三維非均質含水層污染源的推求

Abstract

本研究利用污染源鑑定模式SATS-GWT，針對提升三維非均質含水層污染源的鑑定案例與機率，分別利用蒙地卡羅模擬(MCS)與Latin Hypercube Sampling (LHS)分層取樣法進行分析。在模擬污染擴散在非均質水力傳導場址時，首先以模擬退火演算法進行場址條件模擬，產生一組具有固定平均值、標準偏差、和空間相關係數的隨機非均質水力傳導係數場，作為真實污染場址，並以MODFLOW-GWT模擬得到採樣觀測濃度值。為了分析監測井數目對提升污染源鑑定機率的影響，第一個案例利用蒙地卡羅方法，在與先前相同的水文地質參數條件下，進一步產生另外50組不同的水力傳導場址，透過SATS-GWT模式來進行污染源位置的推求。第二個案例則利用LHS分層取樣法，LHS方法是一種統計取樣方式，透過LHS方法對1000組非均質場址做取樣，並與第一個案例的結果做比較。另外，第三個案例比較了五種不同監測井的採樣位置對於鑑定機率的影響。因此，針對各案例的模擬結果，我們提出在所設計的三維非均質含水層場址中，有效提升污染源鑑定機率的研究分析。

This study presents the analyses of contamination source identification probability in three-dimensional heterogeneous aquifers. A source identification model SATS-GWT is used to estimate the source information using Monte Carlo simulation (MCS) and Latin Hypercube Sampling (LHS) method. In the process of the source release simulation in the heterogeneous conductivity field, the computer program sasim, developed based on simulated annealing simulation and available in GSLIB, is first used to generate a random conditional hydraulic conductivity fields with a given mean, standard deviation, and correlation structure parameters. The measured concentrations at sampling points are simulated by MODFLOW-GWT with the assumed release concentration at known source location. For estimating the effect of the sampling numbers on source identification probability, the sasim program is used again to produce 50 realizations of random conductivity fields with known conductivities at some locations as conditioning data. The model SATS-GWT is applied to the contaminated site with one of the random conductivity fields to estimate the source information. Then the probability of obtaining the correct results can be estimated based on those 50 runs of Monte Carlo simulation (MCS). In addition, Latin Hypercube Sampling (LHS) method is also used for 50 conductivity fields drawn from a total of 1000 samples. Finally, a study is made to explore the effect of sampling location patterns on source identification probability. The simulated results regarding to the number of sampling points and the patterns of the sampling location for effectively determining source location in three-dimensional heterogeneous aquifers are concluded.