Changes in seasonal cycle of surface ozone over Japan during 1980–2015

Abstract

High concentrations of surface ozone (O3) have adverse influences on vegetation and human health. To assess tropospheric ozone, it is necessary to understand both their long-term and their seasonal behaviours. Seasonal cycles of surface O3 at remote sites have shifted earlier in the last couple of decades. However, at other monitoring sites, the long-term behaviours of seasonal cycles of surface O3 have not been discussed due to a lack of continuous air quality data records. In Japan, continuous air quality monitoring records with high temporal and spatial coverage are available. We fitted a third-harmonic function to monthly mean O3 data in the records to investigate the changes in the seasonal cycle of surface O3 in Japan over a 36-year period (1980–2015) in terms of the change in the annual maximum O3 date (mO3d). Temporal changes in the mO3d for the entire period at each monitoring station were classified into six clusters by applying the k-means method. The linear trends of the mO3d in four of these six clusters were significant, with magnitudes of −0.15 to +0.65 days year−1; however, the temporal patterns of the mO3d fluctuated by about 10 years in all clusters. Upon dividing the 36 years into three decadal analysis periods (1982–1991, 1992–2001, 2002–2013), the temporal patterns of the mO3d in these three periods varied depending on the region in Japan. In one region, the mO3d continuously shifted to a later point in the year through the three decades. Meanwhile, in the other regions, the mO3d shifted to a later point in the year or were delayed in the first two analysed decades, but then these patterns were reversed in the last decade. The temporal patterns of the mO3d could mostly be explained by the interrelationship between the monthly mean O3 mixing ratios in April and May, or the months surrounding May (e.g. March and June, and April and June). Furthermore, although similar increases in O3 mixing ratio in April were found in the first and last analysis periods, the chemical processes behind such increases were deemed to be different. This implies that photochemical production and weakening NO titration were the main factors contributing to the increases in daytime O3 for the first analysis period and to the night-time O3 mixing ratio for the last analysis period.