考量水文地質架構之數值模式率定原則建立與應用

Abstract

地下水數值模式為地下水分析及管理之重要工具，在建立數值模式中，參數檢定為其重要之步驟。傳統上地下水參數檢定可分為自動化參數檢定及人工參數檢定；自動化參數檢定通常是結合地下水流模式與優選法，雖然可使檢定誤差達至最小且計算過程較快，惟其檢定過程是個半黑盒過程，建模者難以在檢定過程中進一步了解系統架構。人工參數檢定之檢定過程需要以人為方式決定參數值，較為費力費時且以往並無可遵循的原則，需具有模擬經驗者使能進行，惟人工檢定建模者可在過程中檢定過程中了解或修正水文地質架構，有助於進行後續結果分析。 有鑑於此，為了有效將人工參數檢定的專業知識轉移給他人，本研究以地下水流場之水力特徵為基礎，發展出以釐清水文地質架構與參數為導向之地下水數值模式檢定原則，參數調整原則如下：一、參數檢定順序；二、因應模擬水位趨勢的調整原則；三、因應模擬水位振幅的調整原則；四、阻水層範圍與垂向透水係數之調整；五、抽水量之調整；六、模擬水位乾枯之處理。 藉由上述檢定原則，檢定者可更深入掌握研究區域之水文地值架構。其中，以屏東平原內埔(1)之第一阻水層(T1)範圍為例，內埔(1)本身位於第二含水層(F2)，西勢(1)位於相對上游(西北方約10公里處，且位於第一含水層)，依地質調查所之地質圖，兩站間存在一延伸性良好之阻水層，從觀測上兩者水位相關性高，經數值模式數值試驗顯示，過大之阻水層範圍阻礙上下兩分層之水位連動，因此應打破地質圖延伸性良好之假設，縮小阻水層分布範圍。本研究已依據提出之檢定原則進行屏東平原地下水數值模式之建模與檢定，藉由數值試驗本研究可更深入掌握本區域之水文地質架構，供後續地下水管理研究之參考。

Groundwater numerical model is an important tool for groundwater management. To obtain a reliable model, model calibration is required. Conventionally, the approaches for model calibration can be classified into two categories: automatic calibration and manual calibration. Automatic calibration is normally based on the optimization method which can decrease the value of model error with high efficiency. However, the automatic calibration, which is a black box process, is difficult to be interpreted to researchers for understanding the groundwater system. In the other hand, manual calibration requires researchers to adjust model parameters and the model structure manually. This manual process consumes a lot of time and the efficiency is low. The advantage of manual calibration is that the process can includes more physical information such as grill’s log and geophysics. The manual calibration probes can help researchers understand the features of the studied groundwater system. In this study, a set of calibration principles for manual calibration are proposed and a revision of hydrogeological structure is also made based on the principles of manual calibration. This study develops model calibration principles considering hydrological structure to incorporate the advantages of both optimization and manual calibration. This model also has a benefit of transferring the expertise of manual calibration to other researchers. In this model, the hydraulic characteristic of the groundwater flow field is used as the physical base. The calibration principles are defined to clarify hydrogeological structure and associated parameters. These principles are: 1. The sequence of parameters for calibration. 2. The principles of parameter adjustment for the simulated water level trend. 3. The principles of parameter adjustment for the simulated water level fluctuation amplitude. 4. The principles of parameter adjustment for impervious layer range and vertical permeability coefficients. 5. Pumping volume adjustment. 6. The handling of dry cells in the simulation model. The developed methodology is applied to the first-impervious layer (T1) of Neipu (1) in Pintung Plain. Neipu (1) is at the second-aquifer (F2). Xishi (1), in the first-aquifer, is located in the northwestern upstream area with about 10 km in distance. According to hydrogeological map of Central Geological Survey, there is an impervious layer with favorable extensibility between two station. This impervious layer should hinder the groundwater flow. However, the water level observation shows high correlation between two stations. Therefore, the impervious area and the extension of the impervious layer maybe different to what was assumed. This study establishes and calibrates the groundwater numerical model in Pintung Plain using the propose calibration principles. The results show more clarification and details of the hydrogeological structure of the studied area. This improved hydrogeological information can be a good reference for future groundwater management.