運用土地利用回歸模式評估高雄都會區細懸浮微粒及其元素濃度變異情形

Applying Land Use Regression Models to Evaluate Concentration Variations of PM2.5 and Elemental Composition in Kaohsiung Metropolis

林虹儀(Hung-I Lin)

指導教授 ： 吳章甫

國立臺灣大學/公共衛生學院/環境衛生研究所/碩士(2012年)

https://doi.org/10.6342/NTU.2012.10762

摘要

前言：空氣汙染物中的懸浮微粒與人類不良健康效應有關，而懸浮微粒的大小、形狀、化學組成都會影響其毒性表現，許多有毒的化學物質較多存在於細懸浮微粒(PM2.5)或粒徑更小的微粒當中，而細懸浮微粒由於粒徑較小可進入到人體肺泡交換氣體區，對人體造成更大的傷害，因此不容忽視。本研究利用土地利用回歸模式評估大高雄都會區細懸浮微粒及其元素組成濃度空間分布情形。
方法：本研究於2010年在高雄都會區，依地理環境與交通狀況選擇採樣點，採樣地區包含岡山區、橋頭區、楠梓區、左營區、仁武區、三民區、鹽埕區、苓雅區、鳳山區、大寮區、前鎮區、小港區、林園區。水平方向選取20個家戶(交通、都市背景採樣點各10個)、垂直方向10個家戶(中、高樓層各5個)進行細懸浮微粒監測。另外為校正季節差異，於採樣區域中設置一個連續監測點，於採樣期間架設採樣器進行每兩周一次長期監測。各家戶點採樣方法為使用Harvard PM2.5 Impactors、37-mm鐵氟龍濾紙結合幫浦進行為期兩季每次採集14天的戶外濃度量測，濾紙秤重後以能量散射X光螢光分析儀(energy dispersive X-ray fluorescence, ED-XRF)分析樣本元素組成。所得結果再以土地利用、道路長度、人口數、樓層高度作為預測變項，結合PM2.5以及元素年平均濃度，建立土地利用回歸模式。
結果：兩季平均經連續監測點濃度校正後汙染物濃度分為PM2.5：43.23±12.96 μg/m3，鋁：84.28±24.76 ng/m3、矽：279.58±36.85 ng/m3、硫：3236.46±852.40 ng/m3、鉀：328.93±91.56 ng/m3、鈣：110.07±22.38 ng/m3、鈦：14.19±2.41 ng/m3、錳：23.79±5.59 ng/m3、鐵：270.73±75.94 ng/m3、鎳：9.10±2.02 ng/m3、銅：10.28±3.27 ng/m3、鋅：174.94±68.38 ng/m3、鉛：55.31±24.58 ng/m3。水平空間採樣結果，交通採樣點PM2.5 連續採樣點濃度校正後年平均濃度為49.95±12.96μg/m3，顯著高於都市背景採樣點之35.76±8.34μg/m3，元素濃度則無此差異性。垂直空間分布結果，各樓層PM2.5連續採樣點濃度校正後年平均濃度隨著高度上升而濃度下降，濃度值分布為低樓層56.35±13.64μg/m3>中樓層30.07±7.16μg/m3及高樓層27.92±10.06μg/m3，元素組成部分並無空間垂直分布之差異。另一方面，使用逐步回歸分析結合土地利用資料與採樣結果建立PM2.5及其元素土地利用回歸模式，所得R2值範圍從0.234至0.799，而大部分元素土地利用回歸模式可解釋濃度空間變異程度優於PM2.5 土地利用回歸模式(Adjusted R2=0.397)，這顯示本研究使用之土地利用資料對於大部分的元素有較高解釋空間濃度變異的能力，然而鉀及鋅土地利用回歸模式可解釋濃度空間變異程度較差(鉀：0.234、鋅：0.259)，這顯示並非所有元素都可建立該汙染物之土地利用回歸模式。
結論：本研究在高雄都會地區建立細懸浮微粒及其元素組成之土地利用回歸模式，透過4至7個土地利用變項，即可有效預估50%以上汙染物(鋁、矽、硫、鈦、錳、鎳)濃度空間分布變異。未來可將此土地利用回歸模式應用於高雄都會區其他未採樣之地區，進行汙染物濃度的估計。

關鍵字

土地利用回歸模式 ； 地理資訊系統 ； 垂直分布 ； 細懸浮微粒 ； 元素

並列摘要

BACKGROUND: This study was designed to develop land use regression models to estimate ambient concentrations of PM2.5 and elements in Kaohsiung Metropolis and evaluate their spatial variation.
METHODS: Based on population density, micro-environmental characteristics, and building storey heights, we selected 30 locations to monitor PM2.5 and elements across Kaohsiung Metropolis (thirteen districts: Gangshan, Ciaotou, Nanzi, Zuoying, Renwu, Sanmin, Yancheng, Lingya, Fengshan, Daliao, Qianzhen, Xiaogang, and Linyuan), including 10 traffic and 10 urban background sites. One additional continuous monitoring site was selected for annual adjustment. Each sample was collected by the Harvard PM2.5 Impactors with a 37-mm 2-μm pore size Teflon filter and a sampling pump over a two-week period. To obtain data for different seasons, two sampling periods per site was conducted. The PM2.5 filter samples were further analyzed by energy dispersive X-ray fluorescence (ED-XRF) after weighing to acquire concentration of thirteen elements. LUR models were constructed by integrating land use information, traffic related data, population, and elevation as predicted variables with adjusted annual average concentrations in stepwise procedure for PM2.5 and 13 elements.
RESULTS: The adjusted annual average concentrations were 43.23±12.96 μg/m3 for PM2.5, 84.28±24.76 ng/m3 for Al, 279.58±36.85 ng/m3 for Si, 3236.46±852.40 ng/m3 for S, 328.93±91.56 ng/m3 for K, 110.07±22.38 ng/m3 for Ca, 14.19±2.41 ng/m3 for Ti, 23.79±5.59 ng/m3 for Mn, 270.73±75.94 ng/m3 for Fe, 9.10±2.02 ng/m3 for Ni, 10.28±3.27 ng/m3 for Cu, 174.94±68.38 ng/m3 for Zn, and 55.31±24.58 ng/m3 for Pb. Horizontal variation showed different trends of PM2.5 and its elementals. The adjusted annual average PM2.5 concentration was significant higher at the traffic sites (49.95±12.96μg/m3) as compared to that at the urban background sites (35.76±8.34μg/m3). The vertical profile of pollutants showed that PM2.5 concentration decreased with the height, but not the elemental concentrations. The adjusted R2 values of the model were 0.397 for PM2.5, 0.655 for Al, 0.702 for Si, 0.533 for S, 0.234 for K, 0.40 for Ca, 0.619 for Ti, 0.558 for Mn, 0.490 for Fe, 0.799 for Ni, 0.483 for Cu, 0.259 for Zn, and 0.378 for Pb. Better predictability of elements, except for K and Zn, was generally found than that of PM2.5.
CONCLUSION: We were able to develop land use regression models by combining household outdoor sampling data with geographic variables to evaluate concentration variations of PM2.5 and its elemental components in Kaohsiung metro. These PM2.5 and elements LUR models could be apply to other un-sampled locations in the study area to estimated ambient concentrations.

並列關鍵字

Land use regression ； Geographic information systems (GIS) ； Vertical distribution ； Fine particles (PM2.5) ； Elements

參考文獻

Beelen R, Hoek G, Pebesma E, Vienneau D, et al. 2009. Mapping of background air pollution at a fine spatial scale across the European Union. The Science of the total environment 407(6): 1852-1867.

Bencs L, Ravindra K, de Hoog J, Spolnik Z, et al. 2010. Appraisal of measurement methods, chemical composition and sources of fine atmospheric particles over six different areas of Northern Belgium. Environmental Pollution 158(11): 3421-3430.

Brauer M. 2002. Air Pollution from Traffic and the Development of Respiratory Infections and Asthmatic and Allergic Symptoms in Children. American Journal of Respiratory and Critical Care Medicine 166(8): 1092-1098.

Brauer M, Hoek G, Vliet Pv, Meliefste K, et al. 2003. Estimating long-term average particulate air pollution concentrations: application of traffic indicators and geographic information systems. EPIDEMIOLOGY 14: 228-239.

Briggs DJ, Collins S, Elliott P, Fischer P, et al. 1997. Mapping urban air pollution using GIS: a regression-based approach. International Journal of Geographical Information Science 11(7): 699-718.