Bayesian modeling of air pollution extremes using nested multivariate max-stable processes

Area of Science:

• Environmental Science
• Statistics
• Public Health

Background:

• Assessing public health risks necessitates understanding the dependence among peak air pollutant concentrations across regions.
• Existing models may not fully capture the complex spatial tail dependence of extreme pollution events.

Purpose of the Study:

• To introduce a novel class of multivariate max-stable processes for analyzing multivariate spatial dependence of air pollution extremes.
• To develop a hierarchical, tree-based model facilitating Bayesian inference and interpretable characterization of pollution data.

Main Methods:

• Development of a new class of multivariate max-stable processes with a hierarchical, tree-based formulation.
• Utilizing latent nested positive stable random factors for conditional independence.
• Application of Bayesian inference for model fitting.

Main Results:

• The proposed nested multivariate max-stable model effectively captures complex tail dependence structures.
• Demonstrated success in modeling air pollution concentrations and temperatures in the Los Angeles area.
• The hierarchical structure provides a convenient and interpretable characterization of spatial dependence.

Conclusions:

• The new multivariate max-stable process is a powerful tool for analyzing extreme air pollution data.
• Accurate modeling of spatial dependence is crucial for robust public health risk assessment.
• The model offers improved interpretability and facilitates Bayesian analysis in environmental studies.