Modelling the concentration of air pollutants is a complex task because their dynamics can be simultaneously related to anthropogenic sources, the morphological conformation of the territory, and meteorological conditions. With the aim to explore the role of temporal and weather-related variables on the local variability and the dependence among urban areas, we consider a multivariate Gaussian additive model with dynamic covariance matrix. We illustrate our proposal by analysing daily concentration of PM_10 in urban areas in the same region.
