Spatiotemporal Model for Short-Term Predictions of Air Pollution Data

Recently, the interest of many environmental agencies is on short-term air pollution predictions referred at high spatial resolution. This permits citizens and public health decision-makers to be informed with visual and easy access to air-quality assessment. We propose a hierarchical spatiotemporal model to enable use of different sources of information to provide short-term air pollution forecasting. In particular, we combine monitoring data and numerical model output in order to obtain short-term ozone forecasts over the Emilia Romagna region where the orography plays an important role on the air pollution; thus, the elevation is also included in the model. We provide high-resolution spatial forecast maps and uncertainty associated with these predictions. The assessment of the predictive performance of the model is based upon a site-one-out cross-validation experiment.