Bayesian Spatial Quantile Regression.

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Biostatistics

Background:

• Tropospheric ozone is a regulated criteria pollutant linked to adverse health effects.
• Ozone levels are sensitive to weather, raising concerns about climate change impacts on air quality and public health.

Purpose of the Study:

• To develop a Bayesian spatial model for predicting tropospheric ozone concentrations.
• To analyze spatial and temporal ozone trends and forecast future concentrations under climate change scenarios.

Main Methods:

• Developed a spatial quantile regression model to predict ozone, accommodating non-normal distributions and site-specific variations.
• Implemented an approximate method for large datasets, applying it to Eastern U.S. summer ozone data (1997-2005).
• Utilized deterministic climate models for future ozone projections.

Main Results:

• The model successfully predicts ozone concentrations under varying meteorological conditions.
• Analysis revealed spatial and temporal trends in ozone levels across the Eastern U.S.
• Projections indicate that increased daily average temperatures will significantly raise ozone levels, especially in the Industrial Midwest and Northeast.

Conclusions:

• The developed Bayesian spatial model provides a robust tool for ozone prediction and climate change impact assessment.
• Future climate change, particularly rising temperatures, is projected to exacerbate ozone pollution in key regions of the Eastern U.S.
• Findings highlight the need for strategies to mitigate ozone's public health impacts in a changing climate.