CMStatistics 2021: Start Registration
View Submission - CMStatistics
Title: Constructing mechanistic spatial models from Ornstein-Uhlenbeck processes Authors:  Nathan Wikle - University of Texas at Austin (United States) [presenting]
Ephraim Hanks - Penn State (United States)
Corwin Zigler - University of Texas at Austin (United States)
Abstract: A mechanistic model is developed to analyze the impact of sulfur dioxide emissions from coal-fired power plants on average sulfate concentrations in the central United States. A multivariate Ornstein-Uhlenbeck (OU) process is used to approximate the dynamics of a linear space-time SPDE. The distributional properties of the OU process are leveraged to specify novel probability models for spatial data (i.e., spatially-referenced data with no temporal replication) that are viewed as either a snapshot or a time-averaged observation of the OU process. Air pollution transport dynamics determine the mean and covariance structure of our atmospheric sulfate model, allowing us to infer which process dynamics are driving observed air pollution concentrations. We use these inferred dynamics to assess the regulatory impact of flue-gas desulfurization (FGD) technologies on human exposure to sulfate aerosols. Extensions of this methodology are discussed, including its potential applicability to methods for causal inference with interference.