Land subsidence induced by groundwater withdrawal: from conceptual model to Hydromechanical model in Chousui River Alluvial Fan, Taiwan

The Choushui River Alluvial Fan (CRAF) is facing serious land subsidence problems. In recent years, the main subsidence areas have gradually moved inland, causing security issues for the Taiwan High-Speed Rail (THSR) in the Yunlin county. Although pumping groundwater along the high-speed railway is forbidden, the problem of sinking land has remained. The causes of land subsidence can be multiple and complex. There are discussions about the causes of land subsidence in the Chousui River Alluvial Fan. This study aimed to develop a hydromechanical land subsidence model in this groundwater basin. The modeling process is divided into two broad parts: the groundwater flow model and the hydromechanical/land subsidence model. The development of the model was focused on the severe area of land subsidence in this basin; for this situation, the so-called “site specific model” was developed. Another reason this study used the site-specific model is that the Taiwan government has installed an integrated land subsidence monitoring system in the severe area, including Multilevel Compacting Well (MLCW), GNSS, and Groundwater observation well. This abundant data can be used when calibrating the groundwater flow and mechanical model. The modeling process starts with the construction of site-specific conceptual modeling derived from the basin scale conceptual modeling. This process revealed that the site consists of four aquifers and four aquitard layers with various thicknesses. The next process was creating a numerical groundwater model that began with creating a grid of the model domain. The model consists of fifty rows and fifty columns with ten by-ten meter grid cells and eight layers representing four aquifers and four aquitards. For perimeter boundary conditions, the model has specified head boundary conditions on the east and west part and no-flow boundary conditions on the north, south, and bottom of the model. the hydraulic and mechanical for the initial input of the model were generated using the previous study in this basin. Groundwater flow calibration processes were done using the PEST package. The model was evaluated using a multi-criteria performance meter: R-squared, root mean squared error (RMSE), mean absolute error (MAE), and Nash Sutcliffe Error (NSE). The calibration process for the groundwater flow model shows an excellent result for both mechanical and groundwater flow. The next step is modeling simulated subsidence using scheduling pumping using a different pumping rate scenario. This simulation aimed to reduce subsidence using calibrated pumping rate value but the difference in time of pumping. The result shows a significant subsidence reduction with scheduling pumping in a certain well. Any stakeholder can consider this result to reduce subsidence in the Chousui River Alluvial Fan.