The dynamic coregionalization model with application to air quality remote sensing

In this paper, we discuss the dynamic coregionalization model and its capability for model selection inference and interpretation in relation to spatio- temporal dynamic calibration and mapping of daily concentration of airborne particulate matter. To do this, we consider the problem of joint modelling ground level concentration data and satellite measurements of aerosol optical thickness (AOT), which are rarely available. The maximum likelihood estimation for the large data set related to the ”padano-veneto” region, North Italy, with missing data is covered by the stable EM algorithm and implemented on a small size computer cluster.