부산대학교 도서관

With the introduction of the 8-hour ozone standard and the associated regulations regarding how missing data are handled in this computation, data completeness becomes vitally important in gathering unbiased ozone readings for regulatory purposes. For this analysis the relationship between the calculations of 8-hour ozone concentrations with no missing data were compared to the calculations of 8-hour ozone concentrations which had missing data blocks of 2 and 3 hours. To determine which hours of the day are most sensitive to missing data biasing the 8-hour ozone calculation, 2-hour and 3-hour blocks of missing data were iterated through the full range of hourly readings. This procedure was also applied on a monthly basis to determine which months were most susceptible to data biasing. Hourly ozone data from 1990 through 1997 was collected from the Aerometric Information Retrieval System (AIRS) for 8 different ozone sites in the Southeast. These 8 years (1990--1997) of hourly 1-hour ozone monitoring data were reduced to the peak exceedance days that occurred during this period and five statically based hypothetical days. It was found through this analysis that missing 2 consecutive hours of data between the approximate hours of 0800 through 1100 and again from approximately 1600 through 1900 will bias the calculation of the 8-hour ozone concentration artificially high. Under episodes where ozone production is highly conducive, the magnitude of the positive bias can be as great as 10 to 17 ppb for an 8-hour average. A negative bias was found to exist between the hours of approximately 1100 to 1600.