TR Design Concepts/Tools

Introduction

The overall approach used to assess the effect of riffle-pool relief on river rehabilitation for the Trinity River was to (a) design six synthetic river digital elevation models using AutoCAD Land Desktop®with different riffle-pool sequence configurations for a given testbed regulated river reach, (b) conduct 2D modeling of the six different synthetic designs at two significant discharges, (c) extrapolate hydraulic predictions through physical-habitat curves and sediment transport regime equations, and (d) extract and compare performance indicators to determine the relative merits of the designs. Two configurations (designs 1 and 2 below) used a high riffle-pool amplitude, three used a low one (designs 4-6) , and one (design 3) had an intermediate value with hybrid features.

Design Hypothesis Concept

A key feature in SHIRA is the adoption of a scientific hypothesis testing methodology for design development (e.g. Brown and Pasternack, 2009) and for long-term monitoring of project outcomes (e.g. Wheaton et al., 2010). A design objective is a specific goal that is aimed for when a project plan is implemented. To achieve the objective, it has to be translated into a design hypothesis. According to Wheaton et al. (2004b), a design hypothesis is a mechanistic inference, formulated on the basis of scientific literature review and available site-specific data, and thus is assumed true as a general scientific principle. Once a design hypothesis is stated, then specific morphological features are designed to work with the flow regime to yield the mechanism in the design hypothesis. Finally, a test is formulated to determine in design evaluation using 2D modeling and after real-world implementation whether the design hypothesis was appropriate for the project and the degree to which the design objective was achieved. The six alternative designs for the Lewiston Hatchery Reach were created with diverse features that have many specific design hypotheses. For each design shown in this site, all of the design hypotheses involved in the design are transparently listed.

Design Concepts

For each design element, a suite of concepts and objective design tools aided the creative process of conceptualizing landforms and articulating their value toward hydraulic, geomorphic, and fisheries objectives in the form of design hypotheses. Alternative designs were guided by the extensive scientific knowledge for riffle-pool assemblages and the less abundant knowledge for sub-width channel features that has been rapidly emerging through studies of the structure, organization, and function of morphological units, biotopes, and mesohabitats.

Research to date suggests that flow-dependent width variations between riffle and pool units is a prerequisite for sustainable (e.g. resilient during floods) riffle-pool units, assuming sediment supply is not limiting and in the absence of extreme bed-material grain-size differences between riffles and pools (Carling and Wood, 1994; MacWilliams et al., 2006; Wheaton et al., 2010; Thompson, 2011). However, a key constraint imposed by the setting was a minimal opportunity for channel widening (or implementation of width undulation), due to the presence of bedrock on river right and the hatchery on river left (Brown and Pasternack, 2008). Therefore, adding gravel/cobble fill and adjusting the amplitude and relief of morphologic units were the primary opportunities to enhance riffle-pool units.

Although substantial width undulation would be desirable for riffle-pool resilience even in confined settings (White et al., 2010), the top reaches below most large dams are channelized, constricted, and at least armored, if not scoured to bedrock and the flow regime is such that higher valley wall oscillations are no longer engaged. Furthermore, the standard construction method for gravel placement to build the kind of designs evaluated in this study uses front loaders (Sawyer et al., 2009) and involves bulk placement of the sediment mixture; it is uncommon to design and install different surficial bed-material facies.

For plan view in-channel landform patterns, some of the relevant design concerns included the lateral distribution of mesohabitats, stage-dependent resilience of microhabitat patches, sediment routing through pools, knickpoint migration through riffle crests, the resilience of riffle-pool relief, and accessibility of pools preferred for recreation fishing. Some specific plan view morphometric characteristics that were targeted for careful design and evaluate included degree of pool constriction (in an attempt to instill some riffle-pool width variation in support of stage-dependent maintenance of relief), plan view shape of riffle exit (e.g. horseshoe shaped, straight, or convex), diagonal riffles to promote lateral streamlines, partial riffle-crest notches/chutes (for bypassing excess flows to protect riffle crests), central bars (for stage-dependent microhabitat resilience), and localized forced pools adjacent to anthropogenic hard points. There exists little peer-reviewed scientific and engineering guidance for the design of these complex spatial patterns at the sub-width scale (e.g. Pasternack et al., 2004), but grey literature government reports and design handbooks were consulted for ideas. Evaluation and selection of these features hinged on 2D hydrodynamic modeling (as detailed below).

Design Tools

Design tools were utilized in a hierarchical fashion, where simpler analytical and empirical tools were utilized before more sophisticated tools such as 2D modeling were employed. Hierarchical design has been a necessary component of applied river rehabilitation thus far, because it considers efficiencies in cost and effort. Initially, riffle and bar analogs for channel morphology were generated through fuzzy visualization from similar morphologic unit scale features in unregulated rivers. Next, these analogs were scaled to the existing site topography and overlaid on the existing topography. The spacing and location of riffles were determined somewhat by existing pool locations and more so by fixed forcing elements, but this was varied between designs to the extend feasible. Analytical and empirical testing of riffle crests was performed for low-flow conditions to determine if riffle crests had gross hydraulic properties consistent with ecological and geomorphological goals and objectives. Finally, 2D modeling of final design surfaces was undertaken to evaluate each design with respect to performance indicators and spatial mechanisms.

For riffle-pool relief, some of the relevant design concerns included base flow riffle and pool habitat suitability and quality, stage-dependent riffle scour potential, knickpoint migration through riffles, and the resilience of riffle-pool relief. Early in design, initial sizing of riffle crest geometry was done iteratively using a low-order predictive tool in the hierarchical design context. Because a riffle crest functions like a weir and causes a backwater effect (Clifford et al., 2005; Pasternack et al., 2008), the Cipoletti weir equation was used as a design tool for estimating flow depth for each riffle in each design to give a quick and crude estimate of hydraulics during design development without having to run a hydrodynamic model (USBR, 1953). Once depth was estimated, the mass conservation equation was used to estimate riffle-crest velocity. Estimated riffle depths and velocities for the regulated baseflow typically present during salmon spawning, embryo incubation, and fry development were used to get a sense of habitat quality and gravel scour potential at that flow.

Once alternative designs were finalized, the two slope-detrended, riffle-pool relief metrics defined earlier (riffle-pool amplitude and the asymmetry index) were computed and analyzed. First, longitudinal profiles for the baseline and design DEMs were slope-detrended using linear regression (Richards, 1976a,b; McKean et al., 2008). Then the relief metrics were computed for each individual pool-riffle pair (in that sequence). Next, metrics were averaged among pool-riffle pairs for each design. Finally, the metrics for each design were compared to each other and those for the baseline DEM.