Two-stage channel systems: Part 1, a practical approach for sizing agricultural ditches

Two-stage channel systems: Part 1, a practical approach for sizing agricultural ditches G.E. Powell, A.D. Ward, D.E. Mecklenburg, and A.D. Jayakaran

Key words: bankfull—best management practice—drainage—floodplain—geomorphology Rural watersheds in the Midwest region of the United States are dominated by agricultural land uses that often incorporate subsurface drainage improvements. Subsurface drainage systems discharge into headwater channels that have been deepened and straightened to facilitate the flow of water from drainage outlets and to lessen flooding of agricultural fields (figure 1A). Often, these modified headwater systems exhibit geomorphic features such as a main channel with a series of bars and benches (Landwehr and Rhoads 2003; Jayakaran et al. 2005). However, rarely do these systems exhibit out-of-bank flows onto a broad floodplain and, in many cases, when stable benches form they are periodically removed by human maintenance activities (figure 1B). In contrast to agricultural ditches and modified rural headwater channels, natural streams and rivers in the region often have a main channel and an active floodplain that is flooded several times or more annually. Bankfull dimensions of these natural streams are sized and maintained by fluvial processes

associated with channel-forming discharge concepts (Powell et al. 2006a). The main channel concentrates frequent low flow events, thus reducing sediment disposition. For larger flow events that spill out of the main bankfull channel, energy is dissipated, and there is a balance between aggradation and degradation. Simon (1989) and Rosgen (1996) describe evolution cycles of stream systems that outline how unstable systems might adjust to achieve equilibrium. Their evolution models illustrate that incised and over-wide streams are unstable channel systems. In the Midwest region, constructed agricultural drainage systems are usually incised and often overwide. Fluvial processes attempt to adjust the unstable channel shape to achieve an equilibrium state. As shown in figure 1A, in the lower part of the cross-section, vertical accretion processes have formed a smaller inset channel by building small floodplains within the confines of the ditch (henceforth called benches). On many ditches, channel maintenance work is conducted periodically in response

Steps in Sizing a Two-Stage Channel System Two-stage channel sizing, construction, and assessment procedures include nine steps: (1) project identification, (2) data collection, (3) data analysis, (4) hydrologic evaluation, (5) conceptual channel sizing, (6) project assessment, (7) final sizing and design, (8) construction, and (9) monitoring and performance evaluation. Step 1: Problem Identification. The initial step for any design project is to evaluate the situation and identify problems and potential solutions. Identifying channel problems involves evaluating physical, hydrological, ecological, and chemical aspects of the channel and watershed. Streams are part of a complex system. Failure, recovery, and sustaining dynamic equilibrium might depend

Copyright © 2007 Soil and Water Conservation Society. All rights reserved. Journal of Soil and Water Conservation 62(4):277-286 www.swcs.org

Abstract: Outlined is a practical approach to size and modify agricultural drainage channels to two-stage geometry to maintain drainage function and capacity while increasing channel stability. Two-stage channel systems consist of an inset channel and small floodplain (benches) within the ditch confines. The two-stage channel sizing procedure includes nine steps: (1) project identification; (2) data collection; (3) data analysis; (4) hydrologic evaluation; (5) conceptual channel system sizing; (6) project assessment; (7) design and/or final sizing; (8) construction; and (9) monitoring and assessment of performance. Channel width and depth dimensions are determined based on a weight-of-evidence approach that considers geomorphology measurements at the project site and throughout the watershed. The authors have developed spreadsheet tools to aid in evaluating the geomorphology of one and twostage channels. Constructing a two-stage channel requires more excavation than traditional ditch maintenance, but benefits include improved conveyance capacity, a channel geometry that will be more self-sustaining, and improvement to in-stream habitat.

to concerns that subsurface drainage function is diminished due to a reduction in outlet depth or conveyance capacity. This work includes removal of woody vegetation, weeds, and deposited sediment, and channel erosion stabilization associated with bank failures and toe scour (Fausey et al. 1982). Rather than removing stable benches to improve conveyance capacity, an approach has been developed that widens the top portion of the cross-section to provide larger benches. The approach for sizing two-stage systems (figure 2) consists of: (1) an inset channel to convey the bankfull discharge, (2) a floodplain for the inset channel, and (3) sufficient capacity above the benches to reduce the likelihood that flow will overtop the ditch banks and flood surrounding land. The primary intent for establishing a two-stage geometry is to have a stable ditch system working in harmony with natural fluvial processes so that sediment transport is in balance. The two-stage form is more likely to be self-sustaining, and water quality and ecology might improve (Powell 2004; Powell 2006a).

George E. Powell is an engineer for Brockway Engineering in Twin Falls, Idaho. Andrew D. Ward is a professor in the Department of Food, Agricultural and Biological Engineering at the Ohio State University in Columbus, Ohio. Daniel E. Mecklenburg is an ecological engineer for the Soil and Water Conservation Division of the Ohio Department of Natural Resources in Columbus, Ohio. Anand D. Jayakaran is an assistant professor at Clemson University, Georgetown, South Carolina.

JULY | AUGUST 2007    volume 62, number 4

277

Figure 1 A Minnesota agricultural channel (A) before and (B) after reconstruction maintenance.

A

on upstream, downstream, and/or watershed and stream characteristics, as well as factors in the vicinity of the point or reach of interest. Examples of agricultural channel failures include bank instability, cut banks, sediment deposition, restricted drainage outlets, inad-

278

journal of soil and water conservation

equate subsurface drainage, and insufficient capacity (Fausey et al. 1982). In both natural and modified channel systems, a response is made to real or perceived problems that the system is causing humans. Often the channel system is sim-

Copyright © 2007 Soil and Water Conservation Society. All rights reserved. Journal of Soil and Water Conservation 62(4):277-286 www.swcs.org

B

ply making a natural adjustment, such as meander migration or building benches, that aids recovery or self-sustainability. In these cases, human “solutions” that do not consider geomorphology are likely to be temporary and detrimental to the channel system. Inadequate knowledge of channel equilibrium prior to any modification, such as cleanout or construction of an alternative geometry, increases the risk of failure. Consideration must be given to land use changes on the watershed. Land use changes, such as urbanization or conversions of forests to agriculture, will usually increase peak discharges, the frequency of discharges, and the volume of runoff (Ward and Trimble 2003). Without adequate space to adjust, the inset channel will tend to down-cut, widen, and/ or change its meander pattern. A detailed hydrology discussion is presented in step 5. Bank stability should be analyzed through geotechnical engineering approaches. Simon and Langendoen (2006) have developed a useful, Microsoft Excel-based, spreadsheet tool to estimate the stability of channel banks and associated factors of safety. Step 2: Data Collection. Following project identification, geomorphic data should be collected at the project site, at a reference site if possible, and throughout the watershed. A detailed study of the project site is always required. Measurements throughout the watershed provide information regarding the channel dimensions based on drainage area. It is recommended that channel geomorphology measurements be made using procedures that are consistent with those presented by Harrelson et al. (1994). Collecting data at the project site includes measuring the inset channel width and depth, and any associated benches. The channel pattern is the meandering pattern of the inset channel. Bed slope, water surface slope, and floodplain slope should all be measured. A pebble count of the bed material should be conducted to estimate the median bed particle size (Wolman 1954). In aggraded systems with deposits of fine material that are more than a few tenths of a meter in depth it might not be possible to use conventional methods to conduct a profile survey. For these cases, it is recommended that frequent cross-sectional information be obtained and data at the cross-sections then be used to also create profiles of the channel bed, sediment depositions, the water surface elevation, and the benches.

Figure 2 Two-stage channel typical cross section outlining the inset channel width and the flooded width.

Floodplain width

Channel width Channel depth

(Leopold and Maddock 1953; Rosgen 1996). Dimension measurements collected throughout the watershed are plotted to generate regional curves. Typically, data plotted against a log-log scale will generate an obvious relationship that can be fit with a power regression curve with high correlation (figure 3). Several regional curve measurements should be collected for each drainage area log-scale. Reconnaissance is required to locate well-defined channel dimensions throughout the project watershed. Regional curve measurements should be taken at drainage areas that are smaller, larger, and similar to the project’s drainage area that range through several magnitudes (Ward and Trimble 2003). If a USGS gaging station is located within the project water-

Figure 3 Regional curve showing bankfull channel dimension related to drainage area.

1,000

Measurement (m or m2)

A = 0.81DA0.77 R2 = 0.97 100

10

W = 0.97DA0.47 R2 = 0.97

1

0.1 0.1

D = 0.26DA0.30 R2 = 0.93 1

10

100

Drainage area (km2)

1,000

10,000

JULY | AUGUST 2007    volume 62, number 4

Copyright © 2007 Soil and Water Conservation Society. All rights reserved. Journal of Soil and Water Conservation 62(4):277-286 www.swcs.org

If the project site does not have an inset channel and benches, detailed measurements of the project site should still be performed. In addition, a reference site can be selected to provide additional information on similarly-sized stable channels (Rosgen 1996). The reference reach should be similar in all respects to the project site; ideally, the only exception would be evidence of self-forming bench features. In many situations, finding a suitable reference site is not practical or requires significant time investments to find an adequate location, so more dependence has to be placed on other measurements. Regional curves are often used in stream geomorphology and are empirical relationships between bankfull channel dimensions, or bankfull discharge, and drainage areas

shed, it is recommended that regional curve measurements be extended to include the gage station. Often, an abbreviated “rapid regional curve” may be adequate. This method differs from common approaches to developing regional curves because only the bankfull width and depth of main or inset channels are measured at sites that exhibit distinct fluvial features associated with connected floodplains. Regional curve measurement locations do not have to be in natural streams, and detailed geomorphic surveys need not be performed at all sites. Sites should be in the project watershed and should have similar characteristics to the project site. Step 3: Data Analysis. The following site specific data are needed to size a channel system: (1) drainage area, (2) channel slope, (3) inset channel dimension, (4) regional curve, (5) d50 and d84 bed material, and (6) bankfull discharge and an index of the frequency of out of bank discharges. Additional useful information is knowledge of the unit stream power, the effective discharge, and the likelihood of a mass bank failure. Drainage areas are required for the project site, reference reach, and all locations where measurements were made for the regional curve. The channel slope (figure 4) is determined from a profile survey of the bed and the water surface. Often, the average channel slope can be better estimated from changes in the water surface elevation. This is particularly true where sediment deposits make determining the channel elevation difficult and in short reaches with pools and riffles. In agricultural ditch systems, the dimensions of the inset channel usually vary throughout the project reach, there are often several grade breaks on the banks, dominant benches are not always evident, and maintenance activities and bank slumping make it difficult to identify the fluvial features. These factors make it useful to compare measured representative bankfull widths, depths, and cross sectional areas to estimates from the regional curve (figure 5). Determination of a representative channel cross-section aids in estimating the channel bankfull discharge and provides the framework necessary for any channel modification project. Figure 5 illustrates how, for the case reported, the cross section dimensions (dashed line) agreed with the estimated channel dimensions from the regional curve (solid line) for

279

Figure 4 Channel profile, or slope, of the existing bed, water surface, bench, and top of ditch.

96 94 Top of ditch

Elevation (ft)

92 90 88 86

Bench

0

200

400

600

800 1,000 1,200 1,400 1,600 1,800 2,000

Station or channel distance (ft) the inset channel. If there is poor agreement, or the project site does not exhibit benches and an inset channel, further investigation might be needed before a decision is made on whether to construct a two-stage system. Factors such as low unit stream power, high base flows associated with groundwater or a

point discharge, or coarse bank and sediment loads (primarily sand or coarser) might limit the formation of benches. In other cases, the lack of benches might be associated with ditches too narrow for benches to form. A Wolman (1954) pebble count provides information on the particle size distribution

Figure 5 Existing geometry shown in dashed lines and proposed two-stage channel dimensions based on the regional curve shown in solid lines.

31

Elevation (m)

30

29

28

27 0

4

8

12

Distance (m) Note: The existing main channel has a similar geometry to the projected geometry.

280

journal of soil and water conservation

16

Copyright © 2007 Soil and Water Conservation Society. All rights reserved. Journal of Soil and Water Conservation 62(4):277-286 www.swcs.org

82

Water surface

Existing channel bed

84

of the bed material (figure 6). Channel systems in dynamic equilibrium often have mean bed shear stresses (tractive force) associated with the bankfull discharge that move particle sizes similar to the measured median bed particle size (Knighton 1998; Wilcock 2001; Ward and Trimble 2003). The tractive force (kg m–2) was calculated as the product of the specific weight (1000 kg m–2), the slope of the water surface as a fraction, and the mean bankfull depth (m). Channel systems that are in equilibrium should have a d50 or a d84 that falls bed material size within the upper and lower limits shown on figure 7. The d50 and d84 are the bed materials sizes that correspond to 50% and 84% of the bed material being finer than these sizes, respectively. The authors have developed the STREAM spreadsheet tools to assist in data analysis by reducing the number of hand calculations required and providing a uniform presentation of all collected data (Powell et al. 2006b; www.dnr.state.oh.us/soilandwater/streammorphology.htm). These tools allow users to input measured data, including meander pattern, bed and water profile, channel cross sections, pebble count, regional curves, and hydrology estimates, and then provide information on most of the factors that are needed to size a two-stage channel design. Step 4: Hydrologic Evaluation. Included in the STREAMS spreadsheet tools are USGS empirical regression models for predicting peak discharges in urban and rural ungaged watersheds in Ohio (Koltun and Roberts 1990; Sherwood 1993). Other hydrologic estimates can be entered into the spreadsheet in place of the Ohio USGS estimates. Regardless of location, pertinent gage data should be used to first calibrate the hydrology method that is used. If actual flow data are not available for a location near the project site, we recommend calibrating the hydrology method with data from the nearest USGS gage. In Ohio, the bankfull channel usually conveys a discharge exceeding the average annual flow but not large enough to convey the two-year recurrence event (Powell et al. 2006a). However, in ditch systems with low stream power and dense herbaceous vegetation, the bankfull discharge for the inset channels is often much smaller. One study in Ohio indicated bankfull discharge was met or exceeded 39 days per year from 23 events and had recurrence intervals

Figure 6 Wolman pebble count results showing number of particles counted in each range bin size and the cumulative percent finer than each bin size.

Silt/clay

Sand

Gravel

Cobble

Boulder

18

90%

16

80%

14

70%

12

60%

10

50%

8

40%

6

30%

4

10%

2

0% 0.01

0.1

1

10

100

Cumulative %

1,000

0 10,000

# of particles

Particle size (mm) Note: The d50 and d84 are shown by dashed lines and are the particle sizes that correspond to 50% and 84%, respectively, of the bed material being finer than these sizes.

between 0.25 and 0.5 year on the partial duration series (Jayakaran et al. 2005). Crowder and Knapp (2005) also report that the bankfull discharge in the Midwest United States is

substantially less than the frequently quoted recurrence interval of 1.5 years. Fortunately, hydrology evaluation of bankfull discharge recurrence intervals is only one aspect of a

Figure 7

Mean diameter bed material (mm)

Mean bed material moved by tractive forces on a streambed.

1000 Upper limit 100

10 Lower limit 1

0.1 0.01

0.1

1

Tractive force, T (kg m–2) Notes: Based on Lane (1955). Upper and lower limits by Ward. Source: Ward and Trimble (2003).

10

100

JULY | AUGUST 2007    volume 62, number 4

Copyright © 2007 Soil and Water Conservation Society. All rights reserved. Journal of Soil and Water Conservation 62(4):277-286 www.swcs.org

20%

Number of particles

Percent finer than

100%

weight-of-evidence approach. Sizing fluvial channels should never be based solely on a discharge associated with a specific recurrence interval. In addition to providing adequate bench width, the second stage of the channel system should be able to transport a design flow that will prevent frequent flooding of adjacent areas. The maximum design flow is typically based on economic criteria, including a cost-benefit analysis such as loss of crops, flooding, or maintaining capacity flow. Modification of a channel system to a two-stage geometry will usually provide a greater increase in conveyance capacity than maintenance activities to remove sediment deposits on the bed and benches. Examples of hydrologic peak flows in a trapezoidal and two-stage channel are shown in figures 8A and 8B. In this example, discharges with a recurrence interval of about five years or larger have a lower stage in the two-stage system than in the trapezoidal channel. In contrast, the more frequent discharges have a slightly higher stage in the two-stage system. For all discharges the difference in stage for the two systems illustrated is less than about 0.3 m. Step 5: Conceptual Channel Sizing. A conceptual geometry is developed based on the weight-of-evidence collected through the previous sizing steps and flooding considerations. Sizing a two-stage channel involves first determining the inset channel geometry, which defines the bench height, then sizing the flooded width at the bankfull elevation of the inset channel, and finally determining the channel side slope for the second stage. The flooded width includes the width of the fluvial benches and the channel width. The project drainage area is applied to the regional curves developed in step 3 to provide estimates for bankfull width, depth, and cross-sectional area. Estimates from the regional curve need to be compared with the actual measured fluvial features at the project site, the reference reach, the hydrologic estimates, the shear stress depth, and the estimated bankfull/effective discharge depth. If there is good agreement between all factors, then the likelihood of success is high and the project should proceed. The design of the new flooded width in the channel is a function of the top width of the inset channel (figure 2). In systems with cohesive bank materials that can readily be vegetated with grass, our experience is that

281

Figure 8 Capacity to carry different recurrence interval storm runoff volumes of (A) a traditional trapezoidal channel and (B) a two-stage channel.

4

Depth (ft)

3

2

1

4

Depth (ft)

3

2

1

0

the ratio of the flooded width to the inset channel width should be at least three. The bench elevation corresponds to the height above the channel bed as estimated by the inset channel depth; the existing bank will be excavated at the bench elevation. In cohesive soils, the inset channel side slopes typically form at one-to-one slopes. In the second stage of the channel, side slopes are typically formed at two-to-one slopes. The side slopes need to be stable to avoid slump failures during high flow events. A geotechnical engineering analysis should be conducted to ensure that the bank stability has a factor of safety of at least 1.5 (Simon and Langendoen

282

journal of soil and water conservation

100 50 25 12.5 6.3 3.1 1.6 0.8 0.4 0.2

2006). The second stage can also be sized to accommodate the maximum design flow rate used for the existing trapezoidal channel. Typically, provision of adequate floodplain will result in a second stage with more available conveyance capacity than existed prior to channel modification. Ideally, sediment supply and sediment transport should also be considered, but measured information is rarely available. The Contrasting Channel spreadsheet that is part of the STREAM spreadsheet tools includes a procedure to estimate effective discharge and depth of flow associated with the effective discharge (Powell et al. 2006b).

Copyright © 2007 Soil and Water Conservation Society. All rights reserved. Journal of Soil and Water Conservation 62(4):277-286 www.swcs.org

0

100 50 25 12.5 6.3 3.1 1.6 0.8 0.4 0.2

An important consideration is the likelihood of success or failure. Table 1 illustrates how the evidence obtained might be used to determine whether the establishment of a two-stage system might result in a more stable and self-sustaining system. The weight-of-evidence approach outlined in table 1 is based on the assumption that rapid land use change is not occurring and the channel is located in a region where stable benches might form. The approach has been tested against case studies presented in a companion paper (Powell et al. 2007) and ditch systems studied by Jayakaran and Ward (2007). If the evidence for a particular factor suggests that confidence in success is low, a two-stage approach might still be appropriate. However, further measurements and/or analysis might be required. It is important to recognize there is considerable uncertainty in most of the measurements and methods used to make any channel system assessment. A practical “weight-of-evidence” approach is presented in this paper, and useful alternative approaches have been reported by other authors (Montgomery and Macdonald 2002). Step 6: Project Assessment. Results from the previous five steps need to provide sufficient detail to obtain a cost-benefit estimate of the project. Each aspect of the two-stage design should be examined in the context of the specific project conditions. Reexamining project objectives and identified problems will assist in validating the channel design. A final project assessment and design presentation should occur with all stakeholders participating, and should include but not be limited to, outlining all previous steps, costs, and post-construction management plans. The two-stage channel concept involves widening the existing channel; the land owner must be willing to permanently have a small width of potentially farmable land excavated. Some farmers agree to compensation for the loss of this land. An integral part of project assessment is the establishment of a post-construction management plan that should address landowner, engineering, and environmental concerns. In some cases, no modification or maintenance activities will be needed. In other cases, for example where the subsurface drainage outlets are well below the proposed floodplain elevation, a conventional removal of deposited sediment might best meet the needs of the stakeholders. Ditches that exhibit rapid

Table 1 Weight-of-evidence evaluation for sizing two-stage channel systems. Weight-of-evidence categories and observed and calculated evidence

Confidence in success (score)*

1. Inset channel dimensions High (5) High (5) Moderate (3) Low (1)

B. Unstable benches observed Observed width and depth or cross-sectional area within 25% of regional curve values More than a 50% difference

Moderate (3) Low (1)

C. No benches observed Distinct grade break within 25% of regional depth Distinct grade break within 50% of regional depth No grade break or normally a deep baseflow in the ditch Width in lower third of ditch less than 125% regional bankfull width

Moderate (2) Low (1) Unlikely (0) Unlikely (0)

2. Particle at motion Measured d50 or d84 at riffle within upper and lower limits on figure 7, calculated based on bankfull depth Fine materials deposited over coarser material within the upper and lower range Extensive aggradation of fines, low bed slope, and small watershed area (