Particulate matter 10 µm (PM10), 2.5 µm (PM2.5) datasets gathered by direct measurement, low-cost sensor and by public air quality stations in Fontibón, Bogotá D.C., Colombia

Abstract

Concentration of particulate matter directly affects air quality and human health. Three sources of information were used in this work to generate datasets on this matter at the Fontibón county in Bogota D.C., Colombia. The first source was a Davis AirLinkⓇ low-cost sensor air quality readings for PM2.5, PM10 and meteorological variables. The sensor was installed in the referred area, collecting air quality readings for PM2.5, PM10, as well as temperature, relative humidity, dew point, wet bulb, and heat index as meteorological variables during the months of May to August 2022. The second source was collecting by direct measurement the PM10 particles using a TischⓇ Hi- Vol equipment, evaluated the concentration of particulate matter PM10 in the same place for 27 days. Finally, raw data was provided by the Bogotá’s Environmental District Bureau (SDA), validating in this work the data readings for the years 2021 and 2022 from the two meteorological stations located in the same county, named “Fontibón” and “Móvil Fontibón”, including Air quality data for PM2.5, PM10, Carbon Monoxide (CO), Ozone, Nitrogen Dioxide (NO2), Sulfur Dioxide (SO2) and the meteorological variables wind speed, wind direction, temperature, precipitation, relative humidity (RH) and Barometric pressure.A Machine Learning model was made to perform the mining and completeness of the missing data with an iterative imputation and with a regression model, and the Pearson, Spearman and Kendall correlation coefficients were calculated, using Python language.

Keywords

• Particle pollution
• Data mining
• Data science
• Machine learning
• Meteorological variables