SPATIOTEMPORAL STOCHASTIC OPEN-CHANNEL FLOW .2. SIMULATION EXPERIMENTS

Spatiotemporal solutions for open-channel flow are obtained in a stoch astic setting using field data on parameter variability. Statistical d escriptions of the how variables are estimated through Monte Carlo sim ulation using finite difference equations for a 10-km reach of the Col umbia River. Results indicate considerable uncertainty in predicted fl ow behavior: ensemble coefficients of variation at different space-tim e locations ranged from 0.18-0.60 for flow velocity and from 0.04-0.13 for flow depth. The band widths between the 16% and 84% quantiles wer e typically 0.6-1.4 m/s and 5-7 m, respectively, for velocity and dept h. Probability distributions for predicted velocities were found to be gamma, lognormal, or Weibull, whereas those for depth were normal, ga mma, and, in a few cases, lognormal. The various quantiles of the pred icted variables are associated with notions of risk, reliability, and variability that influence engineering decisions. Sensitivity of the l evel of uncertainty in predicted flow variables to the level of uncert ainty in the parameters is investigated for a generalized stream syste m through fractional factorial analysis of coefficients of variation. Uncertainty in predicted how velocity was most sensitive to the uncert ainty in the channel cross-section geometry, particularly scale and sh ape parameters for flow area. Uncertainty in predicted how depth was p redominantly sensitive to the uncertainty in channel bed slope.