| Popis: |
With continuing urbanisation and business-as-usual emission regulation policies, mortalities due to urban air quality will continue to rise in the period up to 2050. Poor urban air quality has significant impacts across the three capitals of sustainability (environmental, economic and social effects). Our understanding of and ability to model air pollution in urban areas is integral towards the development of both short- and long-term solutions to this problem. This thesis covers the development of the model capabilities required to study urban air pollution at the microclimate scale; with a focus on five key elements: 1) morphology, 2) emissions, 3) chemistry, 4) atmospheric stability and 5) trees. Following the conceptual methodology outlined in this thesis, these elements are systematically investigated within the context of the inherently multi-scale and heterogeneous nature of urban environments. Large-eddy simulation (LES) models are capable of resolving the energetic turbulent scales of the unsteady urban flow field up to the highest possible resolutions (O(1 m, 0.1 s)), which is integral to accurately capturing the dispersion of pollutants within the urban canopy layer (UCL). The application of the urban LES model, uDALES, to study pollution dispersion within realistic urban morphologies is validated against wind tunnel data. uDALES is coupled to a traffic microsimulation model to capture realistic vehicular emissions and the LES model is further extended to resolve the chemical reactions of the null cycle. In terms of atmospheric stability, the incorporation of thermal effects into urban LES results in slow transients and significant changes in the modelled boundary layer dynamics. A comprehensive study is therefore conducted to formulate best practice guidelines for the modelling of non-neutral urban boundary layers in statistical steady states. A novel method is devised in which a developing convective or stable boundary layer can be `frozen' in time. Trees are an integral part of the urban form and their presence within the UCL affects the transfer of momentum, radiation, temperature, moisture and pollutants. A minimal model for trees is presented that captures all of these processes whilst minimising the number of additional free parameters introduced to the problem. A bulk parametric study is used to investigate the sensitivities of the resulting energy balance system, with a focus on the attribution between tree's transpirational and shading cooling effects. Trees are shown to have both positive and negative effects on urban air quality depending on the incorporation of thermal effects and the roadside-background concentration ratio of the considered pollutant (a reduction in PM2.5 of 40% under neutral conditions and an 8% increase in NOx under convective conditions are found). In the vicinity of high emissions, the drag imposed by the tree canopies (as well as the elevated buoyancy sink and reduced surface heat flux) are shown to lead to degradations in air quality. However for pollutants that exist more homogeneously within the urban environment, the deposition of pollutants onto the tree canopy can lead to local reductions in pollutant concentrations. The discussed model developments are showcased through a comprehensive case study over South Kensington, London, UK. Incorporating all of the five elements of urban air quality into one investigation presents a significant contribution towards realistically modelling and fully resolving air pollution in the urban microclimate. LES results are systematically compared to gain a fundamental insight into the change in air quality due to the switch between e.g. neutral and convective conditions (16% reduction in pedestrian-averaged NOx concentrations) and with and without trees (maximum increase in pedestrian-averaged NOx concentrations of 87.9% and maximum decrease of 15.4%). Comparable simulations are also run over the case study region using the operational air quality model SIRANE. This enables an evaluation of the assumptions, simplifications and parametrisations used in operational models and makes use of the high resolution capabilities and numerical control of uDALES. SIRANE is generally shown to provide a good prediction of canyon-averaged NOx concentrations over a range of different conditions. The assumption of photostationarity is shown to lead to over- (average fractional bias, FB = -0.20) and under-predictions (FB = 0.11) of reactive NO and O3 respectively. It is also found that the use of canyon-average concentrations in SIRANE results in a systematic under-prediction of pedestrian-level exposure. Open Access |
