The investigation of fine suspended particle PM2.5 levels between automatic and manual analysis

• Main Authors: Chih-Hsiang Hsu, 許志祥
• Other Authors: How-Ran Chao
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/67628660372010634734

碩士 === 國立屏東科技大學 === 環境工程與科學系所 === 103 === Fine particulate measuring instruments that employ BAM technique have the advantage of taking hourly data automatically to help build up air quality forecast and warning system. However, BAM is not as accurate as FRM manual sampling analysis. To overcome this disadvantage, Environmental Protection Administration deploys a manual sampling method standard to build a correlation model for these two methods, and uses regression equations to modify auto-collected data. This research is an investigation of fine suspended particle PM2.5 levels between automatic and manual analysis, based on data collected from the fourth quarter of 103 to the first quarter of 104 in southern air quality monitoring stations. The result shows no evidence of a fixed higher or lower ratio between automatic and manual values. Though most automatic values are higher than manual values, their relative deviation is correspond to PM2.5 level observed at monitoring stations, especially at lower-level stations. However, once the equation is applied, automatic values with higher deviation are modified to approximate manually-analyzed values. Relative difference between each station is modified as well, but not much. For some stations, there is no apparent effect on the outliers. Data in different years show dissimilar results of regression modification, while the ratios between automatic and manual values are both lowered. Ideally, the expected value of modified automatic value should be close to manually analyzed value, rather than just being lowered. This research shows that there is still room for improvement to automatic regression equation.