Quantifying the parameter uncertainty in the cross-sectional dimensions for a stream restoration design of a gravel-bed stream

Abstract

Public agencies spend significant funds on stream restoration projects to improve the quality of impaired stream reaches. Many sources of uncertainty can potentially influence project outcomes, such as knowledge gaps in our understanding of fluvial systems as well as the natural stochasticity of parameters involved with the channel design process. A two-phase uncertainty analysis was performed on a two-stage channel stream restoration design for Stroubles Creek in Blacksburg, Virginia. Monte Carlo simulation was used to calculate the distribution of possible design outcomes for cross-sectional dimensions. Outcomes included width and depth of the main channel (stage 1) and width and depth of an inset floodplain (stage 2). The analysis incorporated stochastic uncertain parameters (bankfull discharge and grain-size distribution) and knowledge uncertain parameters (Manning9s <i>n</i> and critical Shield9s number). The results indicate that the design width can vary on average by 300% with respect to the deterministic solution. Design discharge was the most sensitive parameter for defining the stage 1 channel, while Manning9s <i>n</i> was the most sensitive for stage 2. The range in statistically probable design outcomes emphasizes the large uncertainty in channel design and suggests the potential for the channel planform and cross section of restored streams to evolve over time as established riparian vegetation matures. Additional uncertainties in need of future evaluation include (1) longitudinal variability of stream morphology, (2) design of instream structures, (3) temporal variability, and (4) knowledge errors in design models and measurements.