Stochastic Modelling Applied to Air Quality Space-Time Characterization

Abstract

Atmospheric pollution directly affects the respiratory system, aggravating several chronicle illnesses (e.g. bronchitis, pulmonary infections, cardiac illnesses and cancer). This pertinent issue concerns mainly highly populated urban areas, in particular when meteorological conditions (e.g. high temperature in summer) emphasise its effects on human health. The proposed methodology aims to forecast critical ozone concentration episodes by means of a hybrid approach, based on a deterministic dispersion model and stochastic simulations. First, a certain pollutant's spatial dispersion is determined at a coarse spatial scale by a deterministic model, resulting in an hourly local trend. Afterwards, spatial downscaling of the trend will be performed, using data recorded by the air quality (AQ) monitoring stations and an optimization algorithm based on stochastic simulations (Direct sequential simulation and co-simulation). The proposed methodology will be applied to ozone measurements registered in Lisbon. The hybrid model shows to be a very promising alternative for urban air quality characterization. These results will allow further developments in order to produce an integrated air quality and health surveillance/monitoring system in the area of Lisbon.