资源环境科技发展态势分析平台

Predicting the frequency and aerial extent of flooding in river valleys is essential for infrastructure design, environmental management, and risk assessment. Conventional flood prediction relies on assumptions of stationary flood distributions and static channel geometries. However, nonstationary flow regimes are increasingly observed and changes in flow and/or sediment supply are known to alter the geometry and flood conveyance of alluvial channels. Systematic changes in flows and/or channel geometry may amplify or attenuate the frequency and/or extent of flood inundation in unexpected ways. We present a stochastic, reduced complexity model to investigate such dynamics. The model routes a series of annual peak discharges through a simplified reach-averaged channel-floodplain cross section. Channel width, depth, and slope are permitted to adjust annually by a user-specified fraction toward equilibrium geometries predicted based on each year's peak discharge and sediment supply. Modeled channel adjustments are compared with empirical observations for two rivers in Minnesota, USA that have experienced multiple large floods over the past 6 years. The model is then run using six hypothetical scenarios simulating nonstationary flow regimes with temporal adjustments in the mean and/or variance of the governing peak-flow distributions. Each scenario is run repeatedly while varying parameters that control the amount of fractional adjustment that channel geometries can make annually. Results indicate that the intra-annual mean horizontal width of floodplain inundation primarily depends on the governing peak-flow distribution's coefficient of variation, but the intra-annual frequency of floodplain inundation (i.e., the fraction of modeled years with inundation) primarily depends on the amount of channel adjustment permitted annually.