Particulate Matter 2.5 Prediction in Osorno, Chile, Using a Discrete Markov Chain Model

Abstract

Air quality is of growing concern globally due to its impact on human health. One of the most important air pollutants is particulate matter, principally fine particulate matter (PM 2.5). This study was carried out in the city of Osorno, Chile, where high levels of PM 2.5 are recorded, specifically during autumn and winter. A large database was assembled of daily PM 2.5 concentrations, precipitations, ambient temperature and mean wind speed in the months of April to September from 2013 to 2023. Using a discrete Markov chain model, the evolution and projection of PM 2.5 concentrations associated with existing weather conditions (environmental variables) were analysed. The analysis showed that the combination of low PM 2.5 concentration, cold temperature, presence of precipitations and wind is the commonest and most stable state, while states with high PM 2.5 concentration are more probable when conditions of cold temperature without rain or wind occur. The transition matrix drawn up enabled us to identify patterns of change and recurrence, showing the importance of weather factors in the accumulation or dispersal of pollutants and the associated probabilities. The results provide a basis from which to project air quality conditions and plan preventive measures. This work will provide tools to improve air quality management in Osorno, and thus the health of the inhabitants. It also opens up opportunities for future investigations to adjust and refine the predictions, based on more recent data and additional factors.