Efficient Calibration and Validation of Physical Stormwater Quality Modelling by Meta-model Based Approach

Model calibration and validation is still challenging for applications of physical stormwater quality model in the field of urban drainage modelling. In this context, this study aims to develop a new meta-model based framework for efficient calibration of complex and computationally intensive physically-based models. The proposed approach is applied to the physical FullSWOF-HR model for optimizing the washoff parameters. According to the average rainfall intensity, eight rainfall events are categorized into three groups for parameter optimisation, such as three light rains, three moderate rains and two heavy rains. After upscaling the original model, 77 parameter sampling experiments can be defined by a convergence analysis. Applying these 77 parameters series in FullSWOF-HR simulation runs, the interpolating polynomial of the original model is then generated by using the adaptive stochastic collocation method, which adopts sparse grid algorithm and selects the important parameters adaptively and automatically. Calibration process of the meta-model is based on the Markov chain Monte Carlo (MCMC) method. The optimized parameters are verified with the original model and then validated for different rainfall events. These promising results show that the proposed meta-model based approach can efficiently calibrate parameters for complex physical stormwater quality models. Our ongoing work focuses on the sensitivity/uncertainty analysis with this new meta-model based approach.