Screening method for outliers in photochemical oxidant monitoring using potential ozone

Abstract

We evaluated a screening method for outliers in the routine monitoring data of photochemical oxidant (O_X) using potential ozone (PO). First of all, we demonstrated that the inter-site correlations of PO were high over a wide area around the investigated area, and that the correlations were higher with the shorter inter-site distances. In addition, the correlations in PO were higher than in O_X especially in the area with dense NO_X sources, implying similar PO data between the two neighboring sites independent of the strength of the NO_X emissions around the sites. A case study showed a good consistency in the hourly PO data between site A and site B spacing of about 3 km, however, some of the data shifted from the one-to-one correlation. There was a concern with the monitoring data of O_X or NO_X at site A or site B. Further comparison with another neighboring site (site C) determined that the O_X data at site A seemed to have a problem. In the same manner, we could detect possible outliers in the other data set in a different area. Accordingly, the comparison of the PO data between two or more sites could provide an efficient screening method for the outliers in NO_X as well as O_X.