水鳥種數時序分析與空氣品質影響之研究

外文標題: 
Time series analysis of waterfowl species and air quality impact study
校院系所: 
中興大學 水土保持學系所
指導教授: 
陳鴻烈
出版年份: 
2009年
主題類別: 
摘要: 

本研究以大肚溪口鳥類調查資料計算水鳥種數 (Number of species),並利用相乘性分解模式與相加性季節變動模式進行大肚溪口水鳥種數之時間數列分析,以瞭解78至96年間大肚溪口水鳥種數變化情形。兩模式之水鳥種數長期趨勢 (T) 都顯示隨著時間 (t) 而減少。在相乘模式中,其間的關係可表示為 T = 61.575586 - 0.04532t;相加模式中則可表示為 T = 61.638312 - 0.046083t,其中 t = 1, 2, 3, … 228。在季節變動 (S) 方面,結果顯示S於每年4月及11月分別為2個高峰,4月為冬候鳥之高峰,11月為夏候鳥之高峰,其中4月的高峰最高,故大肚溪口的水鳥以冬候鳥為主。在循環變動 (C) 中發現,民國78至95年6月間為一個大循環,期間約每間隔二年為一個小循環。而不規則變動 (I) 則顯示兩模式皆呈隨機不規則變化。由上述可知,兩模式之結果大致相同,且大肚溪口水鳥種數有明顯下降的趨勢。 在地理統計方面,距離類別中之距離平方成反比預測法比距離反比權重預測法正確。Surfer當中最佳預測方法為8.0版的Local Polynomial interpolation為最佳。在距離類別中之Arc View預測結果皆優於Surfer。本研究五大類預測方法中,以ArcView方法中之Local Polynomial interpolation預測結果最接近實測值,故在本研究中此法為最佳方法,並以此方法推估水鳥樣區中心之空氣污染濃度,且以此數據為基礎進行水鳥種數及空氣污染參數之動力分析。 在動力分析方面,於敘述統計結果得知此水鳥樣區許多空氣污染指標呈現具有污染潛勢,包括NO2、NOx、PM10、 SO2、O3、CO等參數均指出曾發生過污染事件,污染可能會導致水鳥族群的變化;而在單迴歸分析結果顯示,與水鳥種數為顯著迴歸參數包括 NO2(avg)、NOx(avg)、PM10(avg)、O3(avg)、NO2(max)、NOx(max)、O3(max)等7個參數,顯示出空氣污染狀況確能部分解釋水鳥種數的變化情形;複迴歸之分析結果,顯示氮氧化物相關參數中,可以NO2(avg)與NO(max)作為代表解釋水鳥變化,而所有參數中又以PM10(avg)對水鳥種數之變化影響最大;在主成份分析方面,第一主成份其組成幾乎涵括所有重要空氣污染參數,其中最重要參數為 NOx(avg)。而複迴歸模式與主成份迴歸選擇結果仍均顯示透過逐步(向後)選擇迴歸分析,可得最佳模式。

外文摘要: 

Waterfowl is one of the most important ecological bio-indexes and human activities would greatly affect amounts of waterfowl with its environmental habitat by pollution. Air quality probably is one of the key factors for waterfowl secession. Therefore related to this issue, this study focus on following three objectives: (1) to apply time series models to understand temporal variation of waterfowl species in Da-du estuary, at central part of Taiwan, (2) to simulate air pollution data of the habitat by the 26 geostatistical methods and finally (3) to discuss the mechanism of species variation with air quality by multivariate analysis. In time series analysis, a multiplicative decomposition method as well as additive decomposition method has been adapted to determine and evaluate the species secession, including long-term, seasonal, circular, and irregular changes. Geostatistics was applied to estimate air quality data, categorized by 5 groups of methods, including of mathematical, distance, polygon, Arc View and Surfer method. The estimated values were evaluated by error sum of square (ESS). For effectiveness analysis of relationship between air quality and waterfowl species, descriptive statistics, simple regression, multiple regression and principal component analysis were adapted. In the results of time series models, the two methods indicated similar outcomes: (1) both were shown a long-term trend (T) decreased with time (t), as following equations, T = 61.575586 - 0.04532t for multiplicative decomposition method, T = 61.638312 - 0.046083t for additive decomposition method; (2) both models all indicate there are two seasonal high peaks in April and November each year; and (3) a circular change, every two years. Both methods were reasonable well to present the waterfowl species changes numerically. For geostatistical method selection, the polynomial method in Kriging group exhibited the best result and by this method we estimate the air quality for the study waterfowl habitat. In the results of multivariate analysis, PM10 and NOx are the most related parameters for the mechanism of waterfowl species variation.