EcoSta 2018: Registration
View Submission - EcoSta2018
Title: Hierarchical Bayesian autoregressive models in South Korea ozone Authors:  Sanghoo Yoon - Daegu University (Korea, South) [presenting]
Dain Park - Daegu University (Korea, South)
Abstract: Environmental has a property of space-time. Having both the spatial and temporal dimensions adds substantial complexity to environmental data analysis. We model daily maximum 8-hour ozone concentration data obtained from $n=32$ sites in South Korea for analysis between 2013 and 2017. Maximum temperature, relative humidity, and wind speed were considered as covariates. The data on the square root scale seems most attractive in terms of both symmetry and stabilizing the variance. Independent Gaussian process model and autoregressive model are specified within a hierarchical Bayesian framework and Markov Chain Monte Carlo techniques. These space-time models allow accurate spatial prediction of a temporally aggregated ozone summary along with its uncertainty.