移動污染空間性影響及風險評估

Abstract

汽機車所排放污染物是都巿空氣污染主要來源，且會隨著車流、車行里程及人口在不同區域造成不同程度的影響與風險，本研究因而發展適當的方法評估移污所造成的空間性影響及風險。且應用於案例區台北市。 移污空間分佈不易估算，雖有人提出道路密度(RND)法，但RND相似不代表車流大，因而會低估高污染區，實際誤差可能很大，本研究因而建立車流污染強度密度(VFPID)法改善之，但VFPID法未考量行駛距離不同亦會有不同排放量，故以車行里程污染強度密度(VTPID)法進一步改善之。各方法估算前首先依耗油量估算總污染排放量且將案例區劃分為多個網格，再分別依各網格RND、VFPID及VTPID所佔的比例分配總污染排放量至各網格中，並以失能人年DALY值來表示風險的大小。三個方法所得之移污分佈，再配合人口分佈，即可分析移污空間性風險。由於移污亦受街道形式影響，本研究亦因而以OSPM街谷模式模擬街道之移污濃度，再配合街道旁的人口，分析案例區街道移污對街道旁住戶的暴露風險。 結果顯示依RND法所分配的區域排放量與VFPID法的差異最大可達10000噸。而VTPID結果亦與VFPID結果差異可達5000噸；而由OSPM模擬結果可發現在不同型態街道會造成不同污染濃度，甚至與前三個方法產生不同的結果，例如敦化南路一段的車流雖然大於北安路，但街道寬度比北安路寬58公尺，OSPM模擬的CO濃度反而較後者低。依各方法結果及人口分佈所計算的風險值，以PM10的DALYs最高，NOx、CO次之，SOx最小，一般車流較大時人口亦較多，風險亦較高，如中山等區，但亦有因街道較窄而有較大風險的地區，如大安區。相信所發展的方法可改善移污空間分佈的推估品質，以利於進行相關的決策分析。

Vehicle exhaust emission is the main source of air pollution in metropolitan areas in Taiwan and greatly affects citizens and causes health risks. Different traffic flows, spatial mobile pollution distributions, and population distributions can cause different levels of impact and health risk. This study was thus initiated to develop appropriate methods for assessing the spatial impact and risk caused by the mobile pollution. The methods were also applied to Taipei City, the case study area for this study. It is not easy to evaluate the spatial distribution of the mobile pollution. Although a method called the Road Network Density (RND) method had been previously proposed, similar RNDs do not indicate similar traffic flows and thus may underestimate the pollution in high traffic-flow areas, and the error may be quite significant. Therefore, the Vehicle-flow-based Pollution Intensity Density (VFPID) method was proposed in this study to improve the problem. However, the VFPID method does not consider different traveling distances that can cause different emissions. The Vehicle-travel-mileage-based Pollution Intensity Density (VTPID) method was thus proposed. Before implementing the three methods, the total mobile pollution (TMP) emission was estimated according to the amount of gasoline consumed, and the entire study area was divided into grids in the same size. The TMP emission was allocated to each grid according to its RND, VFPID, or VTPID ratio. With the pollution distributions determined by the three methods and the population distribution of the city, spatial effects and health risks caused by mobile pollution were estimated. And the Disability Adjusted Life Years (DALYs) is used to express the risk level. Furthermore, since similar emission in different types of roads would give different distributions of mobile pollution, the Operational Street Pollution Model was adopted to simulate air pollutant concentrations on road sides, and the results obtained were used to assess the exposure risk on the residents living in the road sides. The results show that the pollution distribution estimated by the RND method can underestimate the pollution up to 10,000 tons when compared with that estimated by the VFPID method. Also the difference between the VTPID and VFPID results can be up to 5,000 tons. According to the OSPM simulation results, it can be observed that different types of streets cause varied pollution concentrations. Significantly different results were observed also when compared to those from the other three methods. For example, the traffic flow of the Section 1 of the Dunhua-South Road is more than that of the Bei-an Road, but the street width of the former is 58 meters wider than the latter street, and subsequently the simulated CO concentration is lower in the former street. According to the risks estimated from the obtained pollution and population distributions, the risk estimated for PM10 is the highest, the SOx risk is the lowest, and NOx and CO have median risks. When the traffic flow is large and population density is high, the associated risk is high too, such as the Zhongshan district. However, some areas have narrow streets that cause high risk, such as the Da-an District. It is believed that the proposed methods can improve the estimation of spatial mobile pollution distribution and subsequently facilitate further decision-making analyses.