Integrated model for the fast assessment of flood volume: Modelling – management, uncertainty and sensitivity analysis

Abstract

The specific flood volume is an important criterion for assessing the performance of sewage networks. It has been shown that its value is greatly influenced by the layout of the sewers in the catchment area, which is usually expressed by a fractal dimension. Currently, only mechanistic models (such as SWMM) enable the determination of the impact of the layout of the sewers on flooding volume, but they require additional and robust calculations. In the presented study an integrated tool has been proposed that includes: a flooding volume simulator based on rainfall data, catchment and sewage network characteristics, sewers layout expressed by fractal dimension. Alogistic model can be applied for fast flooding volume estimation as an alternative approach to SWMM, design and upgrade sewer layout even with limited access to data (spatial planning, architectural concepts, etc.). Using the random forest (RF) method, a likelihood function simulator was developed, which enabled the analysis of interactions and optimal selection of combinations of SWMM model parameters for calibration. It has been shown that the higher the fractal dimension and retention coefficient (the ratio of surface to sewer retention), the greater the influence of SWMM parameters on the specific flood volume.