The Impact of Temporal Rainfall Pattern Uncertainties on Water Quantity and Sediment Transportation Results of an Integrated Flood Simulator

Abstract

Meteorological dynamics are characterized by significant non-linearities and therefore, we cannot predict the long-term rainfall behavior. Hence, either the design of a new hydraulic work or the analysis of a current hydraulic system is performed using numerical models and simulating the impact of a synthetic storm at the ground, for a specific return period which is derived using the Intensity-Duration-Frequency curves of an area. However, the methodology used to produce synthetic storms, as well as the parameters incorporated in each methodology, are a source of uncertainty at the model output. In this work we investigate these uncertainties assuming water quantity metrics, such as the inundated area and water depths, and water quality metrics, such as the impact of sediment motion on the landscape transformation. To achieve this analysis, we use the well-known hydraulic software HEC-RAS, and specifically the Rain-on-Grid approach in a small catchment located in Xanthi (Greece). First, we perform a Morris-based Sensitivity analysis for the parameters of the Alternating Block Method (ABM), which is used to produce a synthetic storm, to screen the most influential parameters to the model output. Then, we perform an interval analysis for several rainfall disaggregation methodologies, to define the uncertainty band at the model output. It is found that the synthetic storm pattern and the parameters used for rainfall disaggregation create significant uncertainties. Moreover, it is found that the variables with quantitative vs. the corresponding with qualitative nature have different dynamics in terms of uncertainties produced by the model.