Where air quality has been impacted by weather changes in the United States over the last 30 years?

Previous studies have contributed to the understanding on the impacts of weather on air pollution. Most of these investigations focused primarily on future climate scenarios, laboratory experiments, or weather trends-related air pollution at regional scale. In particular, an important limitation of the trends studies is that the estimated weather impact on air pollution may be underestimated or overestimated given that the regional scale does not capture the site-to-site variation of air pollution and weather. The primary objective of this research is to addresses this gap by quantifying weather-associated changes in air pollution at site location (air pollution sites) in the U.S between 1988 and 2018. We quantified the past weather-related increases CO, NO2, O-3, nitrate, organic carbon, silicon, sodium, sulfate, and SO2 concentration using Generalized Additive Models (GAMs). We used a framework that derives "penalties" (weather penalty, in mu g/m(3), ppm or ppb per year) for each season (warm and cold). Three pollutants presented significant results (weather penalties), including CO, NO2, and O-3. Our findings show significant spatio-temporal variation of climate impact on CO, NO2, and O-3. For example, in the warm season we estimated a total penalty over the study period for the sites with the highest penalty on CO (site in Boise, Idaho), NO2 (site in New York City), and O-3 (site in Tucson, Arizona) of 6.18 (95%CI: 0.30; 12.0) ppm, 182.04 (95%CI: 39.33; 324.72) ppb, and 0.09 (95%CI: 0.030; 0.150) ppm, respectively. In the cold season, the estimated total penalty for the sites with the highest penalty on CO (site in Los Angeles, California), NO2 (site in Washington, Pennsylvania), and O-3 (site in Decatur, Illinois) was 12.01 (95%CI: 1.50; 22.50) ppm, 285.03 (95% CI: 14.37; 555.69) ppb, and 0.066 (95%CI: -0.120; 0.030) ppm, respectively. Our results should be of interest to policy makers to create future strategies related to environmental health and climate change. Climate models typically show large variation in projections of the key variables influencing pollution, and our model based on local scale allowed us to identify statistically robust results.