Spatial-temporal model for ambient air pollutants in the state of Kuwait

In this paper we consider dynamic Bayesian models for four different pollutants: nitric oxide(NO), carbon monoxide(CO), sulphur dioxide(SO2) and non-methane hydrocarbon (NCH4) recorded daily in six different stations in Kuwait from 1999 to 2002. The structure of the models depends on time, space and pollutants dependencies. The approach strives to incorporate the uncertainty of the covariance structure into simulated models and final inference; therefore, hierarchical Bayesian model is applied. Association between level of pollutants and different meteorological variables, such as wind speed, wind directions, temperature and humidity are considered. The models will decompose into two main components: a deterministic part to represent the observed components term and a stochastic term to represent the unobservable components. Our analysis will start with basic model and gradually increase its complexity. At each stage the efficiency of the model will be measured. The resulting models subsequently are tested by comparing the output terms and by comparing and the predictions with the real observations. Copyright (c) 2006 John Wiley & Sons, Ltd.