Uncertainty Analysis and Hydrologic and Hydraulic Model Linkage in the Watershed Modeling System

While the process of modeling and mapping a 100-year floodplain is inherently full of uncertainty, engineers are still required to determine a single line boundary for flood insurance rate and other floodplain studies. This is despite the fact that a good engineer knows that, within reasonable limits, different modeling parameters can be chosen that result in shifting the floodplain boundary either in or out. Because of the uncertainty associated with modeling input parameters it is not difficult to come up with multiple answers that are reasonable in approach, and yet quite different in results. The Environmental Modeling Research Laboratory (EMRL) at Brigham Young University (BYU), in conjunction with the Army Corps of Engineers (USACE), has developed modeling tools that allow for Monte Carlo type simulations to be run on developed hydrologic (HEC-1), hydraulic (HEC-RAS), and floodplain delineation models in order to determine a probabilistic flood plain boundary. By linking the models so that the result of one model can be used automatically as input to the next, several simulations can be run which explore the distribution of probable values on such important parameters as rainfall, watershed losses, discharge, and channel roughness. The map resulting from these simulations shows the probability of flooding within certain regions. The 10, 50, 100, or any other percent probability of flooding can easily be identified. This allows the engineer to create a visualization of the floodplain that is “honest” in terms of indicating the inherent uncertainty in the processes that are used to develop a floodplain map. By taking this process one step further and examining, through Monte Carlo type simulation, the entire space of probable values for each return period, a more certain boundary for the 100-year (or any return period) floodplain can be determined. The process of exploring the range of possible values over all probabilities is done in a fashion similar to the Corps of Engineers Flood Damage Assessment (HEC-FDA) program. The result of such a process results in a spatial map of annual exceedance probabilities. This map is determined by dividing the number of times a particular point is flooded from a simulation by the total number of simulations. Contouring the 0.01 exceedance probability results in a more certain 100-year floodplain that incorporates the uncertainty inherent in the modeling parameters. Such a line is not the result of a single set of modeling parameters and could not be reproduced by a single simulation; rather it is the composite of all the simulations. This paper and presentation will focus on the latest developments of the Watershed Modeling System in addressing uncertainty for hydrologic and hydraulic modeling.