Urban Air Quality: Monitoring and Modelling

Mass Concentration of ambient particulate matter with an aerodynamic diameter less than 10 µm (PM₁₀) are reported for the first time for a range of sites in Dublin City over a 6 month period from January 1^st 1996 to June 30^th 1996. PM₁₀ gravimetric mass concentration measurements are made with low flow Partisol 2000 air samplers using an impaction type PM₁₀ inlet and 47mm diameter glass fibre filters. In addition, much finer time resolution measurements (minimum sampling frequency of 30 minutes) are made using a tapered element oscillating microbalance (TEOM) PM₁₀ mass monitor. These PM₁₀ mass concentrations methods are also compared with mass concentration inferred using the standard black smoke method. Analysis of the ambient mass concentration data with reference to traffic density and meteorological influences are presented. Results for the first six months of 1996 show that the average PM₁₀ values range from a high of 49 μg m^-3 at the Dublin city centre site to 14 μg m^-3 at one of the suburban sites. Intercomparison between PM₁₀ and black smoke mass concentrations show that the relationship is site specific. Statistical analysis between PM₁₀ levels and car traffic number show a positive correlation while a weak negative correlation is found between PM₁₀ levels and rainfall amount, wind speed and air temperature.

A passive sampling device, developed for personal monitoring of airborne dust levels in industry, has been tested as a site sampler in the urban environment. The device weighs approximately 15g and the essential sampling element is a small disc of electret material (polymer carrying a permanent electric charge). During use it captures particles by electrical attraction, at a rate that depends upon their electrical mobility, but which is independent of the rate at which air flows past the sampler. It collects measurable quantities of particulate, though the sample size tends to be small and correlation with results from conventional samples has not yet been established. Samplers have been exposed to urban particulate for periods of up to seven days, without the electret suffering unacceptable loss of electric charge. It has been shown to be potentially useful for long-term monitoring, a situation in which dispensing with a power source is particularly useful. Being small, the passive sampler is easy to hide or camouflage. It has potential for multiple simultaneous site sampling and for monitoring personal environmental exposure.

Concentrations of PM₁₀ particulates have been compared to the concentrations of oxides of nitrogen and sulphur dioxide at the fourteen Automatic Urban Monitoring Network (AUN) sites operating during 1993, 1994 and 1995 using fully ratified data.

Factors which are considered include diurnal variations in concentrations of the various substances, ratios between concentrations of PM₁₀ and the other substances and differences between relationships in summer and winter and between weekdays and weekends. In addition temporal patterns of concentrations are considered. Variations between different cities is demonstrated. Differences in the seasonal size distribution are identified. It is shown that there is a good degree of consistency in concentrations of PM₁₀ across urban areas at background locations and that there is evidence for long range transport of PM₁₀.

The Building Research Establishment (BRE) has been using diffusive samplers for the study of VOCs in indoor and outdoor air since 1989. The Perkin Elmer type sampler packed with Tenax TA adsorbent is used for the diffusive sampling of C₆–C₁₆ organic compounds. This method was applied in a major study of relationships between the environment and child health carried out during 1990–1993 in the Avon area of the UK. The present paper reports results of an investigation into the repeatability of the sampler in outdoor air and measurements of 6 aromatic hydrocarbons inside and outside a home over a 5 year period and inside and outside an office building over a 12 month period. Both the home and the office were located in Hertfordshire, England. Concentrations of VOCs recorded are similar to those found in the Avon area. Higher concentrations of each of the six compounds were recorded inside the home than outside, whilst greater amounts of benzene and toluene were found inside the office than outside. Seasonal variations in concentrations are observed and measurements recorded outdoors are similar to those recorded by other workers.

Diffusive sampling of Volatile Organic Compounds (VOCs) onto thermal desorption tubes, followed by gas chromatography, is an established technique for area or personal monitoring of typical workplace concentrations and there has been increasing interest in extending the application to environmental levels, particularly for benzene, toluene and xylene (BTX). Diffusive sampling rates for BTX on Chromosorb 106 and Carbograph-1 (a graphitised carbon) were measured over periods of 1–4 weeks in field validation experiments using ambient air and parallel pumped sampling (the reference method) at the HSL site in central Sheffield. The reference method was also used to investigate the possible bias of an open-path spectrophotometer (OPSIS) used by Sheffield City Council. A bias for BTX was suspected from results of a two-week initial exercise in which several diffusive samplers were placed close to the light path. In the full field validation of the diffusive samplers carried out subsequently, the significant bias of BTX concentrations reported by OPSIS were confirmed when compared with concurrent results from the reference method. OPSIS gave benzene and toluene values up to eight times higher than expected from the GC measurements. Xylene discrepancies were smaller, but in one 3-day peak episode, OPSIS demonstrated a negative correlation with GC.

The acquisition of a comprehensive air quality dataset for a small city environment is described for use in statistical modelling of dispersion processes and micro-scale assessment of polluted zones. The dataset is based on a nitrogen dioxide diffusion tube survey for Cambridge where up to 80 roadside and background sites have been monitored continuously over two years, using a two week exposure period. Site categories are defined by their function within the urban landscape. Spatial and temporal features of the data set are explained in terms of urban location, street geometry, meteorology and traffic behaviour. The highest levels of NO₂ are found in central canyon streets which are narrow with enclosing architecture and slow-moving traffic. In contrast lower levels are found for the wider, more open radial routes where traffic is free-flowing. The influence of street geometry on NO₂ levels for central streets is demonstrated, where canyon sections adjacent to open sections having the same traffic flow record higher concentrations. Whilst all roadside sites are affected by a photochemical pollution ‘episode’, the greater potential for elevated NO₂ concentrations within the canyon sections is significant. The close proximity of low background levels of NO₂ to roadside ‘hot-spots’ is important for public exposure assessment. The variation in background levels across the urban landscape is very small and unrelated to location; whether central, suburban or outer city. Seasonal variation, not seen in roadside data, is clearly apparent in background data with a winter maximum and summer minimum.

The Helsinki Metropolitan Area Council (YTV) is responsible for air quality monitoring in the Helsinki area. Air quality has been monitored periodically since the late 1950s. An automatic SO₂ monitoring network was constructed in 1975 and TSP measurements were added in 1978. Since then the network has been expanded and currently five automatic multicomponent stations form the basis of the network monitoring SO₂, NO, NO₂, CO, PM₁₀ and O₃ concentrations. Manual TSP and PM₁₀ measurements are also conducted. Mobile monitoring units are also being used as well as special measurement campaigns. The effects of air pollution on nature are studied in bioindicator monitoring. An air quality index is used in order to inform the public of the current air quality situation. Changes in air quality are reflected in monitoring strategy. SO₂ concentrations have decreased in the past two decades. Annual averages in 1995 were at or below 5 μg/m³. Traffic is the major source for pollutants even though catalytic converters have lowered traffic emissions somewhat. The highest annual average NO₂ concentration at an urban site was 49 μg/m³ in 1995, and there has been no clear change in NO₂ levels. There has been a decreasing trend in CO concentrations. Maximum annual TSP and PM₁₀ averages in 1995 were 92 and 32 μg/m³, respectively. The highest average lead concentration was 0.01 μg/m³. Elevated concentrations are experienced from time to time. During the spring daily TSP and PM₁₀ concentrations can go up to around 300 and 150 μg/m³, respectively. This is caused by resuspension mainly due to street sanding. Also a major winter NO₂ episode occurred in December 1995. The highest hourly NO₂ concentrations reached 400 μg/m³.

A field experiment for the comparison of the efficiency of canisters and adsorption multibed tubes for sampling atmospheric highly volatile hydrocarbons at ppbv levels is described. The canister was passivated by the Summa process and the adsorption tubes were filled with Carbotrap C., Carbotrap B and Carbosieve S-in. The sampling with the adsorption tubes was performed at ambient temperature and at −10°C. The highest concentrations were generally obtained with canisters but these results are very similar to those obtained with refrigerated multibed adsorption tubes. Both methods appear to be equivalent for most of the highly volatile hydrocarbons encountered in moderately polluted urban areas. In contrast, sampling with ambient temperature tubes provides lower concentrations. This study has also shown that K₂CO₃ drying efficiently removes humidity from air samples allowing the obtention of reliable concentration data on highly volatile hydrocarbons at ppbv levels. These drying tubes can easily be re-conditioned and tested for blanks and memory effects, which greatly facilitates the control of external contamination and sample cross-contamination.

The presence of volatile organic compounds (VOCs) from traffic and other sources in urban areas is a cause for concern about public health. Canister, chemical derivatisation, particulate sampling and adsorption sampling techniques were used to measure VOC concentrations of a wide range of compounds (C₆–C₄₀) during a four day campaign in south London with subsequent laboratory analysis of the samples. Compounds quantified included alkanes, mono- and poly-nuclear aromatic hydrocarbons. Also the first sequential measurements of carbonyl compounds (C₁–C₈) in a UK urban area are presented. Results from canister and adsorption sampling methods are compared. A comparison of the results with other urban data is presented and the temporal variations in VOC concentrations were interpreted with reference to the prevalent wind speeds and directions The CALINE4 line source dispersion model was generally successful in reproducing the daytime 12 hour average concentrations of selected VOCs.

Concentrations of nitrogen dioxide were measured in shops and car parks over a 19 week period during the winter. For the shops indoonoutdoor ratios varied from 0.34 to 0.54 with an average value of 0.44. NO₂ levels ranged from 13 to 38 μg m^-3. With little or no contribution from indoor sources, ambient concentrations are the primary factor in determining indoor levels. Concentrations of NO₂ in car parks were similar to those measured at the kerb side. Concentrations in the payment booths were higher than those in shops and ranged from 42 to 60 μg m^-3. Despite ventilation systems in the payment booths ratios of boothxar park NO₂ varied from 0.65 to 0.86.

The experimental site was a Mediterranean type forest located in Viols-en-Laval near Montpellier (France). The principal species studied were Quercus ilex and Quercus pubescens. The determination of biogenic emissions was carried out by the enclosure method, which consists in enclosing an intact branch in a Teflon cuvet. The evolution of global terpenic emissions were recorded on June 1995.

For Quercus ilex it appears that most terpenic emissions take place during the diurnal period. The evolution of terpenic emission rates versus light (PAR), internal temperature within the cuvet, CO₂ exchange and transpiration (H₂O) were carried out. A close relation between terpene emission and light-triggered physiological activities, was always found. It is obvious that there was no simple correlation between the leaf temperature and the terpenic emission. The temperature of the leaves does not seem to play a significant role in the regulation of the monoterpenic emission. This behaviour allows us to conclude that light with its effect on assimilation and stomatal behaviour is the most important factor for monoterpenic emission by Quercus ilex like for Quercus pubescens which is an isoprene emitter.

Inexpensive near-infrared diode lasers are now being used to enable high sensitivity, real time monitoring of gases both in open path measurements of urban air quality and in industrial environments for stack gas emission monitoring. Individual species are detected in a highly selective manner via overtone and combination bands of their vibrational spectra. Operating at room temperature and with simple optical components they can be made into portable instruments, ideal for field measurements. Combined with optical fibres they can be used for accessing remote and possibly dangerous locations. This paper presents current progress on the development of such a system illustrating recent results on ammonia monitoring at 1540 nm and acetylene at 780 nm. Sensitivities of the order of parts per million and below have been attained. Advantages and disadvantages of this approach to air pollution monitoring will be described. Pressure broadening results are also given for the 2v₁ band of ammonia.

Pollutant concentrations relative to the years 1993–1995 measured in the Bologna urban area by a network of automatic stations (S.A.R.A. Sistema Automatico di Rilevamento Ambientale) are analysed to estimate the typical behaviour of the pollutants, especially with regard to periodicities and meteorological dependencies and to attempt prediction for the pollutant daily concentration up to a few days using a statistically based model.

The results obtained show the presence of a very significant weekly periodicity for all the analysed pollutants. A yearly periodicity has been found only for the primary pollutants analysed.

The statistical prediction, using an ARMA model with meteorological variables as transfer functions, shows good predicting capabilities up to one day for carbon monoxide (CO) and sulphur dioxide (SO₂) and up to two days for nitrogen dioxide (NO₂).

A study was carried out to investigate the effects of meteorological conditions on atmospheric Nitrogen Oxide (NO_X (Nitrogen Oxide and Nitrogen Dioxide)) concentrations at a site in Dublin. Data used in the study (meteorological conditions and hourly NO_X concentrations) were compiled from hourly records for the years 1988–1992. The research identified wind speed, air pressure and wind direction as the most important meteorological parameters for understanding the behaviour of extreme NO_X concentrations in Dublin’s air. Daily, weekly and seasonal variation in NO_X concentrations were observed. This work also highlighted the importance of the role played by general synoptic weather conditions over local climatic effects in extreme events.

This paper examines air quality changes that can arise from implementing specific traffic management schemes, based on the findings of current and past impact assessments Air quality changes arising from pedestrianisation, traffic rerouting, traffic calming, and bus priority routes are outlined Case studies are presented where air quality was measured before and after the implementation of a scheme in order to assess how effective traffic management can be in reducing ambient traffic-related pollutant levels Limitations of past and current studies are highlighted. The need for more monitoring studies, and the need for data from those studies to be used to test and validate a range of suitable numerical models, is emphasised

State-of-the-art approaches for urban air-quality characterisation have several drawbacks due to apriori assumptions and/or due to inherent limitations of the concept utilised. For the evaluation of abatement scenarios it is either necessary to embark on extensive monitoring campaigns or to consistently apply numerical models for atmospheric dispersion. The ‘ENVISOR’ methodology applied here is a mixture of the two approaches. It forecasts pollutant concentrations during real episodes and assesses the impact from the construction of a new highway across a large urban domain of 100×100 km². Data from an extensive monitoring network are used to identify real modelling periods and for validating the modelling simulations. The selected periods are aiming to the assessment of ‘annual mean’ or ‘episodic’ conditions. These periods are short-listed according to the abatement scenario under consideration. This approach yields accurate forecasts for the concentration of pollutants after extensive validation tests extended over the whole domain. It is foreseen that the impact from the highway construction will be minimal for photochemical pollution whereas, higher impact will result for inert pollutants due to additional emissions from the highway.

The ANA Air Quality Model (ANA stands for Atmospheric Mesoscale Numerical Pollution Model for Regional and Urban Areas) has been applied over Madrid during a five day period in June, 1995. The domain is 80 × 100 km² and the spatial resolution is 2000 m. The ANA system is driven by a meteorological model REMEST and it includes a detailed emission model for anthropogenic and biogenic sources with 250 m spatial resolution and 60 minutes temporal resolution. Different deposition processes are used such as the Wesely (1989) and Erisman et al. (1994) resistance approaches and the simple aerodynamic resistance. The photochemical processes and the general chemistry is based on the CBM-IV mechanism for the organic compounds and solved by the SMVGEAR method (CHEMA module).

The model uses 14 different landuse types which are obtained by using the REMO module which uses the information provided by the LANDSAT-5 satellite image over the domain. The emission module EMJMA takes into account the point, line and area emissions over the domain. Special importance is given to the biogenic emissions which are obtained by using the satellite landuse classification for caducous, perenneal and mixed terrain. The emission module considers the EPA and CORINAIR emission factors. The results show an accurate prediction of the ozone maxima for the five days and also the general pattern of the ozone observed data. The five day simulation is characterized by a local low pressure over the Madrid Area and high pressures over Spain and West of Europe. The ozone surface patterns show the diurnal cycle and the maxima concentrations up to 140–160 ppb for suburban areas during afternoon hours. The general performance of the model is considered quite good. The computer power requirements continue to be very high for standard workstations. Future progress on parallel platforms should improve considerably the computer time requirements.

The air pollution transport model UGEM (The University of Greenwich Evaluation Model) has been developed to evaluate medium-range transport and deposition of sulphur and oxidised nitrogen from all types of sources of emissions in the UK and to estimate their average annual deposition and concentrations across the UK. The model has been tested for its predictions against the available measurements.

This study was focused on a possibility of applying the UGEM model to the assessment of air quality on a local scale. One parameter in the model is crucial, the local deposition fraction. The effect of this parameter on quality of the model predictions has been studied for different scales of UGEM output, such as the whole territory of the UK, a rural region and an urban area.

The results of the study show that the magnitude of the local deposition fraction should be different for each grid square to reach the best agreement of predictions of concentrations with measurements. Applying a local value of the parameter to each grid square will improve the model predictions of the concentrations in urban areas in particular and will not affect the quality of model predictions of the wet deposition.

An accurate prediction of the transport-reaction behaviour of atmospheric chemical species is required to fully understand the impact on the environment of pollution emissions. Elevated levels of secondary pollutants such as ozone in the lower atmosphere can be harmful to the health of both plants and animals, and can cause damage to property present in the urban environment. Detailed models of pollution mechanisms must therefore be developed through comparisons with field measurements to aid the selection of effective abatement policies Such models must satisfy accuracy requirements both in terms of the number of species represented, and the spatial resolution of species profiles. Computational expense often compels current models to sacrifice detail in one of these areas. This paper attempts to address the latter point by presenting an atmospheric transport-reaction modelling strategy based upon a finite volume discretisation of the atmospheric dispersion equation. The source terms within this equation are provided by an appropriate reduced chemical scheme modelling the major species in the boundary layer. Reaction and transport discretisations are solved efficiently via a splitting technique applied at the level of the non-linear equations The solution grid is generated using time dependant adaptive techniques, which provide a finer grid around regions of high spatial error in order to adequately resolve species concentration profiles. The techniques discussed are applied in two dimensions employing emissions from both point and area sources. Preliminary results show that the application of adaptive gridding techniques to atmospheric dynamics modelling can provide more accurately resolved species concentration profiles, accompanied by a reduced CPU time invested in solution Such a model will provide the basis for high resolution studies of the multiple scale interactions between spatially inhomogeneous source patterns in urban and regional environments.

This paper concerns the role and nature of physical modelling as applied to dispersion studies. The arguments leading to the choice of model scale are discussed. The simulation of atmospheric flows and emissions are described, in particular the similarity conditions linking full scale and model conditions and their implications in terms of facility size and performance. We briefly summarise typical experimental programmes and treat the issues involved in interpreting results. Then we present three examples of laboratory studies. The first illustrates the ability to perform a simple plume rise simulation and the other two demonstrate building affected dispersion simulations. The final discussion centres on applications of physical modelling to urban air quality investigations.

A street canyon model has been formulated based on work published by Hertel and Berkowicz. An outline is given of the theoretical approach used, followed by a modelling of nitrogen oxides and carbon monoxide measurements from sites at Cromwell Road, Central London and Stratford Road, Birmingham. Modelled concentrations were compared with observed mixing ratios for both sites. At Cromwell Road, good agreement was achieved for one month but which was not reproduced as well for the other two months tested. There is uncertainty as to the effect of one of the side streets and whether the general flow is altered during periods of marked solar heating. Also emissions from vehicles may vary from those assumed. The interpretation of the Stratford Road site’s results was less straightforward with complications concerning background pollutant levels and changes in emissions from interrupted traffic flow.

A single compartment model has been constructed for predicting hourly concentrationsof pollutant concentrations arising from vehicular emissions within a typical street canyon The model takes account of traffic densities and composition to estimate pollutant emissions within the model compartment Meteorological data on wind speed and direction are used to define the exchanges of pollutants between the compartment and the surrounding air A parameter is also included to describe the exchange in calm conditions. The pollutant concentrations are then estimated from a steady state mass balance equation for the compartment, assuming conservation of pollutants The model was applied to the prediction of carbon monoxide concentrations in Hope Street, Glasgow Model parameters were fitted using field measurements, together with concurrent meteorological data and traffic flows estimated from traffic census data for Hope Street The model accounted well for the observed variations in carbon monoxide It was found that the model parameters varied seasonally, perhaps due to differences in atmosphericstability which have not so far been included in the model formulation

A two-dimensional steady state numerical simulation has been carried out for a typical street canyon ventilated by a cross-wind. The PHOENICS package from CHAM was used to solve for the air flow above and within the street canyon. The k-epsilon turbulence model was used for turbulence modelling and pollutant sources were added at ground level over the road but not over the pavements. Results for the air flow showed the formation of a longitudinal vortex within the street canyon, as found by other researchers. Pollutant concentrations were predicted with the highest values occurring at the leeward walls of the upwind buildings, and the lowest values on the windward walls of the downwind buildings. The accuracy of these simulations was examined by comparing the predicted results with field observations. Reasonable agreement was obtained, confirming the difference between concentrations on the leeward and windward walls. The results show that the dispersion characteristics can be simulated in terms of structural configurations.

Urban pollution due to roadways is perceived as a major obstacle to implementing lowenergy ventilation design strategies in urban non-domestic buildings. As part of a project to evaluate the use of a computational fluid flow model as an environmental design tool for urban buildings, this paper seeks to address the impact of pollution from roadways on buildings in areas of restricted topography and assess dominant influencing factors and other requirements for testing the flow model predictions. Vertical profiles of carbon monoxide (CO) and temperature at the facade of a building in a Central London street, in addition to above-roof wind speed and direction, were measured over a period of three months. The street has a height-to-width (h/W) ratio of 0.6 and is of asymmetric horizontal alignment. The air flows in the area surrounding the building were modelled using a computational fluid flow model for two orthogonal wind directions. CO concentrations were calculated from the steady-state flow field in order to place point measurements in the context of the flow field, identify persistent features in the measured data attributable to the flow structure and, by comparison with measurements, identify further testing requirements.

Some qualitative and quantitative agreement between measured and modelled data was obtained. Measured CO levels at the building facade and vertical variations of CO were small, as predicted by the model. A wake-interference type flow was predicted by the model for wind speeds >2ms^-1 with formation of a vortex cell occurring for roof-level wind speeds >5ms^-1 for the cross-wind direction, which was reflected in the measured CO levels and facade gradients. A direction-dependent inverse relationship was noted, both in the model and measurements, between above-roof wind speed and facade CO levels although statistical correlations in the time series were poor. CO concentrations at the facade were found to increase with height frequently, as well as decrease, especially for parallel winds. It is expected that mechanical turbulence due to vehicles was largely responsible. In comparison, thermal stratification appeared to play only a minor role in controlling vertical mixing in the street, under low wind speed conditions.

This paper presents a model to forecast concentrations of nitrogen dioxide NO₂ in urban areas. A box model provides a simple and robust description of air quality in urban areas, especially where emissions data are only available on a fairly coarse grid. BOXURB is a box model with some novel features. It can use synoptic observations of the meteorology, or numerical forecasts of wind and cloud. Traffic emissions are adjusted by time and day of week, and space heating emissions by degree hours. The urban heat flux combines an urban heat store adjustment, a space heating term, and the rural heat flux. Pollutants accumulate with the onset of light winds. NO₂ is forecast by subdividing the NO _x into its constituents, NO (nitric oxide) and NO₂. The subdivision is empirical for most of the year, but in ozone episodes the photostationary state is invoked. The forecasts are sent by the Met Office to NETCEN AEA Technology for use in the Department of Environment’s Air Quality Bulletin System.

The Environmental Protection Act 1990 has resulted in a large number of air quality modelling assessments being carried out to predict the effects on ambient air quality of various individual industrial sources as a part of the application process for authorization to operate. Currently there is no standard approach to carrying out air quality modelling assessments in the UK and, consequently, the content and quality of studies can vary enormously. To examine this variation, twenty five assessments which included dispersion modelling were selected and assessed against criteria derived from consideration of the requirements of these studies. The criteria derived also took account of two sets of guidelines. One set has been produced by the Institute of Environmental Assessment and the other by the Royal Meteorological Society. It should be noted, however, that this paper does not set out to produce a “user’s manual” for air quality modelling studies, but rather to examine various studies against set criteria.

Aspects of each assessment which have been examined include: the site description; the objectives of the study; the representativeness and reliability of the input data; the assumptions made; and the communication of the results. Each category considered has been evaluated and marked against a common scoring scheme.

In general the resulting scores were poor with 56% of the assessment scoring less than 50% and only 8% of the assessments scoring more than 70%. Many of the assessments examined would not be capable of being fully audited or reproduced because they contained insufficient details of the input data used or the assumptions made.

In the context of ever increasing concern regarding the effects of air quality there needs to be a clear understanding of the efficiency and effectiveness of the tools which are being used to measure or predict air quality. Also in light of the UK Department of the Environment’s air quality management initiative it is necessary for there to be a clear understanding of the most appropriate ways in which such tools should be applied.