The development of California's water supply system has radically altered the Sacramento and San Joaquin Rivers and their tributaries. Large water supply dams have generally reduced the magnitude, duration, and frequency of peak flows, while simultaneously depriving downstream reaches of a fundamental building block of habitat by trapping the supply of coarse sediment from upstream reaches.

Since it is generally infeasible to remove large water supply dams, restoring geomorphic function and ecological habitat value to the downstream reaches of regulated rivers will require, in many cases, redesigning the channel geometry and floodplain elevations to adapt to the changed hydrologic regime. Also, understanding the relationship between stable channel geometry and the discharge distribution and sediment supply regime is required to define dam releases in a way to achieve fluvial geomorphic objectives. Channel and floodplain reconstruction projects often aim to restore channel migration as well, which has generally been reduced on Bay-Delta tributaries because of the reduction in peak flows and sediment supply, as well as levee construction and channelization.

Questions

The channel/floodplain reconstruction experiments will be designed to answer the following questions:

• How does one scale down a channel below a dam?
• How does width depend on
  • discharge distribution;
  • sediment supply; and
  • lateral migration?
• How far downstream will the effect of reconstruction be felt?

Hypotheses

The following hypotheses have been developed to guide the design of the channel/floodplain reconstruction experiments:

1. The stable channel width can be related deterministically to two measures of the distribution of discharges, one that reflects the total volume of runoff carried by the channel, and another that reflects the variability in the discharge distribution.
• For the case of shifts in the total runoff volume without changes in the variance, we expect the width to vary with the square root of some representative, or dominant discharge, in accordance with the common field observation.
• We further expect that wider channels will correlate with larger discharge variance, all else being equal.
2. Sediment supply can be expected to strongly co-vary with discharge; that is, all else being equal, we expect wider channels for greater rates of sediment supply.
3. Levees reduce effective floodplain width, increasing the geomorphic effectiveness of large magnitude discharges. Thus, levee setback or removal may tend to offset the geomorphic benefits of restoring peak discharges.
4. Bank strength exerts a first-order control on channel width, and is often primarily a function of the type and density of riparian vegetation.

Description of the Experiments

Utilizing Smith's (1998) model floodplain substrate, we will structure a series of experiments to test hypotheses that examine the role of variable discharge and sediment supply in controlling channel width and meander migration rate. In a wide model floodplain basin, we will generate a set of discharge distributions, systematically varying the discharge mean and variance along with sediment supply, and measure the resulting stable channel width and document the pattern and rate of meander formation and migration. We will compare experimental results with predictions of the Johannesson and Parker (1989) meander migration model, examining in particular the relative influence of sediment supply variations on lateral migration rates.

There are large gaps in our understanding of the linkages between the discharge distribution and stable channel geometry and rates of lateral migration. Although there is a strong empirical correlation between the downstream variation in channel width and the discharge that typically has a recurrence interval of 1.5 to 2.0 years (Leopold and Maddock 1953), no theory exists to explain this observation. The only mechanistic theory for channel width (Parker 1978a, 1978 b) assumes that the banks and bed are composed of the same cohesionless material, and it applies only to a single dominant discharge. Similarly, Johannesson and Parker's (1989) theory for meander migration, which has been widely applied in modeling floodplain evolution (e.g. Howard 1992; Larsen 1995), assumes a single steady discharge and makes no allowance for the role of variable sediment supply.

An important constraint on building and testing theoretical models for these processes has been the lack of a methodology for creating laboratory-scale channels that have self-formed stable widths and that form migrating meanders. Model channels composed of sand-sized and larger material do not build bars and banks on the inside of migrating bends and consequently widen until switching to a braided configuration (e.g. Friedkin 1945). Recently, Smith (1998) succeeded in creating migrating channels that, for the first time, maintain a constant width, using a weakly cohesive mixture of silt-sized silica and clay (see photograph). Smith's (1998) meandering channels form many of the morphologic features of large floodplain rivers, including meander cutoffs and scroll bars. In collaboration with Dr. Dietrich, Smith has continued his experiments at the University of California;s Richmond Field Station, investigating the role of variable sediment supply in influencing lateral migration rates.

Smith's innovative floodplain modeling technique offers the possibility of creating a general methodology for physically modeling the dynamics of self-formed channels. Currently, Smith's experimental apparatus is too small to accurately scale the forces driving sediment transport and bank erosion. However, a larger experimental basin, with the capacity to accommodate larger discharges and rates of sediment supply, may be sufficient to reliably reproduce the dynamics of meandering channels at a laboratory scale. Although this effort is somewhat speculative, the potential benefit of such a methodology is enormous for theory and for practical application.

We will use this new technique to investigate the fundamental question of how freely-formed meandering channels respond to changes in discharge and sediment regime. The proposed modeling will strive to provide new tools for solving one of the most problematic and common restoration problems: how to manipulate the distribution of discharges released from upstream dams to drive fluvial geomorphic processes. These experiments also have the potential to improve our understanding of, and ability to predict, rates of lateral channel migration.

Initial condition experiments: The first set of experiments will focus on establishing a reference slope, and it will explore different discharge and sediment feed combinations to help define the design of successive channel-floodplain experiments. This set of experiments will also permit refinements to the discharge and sediment delivery systems.

Uniform discharge experiments: The second set of channel-floodplain experiments will maintain a constant discharge throughout the run. Discharge will be varied over an order of magnitude. For each discharge the sediment supply will be varied through three levels: low, medium and high. For five values of discharge, this will require 15 runs.

Variable discharge experiments: The third set of experiments will vary discharge widely throughout each individual trial, so that the channel experiences a full distribution of discharge magnitudes and durations. The mean and variance of the discharge distribution, and the relative sediment supply rate, will be varied independently. For three means, three variances and two relative sediment supply rates, a total of 18 runs will be required.

Variable floodplain width experiments: The fourth set of channel-floodplain experiments will consider the confining effect of a set of levees that narrow the floodplain by factors ranging from two to five. This set of experiments will use the results of the previous channel-floodplain experiments to select the parameter values most likely to reveal the sensitivity of channel width and migration rate to floodplain width.

Scaled field site experiments: The final set of channel-floodplain experiments will simulate the evolution of a floodplain and channel design taken from an ongoing field restoration project. The floodplain slope and width, discharge and sediment regime, and initial channel geometry will all be scaled from the field condition. The choice of which variables to vary and the range of variation will be determined in coordination with the team implementing the field project that will serve as the case study for this set of experimental trials.