Physically-based urban stormwater quality modelling: An efficient approach for calibration and sensitivity analysis.

Area of Science:

• Environmental Engineering
• Computational Hydrology

Background:

• Physically-based urban stormwater models offer insights into pollution dynamics but face computational challenges for calibration and validation.
• High computational costs hinder the widespread application of detailed urban water quantity and quality models.

Purpose of the Study:

• To develop a novel meta-model based framework for efficient calibration and sensitivity analysis of computationally intensive physically-based urban stormwater models.
• To enhance the usability of complex models for urban water management by reducing computational burden.

Main Methods:

• A meta-model framework was developed and applied to the FullSWOF-HR model.
• Adaptive stochastic collocation with sparse grids was used for parameter node selection.
• Interpolating polynomials generated the meta-model, followed by variance-based Sobol's method for sensitivity analysis and Markov chain Monte Carlo for calibration.

Main Results:

• The meta-model framework successfully performed sensitivity analysis, yielding results consistent with prior research.
• Parameter optimization and validation using the meta-model demonstrated efficiency and accuracy.
• The approach effectively reduced the computational demands of complex stormwater models.

Conclusions:

• The proposed meta-model based approach significantly improves the efficiency of sensitivity analysis and parameter optimization for complex physical stormwater quality models.
• This framework facilitates the broader adoption of detailed water quantity and quality modeling in urban water management.
• The study provides a valuable tool for understanding and managing urban pollution dynamics.