Date of Award


Document Type

Doctoral Thesis

Degree Name

Doctor of Philosophy


Applied Physics and Instrumentation

First Advisor

Dr. Niall Smith


A novel imaging technique based on the combination of differential optical absorption spectroscopy (DOAS) and tomographic reconstruction is presented. This approach, called Tomographic-DOAS, makes it possible to image the spatial distribution of multiple pollutants in an arbitrary region of the atmosphere. Tomographic-DOAS relies on a mathematical model relating measurements of differential optical density to an underlying spatial distribution of pollutants. This model requires the development of an optical system capable of generating multibeam paths or sensing geometries. The optical instrumentation used to construct such sensing geometries is described and a novel scanning beam arrangement that allows a single beam DOAS system to be adapted for this purpose is proposed. To assess the performance that can be expected from such an optical system, a computational fluid dynamic (CFD) model of pollutant behaviour in a typical street canyon is used as a phantom distribution. The imaging process is then modelled and the performance of a number of common algebraic reconstruction algorithms assessed by comparison to this phantom. This assessment is conducted using a new performance metric that combines measures of reconstruction accuracy and speed. The results of this analysis indicate that algorithms such as ART and SART are well- suited to this reconstruction process. However, a new algorithm called WART exhibits improved performance in a number of situations. As part of this assessment, the effect of sensing geometry beam loss and differential optical density noise is simulated. These results indicate that catastrophic degradation of reconstruction quality can occur in some cases of beam path loss and that, in general, noise contamination introduces false high-frequency features in reconstructions and degrades their overall quality.

Access Level


Included in

Physics Commons