Modelling smoke effects on Lisbon air quality

Each summer season wildland forest fires burn a considerable area of south European landscape. Summer 2003 was one of the most severe fire seasons experienced during the last decades in Southern Europe and, due to persistent extreme fire conditions, Portugal suffered the worst forest fire season that the country has faced in the last 23 years, with a total area burned of almost 5 times the average (EC, 2004).

Smoke has to be considered as one of the several disturbing effects of forest fires. Its impacts on air quality and human health can be significant because large amounts of pollutants, like particulate matter (PM10) and carbon monoxide (CO), are emitted to the atmosphere. During 2003 portuguese forest fires there were reports of more than one thousand people (mainly civilian) needing medical assistance due to smoke intoxications, burns, and wounds (EC, 2004). Sattellite images and also air pollutants concentration values measured on the portuguese air quality monitoring network highlighted the impact of forest fire atmospheric emissions.

Lisbon, which suffered the effects of smoke from forest fires spreading north of the urban area, because of its high population density and hence higher risk of human exposure to smoke, is a very interesting case study regarding smoke dispersion and the estimation of air pollutants concentrations.

Smoke dispersion numerical modelling allows to understand how pollutants emitted by a forest fire will be transported and dispersed in the atmosphere estimating the outcoming air pollutants concentration fields. The main purpose of this paper is to estimate the effects of 2003 fires on Lisbon air quality through the apllication of a mesoscale photochemical modelling system, AIRFIRE (Miranda, 2004), and to evaluate its performance comparing results with air pollutants concentration measured values.

The Lisbon airshed, with a population of 3.5 million inhabitants, is the most important urban centre in Portugal. Lisbon is a coastal city built in a complex topographic region, dominated by a large estuary and multiple hills, surrounded by small mountain ranges reaching heights over 400 m above sea level. The urban area has an air quality monitoring networking, which includes several stations, with different typology (urban background or urban traffic), according to location and environmental criteria.

Forest fires spreading north of Lisbon city at September 2003 impacted the urban air quality. A first analysis of measured values of CO and PM10 during September was carried out aiming to identify one period where Lisbon was clearly affected by smoke from forest fires. It was possible to identify the 13rd of September as the most critical day regarding CO and PM10 concentration values in Lisbon urban area. According to European Air Quality Legislation PM10 daily averaged values should not exceed 50 microg.m-3 more than 35 times per year. PM10 concentration values are in general high, but due to forest fires influence this averaged 50 microg.m-3 value was exceeded at all the measuring stations and hourly-averaged values reached 500 microg.m-3.

33 fires occurred at the 13rd of September in Lisbon district, burning an area of about 400 ha of forest stands and shrubs. AIRFIRE was applied to this specific September day, which was very warm, with temperatures reaching 35°C. Winds mainly blew from E quadrant, changing towards SE at the end of the afternnon, when higher concentration values were measured. The simulation domain covered an area of 200 km x 200 km, with a horizontal resolution of 4 km. In the vertical direction the grid consisted of 20 layers non-equidistant with a minimum grid spacing of 20 m near the ground and a vertical extension of 6000. This domain was chosen in order to consider mesoscale circulations, like sea breezes, which are frequent in the Lisbon area.

Forest fire emissions, namely CO, NOx, volatile organic compounds and PM10, were estimated taking into account the availabe information about the fires characteristics and considering fuel loads, combustion efficiencies and emission factors for three main fuel types: shrub, pine and eucalyptus. Besides wildfire emissions the contribution of other pollution sources, like traffic, industry, residential and commercial heating, to air quality degradation was also included.

Hourly wind and concentration fields were estimated by AIRFIRE. Figure 1 presents the surface PM10 hourly-averaged values estimated at 10 pm.

Figure 1 - Surface hourly-averaged PM10 concentration values.

To evaluate quantitatively the obtained results, a statistical analysis was performed based on three parameters: bias, correlation factor and root mean square.

This modelling work allowed the estimation of wind and concentration fields. The complexity of the coastal zone under study justified the use of AIRFIRE. The system presented a good performance for PM10 and CO and a satisfactory one for ozone, revealing the effect of fire emissions on the concentration of these two pollutants. This paper can be a contribution to the understanding of the importance of urban/wildland fires on air quality at the urban level and to the analysis of potential health effects.

References

European Commission (2004): Forest Fires in Europe – 2003 Fire Campaign, Schmuck, Miguel-Ayanz, Barbosa, Camia, Kucera & Liberta (Ed), Official Publication of the European Communities, SPI.04.124 EN.

Miranda, A.I. (2004) An integrated numerical system to estimate air quality effects of forest fires. International Journal of Willdand Fire 13, 217-226.