Evaluating the uncertainty of urban flood model using glue approach

Abstract

This study is an attempt to investigate and analyze uncertainty through the Glue method in parts of Tabriz, Iran, using the SWMM urban flood model. In order to quantify the uncertainty, the initial range of input parameters including curve number, impenetrability percentage, penetrability level’s coarseness coefficient, and impenetrability level’s coarseness coefficient are defined and the GLUE algorithm is used to conduct primary sampling operation from parametric space using Latin hypercube sampling method. Considering the simulation results and the observed values of synchronized events, about %20 of the total outputs and generated parameters series have been excluded as acceptable simulations. The results obtained from analyzing posterior distribution diagrams illustrated that the acceptable posterior for CN parameters are included in the same primary range for datasets. The second priority includes N prev parameter; however, parts of the primary given range have been obtained as the acceptable posterior range for %Imperv and N Imper parameters which displays the high sensitivity of these parameters. Finally, the proper ranges for the abovementioned parameters have been computed as 85.62–88.13. 0.0275–0.0325, 54.11–76.93, and 0.0118–0.0183, respectively.