Spatio-temporal joint modelling on moderate and extreme air pollution in Spain

Abstract Very unhealthy air quality is consistently connected with numerous diseases. Appropriate extreme analysis and accurate predictions are in rising demand for exploring potential linked causes providing suggestions the environmental agency public policy strategy. This paper aims to model spatial temporal pattern of both moderate extremely poor PM $$_{10}$$ 10 concentrations (of daily mean) collected from 342 representative monitors distributed throughout mainland Spain 2017 2021. We first propose compare a series Bayesian hierarchical generalized models annual maxima concentrations, including fixed effect altitude, temperature, precipitation, vapour pressure population density, as well spatio-temporal random Stochastic Partial Differential Equation (SPDE) approach lag-one dynamic auto-regressive component (AR(1)). Under WAIC, DIC other criteria, best selected good predictive ability based on four-year data (2017–2020) training last-year (2021) testing. bring structure establish joint mean provide evidence that certain predictors (precipitation, density) influence comparably while (altitude temperature) impact reversely different scaled concentrations. The findings applied identify hot-spot regions using excursion functions specified at grid level. It suggests community Madrid some sites northwestern southern likely be exposed severe pollution, simultaneously exceeding warning risk threshold.