The Flume Study
Too often our floodways are cleared of native vegetation for fear that trees and shrubs will block or redirect flood flows that may damage property. However, native vegetation can protect soil erosion and can be designed to have zero impact on the movement of flood waters. Engineers need quantitative data about the behavior of different kinds of riparian plants, in order to incorporate their use in floodways. In a recent study at the J. Amorocho Hydraulics Laboratory at UC Davis Large Flume (Kavvas and others 2009), multiple depths and velocities of flows were tested on four species of flexible stem riparian plants and for bare soil. Their results indicate that riparian vegetation can be beneficial to floodway designs.
Background of Flow Properties
The velocity of the flow in any stream or river is primarily determined by gravity operating through the slope of the channel – water runs faster down a steep slope than a gentle slope. Resistance to flow, or hydraulic roughness of the channel (the texture of the surface of the channel, banks, and floodplain due to type of vegetation structure, geomorphology and topography, and size of sediment) modifies the effect of gravity and slows the flow, resulting in the velocity that we see. Because water molecules are in a fluid state, literally anything in the channel can locally modify their velocity and redirect their flow for short distances. We call this hydraulic turbulence and it is seen as waves, and boils on smooth river water from upward oriented turbulence. Hydraulic turbulence is the mechanism of resistance to flow especially at higher velocities.
Flows in open river channels are turbulent, with vectors moving in all directions within the general slope induced flow. At any point in the channel or on the bank, the velocity is variable over time spans of seconds and minutes. This is caused by turbulence in the flow. As velocity increases so does the turbulence. On the bed of the channel, or on the surface of the floodplain, the turbulence dislodges sediment and entrains it into the flow. As velocity increases, the force of the turbulence increases as well, such that larger diameter sediment is kicked up and entrained. The turbulence in such a flow will keep particles as large as cobbles entrained for great distances, resulting in the placement of cobbles onto the floodplain that we see after a large flood.
Flume tests showed that at the higher velocities (4-6 feet per second, fps) the plants bent over with the flow, becoming more streamline as the flow velocity beneath the plant canopies dropped, while the flow velocity increased over the top of the canopies. Surface erosion of the soil was minimal under the plant canopies. However, the flume tests on bare soil with no plant canopies to protect the surface, resulted in dramatic increases in soil erosion off the surface at about 4 fps. As velocity increases over the bare soil, lift-forces cause soil particles to rise into the water column while hydraulic turbulence also increases in intensity, thereby kicking up soil particles and entraining them in the flow. Therefore, both increases in velocity and hydraulic turbulence cause greater erosion on bare soil.
Comparison of soil surface erosion depths under different flood flow velocities for four California native plant canopies and a bare soil bed. Used with permission from Kavvas and others 2009.
Distribution of velocity differences across channel-floodplain (x-section)
During a flood the velocity the floodway is not uniform. The velocity is fastest and deepest in the main channel. On the adjacent floodplain velocity is slower and variable as one proceeds along a transect. In fact, where eddies occur, the flow may be directed upstream for short distances. Eddies and low-flow backwaters are relatively slow velocity compared to other areas in the floodway. The slow and fast velocity areas are determined by the geomorphology of the floodway. Thus trees and brush growing in these low velocity areas have low, or no impact on the hydraulic roughness of the floodway. Hydraulic models of the flows within the floodway can tell us how fast and where the flow velocity is distributed.
If the velocity of flow changes at any point along a stream, effects are felt both upstream and down. Slowing the velocity will cause the flow upstream to “stack-up” on top of the slower water. This raises the elevation of the stream and could force water over the top of a levee. On the other hand, increasing the velocity at a point along a stream can lead to water stacking up downstream. From Knighton p 99: “At a given point on the bank, over time (seconds and minutes), velocity will naturally fluctuate, causing the elevation of the flow to fluctuate also.” The hydraulic model for the Feather River shows plus or minus 0.1 ft. accuracy; this is roughly the same variability that we observe while standing on the bank!
Flow depth and roughness
Flow depth greatly influences the effects of channel, floodplain, and vegetation roughness. Central Valley Rivers function as floodways with confined flows between levees that results in flow depths over the floodplain that are unnaturally deep – 15 to 20 feet, or more on the lower Feather River. Two phenomena are occurring at these great depths: 1. Under deep flows, proportionately less of the flow is in contact with the channel and floodplain, compared to low shallow flow. 2. The weight of the water column under deep flows exerts a tremendous force upon the channel, floodplain, and vegetation causing sediment erosion and pressing flexible-stemmed plants to the bottom. For example, one early sign of an unsafe levee is that erosion is taking place at its base where the hydraulic forces are greatest.
Plant stems and roughness and velocity - Flexible stems vs. rigid stems
Vegetation is composed of many different species of plants, each with its characteristic growth form and range of stem diameters. Thus, trees have a main trunk that is many inches in diameter, shrubs have many stems of smaller diameter, and herbaceous plants have stems less than an inch in diameter that wither and die at the end of each growing season. Each growth form is composed of stems of a specific range of diameters. Clearly, a single tree trunk that is many inches in diameter is not at all flexible. A tree truck will resist flows causing hydraulic roughness to remain the same or to increase as flows become faster and deeper. On the other hand, shrubs are composed of many stems that are of smaller diameter than a tree trunk. These shrub stems are more flexible and can bend as depth and velocity of flow increases. Vines have stems typically less than one inch in diameter and will bend even under low flow velocities.
As stream flows increase in depth with larger volumes of water compressed between two levees, the weight of the deeper water impacts the channel, the floodplain sediments and the vegetation growing on them. Flood flows can be 15 to 20 feet deep at maximum design discharge in the Feather River floodway (historically flows were much shallower when the river could spread across the natural floodplain during a flood.) The weight of the deep water is more effective than shallow water at mobilizing sediments and it will press down to the bed flexible stems of vegetation. Tall, non-flexible trees and shrubs may stand upright during a deep flood. However, flexible shrubs (rose, blackberry, sandbar willow, etc.) will be pressed to the bottom. This fact/phenomenon has recently been quantified by the flume study at UC Davis.
The Flume Study quantifies the impacts of flexible stem plants upon hydraulic roughness and bed erosion. Plant growth forms composed of small diameter stem – less than one inch in diameter; rose, blackberry, sandbar willow, mulefat – will bend under the force of moving water at relatively slow velocities ( 2- 5 fps). Direct measurements of velocity above, below and within the plant canopies in the flume reveals important characteristics of the bendability of stems and their impacts upon hydraulic roughness and bed erosion.
Kavvas and others 2009 showed quantitatively the velocities that 4 species of riparian plants bend over in response to flows. They also showed that water velocity slightly increases above the plants and decreases underneath the plants. The slower velocity of water beneath the plants decreases soil erosion.
As depth and velocity increase, flexible stemmed plants bend with the flow and hydraulic roughness (Manning’s n) decrease as the flow passes over them. Measurements of velocity in the flume show that velocity increases over the prostrate plant stems. Under the plant stems, velocity decreases by over half, thereby protecting the bed from sediment erosion.
Vertical distributions of flow velocity averaged over the three replicate wild rose canopies. The blue diamonds reflect the velocity profile as the flow entered the wild rose canopy. The red squares represent the velocity profile 18 feet into the canopy, and the yellow triangles represent the velocity profile at the downstream end of the wild rose canopy.Used with permission from Kavvas and others 2009.
Velocity Profiles for Sandbar willow. The blue diamonds represent the velocity profile as the flow first enters the canopy of the Sandbar willow; the red squares represent the velocity profile after the flow has traveled 18 feet into the canopy; and the yellow triangles represent the velocity profile as the flow leaves the Sandbar will canopy. Used with permission from Kavvas and others 2009.
Vegetation can direct flows
When a dense stand of non-flexible trees or shrubs grow adjacent to an open area composed of only herbaceous plants, flows will deflect off the dense stand and into the open area of low hydraulic roughness. This phenomenon can be used to protect structures and focus sediment transport, as at O’Connor Lakes (see discussion below).
Engineers have several mathematical formulae to describe flow velocity. Resistance to flow by the channel boundary (bed and banks and vegetation) is usually described by a roughness coefficient in a formula. The Manning’s equation is the equation used by flood control engineers in California:
v = 1.49((R 2/3 s ½)/n)
where: v = velocity; R= hydraulic radius (R= A/p, where A= Channel cross-section area, and p= wetted perimeter of channel); s=slope of energy gradient; n=resistance (or roughness) coefficient.
(These parameters are all typically averaged across a stream cross-section. Recently, two-dimensional hydraulic models have been developed that sub-divide the channel cross-section into smaller, more realistic units. See Hydraulic Modeling).
In the formula, n is a coefficient that describes the resistance to flow as a function of flow velocity (v) and depth (R) of flow. The Manning’s equation describes the interactions of river flow velocity, flow depth, and channel slope and roughness. Thus, as depth increases, roughness (n) will decrease and velocity will increase because less of the flow is in contact with the perimeter (channel and floodplain).
The absolute value of n can be used to describe the flow resistance caused by different vegetation structures. For example, a grove of dense trees with relatively large diameter, non-flexible stems and trunks will resist flow to a much greater magnitude than a stand of flexible stemmed sandbar willows covering the same area. In this example the grove of trees might have a roughness coefficient of n=.07, while the more flexible sandbar willows might have a roughness coefficient of n=.05. The O’Connor Lakes hydraulic modeling exercise that resulted in the final restoration planting design is a good case study that demonstrates how Manning’s n was used to create the restoration planting design.
How the flume is not a river
Flume results are presented with Manning’s n as a function of Reynolds number (Reynolds number = velocity x depth/viscosity of water). The flume is much shallower than a river, hence Reynolds number will be even larger in the real world. In the real world, flood flows are much deeper – 14 feet deep in the Feather River – compared to the flume – 5 feet. Referring to the figure below, we see that the canopy Manning’s n for all species are decreasing and about to intersect with the rising Manning’s n for bare soil. Increasing the axes of this figure out to the Reynolds number that would occur at 14 feet deep, we see that the vegetation canopies would have a minimal impact on hydraulic roughness because they would likely be pressed to the bed.
Manning’s roughness coefficients as a function of Reynolds number under various California native riparian vegetation canopy conditions.Used with permission from Kavvas and others 2009.