EM estimation of the dynamic coregionalization model with varying coefficients

The satellites from NASA's Earth Science Project Division, like AURA, produce data for the concentration of various airborne pollutants. Calibrating satellite data using ground level monitoring networks and other meteorological and land characterizing variables is mandatory. To do this, it is important to use an approach which is able to manage large datasets coming from different sources, structural missingness and spatial and temporal correlation. In this paper, we extend the Dynamic Coregionalization Model introduced in Fassó and Finazzi (2011) to the case of space-time varying coefficients in order to increase the model flexibility and to make it suitable for large regions such as Europe.