文摘
In this analysis, we characterize urban and rural ozone (O<sub>3sub>) trends across the US for the periods 1980–2008 (29 years) and 1994–2008 (15 years) using three exposure metrics, which summarize daily O<sub>3sub> concentrations to reflect different ways O<sub>3sub> may affect human health and vegetation. We observe that a statistically significant trend at a specific monitoring site, using one exposure metric, does not necessarily result in a similar trend using the other two metrics. The two most common trends among the monitoring sites are either a continuation of negative trending over the 29-year period or a shift from negative to no trend status, indicating a leveling off of the trending. Very few sites exhibit statistically significant increases in the exposure indices. In characterizing the statistically significant changes in the distribution of hourly average O<sub>3sub>, we observe subtle statistically significant changes in the lower part of the distribution (i.e., below 50 ppb) that are not necessarily captured by the trending patterns associated with the three exposure metrics. Using multisite data from 12 metropolitan cities, we find that as the frequency of higher hourly average concentrations is reduced, the lower hourly average concentrations also move upward toward the mid-level values. The change in the number of the hourly average concentrations in the lower range is consistent with decreased NO scavenging. We recommend assessing possible subtle shifts in O<sub>3sub> concentrations by characterizing changes in the distribution of hourly average concentrations by month. Identifying statistically significant monthly changes in the mid- and low-level hourly average concentrations may provide important information for assessing changes in physical processes associated with global climate change, long-range transport, and the efficacy of models used for emission and risk reductions. Our results indicate that it is important to investigate the change in the trending pattern with time (e.g., moving 15-year trending) in order to assess how year-to-year variability may influence the trend calculation.
