2008-08-012024-05-17https://scholars.lib.ntu.edu.tw/handle/123456789/681096摘要：致癌毒物─砷對環境與人類健康之影響係為社會所關切的議題，飲用含有高濃度砷之地下水，已經被證實與烏腳病有相關連，且曾經普遍存在於台灣西南沿岸地區。經濟部於1992年在雲林沿海地區開鑿21口地下水質觀測井，並開始觀測每季的水質變化。本研究收集此區的地下水質、地下水位、及雨量資料，首先以動態因子分析 (dynamic factor analysis)來探討影響地下水中砷濃度變化的主要因子。動態因子分析具有縮小維度的技術並可以提供(1)於時間序列的資料中是否存在任何基本的共同趨勢，(2)是否有交互作用存在，(3)是否受其他變量之影響。另類神經網路 (artificial neural networks, ANN)具有人腦運算及解決非線性問題之技術，用來模擬環境和水文地質中複雜的非線性水質變化關係。我們預期探討影響地下水中砷濃度變化的主要成因和地下水質及水文資料間複雜的相互作用關係，並建立砷濃度變化類神經網路推估模式。動態因子分析所得到的主要影響變量，將被應用於類神經網路中來架構水質預測模以預測砷濃度之變化，改善ANN模式不穩定性及過度訓練情況，擴大類神經網路之應用領域，提供可靠砷濃度推估結果，對於了解此區域內砷濃度變化有很大助益。對區域內砷濃度進行推估工作，提供區域內砷濃度變化，以維護當地民眾使用地下水之安全。同時依據此結果可減少居民誤飲用含高濃度砷地下水之危險，達地下水有效管理及利用之目的。<br> Abstract: With the great concern for the potential effects of arsenic on human health and the environment, there is a growing need for efficiently determining and modeling the presence and amount of arsenic in the ecohydrogeological systems. Drinking high Arsenic (As) concentrations from groundwater have been verified to be associated with blackfoot disease, which was once common in the southwestern coast of Taiwan. Ministry of Economic Affairs installed 21 groundwater observation wells distributed in the coastal area of Yun-Lin County, Taiwan, in 1992 to monitor the groundwater quality. Data of groundwater quality, groundwater level, and rainfall were collected for this study. In order to investigate the main factors affecting the fluctuations of As concentrations, dynamic factor analysis (DFA) is applied to the hydrological and water quality data obtained from an arsenic pollution area. Dynamic Factor Analysis, a dimension-reduction technique, provides information about whether (i) there are any underlying common patterns in the response time series, (ii) there are interactions between the response time series, and (iii) these are affected by the explanatory variables considered. Furthermore, the artificial neural network (ANN), a technique for the human brain’s problem–solving process, is applied to effectively construct the complex nonlinear relationships in the environmental and hydrogeological modeling. We expect to investigate underlying common patterns of As concentrations and interactions between the groundwater parameters and hydrological data, and to develop artificial intelligence techniques for predicting the As concentrations in hydrogeological systems. The Dynamic Factor Analysis could reduce the unsteady and over-fitting problems in the traditional ANN modeling. Moreover, the successfully developing prediction and management methods will elevate the efficiency in management practices for improving groundwater quality and decreasing contaminant areas. The results would further improve the realization of the spatial-temporal distribution of arsenic and decrease the risk of ingesting the high arsenic groundwater to reach the goal of efficiently controlling and usage of the groundwater.砷動態因子分析類神經網路Arsenicdynamic factor analysisartificial neural networks