Statistical analysis of atmospheric pollution in an urban area (Spanish text).

摘要

The chemical laboratory of the Valencia city Council, using both automatic and manual information gathering networks, has been measuring atmospheric pollution and meteorological parameters.; The manual network has a total of fourteen stations distributed at various points around the city. This collects daily information.; The automatic network, on the other hand, has five stations situated throughout the city and collects information every fifteen minutes.; Information from both networks is available for the period between the 1st of January 1995 and the 28th of February 1997.; A first descriptive analysis allowed us to detect a large quantity of missing data, which meant that weekly averages had to be collected for the information from the manual network and daily averages for the data from the automatic network. These averages have been calculated using various programmes using clipper language.; It was observed that, in the majority of cases, the levels fell within the legally accepted norms.; The objective of this Thesis is to statistically analyse the spatial-temporal behaviour of the different polluting agents.; To enable this analysis an extensive bibliographical revision of the different statistical techniques applicable to environmental data was made. This determined the methodology used in attaining the stated objective. An extensive bibliographical revision was also made of the different multivariant techniques used in the area of atmospheric contamination.; Since the automatic network consists of only five stations it has been possible to study the spatial behaviour in a merely descriptive manner. However, it has been possible to do an exhaustive study of the temporal behaviour of the different series of polluting agents given that each of these agents presents very similar behaviour in all of the stations. To explain this in greater detail different multivariate techniques have been applied which demonstrate the existing relationship between the levels of pollution and the day of the week and season of the year, also relating these factors to the meteorological parameters. This was repeated in all of the stations.; Although the manual network consists of fourteen stations, sufficient data was collected for a temporal study of only one polluting agent. It was therefore decided to work with this agent, studying the temporal model. Similar behaviour was observed in all of the stations. Once the temporal component was eliminated the spatial behaviour was studied using the variogram.