請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/87944
標題: | 以資料科學方法評估環境衝擊影響:以觀塘生態區為例 Data Science Approach in Environmental Impact Assessment: Case Study in Guantang Ecosystem |
作者: | 吳浚瑋 Chun-Wei Wu |
指導教授: | 廖世偉 Shih-Wei Liao |
關鍵字: | 結構方程式,R,lavaan,生態,資料科學, SEM,lavaan,R,Ecology,Data Science, |
出版年 : | 2023 |
學位: | 碩士 |
摘要: | 本研究旨在針對桃園觀塘生態區對三接天然氣接收站施工前後的數據做分 析,用以探討是否施工對於生態區有負面影響。根據低潮帶測站的環境參數與生 物參數,建構出具良好適配度的觀塘潮間帶藻礁生態系結構方程模型(RMSEA =0.046, CFI = 0.955, P-value of Chi-Square = 0.057), 具有解釋及預測藻礁生態系環 境因子與生物因子之間的因果關係。模型顯示關鍵的環境因子是水質的河川淡水 與營養鹽注入效應,營養鹽注入效應,微棲地的礫石覆蓋面積,沙覆蓋面積變異 度,及底泥重金屬銅,這三項環境因子對生物的分佈最有影響力。造礁藻喜好生長 於礫石多的地帶且能依附於礫石,同時,並且造礁藻與非造礁藻兩者覆蓋率呈空 間上的競爭關係。底表軟體動物、底表節肢動物喜好沙覆蓋變異程度小的區域。 底內節肢動物傾向分佈於底泥重金屬 銅 濃度高的區域。底泥銅濃度都在其濃度 的環境容許值之下,因此無需顧慮銅會對藻礁生態系有負面影響,但須考量生物毒 素累積效應。底內動物之多毛類分佈有利於底內星蟲動物、底內雙殼動物及底內 節肢動物之分佈,這顯示藻礁礁體為一多孔隙的立體構造物,提供底內多毛類、星 蟲、雙殼類、節肢動物共棲的空間。整體而言, 有些環境參數, 如 NO3-,呈季節變 化,Cu 濃度呈年間變化;有些則有樣區間的差異,例如沙覆蓋變異度。施工迄今,水質 參數、硝酸及銨氮營養鹽、各項重金屬含量及微棲地上礫石覆蓋面積比例或與時 間有關,或與河川輸出、農業相關,並無與施工有關的變化。最後,我們也探討了 關於 SEM 模型的修正方法以及模型識別,分別探討了兩個修正方法分別為移除非顯著影響路徑以及調整潛在變項成分組成,以及三種模型識別的要素分別為 (1) 模型的自由度必須至少為零。(2) 每個潛在變項(包括殘差項)必須被賦予一個尺 度,這意味著殘差項的路徑係數和其中一個潛在變項的因子負荷之一應被固定為 1,或者潛在變項的變異數應被固定為 1。(3) 每個潛變量應至少有兩個指標。藉 以探討更有效的 SEM 模型調教方法以及模型組成架構 This study aims to analyze the data from the Guam Tang Ecological Area before and after constructing of the Third Liquified Natural Gas Receiving Station to investigate whether the construction destroys the ecological area. Based on the environmental and biological parameters of the low intertidal zone stations, a close-fit Structural Equation Modeling (SEM) model of the Guan Tang intertidal zone coral reef ecosystem was constructed (RMSEA = 0.046, CFI = 0.955, P-value of Chi-Square = 0.057), which provided a causal relationship between environmental and biological factors for explaining and predicting the ecosystem. The model revealed that key environmental factors included the effects of river freshwater and nutrient influx, nutrient influx, the gravel habitat of microhabitats, sand cover variation, and sediment heavy metal. These five environmental factors had the most significant impact on the distribution of organisms. CCA preferred areas with more gravel and could attach to them, while there was spatial competition between CCA and non-CCA regarding coverage. CCA, epifauna- mollusk, and epifauna-arthropod preferred areas with low variation in sand coverage. Epifauna-arthropods tended to be distributed in areas with high sediment heavy metal Cu concentrations. The sediment Cu concentrations were within the effects range low, so there was no concern about Cu harming the coral reef ecosystem, but the accumulation of biotoxins should be considered. The distribution of infauna-polychaete effects the distribution of infauna-sipuncula, infauna-bivalves, and infauna-arthropods, indicating that the coral reef structure provided a three-dimensional porous structure for cohabitation of infauna-polychaete, infauna-sipuncula, infauna-bivalves, and infauna-arthropods. Overall, some environmental parameters, such as NO3-, showed seasonal variations and Cu concentrations showed annual variations. Some parameters exhibited differences between sampling areas, such as sand cover variation. Since the construction, water quality parameters, water quality data, various heavy metal contents, and coverage area proportion of microhabitats were either related to time or river discharge and agricultural components and not related to construction-related changes. Finally, we also explored the modification methods and model identification method for the SEM model. These were investigated to explore more effective methods for SEM model modification and model composition structure. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/87944 |
DOI: | 10.6342/NTU202300830 |
全文授權: | 同意授權(限校園內公開) |
顯示於系所單位: | 資訊工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-111-2.pdf 授權僅限NTU校內IP使用(校園外請利用VPN校外連線服務) | 2.56 MB | Adobe PDF | 檢視/開啟 |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。