Statistical modelling of air quality in Aberdeen

GE Environmental Sciences

Doherty, Cillian Francis ; Scott, Marian

University:University of Glasgow

Department:School of Mathematics and Statistics

关键词: Time series analysis, spatial modelling, air quality, Aberdeen, DEFRA, AURN, nitrogen dioxide, inverse regression, SEPA.;   

Others  :  http://theses.gla.ac.uk/8357/7/2017DohertyMScR.pdf

来源: University of Glasgow

【 摘 要 】

This thesis focuses on modelling air pollution in Aberdeen. It takes into account how traffic and meteorological variables affect the Nitrogen Dioxide concentrations at a number of different sites throughout the city during the year 2014. The aim of the thesis is to build a regression model of spatial and temporal concentration variations and use inverse regression to develop a tool to identify control mechanisms that will help manage Nitrogen Dioxide concentrations in an urban setting. This is of particular importance to the Scottish Environment Protection Agency (SEPA).Chapter 1 focuses on the motivation for carrying out such a study, as well as the aims and objectives. The data are introduced in this Chapter. These include data from different AURN (Automatic Urban Road Network) sites in Aberdeen, as well as diffusion tube data, traffic counts from different locations as well as meteorological data recorded at Dyce Airport.Chapter 2 covers the temporal modelling of air quality in Aberdeen using time series analysis. Time series methodology is explored which includes an initial exploration of the model variables using linear regression; followed by residual diagnostics; time series regression; the definition of autocorrelation function (ACF), partial autocorrelation function (PACF) and stationarity; the exploration of seasonality and harmonic regression, and ends with generalized additive model methodology. This spans from 2006-2015.Chapter 3 investigates the spatial modelling of air quality in Aberdeen. This is done through numerical and graphical summaries. Methods used to explore NO2 data are presented. This includes geostatistical modelling. Two different models are investigated. Model parameters are estimated, using maximum likelihood estimates and restricted maximum likelihood estimates. This is followed by prediction of future values, using a statistical technique known as Kriging.Chapter 4 uses inverse regression to estimate road traffic flows required to achieve compliance with national air quality objectives. This Chapter also presents the usefulness of inverse regression.Chapter 5 ends with a discussion on what further work can be done, and any conclusions for this thesis. It looks at the strengths and weaknesses of each Chapter in turn.