SENSITIVITY OF PREDICTED IRRIGATION-DELIVERY PERFORMANCE TO HYDRAULICAND HYDROLOGIC UNCERTAINTY

A stochastic simulation model is developed which treats selected hydra ulic and hydrologic input parameters as random variables in predicting the performance of an irrigation-water-delivery system. The model is applied to a hypothetical earthen canal network representative of fiel d conditions in the upper Nile valley in Egypt to investigate the sens itivity of the relative variability in predicted system performance to the relative variability in the input parameters. The methodology com bines a model of steady spatially-varied canal network flow with stati stical models that generate possible realizations of the random hydrau lic and hydrologic parameters through Monte Carlo simulation. System p erformance is assessed by statistical analysis of predicted performanc e measures for adequacy, efficiency, dependability and equity of water delivery. Though the magnitude of the relative variability will vary for the particular system conditions, results from this study indicate the degree to which the coefficient of variation, CVomega, in predict ed system performance is sensitive to changes in the CVomega of the re spective input parameters. Results show that sensitivity to the CVomeg a in Manning hydraulic resistance and channel bed slope was low; sensi tivity to the CVomega, in irrigation application efficiency was low to moderate; sensitivity to the CV in upstream water supply level was mo derate to high; and sensitivity to the CV, in channel cross-section ge ometry and potential crop evapotranspiration was high. These results p rovide insight into the stochastic nature of irrigation canal network flows and indicate the comparative value of data describing the statis tical space-time variability of selected parameters.