| 英文摘要 |
This project aims to conduct intercomparisons between manual and automated instruments for fine particulate matter (PM2.5) during precision and regional periods. The performance of manual instruments showed no significant difference statistically (p<0.05) from each other during the precision intercomparison period. The automated instruments were found no equivalence to manual instruments based on the assessment made for the equivalence of class III candidate instruments to the U.S. Federal Equivalent Method (hereinafter refer to as “FEM criteria”). However, the measurements of automated instrument can be converted to that of the manual instrument through linear regression modeling.
During the precision intercomparison period, the absolute deviation in percentages of PM2.5 mass concentration between the average of the two automated instruments Met One BAM 1020 (hereinafter refer to as “Met One”) and Thermo R&P 1405-F FDMS (hereinafter refer to as “Thermo”) and the average of the two manual instruments were both at 21%. Since the intercept for Thermo and the slope and intercept of the linear regression for Met One failed to comply with the FEM criteria, both automated instruments are considered a failure in complying with the FEM criteria although other items may pass. During the regional intercomparison period, the absolute deviation in percentages of PM2.5 mass concentration between the average of Met One and manual instruments were 18% at the Hsinchuang station, 29% at the Lunbei station, and 11% at the Cianjhen station, respectively. In contrast, those of Thermo and manual instruments were 18% at the Hsinchuang station, 19% at the Lunbei station, and 13% at the Cianjhen station, respectively. At the Cianjhen station, Met One met all the requirements of the FEM criteria while Thermo failed by a narrow margin. The performance of the two automated instruments did not comply with the FEM criteria for the other two stations. The difference of 24-hour averages between Thermo and BGI PQ 200 manual instrument is related to the volatilized species concentration measured by Thermo; however, the measured volatilized species concentration cannot account for this difference completely. Among the volatilized species measured by the manual instrument, the semi-volatile NO3- concentration was found to play the most important role. Deposited particulate NO3- on the Thermo PM2.5 measuring filter will evaporate when the environmental temperature is lower than 300C as the measuring temperature of Thermo is set at 300C. During the study period, the highest seasonal averages in PM2.5 mass concentration at the Hsinchuang, Lunbei, and Cianjhen stations were 42±21 μg m-3, 42±17 μg m-3, and 83±24 μg m-3, respectively. Mean PM2.5 mass concentration of 37±8 μg m-3 was found at the remote Kinmen Island for five days’ collection in May. The dominant PM2.5 chemical species in the three stations and Kinmen is SO42- followed by the modified NO3- or modified organic carbon.
For the international aspect, Europe and the United States of America (USA) reached a consensus to control black carbon from emission sources. A Model Attainment Test Software (MATS) developed by the USA Environmental Protection Agency features the combination of PM2.5 mass and species concentrations from manual instrument with simulation results from a grid-cell model. MATS is capable of estimating future PM2.5 mass concentration upon selecting various source control scenarios. For the future PM2.5 speciation monitoring network in Taiwan, this project proposes to install six stations located in the urban area of the northern, middle, and southern part of Taiwan, and northern coastline, Hualien Taroko National Park, and Kinmen Island, respectively, to build up a preliminary PM2.5 speciation monitoring network. The intercomparisons between regular manual and automated instruments are with satisfactory slopes and correlations in the 30 air quality monitoring stations. The average measuring ratio is at 0.82±0.09 between manual and automated instruments in the 30 air quality monitoring stations ended on August 29, 2013. However, a better way to conduct the conversion of automated measurements to manual values is through the establishment of a linear regression model.
|