Markov chains modelling of particulate matter (PM10) air contamination in the city of Ruse, Bulgaria

摘要

High levels of air pollutants PM10 are a problem of great importance for human health. During the months from April to September of the period 2010 - 2019 the levels in Ruse remain within the norm in 95% of the days. During the other, "cold" months of the year, only 58% of the days have values below the daily norm of 50μg/m~3. When planning their activities, it is useful for people to have forecasts for PM10 levels in the coming days. Markov chains allow such predictions to be given in a tabular form, convenient to use, without the need for calculations. The data for the "cold" months are modelled using three Markov chains with different degrees of discretization of the original values, respectively with 12, 7 and 3 possible states. The latter, with states: {in the norm}, {slightly above the norm} and {a strong excess of the norm}, can be used without official data on the exact PM10 levels. Determining the condition of the PM10 pollutant today in this case can also be done on the basis of a personal assessment of the purity of the air over the city at the moment. The measured levels in the period 01.01.2020 - 31.03.2020 were used as test data. They show consistency of the measured levels in 2020 with all three Markov chains considered. The obtained tabular values can be used to predict PM10 levels in the following years, in the months from October to March.