9:30 AM - 9:45 AM
[AHW26-03] Modeling Surface–Subsurface Water Dynamics in Southern Taiwan Using coupled land surface and groundwater model - A Case of southern Taiwan
Keywords:High-resolution hydrologic modelling, Data-driven parameter estimation, ParFlow-CLM, Surface–groundwater interaction
In 2021, Taiwan experienced its most severe drought in a century, causing widespread water shortages, irrigation suspensions, and industrial disruptions. Simultaneously, the rapid growth of the high-tech industry has increased water demand in southern Taiwan, highlighting the need to better understand surface–subsurface water interactions for effective water management.
Traditional groundwater models often simulate only the saturated zone and primarily focus on downstream areas. These models typically rely on external estimates for key parameters, such as groundwater recharge and surface water–groundwater exchange, introducing substantial uncertainty due to parameter assumptions. To overcome these limitations, this study employs the integrated hydrologic model ParFlow-CLM, which couples surface and subsurface water processes by solving the 3D Richards equation for subsurface flow and applying a 2D kinematic wave approximation for surface water dynamics. This approach allows for the simulation of recharge and river–aquifer interactions directly within the model, reducing the reliance on uncertain external estimates. Additionally, ParFlow-CLM leverages parallel computing capabilities and CUDA-based GPU acceleration to significantly reduce computation time.
This research adopts a data-centric modeling approach to minimize the uncertainties associated with subjective parameter assumptions. Key innovations include the use of machine learning techniques to estimate bedrock depth, the application of groundwater-level fluctuations to infer pumping rates and storage coefficients, and the utilization of the Bayesian Maximum Entropy (BME) geostatistical approach to estimate hydraulic conductivity and lithological distributions.
The model is implemented in the Pingtung Plain of southern Taiwan, with a 250 x 250-meter grid and hourly simulations to capture high-resolution hydrologic dynamics, providing insights into groundwater–surface water interactions and helping identify potential zones for hyporheic flow development and siting for Managed Aquifer Recharge (MAR).
The results contribute to a more reliable foundation for water resource management in the region, supporting strategies to mitigate water shortages amid growing demand and climate change.
Traditional groundwater models often simulate only the saturated zone and primarily focus on downstream areas. These models typically rely on external estimates for key parameters, such as groundwater recharge and surface water–groundwater exchange, introducing substantial uncertainty due to parameter assumptions. To overcome these limitations, this study employs the integrated hydrologic model ParFlow-CLM, which couples surface and subsurface water processes by solving the 3D Richards equation for subsurface flow and applying a 2D kinematic wave approximation for surface water dynamics. This approach allows for the simulation of recharge and river–aquifer interactions directly within the model, reducing the reliance on uncertain external estimates. Additionally, ParFlow-CLM leverages parallel computing capabilities and CUDA-based GPU acceleration to significantly reduce computation time.
This research adopts a data-centric modeling approach to minimize the uncertainties associated with subjective parameter assumptions. Key innovations include the use of machine learning techniques to estimate bedrock depth, the application of groundwater-level fluctuations to infer pumping rates and storage coefficients, and the utilization of the Bayesian Maximum Entropy (BME) geostatistical approach to estimate hydraulic conductivity and lithological distributions.
The model is implemented in the Pingtung Plain of southern Taiwan, with a 250 x 250-meter grid and hourly simulations to capture high-resolution hydrologic dynamics, providing insights into groundwater–surface water interactions and helping identify potential zones for hyporheic flow development and siting for Managed Aquifer Recharge (MAR).
The results contribute to a more reliable foundation for water resource management in the region, supporting strategies to mitigate water shortages amid growing demand and climate change.