Spatio-temporal modeling for real-time ozone forecasting

Accurate assessment of exposure to ambient ozone concentrations is important for informing the public and pollution monitoring agencies about ozone levels that may lead to adverse health effects. A practical challenge facing the US Environmental Protection Agency (USEPA) is to provide real-time forecasting of current 8 h average ozone exposure over the entire conterminous United States. Such real-time forecasting is now provided as spatial forecast maps of current 8 h average ozone defined as the average of the previous four hours, current hour, and predictions for the next three hours. Current patterns are updated hourly throughout the day on the EPA-AIRNow web site. Our contribution is to show how we can substantially improve upon current real-time forecasting systems. We introduce a downscaler fusion model based on first differences of real-time monitoring data and numerical model output. The model has a flexible coefficient structure with an efficient computational strategy to fit model parameters. This strategy can be viewed as hybrid in that it blends offline model fitting with online predictions followed by fast spatial interpolation to produce the desired real-time forecast maps. Model validation for the eastern US shows consequential improvement of our fully inferential approach compared with the existing implementations.