Quantifying recent associations between meteorology and multipollutant day types to inform future air quality projections

摘要

Background: Changes to the climate system will impact air quality and related health effects through changes in exposure patterns. Objective: Establish influence of meteorology on multiple pollutants and estimate the potential changes in daily mixtures experienced due to expected changes in climate. Methods: We obtained four years (2011-2014) of daily average CO, NOy, S02, 03, PM2.5 EC, PM2.5 OC, PM2.5 N03, PM2.5 NH4, and S04 and weather data for multiple cities in the Southeastern US. We fit generalized additive models (GAMs) to establish present day associations between daily pollution and weather. Future climate conditions were obtained for the years 2030-2040 from downscaled climate projections and fitted GAMs were then used to predict corresponding responses of daily pollution levels. Self-organizing maps (SOMs) were then used to identify categories of days based on multipollutant conditions (i.e., multipollutant day types (MDTs)) and class frequencies were used to establish differences in present day and future air quality. Results: We found MDTs identified days were conditions ranged from relatively clean days, high single pollutant days (e.g., SO2), to high combination days (e.g., CO, NOy, EC). Classifying days under our future climate scenario revealed that the largest increases in MDTs frequencies occurred on days characterized by moderate-to-high 03 pollution (21% increase), days dominated by relatively moderate-to-high CO, NOy, and EC (10.5% increase), and days with elevated SO2 and OC. The largest decreases occurred for relatively clean days (10.7% decrease); we show that these days transition to a similar profile with higher 03 in the future. Conclusion: We find combining multipollutant day typing (SOM), GAMs, and future climate predictions provides a complementary suite of tools for investigating potential air quality changes driven solely by future meteorological conditions.