A dynamic emulator for physically based flow simulators under varying rainfall and parametric conditions

Area of Science:

• Environmental Engineering
• Computational Fluid Dynamics
• Hydrology

Background:

• Physically based hydrodynamic simulators are computationally intensive, limiting their application.
• Existing surrogate modeling techniques struggle with high-dimensional parameter spaces and fixed rainfall scenarios.
• Traditional Unit Hydrograph (UH) theory has limitations with non-linear model structures due to proportionality and superposition assumptions.

Purpose of the Study:

• To develop an efficient emulation method for physically based hydrodynamic models.
• To overcome the limitations of existing surrogate models in handling large parameter spaces and dynamic rainfall.
• To enable practical applications such as real-time control and flood warning systems.

Main Methods:

• Combines a modified Unit Hydrograph (UH) scheme with polynomial chaos expansion (PCE).
• The emulator is designed to correct for UH theory's linearity assumptions in non-linear systems.
• Validated using a tank-in-series routing structure and 2D Shallow Water Equations (FLOW-R2D).

Main Results:

• The proposed emulator accurately approximates the behavior of physically based hydrodynamic models.
• Achieves high accuracy across a wide range of rainfall inputs and parametric values.
• Significantly reduces computational time compared to traditional simulators.

Conclusions:

• The UH-PCE emulator effectively addresses the computational burden of hydrodynamic simulations.
• It provides a feasible solution for real-time applications and complex scenario analysis.
• Demonstrates high accuracy and applicability in diverse hydrological modeling contexts.