Elsevier

Atmospheric Environment

Volume 37, Issue 37, December 2003, Pages 5197-5205
Atmospheric Environment

Emission and dispersion modelling of Lisbon air quality at local scale

https://doi.org/10.1016/j.atmosenv.2003.09.004Get rights and content

Abstract

The main objective of the paper is the study of air pollution in Lisbon city at local scale through the application of the modelling system developed at the University of Aveiro, linking two numerical tools: (i) the Transport Emission Model for Line Sources (TREM) and (ii) the Local Scale Dispersion Model (VADIS). Furthermore, analysis of the modelling system performance from the point of view of Quality Objectives established by the new European Legislation is one of the principal goals of the present work.

TREM is designed to support quantification of emissions induced by road traffic. The emission rate is estimated as a function of average speed. Different technologies (engine type, model year) and engine capacities are distinguished. The model is particularly designed for line sources and is implemented in a Geographical Information System.

VADIS is an integrated system, coupling a boundary layer flow module with a Lagrangian dispersion module, which was adapted to the simulation of urban air pollution, mainly in street canyon dispersion conditions. This model has the capability to support multi-obstacle and multi-source description, as well as time varying flow fields and time varying emissions.

The modelling system TREM/VADIS was applied to the Lisbon downtown area and results concerning carbon monoxide concentration values are presented and analysed. Also, a comparison of simulated and measured data using European legislation criteria is discussed. This numerical tool has demonstrated a satisfactory performance to calculate the flow and dispersion around obstacles under variable wind conditions, providing important information to be used by decision-makers for air quality assessment.

Introduction

Over the last few decades, road transport became the dominant source of air pollution in urban areas. This fact is mainly related to the rapid increase in passenger and freight transport. At the same time, new regulations on emission standards for industrial sources and power plants lead to a decrease of the contribution of these sources to urban air quality. Consequently, traffic induced pollution is currently one of the major problems of life quality in urban areas. This is particularly true in street canyons, where the combination of large vehicle emissions and reduced dispersion can lead to high levels of pollution (Hester and Harrison, 1997). Study of air quality problems at local scale requires application of an adequate methodology that permits to understand source-receptor relation and to develop a proper strategy to reduce atmospheric pollution (Moussiopoulos et al., 2003).

The continuous increase of hardware capabilities and the optimisation of numerical methods occurred mainly during the last decade allowed Computational Fluid Dynamics (CFD) models to emerge as an accurate tool for predicting urban airflow patterns and air pollutants dispersion on local scale, and specifically in urban environments (Borrego et al., 2002; Hassan and Crowther, 1998; Jicha et al., 2000). The numerical codes applying the standard kε model, in particular, have been widely used within this context. On the other hand, the coupling of boundary layer flow models with Lagrangian dispersion models has been already applied to the simulation of flows around buildings (Lee and Naesslund, 1998; Leuzzi and Monti, 1998). In the scope of SATURN, a subproject of EUROTRAC-2 Project (EUREKA), several inter-comparisons were performed between the results obtained with a set of CFD models and data from a real site, allowing to analyse models constraints and capabilities (Ketzel et al., 2001; Moussiopoulos et al., 2003).

In addition, the recent European Directives on air quality recognise the importance of modelling as a tool in the definition of high pollutant concentration areas that are not in compliance with air-quality objectives. Therefore, atmospheric pollution modelling has lately received an increasing attention from decision-makers, but credible results are also expected. For this purpose, the Directives establish new requirements for air quality modelling including definition of the Quality Objectives as a measure of modelling results acceptability.

The modelling uncertainty defined as Quality Objectives is estimated as the maximum deviation of the measured and calculated concentration levels over the period considered by the limit value (Table 1).

In order to improve the quality of model prediction, it is crucial to identify the error sources and then to decrease the uncertainties (Britter, 1994). One of the sources of model uncertainties is input information, where emission data is always an important issue. Whereas urban air pollution is a subject of research, a precise quantification of pollutant amount emitted by vehicles to the atmosphere is essential. Then, emission models able to compute emission data on different spatial and temporal scale are widely used to provide the required information. Currently, there are a number of emission models that differ by the fleet composition (vehicle categories and age), driving pattern (average speed only or instantaneous speed and acceleration), covered pollutants and type of emissions (hot, cold or evaporative). Selection of an adequate model is an important step to reduce final uncertainty on air quality prediction (Hickman et al., 1999).

The use of a numerical model to estimate pollutants concentration at local scale can be an important contribution to the identification of sensitive urban areas in terms of air quality and evaluation of human exposure to different pollutants. The application of this numerical tool is also important in traffic management and in the definition of strategies for air quality management in urban centres, through the development of future traffic scenarios.

The main objective of the paper is two-fold: the application of modelling tools to an air quality study of a real city case, and analysis of the model performance from the point of view of Quality Objectives established by the new European Legislation. To achieve this objective, a numerical system based in two modelling tools was applied: (i) the Transport Emission Model for Line Sources (TREM) and (ii) the Local Scale Dispersion Model (VADIS). This numerical system was specially adapted to estimate the atmospheric pollution induced by road traffic in urban areas (Borrego et al., 2000b).

Section snippets

Methodology

The methodology to estimate the atmospheric pollution in the Lisbon area is focused on the calculation of traffic emissions and simulation of the dispersion conditions. The two components of the developed numerical system are described below.

Application

In previous works, the importance of road traffic impact on Lisbon air quality at regional scale was analysed, confirming the importance to develop a suitable air pollution abatement strategy (Borrego et al., 2000b). This work intends to contribute to the study of Lisbon air quality in a local perspective. Therefore, the developed modelling system was applied to a specific area during a typical summer day, which was chosen using a statistical meteorological approach. The selected area is

Conclusions

The modelling system TREM/VADIS was developed to predict air quality in urban areas. Emission estimations with high temporal and spatial resolution were provided by TREM and the results analysed in order to define how insufficient input data on fleet composition and driving conditions can affect the emission calculation. Based on Monte Carlo approach, it was demonstrated that different pollutants present different sensitivity to the input data, reflecting the uncertainty of emission

Acknowledgements

This work was partially supported by Portuguese PRAXIS XXI grants and by the EC SUTRA project (EVK4-CT-1999-00013). Also, the work is included in SATURN subproject of EUROTRAC-2.

References (17)

There are more references available in the full text version of this article.

Cited by (0)

View full text