Análisis espacial y temporal de la contaminación atmosférica por pm2.5 en Medellín: patrones y tendencias (2019 2023)

Abstract

This study analyzes the spatial and temporal distribution of fine particulate matter (PM2.5) in the city of Medellín between 2019 and 2023, using spatial analysis techniques and Geographic Information Systems (GIS). Three main methodological approaches were employed: spatial interpolation through Universal Kriging, the Global Moran’s I spatial autocorrelation index, and the DBSCAN clustering algorithm. The interpolation was conducted in QGIS using the SAGA GIS plugin, applying a second-order log-linear trend model to capture global trends. Results show that March and April register the highest PM2.5 concentrations, indicating a marked seasonal pattern during the first half of the year. Universal Kriging provided a more robust spatial representation of these concentrations than deterministic methods such as IDW, as it incorporates the spatial structure of the data through the fitting of semivariograms. Although the global spatial autocorrelation analysis did not reveal significant spatial structures, the LISA analysis identified localized High-High clusters in the central-eastern part of the city. In turn, the DBSCAN algorithm detected a single dense cluster in the same region. These findings highlight the need to strengthen the air quality monitoring network, integrate meteorological and socioeconomic variables, and move towards more advanced predictive models. Ultimately, the results offer valuable insights for land-use planning and urban environmental management in Medellín.