05/29/2018 Congratulate Mina Shahed Behrouz on her MS thesis defense

Mina Shahed Behrouz successfully defended her MS thesis on May 29. Congratulations!

Abstract Among various hydrologic models that are available to simulate urban runoff, the Storm Water Management Model (SWMM) is the most widely used numerical model. A typical SWMM project has about hundreds or thousands of sub-catchments and for each sub-catchment, there are more than 20 parameters associated with six different physical processes. Estimating all of these parameters are practically impossible, so model calibration is a challenging task. Manual calibration is used mostly but requires significant efforts and stops once simulation results are “satisfactory”. Some studies have adopted automatic calibration using a single objective optimization. However, an optimal parameter set obtained for one objective (e.g. peak flow) may perform poorly for another objective (e.g. average or low flow). In this study, SWMM was integrated with OSTRICH (Optimization Software Tool for Research Involving Computational Heuristics) to perform automatic multi-objective calibration. A sub-catchment within Buffalo, NY was selected as a case study. Automatic calibration using single and multiple objectives were conducted and compared. The newly developed OSTRICH-SWMM is proved to be a useful tool for calibrating SWMM models. This study shows multi-objective calibration provides a more robust parameter set than any single objectives and recommends multi-objective calibration for calibration of hydrologic models. A Pareto front is obtained in multi-objective calibration so an optimal solution can be selected according to trade-off. Moreover, this study highlights the importance of determining and considering time delay between simulated and observed values in model calibration.

Advisor: Dr. Zhenduo Zhu Committee: Dr. L. Shawn Matott, Dr. Alan Rabideau