Resumen de Caracterización de la contaminación atmosférica debida a aportes antropogénicos y naturales mediante la aplicación de modelos de mixturas finitas, de Markov homogéneos y otras técnicas de minería de datos
- español
Son cuantiosos los recursos científicos que se dirigen al estudio de las fuentes de emisión de contaminantes atmosféricos en las áreas urbanas. Este estudio puede ser cuantitativo, determinando la contribución de cada fuente a la contaminación ambiente, o cualitativo, para conocer más sobre la composición de las emisiones que afectan a los residentes en las ciudades. En los países mediterráneos, además, la contaminación causada por fenómenos naturales, como el transporte de polvo desde las regiones áridas del Norte de África, también es de primordial importancia. Entre los instrumentos fundamentales de los que se dispone para medir la contaminación atmosférica, se encuentran las redes de vigilancia de la calidad del aire, integradas por estaciones de medida que se sitúan tanto en ambientes urbanos como en el medio rural, con el fin de determinar e informar sobre la calidad del aire que nos afecta. En las ciudades, algunas de estas estaciones de medida se sitúan en emplazamientos fuera del alcance directo de fuentes de emisión, para determinar la contaminación de fondo urbano, representativa de la exposición a la que la población se expone de forma general. Esta tesis ha tenido como objetivos los siguientes:
1. La caracterización exhaustiva de la contaminación atmosférica en entornos urbanos y rurales empleando la información obtenida de las redes de vigilancia de la calidad del aire, desarrollando para ello una metodología general para la gestión eficiente de las redes de monitorización.
2. Mejorar la metodología existente para la estimación del aporte de polvo transportado por las masas de aire cálido desde las regiones norteafricanas.
3. Comparar los niveles de contaminación atmosférica entre diferentes redes de monitorización urbanas, sin influencia industrial y localización geográfica distinta, proponiendo para ello una metodología con la que caracterizar la contaminación atmosférica ambiental y de fondo.
Los resultados de esta tesis, apoyados en cada uno de estos objetivos, están avalados, respectivamente, por las siguientes publicaciones: 1. Gómez-Losada, Á., Lozano-García, A., Pino-Mejías, R., Contreras-González, J. 2014. Finite mixture models to characterize and refine air quality monitoring networks. Science of the Total Environment, 485-486: 292-9. 2. Gómez-Losada, Á.,Pires,J.C.M.,Pino-Mejías,R.2015.Time series clustering for estimating particulate matter contributions and its use in quantifying impacts from deserts. Atmospheric Environment, 117: 271-81. 3. Gómez-Losada, Á., Pires, J.C.M., Pino-Mejías, R. 2016. Characterization of background air pollution exposure in urban environments using a metric based on Hidden Markov Models. Atmospheric Environment, 127: 255-61.
- English
A wealth of scientific resources have been dedicated to the study of the sources of pollutant emissions to air in urban areas. Such studies may be quantitative, determining the contribution of each source of environmental pollution, or they may be qualitative, providing insight into the makeup of the emissions that afect a city's inhabitants. In Mediterranean countries, contamination may also be the result of natural phenomenon, such as the ow of dust from the arid regions of North Africa, and are therefore of primary importance as well. The ow of particulate matter transcends these geographic areas, passing over the Atlantic Ocean and reaching the American coasts. Among the fundamental tools available for measuring air pollution are the air-quality monitoring networks, made up of monitoring stations located both in urban areas and rural environments, with the aim of providing information on the air quality that afects us. In cities, some of these monitoring stations are located on sites that are outside of the direct range of emission sources and thus the determination of the urban background pollution, which is indicative of the generalised exposure of the population to air pollution, is possible. The objectives of this thesis were the following: To exhaustively characterise the air pollutants in urban and rural areas using the information obtained from the air-quality monitoring networks. To this end, a general methodology was developed to efciently manage the monitoring networks; To improve the existing methodology used to estimate the contribution of dust originating in the North African region that is carried by waves of warm air; To compare the air-pollution levels between the diferent urban-monitoring networks unafected by industrial pollution, and between diferent geographic locations, proposing a methodology that can be used to characterise environmental and background air pollution. In order to fulil the First objective, the primary and secondary air-pollution monitoring data were modelled using finite mixture models. Based on the calculation of the first and second moments of these mixtures, hierarchical cluster analysis, imputation using random forests, and principal component analysis were used. This methodological approximation enabled the detection of duplications within the parameters monitored by the monitoring stations, thus allowing these networks to be reconfigured and enabling the economic resources invested in them to be optimised. For the second objective, hidden Markov models (HMM) were introduced and the diferent regimes or PM10 concentration profiles were described in some of the time series (TS) studied, enabling an estimation of the contribution of each of the profiles to environmental pollution. The new method proposed for estimating the natural contribution of PM10 improves upon the reference methodology used in the European Union (monthly moving 40th percentile method) in three ways - it avoids the use of empirical approximation, it applies modelling that is especially designed for the treatment of time-series data, and it allows for obtaining a con_dence interval for the contribution estimations for PM10. For the third objective, hidden Markov models were also used, in this case to define and characterise the environmental and background pollution caused by primary air pollution in diferent urban areas of diferent cities. The attributable fraction for background air pollution was estimated using a new procedure based on the first concentration profile defined by the HMMs in the TS. The ratio and diference between environmental and background concentrations were also studied.
