濁水溪流域地下水位抬升機制及補注量之研究

Investigating the Mechanisms of Groundwater Level Variation and Recharge at Zhuoshui River Basin

張琬渝(Wan-Yu Chang)

指導教授 ： 張斐章

國立臺灣大學/生物資源暨農學院/生物環境系統工程學研究所/碩士(2013年)

https://doi.org/10.6342/NTU.2013.00777

摘要

臺灣受限於地形因素及降雨型態分佈不均，大部份降雨在極短時間內直接逕流入海無法被有效利用，因此地下水因其成本低廉且取用方便等優點，便成為枯水期或缺乏儲水設施地區之重要水源，如何有效的保育及補注地下水資源已成為一大家重視的議題。濁水溪流域於中上游山區及下游沖積扇頂為良好的地下水補注區，然而目前針對濁水溪山區地下水為研究區域的文獻極少，探究此區域地面水與地下水的關係及建立模式有其必要性。本研究以濁水溪流域的中上游山區及下游沖積扇扇頂為研究區域，首先探究影響地下水位變動之因素，著重於分析水文因子對於地下水位的影響，運用統計方法分析累積雨量對於地下水觀測井水位變動之相關性，並嘗試訂定有效補注地下水之累積雨量門檻值，發現透水性較差的井或深井需要較長累積雨量日數才能有效造成觀測井水位抬升；採用非線性分析工具(Gamma Test, GT)篩選眾多雨量資訊，提供類神經網路之最佳輸入項，本研究以具優越非線性映射能力及高度精確性之倒傳遞類神經網路(BPNN)與具模糊規則庫之調適性網路模糊推論系統(ANFIS)建立推估模式，並以模糊推論系統佐以觀測井之地理、地質資訊做一地下水補注機制的特性分析，歸納出三種類型之水位抬升現象的觀測井，模式成果顯示BPNN與ANFIS都有相當良好的推估的表現；本研究亦推估濁水溪山區地下水平均年有效補注量，其推估值為10.42億噸。本研究探討濁水溪流域之降雨及河川流量和當地水文資訊，進一步掌握地下水變動的關係，以期能夠作為防治地層下陷之重要資訊，並提供濁水溪流域之水資源調配管理一參考依據。

關鍵字

地下水位 ； 補注機制 ； Gamma Test(GT) ； 類神經網路 ； 模糊推論系統

並列摘要

In Taiwan, most of rainfalls go straight into the ocean. Rainfall cannot be utilized efficiently due to topographical limitations and non-uniformly distributed rainfall patterns. Therefore, groundwater has become an important water source during drought periods and/or at the areas short of water storage facilities due to the low-cost and easy accessibility of groundwater. How to preserve and recharge groundwater effectively has become an important issue. The mountainous areas and the proximal-fan areas of the Jhuoshui River basin in Central Taiwan have been considered good groundwater recharge areas. However few researches on the recharge mechanisms in the mountainous areas of the Jhuoshui River basin can be found, therefore it is necessary to investigate the relationship between surface water and groundwater and to construct groundwater models at this area.
This study investigates the interactive mechanisms between surface water and groundwater, and the mountainous areas as well as the proximal-fan areas of the Jhuoshui River basin in Central Taiwan is the study area.
This study first investigates the mechanisms that result in the variations of groundwater levels and then focuses on the influence of surface water on groundwater level variations. Statistics methods are adopted to analyze the correlations between cumulative rainfall and groundwater level variation at groundwater monitoring wells, and the effective rainfall thresholds that cause efficient groundwater recharge activities can be identified. The results indicate that it requires accumulated rainfall of several days to make groundwater levels variable at low-permeability wells or deep wells.
This study next adopts the Gamma Test (GT) to select the critical input factors to the ANN models. Then both the backpropagation neural network (BPNN) in consideration of its superior nonlinear mapping ability as well as high estimation accuracy and the adaptive network fuzzy inference system (ANFIS) with a fuzzy rule base are used to construct estimation models for groundwater level variations at groundwater monitoring wells. Finally, this study adopts the fuzzy inference system with spatial and geological information of groundwater monitoring wells to analyze the characteristics of groundwater recharge mechanisms and further classifies three kinds of groundwater monitoring wells with a similar mechanism of water level variation for each type. Results indicate that both BPNN and ANFIS estimation models perform well. This study also estimates the average groundwater recharge over the mountainous areas of Jhuoshui River basin, with an estimated annual amount of 1.04 billion of tons.
In sum, this study investigates the rainfall and streamflow information in the Jhuoshui River basin, and further links the analytical results to groundwater level variations at groundwater monitoring wells. The results of this study can provide valuable information for the prevention as well as treatment of land subsidence and can be a good reference for water resources management in the Jhuoshui River basin.

並列關鍵字

Groundwater Level ； Recharge Mechanism ； Gamma Test (GT) ； Artificial Neural Network (ANN) ； Fuzzy Inference System (FIS)

參考文獻

69. 吳建宏、張天豪、王安培，2009，「模糊決策在洪水期間橋梁安全預警系統之應用－以新海大橋為例」，第17屆模糊理論及其應用研討會。

70. 李品輝，2009，「以類神經網路探討全台蒸發量區域性分類與推估之成效」，國立臺灣大學生物環境系統工程學研究所碩士論文。

72. 林正道，2003，「土石流危險度之模糊迴歸分析和綜合評判」，中原大學土木工程學系碩士學位論文。

73. 林承賢，2012，「以類神經網路建構濁水溪流域地下水位推估模式」，國立台灣大學碩士論文。

79. 洪耀明、萬絢、蘇苗彬、林裕益，2009，「崩塌地地下水位變化之即時預測」，水保技術，第4卷，第3期，第181-189頁。