量化含水層邊界條件的採樣策略

Abstract

對含水層邊界條件進行全面採樣策略的表徵對於可持續地下水管理至關重要。本研究的主要目的是從空間和時間的角度評估在穩態和瞬態流動中使用由沙箱實驗生成的水力傳導係數（K）和比儲水量（Ss）估算值量化邊界條件的採樣策略。為實現這一目標，我們評估了四種錯誤邊界情況和一種真實邊界情況，在每種情況下進行了三種抽水測試方案（PT45、PT15 和 PT10），不包括僅在 PT45 進行的擴展域方案。在這些方案中，分配了恆定水頭和無通量邊界。採用連續線性估計器（SLE）算法進行參數估計後，進行了等高線圖和百分比差異分析。結果表明，瞬態流動在捕捉對邊界條件劃定有害的時間變化方面比穩態流動更有效，從而提供比穩態更詳細的邊界條件識別。不同的邊界情況表明，恆定水頭邊界比無通量邊界產生的 K 和 Ss 估算值更穩定，無通量邊界會引入一些異常。研究強調了在已知或疑似邊界附近布井的重要性，並整合了空間和時間採樣策略用於邊界條件識別和異質性分析。研究還得出結論，K 估算值比 Ss 估算值更適合用於邊界條件量化。通過使用井敏感性分佈圖觀察到，在沙箱中每隔 40 厘米至 60 厘米空間分佈至少 10 口井，可以有效地使用 K估算值來劃定邊界條件。為了進一步了解使用沙箱數據進行邊界量化的複雜性，未來的研究可以集中於調查無通量邊界引起異常的原因，以及為什麼邊界附近的井比模型中心區域的 Ss 敏感性低。

Abstract Comprehensive sampling strategies that characterize aquifer boundary conditions are essential for sustainable groundwater management. The primary objective of this study is to assess sampling strategies from spatial and temporal considerations in both steady state and transient state flow in quantifying boundary conditions using hydraulic conductivity (K) and specific storage (Ss) estimates generated from sandbox experiments. To achieve this, 4 wrong boundary scenarios and a true boundary were evaluated at three pumping test scenarios (PT45, PT15, and PT10) conducted in each scenario, excluding the extended domain scenario that was at PT45 only. Constant head and no flux boundaries were assigned in these scenarios. Contour maps and percentage difference analysis were conducted after employing the Successive Linear Estimator (SLE) algorithm for parameter estimation. Results demonstrate that transient state flow is more effective in capturing temporal variations that are detrimental in delineating boundary conditions than steady state flow thereby, providing more detailed boundary conditions identification compared to steady state. The different boundary scenarios indicated that constant head boundaries yield stable K and Ss estimates than no flux boundaries that introduce some anomalies. The study emphasizes on the significance of well placement near known or suspected boundaries and integrating spatial and temporal sampling strategies in boundary conditions identification and heterogeneity analysis. The study also concluded that K estimates are more suitable for boundary conditions quantification than Ss estimates. Using a well sensitivity distribution figure, it was observed that at least 10 wells spatially distributed in the sandbox at 40 cm - 60 cm apart could effectively help delineate boundary conditions using K estimates. To further understand the complexities of boundary quantification using sandbox data, future research could focus on the investigation of the causes of anomalies by no flux boundaries, and why wells along the boundary exhibited low Ss sensitivity than the central region of the model.