- 學位論文
以空污感測器資料評估台北市細懸浮微粒之時空分布及其與交通流量之關係
Temporal and spatial distribution of PM2.5 pollution in Taipei City and its association with traffic flows assessed by microsensor (AirBox) data
摘要
背景: 台北市現今只有6個環保署監測站，因為監測站數量及分布的不足限制了以監測站的量測值來描述台北市行政區內細懸浮微粒 (PM2.5) 汙染的現象，例如: 解析度較高的PM2.5汙染時空變異特性和PM2.5汙染與交通汙染源之間的關係。近年來開始使用成本低、體積小特性、能夠廣泛設置的PM2.5微型感測器(Airbox)，提供我們一個改進解析台北市行政區內PM2.5汙染現象的機會。 目的: 本研究將探討台北市微型感測器的PM2.5時空分布情形，並且嘗試使用PM2.5微型感測器及車輛偵測器之車流量此兩種資料進一步探討台北市PM2.5濃度與車流量之關係。 材料方法: 本研究使用2020年8月至2021年7月之台北市微型感測器及環保署監測站PM2.5濃度資料，依行政區劃分比較兩者年平均與日平均的差異，並畫出逐時、逐日、逐月的時間趨勢圖，再以地理資訊系統(ArcGIS)空間分析工具的Kriging工具，分別做台北市PM2.5空間濃度推估圖來比較兩者的時空分布差異。本研究同時使用台北市交通局道路交通流量計數資料來探討PM2.5濃度與車流量之關係，我們將台北市各行政區的PM2.5的年平均逐時濃度與各行政區的年平均逐時總車流量做簡單線性回歸以量化車流量與PM2.5濃度之關係，此外將車流量結合土地利用資料來進一步建立PM2.5濃度的土地利用回歸模式(Land Use Regression, LUR)。 結果: 研究顯示台北市微型感測器校正前後的PM2.5濃度與環保署監測PM2.5濃度，在逐時、逐日及逐月均有相似的趨勢，而微感測器相較於環保署監測站能夠呈現更高解析度，例如區內的PM2.5濃度空間變異情形。 PM2.5濃度與車流量之間有顯著相關且線性回歸關係良好，台北市各行政區的R2 的值分布可達0.72至0.89之間，當總車流量每小時增加10000輛時，會使台北市12個行政區PM2.5濃度分別上升0.36μg/m3至1.64μg/m3之間。鑒於其資料完整性，針對士林區及中山區進行土地利用回歸模式建立，士林區年平均逐時土地利用回歸模式 (R2為0.87)能夠反應出士林區100公尺到5000公尺之間環域範圍內各種土地利用情況對PM2.5濃度的影響，包含宗教殯葬設施、交通設施、工業區、河川及水體、建築工地、主要道路及都市綠地及車流量對PM2.5濃度的影響。中山區年平均逐時土地利用回歸模式 (R2為0.66) 能夠反映出中山區100公尺到1000公尺之間環域範圍內各種土地利用情況對PM2.5濃度的影響，包含一般道路、宗教殯葬設施、交通設施、低密度住宅區及車流量對PM2.5濃度的影響。 結論: 本研究利用在小範圍內數量多及密度高的微型感測器(Airbox)的資料，成功解析台北市12個行政區內PM2.5濃度的時間和空間分布特性，也在結合逐時交通流量及土地利用資料後，成功建立台北市12個行政區內PM2.5濃度和交通流量存在相關性和有行政區特色的PM2.5濃度土地利用回歸模式。
並列摘要
Background: There are currently only 6 EPA air monitoring stations in Taipei City. Due to the insufficient number and distribution of air monitoring stations, we can’t use the air monitor data to represent the situation in PM2.5 pollution within the district of Taipei City, for example, high-resolution temporal and spatial variation in PM2.5 concentrations and its association with traffic flows. In recent years, PM2.5 microsensors (Airbox), which are low-cost and small in size, can be widely spread to provide us with an opportunity to improve the analysis of PM2.5 pollution in the district of Taipei City. Objectives: This study will discuss the spatiotemporal distribution of PM2.5 in Taipei City by Airbox, and try to use the two datasets of Airbox and vehicle detectors to further explore the relationship between PM2.5 concentration and traffic flow in Taipei City. Materials and methods: The data of PM2.5 from Taipei City's Airbox and EPA air monitoring stations from August 2020 to July 2021, and compared the difference on the annual average and daily average of PM2.5 between these two exposure data sources in each administrative division by plotting hourly, daily, monthly time trend charts. Further, we applied the Kriging approach of the geographic information system (ArcGIS) to estimate the spatial concentration of PM2.5 in Taipei City to compared differences in the spatial and temporal distribution between the two PM2.5 exposure data sources. We also used vehicle detectors data to evaluate the relationship between PM2.5 concentration and traffic flow. We used simple linear regression to quantify the relationship between the annual average of hourly PM2.5 concentration and traffic flow in each administrative district of Taipei City. And we further develop hourly PM2.5 land use regression model by combing the traffic flow and land use data. Results: There were similar hourly, daily, and monthly trends between EPA air monitoring data, Airbox air monitoring data, and Airbox air monitoring calibrated data. Compared with the EPA air monitoring station, the Airbox can display higher spatial resolution, such as the spatial variation of PM2.5 concentration within the district of Taipei City. Linear regression model showed significant correlation between traffic flow and PM2.5 concentration with the R2 of 12 districts from 0.72 to 0.89. An increase of 10,000 vehicles per hour in the total traffic flow could increase the PM2.5 concentration in each administrative district of Taipei City from 0.36μg/m3 to 1.64μg/m3.In view of its data integrity, we developd the land use regression model in Shilin District and Zhongshan District. The R2 of the annual average hourly PM2.5 land use regression model in the Shilin District is 0.87. This model could reflect the impacts of land use variable from 100 to 5000 meters buffer on the PM2.5 concentration in Shilin District, including folk, transportation facilities, industrial areas, rivers, construction, major roads, urban green spaces, and traffic flow. The R2 of the annual average hourly PM2.5 land-use regression model in Zhongshan District is 0.66. This model could reflect the impact of land use variable from 100 to 1000 meters buffer on the PM2.5 concentration in Zhongshan District, including general roads, folk, transportation facilities, low-density residential areas, and traffic flow. Conclusion: This study successfully found out the temporal and spatial characteristics of PM2.5 concentration in 12 districts of Taipei City using the data of Airbox with a large sample size and high density in a small area. After combining the hourly traffic flow and land use data, we successfully established the PM2.5 land use regression models with the characteristics of different administrative district.
參考文獻
延伸閱讀
- Kao, W. C. (2018). 利用高時間解析度資料進行台北地區細懸浮微粒來源分析與探討該地區細懸浮微粒成份之垂直空間變異 [master's thesis, National Taiwan University]. Airiti Library. https://doi.org/10.6342/NTU201804227
- 賴敏榛（2021）。以空氣品質監測網站數據檢測空氣中細懸浮微粒對大眾生活影響之研究〔碩士論文，義守大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0074-1009202121104200
- Lai, C. H. (2021). 臺北都會地區細及超細懸浮微粒之近道路量測與源解析 [master's thesis, National Taiwan University]. Airiti Library. https://doi.org/10.6342/NTU202102076
- 陳嘉惠、蔡詠名、張國楨（2019）。細懸浮微粒資料精確度校正探討：以臺北地區空氣盒子為例。地理研究，(70)，109-141。https://doi.org/10.6234/JGR.201905_(70).0005
- Chang, J. C. (2017). Application of Positive Matrix Factorization Model for Examining Spatial Variations of Exposure to PM2.5 with Different Height in Taipei Metropolis [master's thesis, National Taiwan University]. Airiti Library. https://doi.org/10.6342/NTU201703688