearly 720 acres, leaving large cottonwoods and native shrubs intact. The Pueblo has encouraged natural establishment or specifically revegetated cleared-bosque areas with a suite of native vegetation such as cottonwood and Gooding’s willow, coyote willow, seep-willow, and New Mexico olive. Remediation of nearly 115 acres of saline and sodic soils was accomplished to facilitate successful planting of native grassland vegetation. In addition, the Pueblo has removed 1,600 obsolete (nonfunctional) jetty-jacks from the abandoned floodplain adjacent to the river. Monitoring is being conducted to document the response of plant and wild- life species to the various riparian restoration activities (see Ayres Associates 2006a). Program highlights include: • Creating over 100 acres of riparian wetland habitat, • Restoring the reach of the Rio Grande traversing the Pueblo, • Restoring 1300 acres of cottonwood bosque by clearing saltcedar and Russian olive thickets; and • Restoring native wildlife habitat throughout the Santa Ana Rio Grande bosque. Restoration Program objectives include providing the following benefits to the Pueblo and the entire Middle Rio Grande Valley: • Preserve the bosque for cultural and recreational uses by tribal members and guests; • Reduce risk of wildfire and protect the Pueblo’s residential communities and economic interests; • Preserve water resources by preventing further declines in the groundwater table; • Enhance economic development (e.g., Hyatt Tamaya Resort) and provide employment for tribal members; • Provide habitat for the endangered Rio Grande silvery minnow and the Southwest willow flycatcher; and • Creating openings in the bosque to enhance wildlife habitat for all species utilizing the Rio Grande corridor. Today, resource managers throughout the Middle Rio Grande Valley look to the Pueblo of Santa Ana for guidance on how to restore and manage their riparian lands.

Design Guidelines and Appraisal of Research Results 163 Channel and Stream Bank Stabilization Measures In 1998, the Reclamation investigated routine bank stabilization measures where active bank erosion persistently threatened the riverside levee on the east side of the Rio Grande about 0.5 miles downstream of the Jemez River confluence. Rather than continue long-term main- tenance, a more permanent solution to the problem was sought in coordination with the Pueblo of Santa Ana. Under their River Maintenance Program, Reclamation restored riverine habitat in the two-mile reach near the Jemez River confluence through the creation of a wider operational channel and floodplain, resulting in reduced water velocities, decreased flow depth, increased width-to-depth ratios, and increased sediment deposition. The project consisted of three phases to be implemented over three to five years. In Phase 1 (completed in 2001), Reclamation realigned the river channel to direct flow away from the deteriorating east-side levee bank. Two portions of the former channel were retained as backwater areas, and bioengineered bank stabilization along the new channel alignment was installed. A long, gently sloped riprap grade control structure [referred to locally as Gradient Restoration Facilities (GRFs)] to restore and hold the vertical position of the river bed, including a 500-foot-long fish-passage apron, was installed by Reclamation approximately four miles upstream of the New Mexico Highway 550 bridge (GRF #1). An adjacent overbank area was lowered to facilitate inundation by flows with a return frequency of two to five years. Phases 2 and 3 of the project consisted of planting 45 acres on bank lines, backwater areas, and floodplain zones with coyote willow, black willow, and Rio Grande cottonwood. In 2003 aquatic and riparian habitat restoration work was completed in the vicinity of the Jemez River confluence. In addition, a program of bar modification that included bar lowering and chute channel formation was conducted from 2003 to 2009. In 2005, the USACE (under the Section 1135 Program) and the Pueblo completed construction of two additional GRFs (GRF #2 and GRF #3) approximately 0.9 and 1.9 miles, respectively, down- stream from Reclamation’s GRF structure. The USACE’s structures consist of a perpendicular sheet pile wall extending approximately two feet above the channel bed and a gently sloped, downstream riprap apron approximately 400 ft long. The apron facilitates upstream passage of small native fish including the endangered Rio Grande silvery minnow. Additionally, a 200-ft-long Bed Sill composed of launchable gravel was installed downstream from the GRFs to provide a transition between the stabilized channel and the downstream reach which is expected to continue to degrade. Bioengineered bank stabilization along the new GRF and Bed Sill channel alignments was also installed. Design of River Bed Stabilization Measures As indicated, the Santa Ana/USACE project consists of GRF #2 and GRF #3 and a Bed Sill (Figure 4.36). The following design information is provided in the Design Letter Report by Ayres Associates (2003). The primary function of the GRFs is to provide grade control for the project reach and halt the degradational trend that has resulted in lowered bed elevations. In addition to providing stabilization of the channel bed and banks, the GRFs also result in greater levels of overbank inundation than the preconstruction condition. Based on 2-dimensional hydraulic modeling, the hydraulic conditions created by the GRFs were designed to mimic those found in natural riffles. The river channel within the active floodplain is dynamic and migrates laterally over time. The GRFs and Bed Sill are designed to account for this potential movement. Additionally, the main channel position of the GRFs was widened slightly from an initial feasibility design to better accommodate lateral changes in the channel alignment. After observing the operation of Reclamation’s GRF #1, the wider section appears to allow temporary bar features and a more braided low-flow channel configuration. The wider section allows for some lateral movement

164 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures of the river channel within the GRF itself. Guide banks were added to the design to maintain a favorable upstream channel alignment (Ayres Associates 2003 and USACE 2008). GRF #2. The increase in water levels generated upstream of each GRF will locally increase the frequency and depth of overbank inundation in comparison to existing conditions. To maximize the benefit created by the GRFs, the upstream floodplain surfaces were lowered and reshaped to ensure that the overbanks were inundated at a 2-year recurrence interval. The overbank elevations of each GRF were designed to ensure a smooth transition from upstream floodplains. The active channel at GRF #2 is roughly 150 ft wide. A relatively narrow (125 ft) bar is present along the left edge of the channel and a natural riffle is located just downstream of the site where flow branches into two paths around a large mid-channel bar. The location selected for the GRF allowed for the extension of an existing riffle, which minimized the size of the structure needed to achieve the desired grade control function. Figure 4.37 shows a representative cross section of GRF #2 (looking downstream). The GRF was designed to tie into the left high bank, rather than into the transient bar feature. As such, the template results in a wider cross section than previously existed. During construction, the small bar located along the left bank was excavated and then recreated on top of the GRF. This Figure 4.36. Plan view of the project reach showing the GRF and Bed Sill locations (Ayres Associates 2003).

Design Guidelines and Appraisal of Research Results 165 accommodates some channel movement and allows for the erosion of the bar surface and depo- sition of new material at any location across the active width. The left bank of GRF #2 was tied into the existing channel bank along the bosque and has a 3H:1V slope. The right bank of the GRF has a milder slope as it ties into the large bar surface that comprises the right overbank. Overbank armor continues across the entire floodplain to the ulti- mate right bank along the bosque. Bank revetment (riprap) on both sides of the structure extends vertically to approximately the 2-year water surface elevation. Willow (Salix exigua referred to locally as sandbar willow or coyote willow) plantings have been added above the revetment to stabilize and protect the upper bank VMSE, in this case. FES lifts were used to stabilize the upper-bank slope. GRF #3. GRF #3 is located approximately one mile downstream of GRF #2 (see Figure 4.36) The GRF is centered along an eroding portion of the right bank adjacent to the recreation trail for the Hyatt Tamaya Resort. This location allowed for the realignment and stabilization of the eroding bankline as a part of the GRF design. The left overbank consisted of a large bar surface that extended 2,000 feet downstream. A smaller, sub-bar surface was present along the left bank near the center of the GRF location. Figure 4.38 presents a representative cross section of the GRF #3 design (looking downstream) in the vicinity of the active channel. The template maintained the existing channel width of approximately 175 feet. The right side of GRF #3 is tied into the eroding right bank of the channel and required fill along the bankline to create a more desirable alignment. The slope of the GRF right bank revetment is 3H:1V and extends vertically to approximately the 2-year water surface elevation. Above the top of the riprap, the right bank was stabilized using FES lifts and willow plantings. The left bank of the GRF tied into the existing bar surface. Overbank armor extends across the bar to the toe of the left bank along the bosque. The revetment carries up the left bank to approximately the 2-year water surface elevation and ties into the jetty-jacks located along the bankline. Bed Sill. The Bed Sill is located approximately 1,800 feet downstream of GRF #3 and just above the New Mexico Highway 550 bridge. The location is in a relatively straight reach of the river where the channel width is roughly 200 feet. This location provides for more efficient grade control. Figure 4.37. Representative cross section for GRF #2 (Ayres Associates 2003).

166 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Figure 4.39 presents a representative cross section of the Bed Sill design. The Bed Sill was constructed at a constant elevation across the channel. This required additional excavation during construction but reduced the amount of gravel required to construct the feature. Along the right bank, a rock toe section of riprap was placed to protect the toe of the slope from scour that could lead to lateral erosion of the channel banks. Along the left bank, the slope has been protected with riprap up to approximately the 2-year water surface elevation, with FES lifts and willows providing stabilization above that elevation. Revetment Components The revetment components of the project were based on hydraulic modeling conducted as part of the project. These components include the main channel riprap and overbank armor used to construct the GRFs and the gravel used in the Bed Sill. Figure 4.38. Representative cross sections for GRF #3 (Ayres Associates 2003). Figure 4.39. Representative cross sections for the Bed Sill (Ayres Associates 2003).

Design Guidelines and Appraisal of Research Results 167 GRF Riprap. The riprap components of the GRFs were designed to withstand the hydraulic forces associated with the 100-year flood (22,300 cfs) and lesser events. During a feasibility study, the results of the 2-D modeling were used to determine the location of maximum shear stress acting on the structures. The main channel and overbank sections of each GRF were ana- lyzed separately. These shear stress values were used to size the main channel and overbank riprap, resulting in median sizes (D50) of 1.0 ft and 0.5 ft for the main channel and overbank, respectively. Bed Sill. The Bed Sill was designed to provide grade control by armoring a portion of the channel bed. It consists of a large mass of gravel placed in an excavated trench across the channel. The material was sized to develop into a riffle as the downstream channel degrades. A median rock size of 2 in. was selected to develop a slope of 0.005 ft/ft, but not to wash away during a 100-year flood event. Enough rock is included in the Bed Sill to armor a 400-foot length of the channel and protect against a downstream degradation depth of 2 ft. The 0.005 ft/ft slope is the average slope associated with natural riffles in the vicinity of the project reach and was also the slope used to design the GRFs. Filter and Bedding Layers The design included a gravel filter layer between the GRF and overbank riprap layers and the underlying substrate materials. The main function of this gravel layer is to prevent the loss of fine material from beneath the riprap. For the invert of the channel beneath the GRFs, filtration and providing for the relief of hydrostatic pressures are not concerns. Consequently a bedding layer was used to protect the underlying soils from the potentially erosive velocities in the void spaces of the riprap. Since the magnitude of these forces is difficult to predict and the protection of the main channel revetment is essential, a dual layer of geotextile fabric and gravel was chosen for providing a protective bedding layer (Figure 4.40). Similar to the invert of the channel, a bedding layer was designed to protect the soils under- lying the overbank armor. In contrast to the channel invert, which includes a dual layer of geotextile fabric and gravel, this bedding layer consists simply of gravel. This is due to the fact Figure 4.40. Granular filter placement over geotextile fabric.

168 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures that the overbank velocities and associated shear stresses are less than those in the main channel. The same gravel used to provide the filter on the bank slopes was used as the bedding under the overbank armor. GRF Transitions to Existing Banks The sloping riprap used to stabilize the banks of the GRF extends beyond the limits of the structure to protect the banks up- and downstream and to provide a uniform alignment of flow into and out of the structure. The top elevation of riprap along the banks of the GRF was set to approximately the 2-year water surface elevation to allow integration of bioengineering methods. Above this elevation the bank was recreated with fill material and then planted with willows for stability and habitat enhancement. The same type of bankline revetment that was used in the GRFs was extended to create the up- and downstream transitions. Typical cross sections for these transitions are shown in Figure 4.41 for both the right and left banks of GRF #2 and GRF #3. A launchable riprap toe was included to provide protection against local scour. In this case, a launchable toe was considered appropriate because the GRF/Bed Sill structures provide sig- nificant vertical control and limit toe scour. Accordingly, the launchable toe provided an additional margin of safety, not the first line of defense. Vegetation Planting During the process of analyzing and revising the project design, one objective was to incor- porate, to the extent feasible, areas for planting vegetation and enhancing ecological and habitat values within the project footprint. In addition, the USFWS made a number of recommendations to prevent and reduce adverse project effects on fish and wildlife resources. Of those recommendations, several were geared toward the establishment of vegetation in the project footprint, including: • Backfilling with uncontaminated earth of alluvium suitable for revegetation with indigenous plant species and • Scarifying compacted soils or replacing topsoil and revegetating all disturbed sites with a suitable mixture of native grasses, forbs, and woody shrubs. Proposals for reducing cost that would also enhance environmental aspects of the project included using combinations of hard and soft (biological) revetment features in lieu of simply riprap on the GRF side slopes, and stabilizing non-riprap side slopes with plant material. As a result of these recommendations and proposals, environmentally sensitive bank-protection features and revegetation components were added to various elements of the project as sum- marized below: Bank Protection. The original design called for using riprap to protect up to the top elevation of the banks. This was revised so that the riprap now extends up to an elevation that is 1 ft below the 2-year water surface level. Above this elevation, the bank was created with excess excavation material. At locations where there are jetty-jacks, the jacks were buried or removed and the slope revegetated. Along banks where there are no jetty-jacks, the soils in the first few feet above the top of the riprap were stabilized using FES lifts (Figure 4.42). FES lifts consist of native alluvium wrapped with two layers of biodegradable coconut fabric, and reinforced with high densities of deeply planted willow cuttings. As these plants mature, their root systems provide stabilization to the soils, replacing the structure of the fabric as it deteriorates over time. Guidelines for FES and brush layering similar to those in Section 4.3.2 were followed. • Vegetated bank slopes—Upper-bank slopes that were not appropriate for FES lifts, such as locations with jetty-jacks, were planted with clumps of native willows and other shrub species.

Design Guidelines and Appraisal of Research Results 169 Figure 4.41. Typical sections of bank transitions (Ayres Associates 2003).

170 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Native species were selected with regard to hydrologic regimes and corresponding depths to water table that occur along the slope of the upper bank. Guidelines for pole planting similar to those in Section 4.3.1 were followed. • Vegetation in overbank areas—Willows are a primary component of the guide banks that are part of the GRFs and Bed Sill and help maintain the desired alignment into the structures while still allowing flood flow to access the entire channel width. Vertical willow plantings (Figure 4.43) were also used because they are well adapted to fluctuating water tables, are resis- tant to erosional forces associated with flood flows, and can withstand anticipated deposition of sediments. Installation Construction of both GRFs and the Bed Sill began in July 2004 and was completed by March 2005. Construction of the GRFs and Bed Sill required diversion of the Rio Grande, installation Figure 4.42. View of FES and willow plantings immediately after installation. Figure 4.43. View of willow-planted FES bankline and planted willow poles in overbank area.

Design Guidelines and Appraisal of Research Results 171 of coffer dams, dewatering of the structure sites, and in-channel excavation as well as installation of access roads and staging and stockpiling sites. Pertinent Data. Pertinent data applicable to the design and construction of the GRFs and Bed Sill are summarized in Tables 4.5 through 4.7. This data is categorized according to each project component and includes physical criteria that govern the dimensional parameters of the structures. The front-end specifications for the project components were provided by the USACE Albuquerque District Office. Many of the front-end specs were in “guide-spec” form and needed specific revision and/or tailoring to fit this project. Construction Issues and Problems. A number of problems arose during construction of the GRFs and Bed Sill: • Bank erosion was an ongoing problem along the diversion channels during construction. In most cases, temporary placement of riprap was needed to counter the erosion. • During construction of the GRFs, the initial rock placement within the main channel portion of the structures was not acceptable and several thin spots were noted throughout the struc- tures resulting in exposure of the granular filter material in several locations. The contractor added rock and reworked several sections. • At GRF #2, the contractor was unclear on the riprap transition along the channel banks to the FES lifts. The plans showed an embankment backfill area between the existing bankline and the limit of the riprap. Rather than placing embankment backfill below the rock, the contractor chose to use additional riprap for the fill which was an acceptable alternative (see Figure 4.41). Performance Monitoring and Maintenance. The Operation, Maintenance, Repair, Replacement, and Rehabilitation (OMRR&R) Manual (Ayres Associates 2006b) provided assistance to the Pueblo GRF #2 Physical Characteristics Length of GRF (in-river portion) 400 ft Width of main channel portion Approximately 260 ft Width of overbank portion Approximately 380 ft Thalweg elevation at crest 5059.8 ft, National Geodetic Vertical Datum (NGVD) 29 Thalweg elevation at Toe 5057.8 ft, NGVD 29 Longitudinal slope 0.005 Sheet pile driving depth 12 ft below final grade Length of buried guide bank 720 ft Length of bank transitions 50 to 70 ft Material Quantities Item Unit of Measure Upstream Transitions Main GRF Downstream Transitions Guide Bank Totals Geotextile fabric SY 1,530 12,940 1,540 NA 16,010 Granular filter CY 250 4,550 280 NA 5,080 Riprap, Type 1 CY 1,230 8,660 1,250 NA 11,140 Riprap, Type 2 CY 80 5,040 320 460 5,900 Vinyl sheet pilling SF NA 9,670 NA NA 9,670 FES LF 240 1,604 244 NA 2,088 Type 1 willows LF 240 1,604 244 NA 2,088 Type 2 willows LF NA NA NA 720 720 Type 1 seed mix SF 1,560 10,426 1,586 NA 13,572 Type 2 seed mix SF 1,560 10,426 1,586 NA 13,572 Type 3 seed mix SF 390 2,607 397 NA 3,394 Note: SY = square yard, CY = cubic yard, LF = linear foot, SF = square foot. Table 4.5. Pertinent data for GRF #2, Rio Grande, Santa Ana Reach, New Mexico.

172 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures GRF #3 Physical Characteristics Length of GRF (in-river portion) 400 ft Width of main channel portion Approximately 240 ft Width of overbank portion Approximately 170 ft Thalweg elevation at crest 5054.0 ft, NGVD 29 Thalweg elevation at toe 5052.0 ft, NGVD 29 Longitudinal slope 0.005 Sheet pile driving depth 12 ft below final grade Length of buried guide bank 385 ft Length of bank transitions 50 to 210 ft Material Quantities Item Unit Upstream Transitions Main GRF Downstream Transitions Guide Bank Totals Geotextile fabric SY 1,330 14,780 1,320 - 17,430 Granular filter CY 425 3,050 410 - 3,885 Riprap, Type 1 CY 1,660 8,450 1,290 - 11,400 Riprap, Type 2 CY 500 2,060 170 230 2,960 Vinyl sheet pilling SF - 7,460 - - 7,460 FES LF 1,020 1,600 352 - 2,972 Type 1 willows LF 1,020 1,600 352 - 2,972 Type 2 willows LF - - - 385 385 Type 1 seed mix SF 6,630 10,400 2,288 - 19,318 Type 2 seed mix SF 6,630 10,400 2,288 - 19,318 Type 3 seed mix SF 1,658 2,600 572 - 4,830 Table 4.6. Pertinent data for GRF #3, Rio Grande, Santa Ana Reach, New Mexico. Bed Sill Physical Characteristics Length Bed Sill 280 ft Width of main channel portion Approximately 240 ft Invert elevation 5051.0 ft, NGVD 29 Length of buried guide bank 540 ft Length of buried grade control 140 to 160 ft Material Quantities Item Unit Bed Sill Guide Bank Totals Granular filter CY 90 - 90 Riprap, Type 1 CY 310 - 310 Riprap, Type 2 CY - 570 570 Gravel CY 3,730 - 3,730 FES LF 560 - 560 Type 1 willows LF 560 - 560 Type 2 willows LF - 540 540 Type 1 seed mix SF 3,640 - 3,640 Type 2 seed mix SF 3,640 - 3,640 Type 3 seed mix SF 910 - 910 Table 4.7. Pertinent data for the Bed Sill, Rio Grande, Santa Ana Reach, New Mexico. of Santa Ana in carrying out its obligation for the operation, maintenance, and replacement requirements for the GRFs, Bed Sill, and other features associated with these structures. Semi- annual inspections are being made of all project features and components. Such inspections are made immediately prior to the beginning of the flood season (normally prior to April 1) and after the flood season. Regular inspections consist of visual observations and field surveys as necessary to confirm elevations of various project features and verify the condition of inundated features. Other inspections are made immediately following each major high-water period, and at such intermediate times, as necessary. Immediate steps are taken to correct dangerous conditions disclosed by the inspections. Regular maintenance repair measures are accomplished during

Design Guidelines and Appraisal of Research Results 173 the appropriate season as scheduled by the Pueblo. The Pueblo is required to obtain all permits applicable to any repair work. Maintenance and repair of damaged project components helps to ensure that project benefits are maintained throughout the life of the project. General repair guidelines are described in the OMRR&R (Ayres Associates 2006b). Performance Assessment. Over the past 10 years (2005–2015) the structures have performed as anticipated. The structures are stable and a good growth of willows and other vegetation has naturally colonized the banks and bar surfaces at the structures. Within a year of construction, the willow branches planted between the FES lifts at both GRFs and the Bed Sill had become well established and have survived and thrived since then. Figure 4.44 shows the eroding bankline before treatment, construction of the willow-planted FES over rock riprap, and the well-established willow-planted FES after construction at GRF #3. Figure 4.45 provides an overview of the finished GRF #3 with the vegetated FES along the right bank, planted willow poles and vegetated right overbank area, and the left bank transition guide bank with the planted willows. Conclusions and Lessons Learned Design. In a follow-up value engineering study after completion of the initial design (USACE 2002b), proposals were made for reducing cost that would also enhance environmental aspects of the project. These included using combinations of hard and soft (biological) revetment features in lieu of simply riprap on the GRF side slopes, and stabilizing non-riprap side slopes with plant material. As a result of these recommendations and proposals, environmentally sensitive bank- protection features and revegetation components were added to various elements of the project. Figure 4.44. Before, during, and after views of willow-planted FES over rock riprap at GRF #3. Figure 4.45. View looking upstream of upstream end of GRF #3 with vegetated FES along right bank, planted willow poles and vegetation in right overbank area, and left bank transition guide bank with planted willows.

174 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Construction. During construction, it was noted that materials were either improperly or inadequately placed or that, in some cases, the size or gradation of bedding materials used in a number of locations were not as specified. These were immediately corrected by regular onsite construction inspections and meetings between the contractor, USACE, and the Pueblo. Post-Construction Monitoring/Maintenance. Ongoing work being conducted along the project reach indicates that the structures and associated riparian and wetland restoration activities are stable and appear to require little or no maintenance (Ayres Associates 2012). Once the GRFs were installed, a program of bar modification that included bar lowering and chute channel formation was conducted and completed by 2009. Some of this work has been conducted in close proximity to or at the GRFs and Bed Sill, which may have an impact on the structures in the future. Continued monitoring will be necessary to ensure that the structures and bank protection remain in a stable condition. Restoration of the sediment supplied to the Rio Grande from the Jemez River as a result of opening of the Jemez Canyon Dam and draining the upstream reservoir pool is currently having an impact. As noted during recent sediment sampling efforts, significant amounts of sand and gravel are currently being transported by the Rio Grande in the vicinity of the GRFs and Bed Sill. Although the supply may be insufficient to cause any significant aggradation or induce active channel migration, the reach and structures must continue to be monitored for any instability. Vertical Stability of the Channel. The effort and cost associated with the design and con- struction of the GRFs and Bed Sill for this project underscore the absolute necessity of establish- ing vertical control of the channel before attempting any stream bank protection treatments (see Section 4.2.3 for further discussion). References Ayres Associates, 2003. Design Letter Report, Supporting Final Design and Preparation of Construction Plans and Specifications for the GRFs, Bed Sill, and Associated Vegetation Planting, Rio Grande Riparian and Wetland Restoration, Section 1135 Project, prepared for Pueblo of Santa Ana, New Mexico. Ayres Associates, 2006a. Middle Rio Grande Santa Ana Reach Geomorphology, Technical Appendix A, In: Section 1135 Program, Detailed Project Report and Environmental Assessment for Aquatic Habitat Restoration at Santa Ana Pueblo, New Mexico, U.S. Army Corps of Engineers Albuquerque District. Ayres Associates, 2006b. Operation, Maintenance, Repair, Replacement, and Rehabilitation (OMRR&R) Manual for GRF on the Rio Grande, Pueblo of Santa Ana Reservation, New Mexico, prepared for U.S. Army Corps of Engineers Albuquerque District, Contract No. W912PP-04-C-0003. Ayres Associates, 2012. Geomorphic Assessment of the Rio Grande in the Vicinity of the RM 205.8 Nick Point Site, prepared for Pueblo of Santa Ana Reservation, New Mexico. Crawford, C. S., Cully, A. C., Leutheuser, R., Sifeuntes, M. S., White, L. H., and Wilbur, J. P., 1993. Middle Rio Grande Ecosystem: Bosque Biological Management Plan, U.S. Fish and Wildlife Service, Albuquerque, NM, 291 pp. Grassel, K., 2002. Taking Out the Jacks—Issues of Jetty Jack Removal in the Bosque and River Restoration Planning, Water Resources Program, University of New Mexico, Publication No. WRP-6, 46 pp. Lagasse, P. F., 1980. An Assessment of the Response of the Rio Grande to Dam Construction, Cochiti to Isleta Reach, a technical report by the Science Research Laboratory, U.S. Military Academy, for the U.S. Army Corps of Engineers, Albuquerque District, Albuquerque, NM. Lagasse, P. F., 1981. Geomorphic Response of the Rio Grande to Dam Construction, New Mexico Geological Society, Special Publication No. 10, pp. 27–46. Lane, E. W., and Borland, W. M., 1953. River Bed Scour During Floods, Transactions of the American Society of Civil Engineers, Vol. 119, Paper No. 2712, pp. 1069–1080. Nordin, C. F., and Beverage, J. P., 1965. Sediment Transport in the Rio Grande, New Mexico, USGS Professional Paper 462-F. Ortiz, R. M., 2004. A River In Transition: Geomorphic and Bed Sediment Response to Cochiti Dam on the Middle Rio Grande, Bernalillo to Albuquerque, New Mexico, Master’s Thesis, University of New Mexico, Albuquerque, NM.

Design Guidelines and Appraisal of Research Results 175 Richard, G. A., 2001. Quantification and Prediction of Lateral Channel Adjustments Downstream from Cochiti Dam, Rio Grande, NM, Ph.D. Dissertation, CSU, Fort Collins, CO. Salazar, C. L., 1998. Morphology of the Middle Rio Grande from Cochiti to Bernalillo Bridge, New Mexico, Master’s Thesis, CSU, Fort Collins, CO. Sixta, M. J., 2004. Hydraulic Modeling and Meander Migration of the Middle Rio Grande, New Mexico, Master’s Thesis, CSU, Fort Collins, CO. USACE, 2002a. Middle Rio Grande Santa Ana Reach Geomorphology, Section 1135 Program, Detailed Project Re- port and Environmental Assessment for Riparian and Wetland Restoration, Pueblo of Santa Ana Reservation, New Mexico, U.S. Army Corps of Engineers Albuquerque District. USACE, 2002b. Value Engineering Study Summary Report, Riparian/Wetland Restoration Project, Pueblo of Santa Ana Reservation, New Mexico, U.S. Army Corps of Engineers Albuquerque District. USACE, 2008. Section 1135 Program, Draft Detailed Project Report and Environmental Assessment for Aquatic Habitat Restoration at Santa Ana Pueblo, New Mexico, U.S. Army Corps of Engineers Albuquerque District. Woodson, R. C., 1961. Stabilization of the Middle Rio Grande in New Mexico, Journal of the Waterways and Harbors Division: Proceedings American Society of Civil Engineers, Vol. 87, No. WW4, Paper 2980, pp. 1–15. Woodson, R. C. and Martin, T. J., 1963. The Rio Grande Comprehensive Plan in New Mexico and Its Effects on the River Regimen Through the Middle Valley, Proceedings of the 2nd Federal Interagency Sedimentation Conference, Jackson, MS. 4.4.3 Humid Region Example Problem Identification Malletts Creek drains an 11 mi2 urban watershed in Ann Arbor (Washtenaw County), MI, and is tributary to the Huron River (Figure 4.46). Malletts Creek has been degraded as a result of water quality and flow capacity problems associated with urbanization. Major water quality problems include high phosphorus loading and sedimentation. Prior to this project, water quantity problems Figure 4.46. Malletts Creek watershed.

176 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures were indicated by high-flow velocities that resulted in both bank and channel erosion. Degraded stream beds and frequent and severe peak flows diminished aquatic habitat. Bed materials are domi- nated by cobble, but bank soils are primarily fine sands and sandy loams with little clay that are susceptible to fluvial erosion (Figure 4.47). A Bank Erosion Hazard Index (BEHI) analysis (Rosgen 2001a, Rathburn 2006) indicated that, of 14 channel reaches, five received BEHI ratings of high, four were very high, and two were extreme [Orchard, Hiltz and McCliment, Inc. (OHM) 2011]. Several sites along the creek presented special erosion, flow conveyance or infrastructure problems which were listed in the basis of design technical memorandum (OHM 2011) (Table 4.8). Reconnaissance data presented by OHM (2011) showed that preconstruction bank heights ranged from 3 to 7 ft except in one reach where banks were up to 40 ft high. Several reaches were incised to the point that little channel-floodplain connection occurred. Reach-mean channel widths ranged from 9 to 50 ft and averaged 25 ft. Infrastructure occurs frequently in the stream (a) (b) Figure 4.47. Bank soils and sediments, Malletts Creek: (a) preconstruction bank erosion, Malletts Creek. Photo courtesy of Harry Sheehan, environmental manager, Washtenaw County, MI. and (b) typical bar sediments, Malletts Creek.

Design Guidelines and Appraisal of Research Results 177 corridor in the form of roadway crossings, utility poles, manholes, storm sewer outfalls, and culverts. The primary bank erosion process is fluvial erosion; limited amounts of mass failure occur where the channel has simultaneously incised and impinged against a high bank. Hydromodification. During the past 40 years, the watershed has been extensively developed with shopping malls, new subdivisions, parking lots, etc. (Lawson et al. 2008). About 37% of the land is covered with impervious surfaces (Water Resources Commissioner undated). Reports of channel reconnaissance indicated moderate levels of conditions symptomatic of urbanization. Much of the Huron River watershed was historically characterized by undulating topography characterized by depressions and internally drained valleys of glacial origin. Richards and Brenner (2004), using numerical hydrologic modeling, argued that the effective size of Malletts Creek watershed had been doubled by anthropogenic features such as storm sew- ers and drains. The Richards-Baker Flashiness Index, developed for wadeable Michigan streams, varies from 0 to over 1, where an index of 0 represents unchanging flow rate, and 1 would indi- cate a highly variable flow. An analysis of mean daily discharges (1999–2010) at the Chalmers Road USGS gage yielded a 3-year average (2009–2011) flashiness index value of 0.723, which is higher than 75% of all similarly sized streams in Michigan and well above the median stream flashiness in a six-state midwestern survey (Fongers et al. 2007). This level of hydromodification, in conjunction with the increasing level of impervious land cover, would be expected to trigger the type of channel erosion that has been reported for Malletts Major Problem Area Subreach (SR) Identified Problems Suggested Improvements Problem Area #1 SR 2 1. Massive debris jam 1. Remove debris 2. Bank erosion/split flow creating island around oak tree 2. Create overflow "control" at split flow and stabilize bypass channel Problem Area #2 SR 4 1. Mass wasting at 17-ft-high steep bank with oak trees at risk of falling into stream 1. Vane arms Boulder toe protection Floodplain shelf Slope 1.1 back to top of grade Stabilize slope with live stakes/fabric Cut inside bank and use as fill Observation area with wood fence Interpretive signage Problem Area #3 SR 6 1. Skewed culvert alignment 1. W-Weir upstream of culvert inlet 2. Bank erosion/split flow 2. Create overflow "control” at split flow and stabilize bypass channel 3. Large debris jam 3. Remove debris Problem Area #4 SR 8 1. Flow blocked upstream of one culvert barrel 1. Remove blockage W-Weir upstream of culvert 2. Bank erosion/split flow on opposite bank 2. Create overflow "control" at split flow and stabilize bypass channel Problem Area #5 SR 8 1. Sanitary crossing 1. Grade control/riffle creation 2. 42" storm outfall 2. Grade control/riffle creation 3. Sanitary manhole in channel 3. Grade control/riffle creation Problem Area #6 SR 11, 12, and 13 1. Stream disconnected from floodplain 1. Floodplain storage Trail relocation Interpretive signage Problem Area #7 SR 14 1. Failed culvert end section at Manchester Road 1. Repair failed end section 2. Large scour hole 2. Install pre-formed scour hole Table 4.8. Tabulation of erosion and infrastructure problems along Malletts Creek Channel prior to construction (from OHM 2011).

178 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Creek. Hydraulic modeling of the 100-year event indicated that maximum velocities occurred at bankfull flow, and that these velocities ranged from 2.75 ft/s to 7.75 ft/s, varying by reach. Modelers attributed higher velocities to encroachments on the channel that prevented overflow onto floodplains at lower discharges. Hand calculations and local soil characteristics produced an estimate of critical shear stress of 0.40 lb/ft2, while HEC-RAS produced reach-averaged shear stresses that varied from 0.27 lb/ft2 to 0.73 lb/ft2 and shear stresses at specific locations as great as 1.66 lb/ft2 (OHM 2011). Water Quality and Ecology. Water quality issues are linked to hydromodification. Lawson and Weiker (2008) analyzed water quality from several sites in the Huron River watershed for 2002–2007, including Malletts Creek, and noted the strong relationship between discharge and loads of suspended sediment [total suspended solids (TSS)] and total phosphorus (TP) for Malletts Creek. They also noted excessively high levels of specific conductance (>800 µS) despite generally acceptable pH and dissolved oxygen levels. OHM (2011) described estimation of the pre-project annual TSS erosion rate (tons/yr) and annual TP load based on the Bank Assessment for Non- point Source Consequences of Sediment (BANCS) procedure (Rosgen 2001b) and the Pollutants Controlled Calculation and Documentation for Section 319 Watersheds Training Manual [Michigan Department of Environmental Quality (MDEQ) 1999], respectively. The analysis yielded estimates of 1,283 tons/year of TSS and 1,091 lb/yr of TP yield from all project reaches, total. Data collected by the Huron River Watershed Council and partners at the monitoring station Malletts Creek at Chalmers Drive (2003–2011) yielded an the average TSS concentra- tion of 18.1 mg/L, with a slight declining trend over that period of time (Lawson et al. 2008). Wet weather monitoring resulted in event concentrations of 33 and 222 mg/L or an average of 127.5 mg/L. Steen (2014) summarized data for the period 2003–2012 and noted that mean flow- weighted TP concentrations for Malletts Creek were >0.1 mg/L, well above the target concentration of 0.05 mg/L. Benthic invertebrate and habitat quality assessments conducted in 1999 revealed conditions were fair to poor (Wuycheck 2004) and conditions were unchanged as of 2011 (Middle Huron Partners and Stormwater Advisory Group 2012). Key factors included flashy hydrology and nonpoint source pollution (sediment and organics). Overall ecological status as assessed by an index based on 10 indicators of “stream health” is only 20 on a scale of 0 to 100 (Steen 2014) based on data from 2003–2012. This places Malletts Creek among the lowest-rated subwatersheds in the Huron River watershed. Infrastructure Issues. As of 2004, there were at least 161 storm water outfalls to Malletts Creek watershed covered by the Phase I storm water runoff permit program (Wuycheck 2004). Several outfalls and culverts needed maintenance (Table 4.8). Stakeholder Concerns. Riparian landowners and concerned members of the public were anxious about tree removals required for project construction. A tree monitoring committee was established by the neighborhood association and a “geosocial network” was used to keep track of trees tagged for preservation and those tagged for removal (Arlinghaus and Arlinghaus 2012). Additional concerns included reduction in privacy due to project construction, since the channel borders several residential back yards. Selection of Treatment Types Restoration Objectives. Multiple restoration objectives were developed for both the creek and its watershed (Goldsmith et al. 2014). A public information component was included to inform and involve citizens about proposed restoration activities. Project goals were to address needed structural repairs (e.g., broken culverts and cracked outlet structures), to reduce phosphorus

Design Guidelines and Appraisal of Research Results 179 loading to downstream waters by 50%, to increase habitat quality for fish and wildlife, to control stream velocities, and to stabilize stream banks. Project measures included stream bank stabiliza- tion, channel bed erosion controls, and storm water detention basins. A total length of 8,700 linear feet of stream channel was treated for stream bank erosion. Both conventional and environmentally sensitive measures were selected for bank stabilization. All stabilization was preceded by extensive removal of invasive woody vegetation accompanied by preparatory grading work. The in-channel work featured in this case study was part of a larger, systemic effort throughout the watershed to retain and treat storm water with a range of management practices. Selected Treatments. Selected treatments included bank grading to create surfaces for revegetation and to increase the channel cross-sectional area to decrease high flow velocities. In some cases, narrow, deeply incised channels were graded to produce two-stage (benched) cross sections. Additional conventional measures included stone bank protection and storm water detention basins placed on the floodplain. Within the channel, the project included rock vanes to deflect high velocities from banks and increase physical habitat diversity, stone grade controls (cross vanes) to limit bed degradation and trap sediments, and coir logs placed along bank toes for erosion control. Protection of upper and middle banks featured ECBs and planting of live willow stakes and seeding. Design Design of erosion controls, both stone and vegetative components, was based on orthodox guidance. A computer modeling study was conducted to predict flow rates and flood elevations, as well as phosphorus levels in the creek subsequent to project implementation. Channel. The channel geometry was modified in both the longitudinal and cross-channel directions to improve stream function. The channel width was widened along its bottom to accommodate increased flow; the sides were also graded back to a more stable 2H:1V angle (Figure 4.48a and b). Where space allowed, the stream bank was “stepped” to provide a shallow terrace or bench above the rock toe to accommodate and slow down intermediate flood flows (Figure 4.48e). Riprap. Conventional rock armor (riprap) was widely used on this project as toe protection. Rock armor was used in higher-velocity (>5 ft/s) reaches and along outside bends. There was no attempt to purposely vegetate the rock with live cuttings using either the “joint-planting” or “bent willow pole” method (see Section 4.3.3). The rock selected was relatively rounded and attractive in appearance (Figure 4.48a, c, and d); however, angular rock is generally preferred for riprap. To compensate for the rounded characteristics of the rock, larger rocks were placed at the bottom and inset into the channel to forestall displacement and undermining during high-flow events. Geotextile fabric was placed on the stream bed under the vanes to prevent subsidence (Figure 4.48i). Coir Logs. Coir logs consisting of coir fiber rolls approximately 1 ft in diameter were staked in place at the toe of the stream bank with lumber stakes (Figures 4.48e and f). These were placed either singly or stacked vertically in a double row. Coir logs were used as an alternative to rock armor in lower-velocity reaches of the channel. Coir logs can be vegetated by inserting live stakes or cuttings; they also lend themselves to natural vegetation by volunteer plants. Coir logs were occasionally protected with rock as well. Rock Vanes and Stone Grade Controls. Rock vanes were placed at strategic locations to deflect impinging flow away from the bank. The vanes were oriented in an upstream direction at roughly 30 degree with the bank (Figure 4.48a and b). Cross vanes served a similar purpose but extended from both banks and created a V-upstream when viewed in plan (Figures 4.48c and d). They

a. Graded bank, stone toe, and upstream-angled rock vane shortly after construction. b. Same view as for (a), two years after construction. Note forked tree in right background. c. Graded bank with coir logs at toe and stone cross vane shortly after construction. d. Stone cross vane two years after construction. e. View shortly after construction. Cross section graded to produce bench. Coir log toe protection, rock vanes. f. Close-up view of channel bank with coir log toe two years after construction. Red arrow indicates about 1.5 ft of bed degradation. Figure 4.48. Key components of Malletts Creek Restoration Project during or shortly after construction and two years later.

Design Guidelines and Appraisal of Research Results 181 g. Cuttings planted in coir fiber ECB showing initial growth shortly after planting. h. Cuttings and ECB two years after planting. One cutting appears to be dead, while the other shows minimal growth. i. Construction crew installing filter fabric on graded bank prior to placing stone. Permission from OHM, Inc. j. Reach similar to the one that includes the bank shown in (i), two years later. Figure 4.48. (Continued). deflected the current into the center of the stream. Rock was placed along the bottom of steeper reaches in a manner that resembled “boulder cascades” to limit bed degradation. Scour analysis was used in cross vane design in accordance with Equation TS14B-23 provided by Shields (2007) (OHM 2011). ECBs and Live Stakes. Both straw and coconut fiber ECBs were used to protect stream banks above either the rock armor or coir logs placed at the toe (Figure 4.48e). Live stakes cut largely from phreatic species such as willow (Salix spp) were inserted through the ECBs to provide additional protection and to help reestablish vegetation on stream banks. ECBs were anchored with stakes placed at 4-ft intervals. Geotextile mesh was used in high-velocity zones, while biodegradable coir was used elsewhere. Installation Public Meetings with Stakeholders. Well-publicized meetings were held with stakeholders, and frequent interactions with homeowners occurred during construction.

182 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Mistakes and Learning Curve. Construction was scheduled for winter months to take advan- tage of frozen ground, but an unusually warm, wet winter complicated construction operations due to muddy conditions. One construction fatality occurred early in the project that involved a mishap with heavy equipment and a pedestrian worker. Work proceeded very slowly at first as workers learned techniques. The contractor used a perennial turf grass seed mix for floodplain reseeding in the park rather than the native grass seed mix that was specified. Plant Materials. Plant materials were imported from Pennsylvania. Primary plant materials were live stakes (see Table 4.9). Dozens of balled and burlap gallon-size plantings were used, pri- marily in the park. The warm spring of 2011–2012 caused problems for plant materials, and cold storage was used to maintain dormancy. Microclimate and timing of plantings made a big difference in survival. Costs. The project was completed during the winter of 2011–2012. The total estimated cost of all restoration activities in the watershed including stream bed and bank stabilization was about $19.1 million. The largest cost components were initial structural repairs to streams ($4.8 million) and detention pond improvements ($4.4 million). The estimated cost of stream bank stabilization was $2.1 million (11% of total). Some funding was received from the state Department of Envi- ronmental Equality—Michigan’s NonPoint Source (DEQ NPS) program for control of sediment and phosphorus. Performance Monitoring and Findings. The project was tested by intense rainstorms shortly after con- struction in March 2012. A severe drought occurred during first summer after completion (2012). Discharges since completion have been moderate relative to the four years prior to construction. A 1-year event overflowed banks at Site 3 (see Table 4.8), but overall flooding is not a major issue within this corridor. The 10-year event only floods one residential lawn. The system performed well during and immediately after intense rainstorms shortly after construction in March 2012. This was a severe test for a stream bank protection system that was likely vulnerable at this time since the live cuttings and seedings (under the ECBs) did not have sufficient time by then to grow and establish fully. The coir logs were still visible during the NCHRP Project 24-39 site visit in 2014 and appeared to have performed quite well. However, bed degradation (up to 1.5 ft) was noted along some coir logs (Figure 4.48f). The rock vanes provide diverse aquatic habitat (see Section 4.2.6). In addition sediment has started to collect behind the vanes. The conventional measures (rock armor) and bio-stabilization techniques (coir logs, ECBs, and live staking) have functioned well together. Stones used for bank protection, toe armor, and vanes were very rounded with a D100 of about 24 inches. As of 2014, Scientific Common Quantity Cephalanthus occidentals Buttonbush 1,489 Cormus amomum Silky Dogwood 1,489 Comus sericea (C. stolonifera) Red Osier Dogwood 1,489 Physocarpus opuifolius Ninebark 1,489 Salix discolor Pussy Willow 1,489 Salix exigua ssp interior Sandbar Willow 1,489 Salix sericea Silky Willow 1,489 Sambucus canadensis Elderberry 1,489 Viburunum dentatum Arrow Wood 1,489 Total 13,400 Table 4.9. Species and numbers of live stakes installed in Mallett’s Creek project.

Design Guidelines and Appraisal of Research Results 183 much of this stone had been moved or dislodged, both along stone toes and at the small drop structures. Some local erosion was noted at one site immediately downstream from a protected bank likely due to a poorly designed transition from the protected bank to a convex bank (point bar). A severe drought occurred during first summer after completion (2012), but no irrigation was provided for plantings. During Fall 2012, stakes were replanted at locations where low survival occurred. However, the high density of initial plantings allowed survival as low as 50% to produce acceptable levels of cover (Figure 4.48j). Several examples of poorly growing stakes were noted in the 2014 inspection, perhaps due to infertile soils (Figure 4.48h). There are no beaver in the watershed, which alleviates herbivory of plantings. However, plantings do face strong competition from invasive exotics (buckthorn, Mack’s honeysuckle, Japanese knotweed, black locust). Many of these invasive plants returned quickly from old root masses. In general, not enough time has elapsed since construction to allow assessment of water quality and ecological effects. Post-construction invertebrate sampling (Spring 2014) showed improvement over pre-project conditions. At the time of the NCHRP Project 24-39 inspection in October 2014 (see Section 2.4), water quality appeared marginal, and no fish were seen. However, bed sediments were primarily clean gravel and cobble. Maintenance. To date, the project proper has not required maintenance. Vegetation in adjacent natural areas in parks is maintained by annual prescribed burning, but this practice does not extend into the stream corridor. Poor alignment of a sheet pile outlet structure that controls flow from a detention basin into the channel has required some adjustment due to undesirable sedimentation. Damage Assessment. Jones and Johnson (2015) present a framework for describing the damage status of stream modification projects in constrained settings (see Section 4.2.5, Table 4.2). The framework is based on widely accepted evaluations of physical habitat quality and stream stability. Their approach was adapted for this study. Based on visual observations made in October 2014, the project was rated as “excellent” in categories of stream bank vegetation, bank stability, and infrastructure protection and rated “good” in the categories of structural integrity, and flow obstruction, and sedimentation. Summary of Conclusions and Lessons Learned The overall impact of the project on several ecological stressors has been positive, based on visual evaluations of the channel and frequency of bank erosion. Public support for the project has generally improved as the project has been implemented. Effort invested in communication with stakeholders is well worth the resources required. Installation of vegetative components is complex, and contingency plans for changes in weather (e.g., cold storage for cuttings or irrigation if needed during period of establishment) should be in place. Close inspection is required to ensure plant materials are handled and installed correctly. Fertility of soils should be assessed before planting, especially if banks are re-sloped. Microclimate and timing of plantings is very influential for survival and performance. Invasive exotic plants are very difficult to control in stream corridors. Naturally rounded stone is aesthetically pleasing, but less stable than angular quarry stone. Recovery of stream corridor aesthetics and physical habitat is typically more rapid than improvements in water quality and ecology. Long-term stability of project features may be at risk if the fluvial system responds to bank stabilization and retention of sediment in storm water detention ponds by bed degradation.

184 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures References Arlinghaus, D. E. and Arlinghaus, S. L., 2012. Geosocial Networking: A Case from Ann Arbor, Michigan, Solstice: An Electronic Journal of Geography and Mathematics, 23(1), http://deepblue.lib.umich.edu/ handle/2027.42/58219 (accessed April 14, 2015). Fongers, D., Manning, K. and Rathbun, J., 2007. Application of the Richards’ Baker Flashiness Index to Gaged Michigan Rivers and Streams, Michigan Department of Environmental Quality, Lansing, MI. Goldsmith, W., Gray, D. H., and McCullah, J., 2014. Bioengineering Case Studies: Sustainable Stream Bank and Slope Stabilization, Springer Verlag, New York, NY. Jones, C. J. and Johnson, P. A., 2015. Describing Damage to Stream Modification Projects in Constrained Settings, Journal of the American Water Resources Association, 51(1): 251–262. Lawson, R., Bouma, D., Olsson, K., Rubin, L., and Powell, E., 2008. Watershed Management Plan for the Huron River in the Ann Arbor—Ypsilanti Metropolitan Area, prepared on behalf of and with funding support from Janis A. Bobrin, Washtenaw County Drain Commissioner, Huron River Watershed Council, Ann Arbor, MI. Lawson, R. and Weiker, D., 2008. Middle Huron Stream Nutrient Monitoring Program: Analysis of Results: 2002–2007, Huron River Watershed Council, Ann Arbor, MI. MDEQ, 1999. Pollutants Controlled Calculation and Documentation for Section 319 Watershed Training Manual, Lansing, MI. Middle Huron Partners and Stormwater Advisory Group, 2012. Total Suspended Solids Reduction Implementation Plan for Malletts Creek, October 2011–September 2016, For the Purpose of Achieving the Total Maximum Daily Load (TMDL) and Removing the Biota Impairment of Malletts Creek, Huron River Watershed Council, Ann Arbor, MI. Orchard, Hiltz, and McCliment, Inc., 2011. Malletts Creek Stream-Bank Stabilization Project, Washtenaw Water Resources Commissioner’s Office, City of Ann Arbor, MI. Rathburn, J., 2006. Standard Operating Procedure: Assessing Bank Erosion Potential Using Rosgen’s BEHI, Michigan Department of Natural Resources and Environment, Nonpoint Source Pollution Unit, Lansing. Available: www.michigan.gov/deq/1, 1607, 7–135. Richards, P. L. and Brenner, A. J., 2004. Delineating source areas for runoff in depressional landscapes: Implications for hydrologic modeling, Journal of Great Lakes Research, 30(1), 9–21. Rosgen, D. L., 2001a. A Practical Method of Computing Streambank Erosion Rate, In Proceedings of the Seventh Federal Interagency Sedimentation Conference, Vol. 2, No. 2, pp. 9–15. Rosgen, D. L., 2001b. Watershed Assessment of River Stability and Sediment Supply (WARSSS), Wildland Hydrology, Fort Collins, CO. Shields, F. D., Jr., 2007. Scour calculations, Technical Supplement 14B in Stream Restoration Design, National Engineering Handbook Part 654, USDA-NRCS, Washington, D.C., CD-ROM. 2007. Steen, P., 2014. The State of the Huron’s River and Tributaries, The State of the Huron Conference. Huron River Watershed Council, Ann Arbor, MI. http://www.hrwc.org/wp-content/uploads/2014/05/Steen_Stateof Affairs.pdf (accessed April 13, 2015). Water Resources Commissioner, Undated. Malletts Creek Restoration Project, Final Report, Washtenaw County Water Resources Commissioner, Washtenaw County, MI. http://www.ewashtenaw.org/government/drain_ commissioner/project-status/malletts_creek/dc_mc_mcrp.html (accessed October 8, 2014). Wuycheck, J., 2004. Total Maximum Daily Load for Biota for Malletts Creek, Washtenaw County, Michigan Depart- ment of Environmental Quality, Water Division, Lansing. 4.5 Appraisal of Results 4.5.1 Advances in the State of Practice In 2005, the results of NCHRP Project 24-19 were published as NCHRP Report 544: Envi- ronmentally Sensitive Channel- and Bank-Protection Measures (McCullah and Gray 2005). After conducting an extensive literature review and evaluation of commonly used environmentally sensitive techniques, McCullah and Gray identified 44 techniques for study. The channel- and bank-protection techniques were grouped into four major categories, including (1) river training techniques, (2) bank armor and protection, (3) riparian buffer and river corridor treatments, and (4) slope stabilization. Technique descriptions and guidelines for their application were developed. The work by McCullah and Gray (2005) served as a starting point and foundation for the present study (NCHRP Project 24-39). Given the objectives of the present study (see Section 1.1.2), which included assessing and eval- uating existing guidelines and development of expanded guidelines for selected environmentally

Design Guidelines and Appraisal of Research Results 185 sensitive stream bank stabilization and protection measures, 16 treatments from the river train- ing and bank armor and protection categories of NCHRP Report 544 were selected for further consideration (see Appendix A for a summary of issues related to hydraulic loading, common reasons/circumstances for failure, and research opportunities for these 16 treatments as identi- fied in NCHRP Report 544). In selecting two of these treatments for rigorous hydraulic flume testing, consideration was given to treatments or treatment components that would have the widest applicability nation- wide. One treatment included a stone (riprap) toe, live siltation, and live staking. The hydraulic data from testing this treatment has broad applicability to either multi-component systems or treatments using just individual components (i.e., using live siltation or willow staking, alone) for stream bank protection. Testing a multi-component system also provided insight on the hydraulic response of the transitions between treatment types having different roughness char- acteristics. The second treatment (VMSE without a hard toe) recognized that many resource agencies tend to discourage, and in some cases prohibit, the use of rock in stream bank protec- tion. Thus, testing the hydraulic limits of a treatment without a hard toe but incorporating FES lifts with live brush layering between the lifts could have wide potential applicability and interest. The laboratory testing task of this project represents a technological breakthrough in develop- ing quantitative hydraulic engineering design guidance for environmentally sensitive treatments. The in-channel (not stream bank) roughness characteristics of living plants have been tested (by others) previously in a laboratory flume, primarily to investigate the influence of vegetation on roughness characteristics under varying flow conditions. However, the testing of the vegetative components of selected stream bank treatments following recommended design criteria, includ- ing fabricating the structural component(s), planting the vegetative component(s) and growing them to maturity, and, finally, moving them to a flume for fully instrumented hydraulic testing at prototype scale under a range of flow conditions represents a “first” in the development of quantitative guidance for environmentally sensitive bank-protection treatments. The detailed results of the laboratory testing task are presented in a standalone format in Chapter 3. Of the two treatments tested in the laboratory, one (live siltation and live staking with a stone toe) met or exceeded all performance expectations. The second treatment (VMSE without a hard toe) exhibited vulnerabilities to damage and soil loss under the same conditions of discharge and longitudinal slope. The laboratory testing task did not represent the only contribution of this study. A synthesis of the current state of practice for environmentally sensitive treatments based on both a literature review and survey of practitioners was completed. The synthesis updated the 2005 findings from the NCHRP Report 544 (McCullah and Gray 2005). Specifically, the literature review concentrated on work completed between 2005 and 2014. While advances have been made in the past 10 years in developing treatment-specific guidance documents for environmentally sensitive bank pro- tection, hydraulic design criteria are still scarce and with few exceptions rely on the literature that was summarized in NCHRP Project 24-19. The available hydraulic criteria were drawn from a variety of sources and varied in quality from qualitative anecdotal rules of thumb to isolated spot measurements of velocity. In this regard, the treatments tested under NCHRP Project 24-39 were designed to respond to specific hydraulic research needs identified by McCullah and Gray in 2005, including: • Live siltation—Research into velocities that this technique can withstand would be helpful. • Live staking—Studies would be valuable regarding the effect live staking has on increasing the ability of other measures to withstand higher velocities and shear stresses. • VMSE—Some uncertainty exists at present as to the exact permissible shear stresses and velocities for VMSE interfaces. Additional research would also be helpful on the nature of the interaction between roots and fabric.

186 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Sixteen site visits to existing field installations of a variety of treatment types in three geographic regions (Southeast, upper Midwest, and the West Coast) were also completed. A highly detailed site visit data form was developed and completed for each site and significant additional effort was applied to gathering design and monitoring information, reports and specifications, cost data, and photographic documentation for each site. Additional effort was devoted to obtaining hydrologic and hydraulic data that supported the design and influenced the level of functionality achieved at each site. While much of this information was qualitative and anecdotal, observations from each site visit provided insight on best management practices, and, in several cases, failure mechanisms and lessons learned that will improve the state of practice. Moreover, the site visit reports and data obtained were assembled into a Compendium to provide easy access for the practitioner to a wealth of experiential information on environmentally sensitive bank-protection treatments. The Compendium is presented in a searchable database format to permit the practi- tioner to employ SQL in searching the database. General hydrologic, hydraulic, and geomorphic considerations and site-specific physical processes that influence the design, installation, and monitoring of environmentally sensitive stream bank protection treatments were presented in a standalone format in this chapter. The site-specific physical process topics include bankfull discharge and conveyance; assessing the stage of evolution for incising channels; analyzing aggradation, degradation and lateral channel stability; predicting meander migration; guidance for protecting the upstream and down- stream “flanks” and “toe” of a stream bank treatment; and estimating toe down requirements and hydraulic stress on a bendway. Geotechnical considerations, guidance for monitoring the success of the vegetative component(s) of environmentally sensitive treatments, and aquatic habitat issues that can influence the design and installation of any treatment were also discussed in this chapter. Detailed, updated design guidelines for three widely used treatments were the focus of this chapter. These include the two treatments tested under Task 7 of this study and an overview and updated guidance for vegetated riprap. These design guidelines were presented in a format that addressed the specific topics listed with the objectives statement of the NCHRP research problem statement and included: • Purpose and advantages (selection), • Hydrologic and hydraulic design parameters, • Materials and equipment, • Construction and installation, • Cost, • Maintenance and monitoring, and • Common reasons for failure. In addition, two detailed case studies of the application of environmentally sensitive stream bank protection measures employed in conjunction with stream channel restoration projects were presented in this chapter. One example involved the application of environmentally sensitive techniques on an arid region perennial river. The second example deals with a smaller stream in a humid region with significant infrastructure issues. The examples illustrated the integration of hydraulic engineering analysis and design with the multidisciplinary approach necessary to achieve project success and included, by example, additional guideline topics such as hydrologic and hydraulic design parameters, performance and longevity, and ecological issues. For the engineer involved in the multidisciplinary design of an environmentally sensitive treat- ment, this chapter also provided current guidance from FHWA on the use of biotechnical treatments in proximity to transportation infrastructure. In addition, for the PE on a design team, aspects of professional liability in environmentally sensitive design were explored.

Design Guidelines and Appraisal of Research Results 187 The following sections highlight relevant observations that contribute to advancing the current state of practice for environmentally sensitive design based on the survey of practitioners and the field site visits conducted under this study. 4.5.2 Observations from the Survey of Practitioners Under Task 2 a survey form was prepared, approved by the NCHRP project panel, and dis- tributed by email to 319 individuals representing all 50 state highway agencies (DOTs) and other federal, state, and local government agencies. Selected Native American nations, consultants, and academic institutions involved in stream bank protection measures were also included in the survey outreach effort. The purpose of the survey was to gather information on the current state of prac- tice with respect to a variety of topics related to environmentally sensitive stream bank protection and included detailed performance-related issues/questions for 16 specific protection treatments (see Section 2.3 and Appendix A). A copy of the survey instrument is included at Appendix B. The survey was constructed to elicit specific information regarding the relative frequency of use of 16 specific environmentally sensitive bank-protection treatments, and their performance. By far, the most widely used was live staking (27 out of 35 responses). The survey responses indicated that vegetated riprap (17 responses) and rootwad revetments (16 responses) were the second and third most common treatments in current practice. While rootwad revetments were not investigated under this research project, live staking was a component of “Tray 1” tested in the hydraulic flume under Task 7 (see Chapter 3 and design guidelines in Section 4.3.1). Updated design guidelines for vegetated riprap were provided in Section 4.3.3. The survey revealed that in general, environmentally sensitive bank-protection measures typi- cally result in satisfactory to good performance at most installation sites. The two most common causes of substandard performance were drought and erosion of the bank slope. These factors were a recurring theme across nearly all of the 16 bank-protection measures listed in the survey. Protection measures incorporating live cuttings tended to indicate that improper plant selection or improper installation resulted in poor performance. Protection measures that incorporated a “hard” or “engineered” component such as rootwad revetment, riprap, articulated blocks, gabions, etc. exhibited failure modes associated with toe or edge scour, undermining, flanking, or geotechnical slope instability. For the treatments (or treatment components) tested under Task 7 and the three treatments for which detailed design guidelines were given in Section 4.3, the following specific comments were received: • Live staking—Performance ranged from “failure” to “excellent.” Most respondents indicated experience with multiple sites and reported generally satisfactory performance. When mortality was indicated, the cause was typically lack of moisture, poor soil conditions, poor quality of cuttings, or a combination of these factors. California respondents indicated that they do not use this as a standalone measure but incorporate a hard armor toe. • Live siltation—Generally satisfactory performance, but not much usage reported. • VMSE—Generally good performance reported by nine respondents. Montana reported some poor performance due to improper installation, while California and Washington, D.C., noted both slope erosion and drought as factors leading to substandard performance. • Live brush layering—Seven respondents indicated experience with multiple sites and reported generally good performance. Montana reported some poor performance due to improper installation, and California noted both slope erosion and drought as factors leading to instances of substandard performance. • Vegetated Riprap—Generally good performance. Slope erosion and drought were the most commonly reported causes of substandard performance.

188 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures 4.5.3 Observations and Lessons Learned from the Field Site Visits Introduction Under Task 6, field investigations were conducted at 16 sites where a variety of environmentally sensitive treatments had been installed and monitored. The sites were grouped in three specific regions as follows: • Southeast (Mississippi)—5 sites (Figure 2.3), • Upper Midwest (Ann Arbor, Michigan)—5 sites (Figure 2.4), and • West Coast (Northern California)—6 sites (Figure 2.5). Priority was given to visit sites where measurements have been made and where those measure- ments can be correlated to original design intent and hydrologic/hydraulic history. The final list of sites included a diversity of stream bank protection measures and geographic locations. The following sections summarize observations from the site visits as they relate to the treatments tested under Task 7 and the three treatments for which detailed design guidelines were given in Sections 4.3.1 through 4.3.3. Selected Observations and Lessons Learned from Site Visits Observations on the installation and performance of the components of “Tray 1” (live siltation and live staking with rock toe—see Chapter 3 and Figure 3.3) are summarized below. Six sites included live staking, three sites included live siltation, and four sites included a stone toe. These treatment components were not necessarily installed together in the combination tested with “Tray 1” under Task 7. An additional three sites (all in California) included a riprap bench or berm as a variation of the stone-toe theme. Live staking: • Willow stakes with split or broken ends will invariably die. Split or broken ends should be cut off. • Willow stakes do poorly when planted in cohesive soils. For success, plant willow stakes in sandy, noncohesive soils that are moist but well drained. If they are not planted deeply, however, the willows will float away or scour out when inundated. • High mortality can be expected if roots are not adequate to anchor plants against buoyant and drag forces during high flows, especially when the plants trap leaves and other debris. • Competition by invasive exotic woody species (e.g., buckhorn, European alder, honeysuckle, black locust) should be anticipated in the monitoring and maintenance plan. • Live staking and other vegetative treatments struggle where soils used to fill an eroded slope are droughty, and some replanting may be necessary. If chimney drain systems are included in the treatment design, they may also depress soil moisture in the root zone. • Straw, curled wood excelsior, and coconut fiber ECBs can be used to protect stream banks above either a riprap armor or coir logs placed at the toe. Live stakes cut largely from phreatic species such as willow (Salix ssp) can be inserted through the ECBs to provide additional protection and to help reestablish vegetation on stream banks. ECBs can be effectively anchored with live stakes placed at 4-ft intervals. Geotextile mesh can be used in high-velocity zones where necessary. • In a humid region, a wide variety of woody species are available to be planted as live stakes. • At the Guadalupe River site in Northern California, it was observed that plantings should be performed with minimal disturbance to the geotextile blanket (holes cut for the vegetation were quite large and could provide a potential point of undermining). The project emphasized the importance of installing geotextile blankets parallel to the flow of water within channels where inundation will occur.

Design Guidelines and Appraisal of Research Results 189 Live siltation: • As a biotechnical practice live siltation can be self-mitigating as it establishes into a row of riparian shrubs. At the Alamitos Creek site in Northern California, cuttings of Baccharis (mule fat) were also used successfully in lieu of willow. • Live siltation can produce a vigorous riparian strip of vegetation (e.g., at the Russian River site in Northern California, where live siltation resulted in a 15- to 20-ft side band of shrubs greater than 12 ft tall with 100% coverage). • At the Russian River site in Northern California, live siltation experienced high flows (11,000 cfs) with 9 ft of flooding over nearby structures (rock vanes). Performance during the 4-year period (2010–2014) after installation was considered excellent. During flooding, the willows attenuated flow velocities and helped to trap and hold sediments on the flood terrace. • At the Guadalupe River site in Northern California, live siltation willow cuttings were staked into the soil and rock above and within log structures at the toe of the bank. Deep, voided areas from previous erosion allowed the stakes to be planted as fill was placed without dig- ging. Willows proved to be capable of withstanding periodic inundation and high-velocity flows, and helped anchor the toe with roots and provide shading and habitat for the treated reach. Rock toe: • Broken concrete should not be used for stone toe in lieu of quarry stone riprap. This material is undesirable because it is not self-launching and can create stream-side hazards particularly when it contains rebar (see Figure 2.6). • At the Goodwin Creek site in northern Mississippi the longitudinal stone toe was completely covered by deposition and excellent vegetation seven years after installation (2007–2014). • At the Hotophia Creek site in northern Mississippi both a stone toe and stone groins (similar to bendway weirs) were installed with willow posts and poles in the upper bank. Fish habitat was monitored for one year before and four years after construction and again 10 to 11 years after construction. Fish numbers, biomass, and species richness increased sharply following construction and these changes persisted over ten years. Inspection in 2014 found the water clear, low current velocity, depths 1 to 3 ft, and numerous fish. Overhanging vegetation at this site provides shade over water. • At the Malletts Creek site near Ann Arbor, Michigan, conventional rock armor (riprap) was used as toe protection. Rock armor was used in higher-velocity (>5 ft/s) reaches and along outside bends. There was no attempt to purposely vegetate the rock with live cuttings using either the “joint-planting” or “bent willow pole” method (see Section 4.3.3). The rock selected was relatively rounded and attractive in appearance. In addition, larger rocks were placed at the bottom and inset into the channel in an attempt to forestall displacement and undermin- ing during high-flow events (see Figure 4.48). The combination of conventional measures (rock armor) and bio-stabilization techniques (coir logs, ECBs, and live staking) at this site has functioned well together. • At several sites on the LAR near Sacramento, California, a unique “stone toe” configuration was installed. Based on hydraulic modeling and slope stability analysis, several mitigation measures were incorporated into project designs. These included a variety of surfaces capable of supporting vegetation; low riverside berms (small constructed floodplains) with varying berm- surface elevations and shoreline configuration; and woody materials submerged in constructed embayments or smaller bank scallops (see Figures 2.9–2.11). Native woody and herbaceous riparian vegetation was planted on the surfaces with the goal of creating a self-sustaining, mixed- canopy riparian forest and riparian scrub habitat. In the 15 years (1999–2014) since project con- struction, this “rock toe” configuration has been monitored on an annual basis and continues to perform as designed.

190 Evaluation and Assessment of Environmentally Sensitive Stream Bank Protection Measures Observations on the installation and performance of “Tray 2” (VMSE without hard toe—see Chapter 3 and Figure 3.4) are summarized below. Two sites included VMSE (specifically, FES lifts) with vegetation inserted in a live brush layering configuration. VMSE: • At the Buttahatchee River site northeast of Columbus, Mississippi, VMSE was installed to pro- tect the secondary upper-bank slope. In addition, rock “refusals” (stone-filled trenches running upslope perpendicular to the channel) were placed at the upstream and downstream ends of the treatment at this site to prevent flanking. In general, plantings at this project may have performed better in more fertile, less droughty soil. It was concluded that it would have been advisable to stockpile the topsoil stripped from the stream bank and reuse it before planting. • A VMSE retaining structure with a 1H:1V sloping face was selected to support a roadway parallel to the stream at the Huron River (Nichols Drive) site near Ann Arbor, Michigan. A rock footing was placed in an excavated trench beneath the VMSE structure. At this site a polymeric geotextile was used for the VMSE. A temporary sheet pile wall was constructed in the river channel running parallel to the bank toe prior to VMSE construction. This allowed for dewatering and excavation of a trench at the bank toe that was filled with stone as foundation support for the VMSE bank retaining structure. • To control roadway runoff at the Huron River (Nichols Drive) site, a sloping drainage course (chimney drain) consisting of a polymeric blanket drain with a high in-plane hydraulic con- ductivity was placed behind the VMSE retaining structure. This drainage blanket intercepts seepage and conveys it to a subdrain system that discharges directly to the river. This measure was an attempt to minimize undesirable saturation of the fill beneath the roadway and of the fill in the VMSE (see potential negative consequences, below). • Based on experience at the Huron River (Nichols Drive) site, it was concluded that close supervision is needed to correctly install chimney drains and VMSE. Also, plant materials used in the VMSE at this site were not dormant when planted. The cuttings were held too long in stagnant water in plastic barrels and inserted into the ground well after the end of dormancy. In addition, the stakes were not buried deep enough. As a result, vegetation has struggled and some replanting has occurred at this site. Contributing to the problem, soils used to fill the eroded slope were droughty, and soil moisture in the root zone may be depressed by the chimney drain system. Poor survival rates (<30%) have required replanting. Grass seeding, live staking, and woody transplants were used in addition to the live, cut branches placed between horizontal lifts in the VMSE structure at this site. The grass seed mix was placed by hydroseeding the top of the stream bank and also directly on the face of the VMSE structure. • During the 2014 visit to the Huron River (Nichols Drive) site it was noted that the site looks visually attractive as observed by the public who use the road frequently to access the Arboretum. However, the vegetation is very sparse and small relative to the vegetation at the River Landing (Nichols Arboretum) site just downstream. Closer inspection during the site visit revealed that the polymeric geotextile used for the VMSE was visible at the soil surface, and it was unsightly. While vegetated riprap was not tested during Task 7, a design guideline overview for this common treatment was provided in Section 4.3.3. Five sites visited during Task 6 included vegetated riprap as a component of the treatment installed. Observations of the installation and performance at several of these sites are summarized below. Vegetated riprap: • At the Huron River (Nichols Arboretum/River Landing) site near Ann Arbor, Michigan, the bank toe was armored with riprap (see Figure 2.6). The riprap was vegetated using the joint-planting method (see Figure 4.30), i.e., live stakes representing 6 to 8 woody tree and shrub species were inserted between openings or interstices in the riprap (sandbar willow, red osier dogwood, ninebark, silky dogwood, native roses). Not all stakes were dormant when planted.

Design Guidelines and Appraisal of Research Results 191 • At the Nichols Arboretum site the live stakes in the riprap have not fared well. Many stakes were damaged when driven through the rock blanket to plant them in interstices. A year after installation only about 30% of the live stakes survived and sent out shoots. The willow bundle method or bent willow pole, which eliminates many of the problems associated with the joint-planting method (see Figures 4.27 and 4.28), should be considered in the future. Red osier dogwood was the most successful species planted, and quite prolific when the site was inspected in 2014. The riprap toe has survived several high-water events but has not been tested by floods approaching peak flows of record. • At several sites on the LAR near Sacramento, California, vegetated riprap was installed on multiple benches for habitat enhancement. Pre- and post-project hydraulic modeling showed no impact on flood capacity and indicated that the lower benches would be submerged for specific environmental flows. At these sites reducing beaver pruning (herbivory) of planted trees (in particular cottonwood) and shrubs on the berm face and low berm was the greatest challenge to meeting shaded riparian habitat goals. Older beaver fence became brittle, providing easier access to the site by beavers. During 2010, linear beaver fence maintenance, and cage installation and relocation continued to be an important management action to minimize beaver damage to vegetation. This strategy proved successful and limited further tree losses or damage at these sites, as well as other sites where cages were previously installed. The observations in this section were taken, primarily, from the field data forms from the site visits performed under Task 6. For more detail and site-specific observations refer to the Com- pendium that accompanies this report.