The objectives of this project include auditing the operation/maintenance of the southern Taiwan PM supersites, intercomparison of automatic and manual sampling methods on sites, evaluation of effective operation rates for each instruments, and analysis of contribution sources to pollution episodes. There were more than ten times auditing every month, including operation conditions of each instruments, comprehensive audition, and calibration audition. Additional flow rate calibration auditions were conducted in September due to the damage by Typhoon and only the EC/OC monitor at Chatou exceeded the operation ranges. Most of the corrections or modifications were done on sites or after the routine monthly auditing meeting. For example, the replacement of filter for EC/OC monitor has been changed from laser correction of 0.88 to 0.90; the correction for shift of zero/span calibration of NH3 monitor and its out of operation range. However, there are some unfinished suggestions, like the inclusion of detail operation parameters into data bank, the split time correction for EC/OC monitor, and the UPS for power breakage. The results of parallel intercomparison show that there are good agreement for PM2.5 mass concentration and nitrate concentration between automatic and manual methods and they both pass the U test. The differences for sulfate concentration are due to the uncertainty of theoretical conversion factor of RP8400S. The overestimation of NH3 monitor is caused by the conversion of other nitrogen-containing species into NO by the NH3 converter. The intercomparison also found that the laser source of OPC has decayed and needed to be replaced. For instruments at all four sites, the effective operation rates are all greater than 90%, except for EC/OC monitor; those for PM10 and PM2.5 were even greater than 96.5% every month. The average effective operation rate for EC/OC monitor is 84% and the lower rate is due to some concentrations less than MDL of monitor. The effective operation rates for instruments at the core site are all greater than 95% except for POM and SMPS+C and their lower rates are due to the missing data. In addition to the effect of poor dispersion, the PM episodes are caused by additional emissions from two anthropogenic sources: mobile emission during rush hours and stationary sources. For the PM episodes in April caused by Asia dust storm, PM10 concentrations increased significantly but not for PM2.5. Most of the ozone episodes at core sites are NOx-sensitive based on the measured NOy/O3 ratios and the ozone precursors are from Kaohsiung city and coastal industrial park. Both sulfate and OC concentrations increased after high ozone episodes, implying the significant contribution of photochemical reactions to secondary aerosol.Based on this year and previous studies, the following suggestions are made: establishment of criteria for normal operation after breakdown, establishment of criteria for operation conditions of gaseous monitors, and the double check of effective operation rates. Parts of both RP8400N and RP8400S are not available any more and substituted instruments should be installed. The Asia dust storm, local dust storm, and rice straw open burnings should be monitored in order to analyze their possible impact on the ambient air quality.

intercomparison measurements、 PM supersite、quality assurance auditing