Spatio-temporal modelling of <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" altimg="si1.svg"><mml:mrow><mml:msub><mml:mtext>PM</mml:mtext><mml:mn>10</mml:mn></mml:msub></mml:mrow></mml:math> daily concentrations in Italy using the SPDE approach

This paper illustrates the main results of a spatio-temporal interpolation process PM10 concentrations at daily resolution using set 410 monitoring sites, distributed throughout Italian territory, for year 2015. The is based on Bayesian hierarchical model where spatial-component represented through Stochastic Partial Differential Equation (SPDE) approach with lag-1 temporal autoregressive component (AR1). Inference performed Integrated Nested Laplace Approximation (INLA). Our includes 11 spatial and predictors, including meteorological variables Aerosol Optical Depth. As predictors’ impact varies across months, regression 12 monthly models same covariates. predictive performance has been analyzed cross-validation study. show that predicted observed values are well in accordance (correlation range: 0.79–0.91; bias: 0.22–1.07μg/m3; RMSE: 4.9–13.9μg/m3). final output 365 gridded (1 km × 1 km) maps over Italy equipped an uncertainty measure. prediction shows procedure able to reproduce large scale data features without unrealistic artifacts generated surfaces. presents also two illustrative examples practical applications our model, exceedance probability population exposure maps.