Study on Groundwater Table Variations Induced by Earthquake in the Choushi River Alluvial Fan.

1999年，集集地震重創台灣，造成人民傷亡、房屋倒塌，更引發許多邊坡產生滑動破壞，也造成西部沖積平原地區許多起液化事件的發生。在國內外許多地震事件中，經常發現地下水位在地震發生前後有異常變化之現象。由經濟部水資源局在鄰近於車籠埔斷層西側的濁水溪沖積扇建立之地下水位監測井網，恰好記錄了濁水溪扇洲於集集地震期間之水位變化。 本研究以濁水溪沖積扇為研究區域，選定好修、文昌、合興、港後、九隆及田洋等六站，利用地層柱資料建立簡化分析剖面，進行富水地層受震之數值模擬，輔以濁水溪沖積扇地下水位監測井網之水位變化資料作驗證，並探討不同地下水受震模式之適用情形。 為能了解富水層受地震力作用下其孔隙水壓變化的關係，本研究首先使用SHAKE程式當中的反卷積(Deconvolution)計算功能推得地下水觀測井基盤位置的地震加速度歷時資料，再利用FLAC程式建立地下水觀測井的地層剖面，輸入地震加速度歷時記錄，藉由數值軟體得到分析結果，以了解地震與地下水位變化之相關機制。 本研究利用有限差分軟體 FLAC 程式建構孔隙水壓力動態反應模式的概念模型，此模型利用地下水觀測井之地層剖面資料，進行合理的土層簡化以利分析，同時以Finn及Biot兩種模式進行分析，進一步根據濁水溪沖積扇的地下水觀測井網，於集集地震發生前後所記錄到的實際地下水位變化量等資料，與數值分析結果進行比較及討論。分析後發現以Finn模式進行模擬的成果較Biot模式優異，不建議以Biot模式進行輸入震波於分析剖面之模型；地下水位變化分為階變反應變化及振盪反應變化，由Finn模式中可看出明顯的階變反應變化，而Biot模式則呈現振盪反應變化。此外，Finn模式的激發量較大，Biot模式的孔隙水壓變化量則是小很多。 比較數值分析結果與實際觀測變化，東西向輸入震波所得模擬結果較南北向輸入震波所得模擬結果更為優異，此與斷層錯動方向有關，車籠埔斷層的主要錯動方向為東向西錯動，因此震波的主要傳遞方向亦為東西向，故模擬結果以東西向輸入震波較佳。 整體而言，以Finn模式進行數值模擬的結果，在富水層二的模擬成效良好，並且與整體趨勢也相當符合，而富水層三次之，富水層四最差，可能是因為深層部分較接近模型底部，受到邊界條件所限制。此外，富水層一的部分，Finn模式的模擬結果與實際地下水位變化量有所差距，但模擬結果與震央距離、斷層距離以及地表加速度尖峰值的整體趨勢相當符合。以Biot模式進行數值模擬的結果，則是在一定的數值範圍中跳動，僅富水層一部分可得到相當不錯的模擬結果，隨著深度增加，其模擬水壓變化並沒有太大的改變。In 1999, Chi-Chi earthquake hit Taiwan, causing civilian casualties, house collapses, and landslides, also caused many liquefaction cases in the alluvial plain of western Taiwan. Groundwater level changes during earthquakes were recorded widely. The groundwater monitoring network established by the Water Resources Bureau to the west side of Chelungpu fault recorded the groundwater level changes during the Chi-Chi earthquake. In this research, records of six wells were chosen for simulation from Jhuoshuei River alluvial fan study area. They are Hao-Hsiu, Wen-Chang, He-Hsing, Kang-Hou, Chiu-Lung and Tien-Yang. A simplified stratigraphic column profile was used for the water-rich strata in the earthquake simulation. Results were comparing with the data of groundwater monitoring network for validation of water level changes using different modes of pore water excitation for the seismic analysis. In order to understand the seismic behavior of water-rich strata, this research uses SHAKE program for computing deconvolution function as the base motion of groundwater monitoring wells. Then FLAC was used with the simplified stratigraphic profile and input acceleration history to obtain analysis results for simulating the mechanism between earthquakes and groundwater level changes. Conceptual model of pore water pressure seismic response is constructed using the finite difference program, FLAC where both Finn and Biot modes of pore water pressure generation were used for analysis. Analysis showed that simulation of Finn mode is better than Biot mode, Biot mode is not recommended to be used in this analysis. There are two types of groundwater level changes, step-like and oscillating changes in response. The simulated water pressure curve from the Finn mode shows reactions of step-like changes, and Biot mode exhibits oscillating changes. In addition, Finn mode triggered a large amount change of the pore water pressure, while in Biot mode the pore pressure change is much smaller. Comparing numerical results with the observed data, the results of input E-W ground acceleration is better than input N-S acceleration. This is related to the east to west direction of the thrust action of Chelungpu fault. Therefore, the direction of seismic waves is more consistent in east-west direction, so the simulation results with E-W ground acceleration is better. Generally speaking, in Finn mode simulation results, the simulation of Aquifer 2 is fairly good and is quite consistent with the overall trend; while Aquifer 4 is the worst. This may be because the deeper part of the models closer to the bottom boundary, which constrains the horizontal and vertical directions motion at the bottom. In addition, in Aquifer 1, there are differences between the simulation results in Finn mode and the recorded groundwater level changes. But the variations of simulated results with the epicentral distance, fault distance and peak ground acceleration are quite consistent with the overall trend. Biot mode results in ground water changes in a small range, and only Aquifer 1can obtain good simulation results, when the depth increasing, the simulation of pore pressure change has not changed much