桃園地區空氣污染物濃度相關性及地理分布

Statistics analysis and spatial homogeneity of air quality in Taoyuan County

邱瑞仙(Jui-hsien Chiu)

指導教授 ： 李崇德

國立中央大學/工學院/環境工程研究所碩士在職專班/碩士(2008年)

摘要

本文以環保署設於桃園縣的空氣品質監測站為評估對象，分析民國93年至94年六個自動空氣品質監測站的懸浮微粒、臭氧、二氧化硫、一氧化碳、氮氧化物各污染物資料，藉由統計分析的方式，解析各監測站測值的時空分布特性。
以Pearson相關係數矩陣分析桃園縣空氣品質測站與污染物濃度的相關性，結果顯示所有測站懸浮微粒濃度和臭氧濃度受到相同污染型態或擴散因子的影響。五權和中壢測站在二氧化硫相關性高，主要是測站污染物排放源或排放時間型態相似。一氧化碳各測站間相關性高，但中壢交通測站與其他各站相關性較低，可能是中壢測站受到車輛排放直接影響大，其他測站則是受到經大氣環境混合的影響。二氧化氮及氮氧化物測站間相關性高的因素為空間傳輸分布上相似。
本文以集群分析方法探討同一個污染物在測站空間分布的範圍，桃園和中壢測站在懸浮微粒濃度空間分布具有ㄧ致性。龍潭、觀音、大園和五權四測站在臭氧濃度相近，二氧化硫則以五權、中壢、大園和龍潭四測站濃度接近，可能是受到6公里內紡織、印染業等固定污染源排放硫氧化物的影響。在氮氧化物的空間分布中，中壢測站與其他各測站最不一致，顯示受到車輛排放直接影響大，這種現象在一氧化碳濃度空間分布也有相同的情形。
本文以各監測站污染物進行探討測站分布的主成分及轉軸分析，懸浮微粒只有一個主成分，顯示桃園縣的微粒濃度受到相同污染型態或擴散因子的影響。中壢測站與其他測站臭氧分布型態較為不同。大園、中壢和桃園測站和其他測站的二氧化硫來源有些不同。大園和觀音測站與其他測站的一氧化氮來源不同，中壢測站與其他測站的二氧化氮來源不同，氮氧化物則只顯示一個主成分，而且所有測站都有中高度權重，可說明所有測站的氮氧化物都來自類似來源。ㄧ氧化碳只顯示一個主成份，但中壢測站較其他測站權重小。
由污染物濃度與環境條件進行主成分分析後得到，發現影響桃園地區空氣品質變化的第一主成分為二氧化氮、一氧化碳、臭氧及風速，第二主成分為二氧化硫、懸浮微粒及溫度。第一主成分顯示車輛排放廢氣的影響，第二主成分主要顯示燃燒化石燃料的影響，可能包含本地和跨境傳輸來源。
本文以相關性分析、集群分析和主成分分析三種方法分析桃園縣各污染物濃度的相近程度，顯示懸浮微粒以相關性分析和主成分分析較接近。臭氧則以集群分析和主成分分析方法較為接近；相同分析方法結果也出現在二氧化氮及一氧化碳。二氧化硫的空間傳輸分布結果主要以相關性分析和集群分析方法較為接近。

關鍵字

桃園縣空氣品質 ； 統計分析 ； 監測站空間分布同質性

並列摘要

This study adopts data from six automated air-quality monitoring stations operated by Environmental Protection Administration (EPA) in Taoyuan County. The analyzed data include various pollutants such as suspended particulate matter (PM10), ozone (O3), sulfur dioxide (SO2), carbon monoxide (CO), and nitrogen oxides (NOx) dated from 2004 to 2005. The objective of this study is to resolve temporal and spatial characteristics of the monitoring stations through various statistical methods.
The correlation between air-quality monitoring stations in Taoyuan County and pollutant concentrations as determined by Pearson Correlation Coefficient matrix (Pearson) indicates that PM10 and O3 from all stations are affected by the same pollution types or dispersion factors. Owing to similar pollution emissions sources or emissions temporal pattern, the concentrations of SO2 from Wuchuan and Jhongli stations are highly correlated. When it comes to CO, the correlations among all stations are high except for Jhongli traffic station. This could possibly be due to direct influence by the exhaust from motor vehicles, while other stations are affected by a well-mixed atmosphere. The correlations of nitrogen dioxide (NO2) and NOx are high among stations due to the resemblance of spatial transport.
Cluster analysis is applied in this study to explore the spatial distribution of a pollutant in different monitoring stations. The PM10 spatial distribution is close for Taoyuan and Jhongli stations. Longtan, Guanyin, Dayuan, and Wuchuan stations all demonstrate they are consistent in O3 level within a reasonable difference. Meanwhile, SO2 concentration reveals a coherence among Wuchuan, Jhongli, Dayuan, and Longtan stations, which might be attributed to sulfur oxides emissions from textile and printing plants within 6 kilometers of the stations. Jhongli monitoring station is distinctly different from others on the spatial distribution of NOx, which shows the direct impact from motor vehicles’ exhaust. The situation is similar for the spatial distribution of CO concentration.
Results from principal component analysis (PCA) with Varimax rotation for pollutants from each monitoring station are as follows. PM10 is found to result in only one major component which implies particulate matter in Taoyuan County is affected by the same pollution types or dispersion factors. O3 distribution is different in Jhongli than others. Sources of SO2 in Dayuan, Jhongli, and Taoyuan stations are somewhat different from other stations. Dayuan and Guanyin stations distinguish themself from others on nitric dioxide (NO) sources, while Jhongli is different from others in NO2. In the meantime, NOx concentrations show only one single component with medium-to-high weightings in all monitoring stations, which illustrating NOx come from the same sources wherever the station is. Although all stations are also grouped in one component for CO, Jhongli monitoring station is rather low weighted in comparison with other monitoring stations.
The PCA is also applied to all Taoyuan stations for various pollutants and environmental factors. The first component is composed of NO2, CO, O3 and wind speed in Taoyuan. It indicates that vehicle exhaust has a significant influence on Taoyuan’s air quality. The second component is associated with SO2 PM10, and temperature, which means fossil fuel burning from both local and cross-border transport affecting Taoyuan’s air quality.
This study applied three statistical methods including Pearson, cluster analysis, and PCA to investigate similarities of pollutant levels for monitoring stations in Taoyuan County. The results indicate that spatial distribution of PM10 is similar from Pearson and PCA methods. The analysis on O3 shows that cluster analysis and PCA are close to each other. An identical result also appears on NO2 and CO. With regard to spatial distribution of SO2, it shows proximity between Pearson and cluster analysis.

並列關鍵字

statistics analysis ； spatial homogeneity of monitoring station ； air quality in Taoyuan County

參考文獻

52. 唐振雄，台北市空氣監測站間污染物濃度分布及特性分析，國立台灣大學環境工程研究所碩士論文，2004。

12. Chen, C. H., Liaw, S. L., Wu, R. S., Chen, Q. L., and Chen, J. L., “Development of a Decision Making Theory and a Decision Support System for River Basin Water Management”, Proceedings of the National Science Council-- Part A: Physical Science and Engineering, Vol. 21, No. 5, pp. 389-409 ,(2006).

32. 柳中明、蘇維中，區域性氣象環境與高臭氧之相關分析，大氣化學，第25卷，第1期，第22-49頁，1997。

37. 李清勝，影響大台北地區懸浮微粒濃度變化之氣象分析，大氣科學，第20卷，第4期，第341-361頁，1992。

1. Calori, G., Finzi, G. and Tonezzer, C., “A Decision-Support System For Air-Quality Network Design”, Environmental Monitoring and Assessment, Vol. 33, pp. 101-114, (1994).