Inference in the context of uncertain complex urban environments for climate change conscious planning and design

This thesis looks at the urban environment as the centre of human habitation. It governs
the comfort of much of the human population and is essential to life itself. In the modern
world, it is governed at many levels and this thesis approaches two of them: modelling a
building’s system and elements of urban city design.
Urban climate, in the UK, is being increasingly affected by climate change and urban
pollution remains a concern. How cities are maintained and designed is being adjusted to
consider these interactions. This thesis looks at the impact of roughness of the cities-scape
on wind speed, considered a factor capable of improving air quality . This thesis will looks
at urban albedo and the impact it has on air temperature at ground level compared with the
general degree of urban density.
Uncertainty is a part of complex systems such as cities which contain many elements
and in order to address this models are used to describe these system. A modeller will not
have access to all information or the time to address every element at a high level of detail.
The Gaussian processes used in this thesis have inherent uncertainty quantification, and they
make estimates that make allowances for inaccuracies. This means conclusions drawn using
this method can be considered more robust to uncertainties in the data.
This thesis will examine empirical data using different methodologies to draw conclusions
about model fitness of the methods used. The case studies that are used are the problem of
emulator construction for the building energy models (BEMs) and two example relationships
of urban weather from the Birmingham University Climate Laboratory (BUCL) and the
urban fabric.
Building energy use, through domestic, office and industrial consumption, is a major part
of how we as a society consumes electricity/gas and this consumption is metered. Building
systems are modelled using physical principles which requires a large amount of information
about constructed systems, user behaviour and the ambient environment which is very costly
justifying alternatives such as statistical modelling. This thesis will showcase how they can
be used to address the issue of climate change for a building energy use, in cooling.