Spatiotemporal Analysis of Variables Affecting Air Quality in Urban Areas of the City of Cartagena, Colombia

Abstract

This article presents the relationship between the Normalized Difference Vegetation Index (NDVI) and atmospheric pollutants (NO2, SO2, O3, CO, PM2.5, and PM10) affecting air quality. To do so, supervised machine learning techniques, specifically Random Forest, are used for data imputation on atmospheric pollutants and classification of the NDVI from 27 satellite images, focusing on part of the urban area, especially in areas where the industrial part of Cartagena is most concentrated. Therefore, the following objectives are pursued: (i) describing vegetation conditions using the NDVI extracted from a 27 PlanetScope image times series, and (ii) associating the concentration levels of some pollutants such as suspended particles with diameters less than 10 and 2.5 micrometers, Nitrogen Dioxide, Sulfur Dioxide, Ozone, Carbon Monoxide, with vegetation information.