運用土地利用迴歸模式結合移動監測評估臺北自行車道細懸浮微粒濃度之空間變異性

Land Use Regression Modeling with Mobile Monitoring to Estimate Spatial Variation of PM2.5 at Bicycle Lanes in Taipei

陳邦華(banghua chen)

指導教授 ： 吳章甫

國立臺灣大學/公共衛生學院/職業醫學與工業衛生研究所/碩士(2018年)

https://doi.org/10.6342/NTU201803470

摘要

近年來，由於大衆的環保意識日漸提升及對健康議題的重視，促使公共運輸系統的普及及自行車通勤人數的增加，而自行車專用道多毗鄰機動車道。當人們選用自行車作爲出勤方式時，就會更加容易暴露於PM2.5(空氣動力學直徑≤ 2.5μm的顆粒物)之下。
本研究針對此相關議題進行實地採樣研究，在2018年6月12日至6月25日間於交通尖峰時刻對臺北市都會區內八條自行車道及配對巷道、四條補充道路進行單人採樣及多人同時採樣兩階段共72小時實地採樣，同時收集採樣路徑周遭的土地利用資料，建立土地利用迴歸模式，並最終用於推估臺北市都會區內PM2.5之濃度分佈。
研究發現，兩階段採樣期間各路徑細懸浮微粒平均濃度分別爲5.3-59.4µg/m3及4.7-73.0µg/m3，同時透過Airvisual Node與TSI DustTrak™ II之模式對比，證明利用簡易式氣懸浮微粒監測儀建立土地利用迴歸模式是可行的。在探討不同時間、空間解析度下模式的表現後，第一階段最終選擇模式R2爲0.83，LOOCV驗證後R2爲0.81，選入的土地利用變項爲工商業區(100m環域)，主要道路面積(25m環域)，周遭四個環保署測站PM2.5濃度值，第二階段最終模式R2爲0.76，LOOCV驗證後R2爲0.75，選入的土地利用變項爲所有道路長度(100m環域)，所有道路面積(500m環域)及周遭四個環保署測站PM2.5濃度值，而後藉此二模式推估臺北市都會區交通尖峰時刻PM2.5濃度分佈圖，呈現了臺北市都會區PM2.5之空間變異，建議自行車出勤者應避開道路、工商業區密集的區域。

關鍵字

土地利用迴歸模式 ； PM2.5 ； 細懸浮微粒 ； 自行車通勤 ； 移動採樣

並列摘要

In recent years, due to increasing public awareness of environmental protection and attention to health issues, the popularity of public transport systems and the number of bicycle commutes have increased. The bike express lanes in Taipei were built adjacent to traffic lanes. Bikers may easily expose to traffic emission of PM2.5 (particulate matter ≤ 2.5 μm in aerodynamic diameter).
This study was conducted through a series of field campaigns to investigate the distribution of air pollutants near the bike lanes in Taipei City. We focus on eight bicycle lanes and paired alleys, as well as four supplementary roads, in Taipei metropolis area from 2018/6/12-2018/6/25. This study collected information on traffic, land use data and population density in Taipei metropolitan area. The regression model was used to estimate the concentration distribution of PM2.5 in Taipei metropolitan area.
The average PM2.5 concentrations of each biking routes during the two sampling periods were 53-59.4µg/m3 and 4.7-73.0µg/m3. By comparing the LUR model of Airvisual Node and TSI DustTrak™ II, it is showed that building LUR models with low-cost sensors is feasible. The R2 of the best LUR model in the stage #1 is 0.78, and the LOOCV validation R2 is 0.77. The selected land use variables are the industrial & business area (100m area buffer), major road area (25m area buffer), the PM2.5 concentration at the air quality monitoring station (AQMS). The R2 of the best LUR model in the stage #2 is 0.75, the LOOCV validation R2 is 0.75. The selected land use variables are the length of all roads (100m area buffer), all road areas (500m area buffer) and the PM2.5 concentration at the AQMS. According to the results of this study, it is recommended that bikers should avoid road-intensive areas and industrial & commercial areas.

並列關鍵字

Land Use Regression Models ； PM2.5 ； Exposure ； Mobile Monitoring

參考文獻

Aggarwal, S., et al. (2012). "Real-time prediction of size-resolved ultrafine particulate matter on freeways." Environmental Science & Technology 46(4): 2234-2241.

Briggs, D. J., et al. (1997). "Mapping urban air pollution using GIS: a regression-based approach." International Journal of Geographical Information Science 11(7): 699-718.

Dons, E., et al. (2011). "Impact of time–activity patterns on personal exposure to black carbon." Atmospheric Environment 45(21): 3594-3602.

Eeftens, M., et al. (2012). "Development of land use regression models for PM2. 5, PM2. 5 absorbance, PM10 and PMcoarse in 20 European study areas; results of the ESCAPE project." Environmental Science & Technology 46(20): 11195-11205.

Hagler, G. S., et al. (2010). "High-resolution mobile monitoring of carbon monoxide and ultrafine particle concentrations in a near-road environment." Journal of the Air & Waste Management Association 60(3): 328-336.