多湖區系統地表地下聯合操作之最佳化研究

Optimizing the Conjunctive Surface and Subsurface Operations of a Multi-Lake System

陳宇文(Yu-Wen Chen) ； 朱宏杰(Hone-Jay Chu) ； 張良正(Liang-Cheng Chang) ； 傅怡釧(Yi-Chuan Fu)

《中國土木水利工程學刊》 21卷3期 (2009/09) Pp. 295-305

https://doi.org/10.6652/JoCICHE.200909_21(3).0006

摘要

台灣近年來新水源開發的重要議題之一為以人工湖為主之水源調配方式，高屏大湖為臺灣第一個區域重要水源規劃案例，其為大型平地人工湖型式之案例，其營運方式較傳統地表水庫，或是僅具備補注功能的人工補注池更為複雜，操作時湖水與地下水間有明顯之出（／入）滲交換機制，本研究將針對平地人工湖形式之高屏大湖，結合遺傳演算法進行最佳操作規線之優選，其中以類神經網路模擬高屏大湖之系統反應。本研究以地下水數值模式（MODFLOW 96）與湖泊模組（LAK2）模擬並產生大量之湖水與地下水交換數據；接著以數值模擬結果訓練類神經網路群（ANNs），以此建立高屏大湖之反應方程式；最後結合遺傳演算法優選高屏大湖之最佳操作策略。研究顯示若不考量規線操作下，未封底案例相對於封底案例，可增加48%的供應量且能補注地下水；在最佳規線操作下可降低尖峰缺水量，相較於無規線操作案例，旬SI可由5.44降低至4.21；若湖區給予不同之打折係數，相較於相同打折係數案例，旬SI可由4.21降低至2.22。本研究發現平地人工湖形式之高屏大湖，透過未封底之地表地下聯合營運，可有效提昇高屏大湖的供水效益。此外，相較於無規線操作，規線操作可以進一步提昇供水效益。

關鍵字

聯合運用 ； 高屏大湖 ； 地下水 ； 類神經網路 ； 遺傳演算法

並列摘要

Artificial lake is considered a new water resources alternative in Taiwan in recent years. Kao-ping Artificial Lake is an example and it is a lake system with five artificial ponds. The groundwater level surrounding the lake is higher than those in the surrounding areas. Therefore, the determination of available water from the system is more complex than those for reservoir operations. This research considers the effects of the water supplied by the lake, the water stored in the lake, and the exchange between the lake and groundwater system in the operations of the Kao-ping Artificial Lake. The research develops an optimal conjunctive model for the Kao-ping Artificial Lake. However, if a numerical model is used to simulate the behavior of Kao-ping Artificial Lake operations, it will increase the computational burden. In order to solve the problem, a Back Propagation Neural Network (BPN) trained by simulation results for MODFLOW 96 and LAK2 is applied to represent the nonlinear dynamic relationship between the lake and the unconfined aquifer. Secondly, the water to be provided by the system at each time step is determined by an optimal rule curve found by a Genetic Algorithm (GA). Results of this study indicate that the conjunctive use model can significantly increase the water supply reliability. In addition, the model also provides an optimal groundwater recharge strategy. The model is shown to be able to reduce the magnitude of the water shortage and increase the resilience of the operating rule curve and is therefore believed to be a promising tool in Kao-ping Artificial Lake's future operations.

並列關鍵字

conjunctive use ； lake ； groundwater ； neural network ； genetic algorithm