Evaluation of a Graph Reconstruction Method of Missing Data in Air Quality: Application to the Aburrá Valley, Colombia

ABSTRACT: Air pollution is an environmental issue that concerns human health all around the world. The air quality is affected by human emissions, meteorological conditions, and topography. The measurement of pollutants is an important task to make better decisions for controlling high pollution con...

Full description

Autores:
Botello Velasquez, Maria Camila
Tipo de recurso:
Trabajo de grado de pregrado
Fecha de publicación:
2021
Institución:
Universidad de Antioquia
Repositorio:
Repositorio UdeA
Idioma:
spa
OAI Identifier:
oai:bibliotecadigital.udea.edu.co:10495/20032
Acceso en línea:
http://hdl.handle.net/10495/20032
Palabra clave:
Environmental quality
Calidad ambiental
Air pollution
Contaminación atmosférica
Meteorology
Meteorología
Measurement
Medición
Data analysis
Análisis de datos
Air quality
Missing data
Data reconstruction
Graph signal processing
http://vocabularies.unesco.org/thesaurus/concept4533
http://vocabularies.unesco.org/thesaurus/concept1946
http://vocabularies.unesco.org/thesaurus/concept185
http://vocabularies.unesco.org/thesaurus/concept5899
http://vocabularies.unesco.org/thesaurus/concept2214
Rights
openAccess
License
Atribución-NoComercial-CompartirIgual 2.5 Colombia
Description
Summary:ABSTRACT: Air pollution is an environmental issue that concerns human health all around the world. The air quality is affected by human emissions, meteorological conditions, and topography. The measurement of pollutants is an important task to make better decisions for controlling high pollution concentrations. However, air quality sensing usually has problems due to machine failures, routine maintenance, among others. As a result, air quality datasets could have missing information that sometimes could represent more than 10% of the data. The correct reconstruction of these missing values plays an essential role in further environmental studies. In this work, we model the reconstruction of missing data as a problem of recovery of graph signals. Therefore, we evaluate the robustness of a graph signal reconstruction method in a dataset of Particular Matter PM2.5 in the Aburrá Valley, Colombia. We observe that 1) the model has better performance during dry months than in wet or transition seasons, and 2) the model could not follow pollution peaks because the algorithm assumes smooth changes in time. This model could be suitable to reconstruct data in the Aburrá Valley in dry seasons for other environmental studies.