Air quality extremes and trends over the United States: Effects of regional climate

Abstract:

We apply a suite of analysis methods, including statistical distribution and correlation, empirical orthogonal function (EOF), linear inverse modeling (LIM), and historical modeling using regional air quality and global chemistry-climate models, to analyze surface ozone (since 1980) and PM2.5 (since 2000) measurements from EPA observation networks. The overarching goal is to understand how regional climate and weather systems affect air quality trends and extreme events. Previous studies documented high or geographically specific ozone episodes and identified contributions from anticyclone, transport, or sub-decadal to decadal time scale patterns pertinent to the events. Here, an ensemble analysis of all events from single day to multi-day episodes in the past three decades places all episodes into a continuum of time and geospatial coordinates. Inter-annual patterns linked to source concentrations and seasonal transport are evident, but anomalies such as unseasonable and persistent anticyclones to winter events over snow cover can also be identified. Overlapping events between ozone and temperature extremes are identified. They tend to occur in eastern and western coast regions with significant local variability. The occurrence frequency of overlapping events decreased from 1980s to 2000s. PM2.5 extreme showed more sensitivity to extreme temperature than drought index. When being divided by two periods (2000-2004 and 2005-2009), the second period had more extreme PM events at lower temperature in winter time. An EOF analysis was conducted to examine how regional and hemispheric climate variability affects the ozone extreme events. A question explored here is how well EOF analysis that links ozone concentrations to climate variables explains the temporal and geospatial variability of extreme ozone events (days and episodes >= 75 ppbv). The seasonal change in controlling weather systems plays a key role in how regional climate affects air quality. We also show the feasibility of long-range (1 month) statistical forecasting of PM2.5. Lastly, regional and global model simulations were investigated to understand the mechanisms and identify required model improvements for simulating air quality extremes and trends.