Stabilizing the streambank with GeoJute

Prairie flowers in June 2005

Wilson Creek Restoration/Revegetation
June 2003 through October 2004

Revegetation efforts began at the Wilson Creek stream restoration in September of 2003 after a majority of construction on channel and floodplain had been completed.  At this stage, Bernheim staff began installation of erosion control measures to conserve soil on the newly constructed stream banks.  Over the next 1 ½ months, fabric was laid as riffle and pool construction was completed.  In all, Bernheim staff covered >3500 linear feet of riffle bank and >2000 linear feet of pool bank.

Riffles were covered using a combination of Anti-Wash Geojute (4’ X 225’ rolls) and 10 oz. burlap fabric (6’ X 300’ rolls).  A  burlap wrap was constructed to stabilize the toe of the stream bank.  The burlap was constructed to be functionally similar to a coir log which can be placed along the toe of restored stream banks for stabilization.  It was thought the relatively tight weave of the burlap fabric would hold soil better than the jute mesh, which has a much greater proportion of open space (» 60% open space).  Burlap was rolled perpendicular to the stream channel along the toe of the bank and secured using six inch sod staples.  Next, the burlap was filled  with soil using a skid loader.  After excess soil, clods, rocks, and other debris were hand raked from the burlap, cover crop and native seed was sown into the soil. The fabric was folded along the long axis to form a “soil log” along the stream bank.  The soil wrap was then tied into a section jute mesh using wooden stakes every four to five feet.  The jute, which was placed over straw and seed, was then secured using sod staples placed every few feet on center.  The coverage of  fabric along riffles varied from six to seven feet wide, depending on the overlap between the burlap and jute mesh.

A similar burlap/soil wrap was constructed in the outside bends of pools.  For pools, 15 ft strips (5 feet wide) of burlap were cut before going to the job site.  Burlap strips were laid perpendicular to the stream; each strip overlapped the adjacent strip by several inches.  Soil was then dumped onto the burlap using a skid loader. The soil depth decreased from a maximum height of eight to ten inches near the water’s edge to one to two inches at the opposite end to promote a smooth transition from burlap on the bank to uncovered soil in the floodplain. Cover crop and native seed was sown into the soil. Burlap strips were then folded along the short axis back over the soil and secured using wooden stakes.  Fabric width in pools averaged between 10 and 12 feet.

Cereal rye and a variety of native species were sown throughout the floodplain in mid-October.  The rye is very cold tolerant and was meant to provide cover over the winter months and subsequent spring.  As the rye died off the following spring, the native seed would germinate and begin growing (many native plants need a period of cold, moist stratification before germination).  In this idealized scenario, soil should not have been exposed for long periods of time and erosion potential should have been low.  

 Soil Loss           
During the first winter following construction of the Wilson Creek Stream restoration, the watershed received abnormally high levels of precipitation.  The wet winter resulted in numerous out-of-bank events (>12), which caused significant erosion of stream banks and floodplain soil and the probable loss of native seed sown on site.

A second factor contributing to erosion on site was the relatively low banks of the restored section of Wilson Creek. Subsequently, even with normal levels of  precipitation, out-of-bank events would have occurred more frequently than in many stream restoration projects.  Increased floodplain access (beyond that of many “restored” streams) was a primary goal of this stream restoration project and bank height was consistent with what University of Louisville researchers believed existed before wholesale human manipulation of riparian corridors.  This in and of itself did not cause erosion, but projects utilizing the floodplain to the extent that the Wilson Creek project did are inherently at higher risk of erosion during the period before vegetation is firmly established.

The lack of established vegetation on-site at the end of the 2003 growing season also exacerbated soil loss.  Floodplain soils were disked after construction to provide an a seedbed conducive to growth, but a large out-of-bank event occurred  less than two weeks after seeding the cover crop/natives resulting in loss of seed and soil.  Even if the first out-of-bank flow had not occurred, seeding, which took place in mid to late October, was too late to achieve adequate herbaceous cover before the coming winter.  Seeding was pushed back to that late date because of extensive construction delays that occurred earlier in the year.  These delays were primarily the result of unusually high rainfall during the summer months. 

Herbaceous Vegetation Establishment - 2004
With the beginning of the 2004 growing season very little herbaceous vegetation existed on the poject site; initial survival of tree, shrub, and rive cane planting was quite high (>90%).  Scouring and top soil losses that occurred during the winter months produced a situation that was not conducive establishment of herbaceous vegetation.  Exposed, scoured subsoil prone rapid drying was the primary substrate in the floodplain.  Sorghum was sown throughout the site at » 60lbs./ac during early June as a warm season cover crop.  The relatively high seeding rate was due to lack of proper seedbed present on site.  Seedbed preparation (disking, harrowing) was hindered due to the recent plantings of over 10,000 bare root tree seedlings on site.  The sorghum, along with native and nonnative volunteers, and seeded species began to fill in as the season progressed.  Unfortunately, vegetation establishment did not occur rapidly enough to prevent bank degradation and floodplain erosion.

Repairs - 2004 
Repair of stream banks and replacement of floodplain soils began in late July, 2004.  Based on erosion observed the previous winter and the larger budget for repairs (we had equal monies for repairs in 2004 and the entire construction in 2003), Bernheim staff used Anti-Wash Geojuteä to cover large swaths (total fabric >110,000 ft2) of the floodplain instead of only a thin strip (total fabric » 12,000 ft2) along riffles.  After bringing the floodplain to grade, areas were seeded with several cover crop species and numerous native floodplain species.  Straw was placed over the seed, and the area was covered with fabric.  Fabric was primarily secured using wooden stakes (18”x1”x 2”) with a hole drilled several inches from the top. As the wooden stakes were driven into the ground along the fabric edges, a nail, three to four inches long, was placed in the hole.  When the stake was driven in the nail firmly held the fabric in place.  Sod staples werestill used, but their primary function was to hold seed and straw in place during an out-of-bank flow.  When these flows do occur, water can move underneath the fabric washing seed and soil away.  Sod staples effectively prevent this from occurring, but do not  effectively secure the fabric.

Additionally, the burlap soil wrap was abandoned due to its ineffectiveness; vegetation did not penetrate the burlap well, especially after the scouring produced in an out-of-bank flow.  Instead we placed large pieces of sod (4’ X 8’) along the toe of the stream bank (Figure 2).   These pieces of sod were eight to 10 inches thick and provided an instant mass of soil, roots, and foliage in the most vulnerable areas of stream bank .  The sod layer provided a much greater level of protection (sod is eight feet wide in every riffle) and appears to be very stable in comparison to the methods of 2003.  

What We’ve Learned

• Wait One Growing Season:  The ultimate challenge is establishing vegetation, thereby stabilizing topsoil, on a frequently, heavily disturbed site.  In the case of the Wilson Creek site the first large, out-of-bank flow occurred less than two weeks after cover crop and native seed was sown.  Because this disturbance occurred before vegetation establishment, seed and proper seedbed (for establishing future seed) was lost.  The scenario could be avoided on stream restoration projects by waiting one growing season before introducing flow into the restored channel.  This would involve building the restored channel and all its attributes, but leaving the connection between the existing and restored channel unbreached thereby giving vegetation an entire year before it could potentially be exposed to flood flow.  When construction equipment reentered the site the following year to introduce flow, only a small percentage of the project area would have to be disturbed to connect the existing and restored channels.
• Cover the floodplain: Erosion control fabric should be used in restoration projects where the channel slope is high enough (>0.005, comment by Art Parola, University of Louisville) to cause floodplain erosion.  This fabric acts as insurance to hold seed and soil in place during out-of-bank flow and accelerates plant growth during dry periods by retaining soil moisture.  Using a layer of straw under the fabric also retains moisture and promotes seed growth.
• Timing:  Revegetation efforts must begin earlier than mid-October to establish vegetation sufficient for protection during the first winter.  This was known in 2003, but construction delays continually pushed the completion date back. The effect of the late completion date on vegetation establishment could have been minimized if better phasing methods were incorporated into the construction process.  For instance, the project site should have been divided into sections.  Then, entire sections could have been completed before moving on to the next area.  This would have guaranteed many areas would have been seeded well before mid-October even with the occurrence of construction delays.
• Utilize a variety of cover/native species: Using a wide assortment of cover/native species adapted to a wide range of conditions is a good way to hedge your bets against issues that can inhibit vegetation growth.  Uncertain hydrology in the restored site, inclement weather (drought, deluge), and insect predation don’t have to spoil revegetation efforts.
• Don’t rely solely on native species:  Stream restoration projects with a high degree of floodplain access are inherently erosion prone.  Revegetation efforts on theses sites  must use species capable of rapid growth (See cover crop section of attached species list).  Since many native species require dormancy periods of varying lengths, annual non-natives species that provide immediate cover are useful, if not integral, to restoration process.  Invasive species should be avoided, but use of non-native cover crops does not have to interfere with the end goal of native dominated ecosystems.   The principle concern should be preventing erosion.
• Exotics management:  Restoration sites in the early stage of succession will certainly have a large non-native component. Consideration of the target plant community should play a vital role when planning for invasive plant management. The goal of the Wilson Creek restoration is a bottomland hardwood forest; a shaded community.  Spending money to control shade intolerant species is not necessarily an efficient use of resources.  Species like Microstegium vimineum, which are shade tolerant and whose seeds are water