Multiple Regression Analysis as a Comprehensive Tool to Model Flood Hazard in Sewersheds

Abstract

Flood modelling in urban areas is usually undertaken using stormwater tools, which are complex and time-consuming in terms of parametrization. To replace them, this research developed a methodology for predicting flooding probability in urban watersheds (sewersheds) through the modelling of peak flow rates from a set of watershed and sewer network-related factors relevant for the occurrence of floods. This was addressed through the stepped integration of Multiple Linear Regression (MLR), Multiple Nonlinear Regression (MNR) and Multiple Binary Logistic Regression (MBLR). A case study of a sewershed in Espoo (Finland) was used to validate the proposed approach and test it for future estimates, enabling the prediction of flooding probabilities under different Climate Change scenarios.