結構方程模式於建立生態水質水文指標之研究

Structural Equation Modeling for Establishing the Multipurpose Hydrologic Indicators

張騌麒(Tsung-Chi Chang) ； 孫建平(Jian-Ping Suen)

《農業工程學報》 55卷3期 (2009/09) Pp. 75-87

https://doi.org/10.29974/JTAE.200909.0007

摘要

自然流態的變化特性對於溪流生態具有相當大的影響性，藉由溪流生態、水質與水文特徵變化關係的探討，可了解不同生物和流態間的連結，顯示生態水質水文指標的建立對於生態流量的管理十分重要。本研究利用主成份分析、結構方程模式等統計方法逐步篩選，挑選出具有代表水文氣候、水質狀況與生物多樣性等多重資訊的水文指標。結構方程模式主要運用於因果關係之分析探討，並具有多項的模式評鑑指標，使研究者可以進一步檢驗研究模式的整體適切性，達到提高研究精確度之目的，相當適合於探討水文指標與魚類生態之關聯性。本研究利用結構方程模式建立自五個水質模式與一個生物多樣性模式，水質模式探討了水文指標與河川污染指標間之因果關係，指出了水文指標與水質情況的相關性，生物多樣性模式則是探討了水文指標與夏農指數之因果關係，指出了反應水文特性、水質狀況之水文指標與生物多樣性之相關性，藉由水質模式與生物多樣性模式分析的結果，則可挑選出具有代表多重資訊之水文指標。此外，本研究亦採用了羅吉斯回歸分析，利用了魚類於五個流域之出現情況，將水文指標與魚類族群分佈情況加以連結，探討特定魚類族群偏好之水文特性。而藉由計算魚類族群在各個水文指標影響下於五個流域之出現機率，則可以了解多重資訊之水文指標與特定魚類族群之相關宜。本研究所得的指標，除了能獲得魚類族群對於特定水文、水質環境需求的資訊，並可提供水利單位作爲生態流量管理的參考，期許在水資源管理層面上可以對溪流生態系統的維護達到更大的幫助。

關鍵字

自然流態 ； 溪流生態 ； 生態水文指標 ； 主成份分析 ； 結構方程模式 ； 生物多樣性 ； 羅吉斯回歸

並列摘要

The variability of the natural flow regime has a significant influence on the river ecosystems. By examining the relationships between stream ecosystems, water quality, and hydrologic characteristics, we can understand the preference of different organisms to flow regimes. The development of the multipurpose hydrological indicators is quite important to the management of ecological flow. By using principal component analysis and structural equation modeling, we select the hydrological indicators that could express the significance of hydro-climate, water quality and fish species diversity. Then the logistic regression analysis model is used to examine the relation between fish and hydrologic indicators by using fish's present conditions. The structural equation modeling provides examination of a set of relationships between one or more independent variables and one or more dependent variables, so the modeler could explicitly capture the unreliability of measurement in the model, which in theory allows the structural relations between latent variables to be accurately estimated.
In this study, the structural equation model is used to build five water quality models and one biodiversity model. The water quality model examines the causality of hydrologic indicators and River Pollution Index, which expresses the relationship between the hydrological indicators and water quality condition, and the biodiversity model examines the causality of hydrological indicators and Shannon Index, which expresses the relationship between hydrological indicators and biodiversity. By analyzing water quality and biodiversity models, a suite of hydrological indicators representing multiple characteristics can be obtained. The multipurpose hydrological indicators selected from this research can not only provide the specific hydrologic and water quality information that fish requires, but also can provide the guidance for water resources authorities to achieve the goal of ecological restoration.

並列關鍵字

Natural Flow Regime ； Stream Ecology ； Ben-hydrological Indicator ； Principal Component Analysis ； Structural Equation Modeling ； Logistic Regression Analysis ； Biodiversity