Use this URL to cite or link to this record in EThOS:
Title: Constructing and validating modelled concentration surfaces for black smoke and sulphur dioxide across Great Britain, 1955-2001
Author: Morris, Chloe
ISNI:       0000 0004 2738 012X
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2013
Availability of Full Text:
Access from EThOS:
Access from Institution:
This thesis was funded as part of the Chronic Health Effects of Smoke and SO2 in the UK (CHESS-UK) project and was aimed at deriving modelled concentration surfaces of historic black smoke and sulphur dioxide in Great Britain for 1955 onwards as a basis for exposure assessment in epidemiological investigations of chronic health effects. The UK’s National Smoke and Sulphur Dioxide Survey provides monitored concentration data from 1955 - 2005 with over 3,000 sites in existence over time. These measured concentrations of black smoke and sulphur dioxide served as a core set of data for the modelling of exposures at 1km scale. Additional data sourced and developed for input into modelling included modelled emission maps (generated with AEA Technology), emission proxies (including land cover and road networks), information on environmental factors (topography and meteorology), population distribution and some ancillary data (e.g. smoke control areas and the Douglas Waller Index based on domestic coal sales). An important product of this work are the data, aimed at modelling historical exposures to air pollution, fully geo-referenced and integrated into a geographic information system. This GIS was used as a platform to test and compare different appropriate methodologies for modelling and mapping air pollution concentrations. Methods evaluated include: ordinary kriging, emission based modelling (dispersion and focalsum); and land use regression models developed using emission proxies. Focus was placed on modelling one target year for each decade (1962, 1971, 1981 and 1991). Modelling methods were piloted for 1971, where ordinary kriging and land use regression outperformed other methods. These were developed further across the full study period with 90% data for model development and 10% reserved for evaluation. The LUR models were ultimately selected as the best consistent approach over the long time period, with model R2s ranging from 0.7 in early years to 0.3 in later years.
Supervisor: Hansell, Anna ; Vienneau, Danielle ; Gulliver, John ; Briggs, David Sponsor: Wellcome Trust
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral