Regional-specific trends of PM2.5 and O3 temperature sensitivity in the United States

Abstract

Climate change poses direct and indirect threats to public health, including exacerbating air pollution. However, the influence of rising temperature on air quality remains highly uncertain in the United States, particularly under rapid reduction in anthropogenic emissions. Here, we examined the sensitivity of surface-level fine particulate matter (PM_2.5) and ozone (O₃) to summer temperature anomalies in the contiguous US as well as their decadal changes using high-resolution datasets generated by machine learning. Our findings demonstrate that in the eastern US, stringent emission control strategies have significantly reduced the positive responses of PM_2.5 and O₃ to summer temperature, thereby lowering the population exposure associated with warming-induced air quality deterioration. In contrast, PM_2.5 in the western US became more sensitive to temperature, highlighting the urgent need to manage and mitigate the impact of worsening wildfires. Our results have important implications for air quality management and risk assessments of future climate change.

Keywords: Climate sciences; Environmental sciences.

Fig. 1. Sensitivities of surface air pollutants concentration to anomalies of summer mean temperature derived from high-resolution datasets.

Results derived from ground-based measurements (scatters) are also shown for comparison. All results are based on data from 2000–2016. Temperature sensitivity of summertime PM_2.5 (a), annual PM_2.5 (b), and summertime O₃ (c). Subplots show the temperature sensitivity in the Chicago (a2–c2), Los Angeles (a3–c3), New York (a4–c4), and Atlanta (a5–c5) metropolitan area. The temperature sensitivities with p < 0.05 are shown in Supplementary Fig. 1.

Fig. 2. Time series and regression coefficients showing positive correlations between air pollutants and summer temperature for all studied regions.

All results are based on data from 2000–2016. Interannual variation of detrended summer temperature anomalies, summertime PM_2.5 anomalies, annual PM_2.5 anomalies, and summertime O₃ anomalies in the contiguous US (a), Southeastern US (b), Northeastern US (c), Western US (d), and Central US (e); ML modeled and ground-based observed regional responses of summertime PM_2.5 (JJA PM_2.5), annual PM_2.5 (Ann PM_2.5), and summertime O₃ (JJA O₃) to summer temperature in the contiguous US (f), Southeastern US (g), Northeastern US (h), Western US (i), and Central US (j). Error bars represent 95% confidence intervals. Inset pie charts show the contributions of five PM components (organic aerosol (OA), sulfate, ammonium, nitrate, and elemental carbon (EC)) to the overall summertime and annual PM_2.5-temperature sensitivities. All regression coefficients are statistically significant (p < 0.05). The p values are included in Supplementary data.

Fig. 3. Changes in the spatial patterns and regional mean values of the climate penalty effects from 2000–2009 to 2010–2016.

a–c ML modeled (colored maps) and observed (scatters) temperature sensitivities in 2000–2009, 2010–2016, and the difference between two periods for summertime PM_2.5 (a1–a3), annual PM_2.5 (b1–b3), and summertime O₃ (c1–c3). d Comparisons of regional sensitivities in 2000–2009 and 2010–2016 (upper panels) and relative changes in percentage (lower panels, bars represent the mean percentage changes; error bars represent 95% confidence intervals) for summertime PM_2.5 (d1), annual PM_2.5 (d2), and summertime O₃ (d3). All regression coefficients are statistically significant (p < 0.05), except the temperature sensitivity of annual PM_2.5 during 2010–2016 in the Central US. The p values are included in Supplementary data.

Fig. 4. Temperature sensitivities in urban and non-urban areas aggregated for CONUS and each subregion in 2000–2009 and 2010–2016.

Error bars represent 95% confidence intervals. Temperature sensitivity of summertime PM_2.5 (a), annual PM_2.5 (b), and summertime O₃ (c). All regression coefficients are statistically significant (p < 0.05). The p values are included in Supplementary data.

Fig. 5. Temperature sensitivity for each 5-year running time window derived from ML-modeled data (2000–2016, 13 time periods) and ground-based observations (2000–2022, 19 time periods).

The shaded area represents a 95% confidence interval. Temperature sensitivity of summertime PM_2.5 (a1–e1), annual PM_2.5 (a2–e2), and summertime O₃ (a3–e3) in CONUS (a), Southeastern US (b), Northeastern US (c), Western US (d), and Central US (e). The p values are included in Supplementary data.

Fig. 6. Number and regional distribution of population exposed to air pollution changes associated with 1 °C change in summer temperature for 2000–2009 and 2010–2016.

Stacked area plots showing the fraction of population living in areas with different levels of temperature sensitivity of summertime PM_2.5 (a), annual PM_2.5 (b), and summertime O₃ (c) during 2000–2009 (a1, b1, c1), 2010–2016 (a2, b2, c2), and box plots showing the temperature sensitivity to which 5%, 25%, 50%, 75%, and 95% of the population is exposed, with light colors represent values in 2010–2016 (a3, b3, c3).