Difference between revisions of "Evaluation of intake fractions for different subpopulations due to primary fine particulate matter (PM2.5) emitted from domestic wood combustion and traffic in Finland"

Revision as of 18:33, 26 April 2012

This page (including the files available for download at the bottom of this page) contains a draft version of a manuscript, whose final version is published and is available in the [Air quality, atmosphere & health http://www.springerlink.com/content/8521j0u04h865751/?MUD=MP]. If referring to this text in scientific or other official papers, please refer to the published final version as: Pauliina Taimisto, Marko Tainio, Niko Karvosenoja, Kaarle Kupiainen, Petri Porvari, Ari Karppinen, Leena Kangas, Jaakko Kukkonen, Jouni T. Tuomisto: Evaluation of intake fractions for different subpopulations due to primary fine particulate matter (PM2.5) emitted from domestic wood combustion and traffic in Finland. Air quality, atmosphere & health. Received: 29 October 2009 / Accepted: 8 February 2011 (c) Springer Science+Business Media B.V. 2011 {doi|10.1007/s11869-011-0138-3}.

Title

Editing Evaluation of intake fractions for different subpopulations due to primary fine particulate matter (PM2.5) emitted from domestic wood combustion and traffic in Finland

Authors and contact information

Pauliina Taimisto, correspondence author

(pauliina.taimisto@thl.fi

(National Institute for Health and Welfare (THL), Kuopio, Finland)

Marko Tainio

(National Institute for Health and Welfare (THL), Kuopio, Finland)

(Systems Research Institute, Polish Academy of Sciences, Warsaw, Poland)

Niko Karvosenoja

(Finnish Environment Institute, Helsinki, Finland)

Kaarle Kupiainen

(Finnish Environment Institute, Helsinki, Finland)

Petri Porvari

(Finnish Environment Institute, Helsinki, Finland)

Ari Karppinen

(Finnish Meteorological Institute, Helsinki, Finland)

Leena Kangas

(Finnish Meteorological Institute, Helsinki, Finland)

Jaakko Kukkonen

(Finnish Meteorological Institute, Helsinki, Finland)

Jouni T. Tuomisto

(National Institute for Health and Welfare (THL), Kuopio, Finland)

Abstract

Domestic wood combustion and traffic are the two most significant primary fine particulate matter (PM_2.5) emission source categories in Finland. We estimated emission–exposure relationships for primary PM_2.5 emissions from these source categories using intake fractions (iF), which describes the fraction of an emission that is ultimately inhaled by a target population. The iFs were calculated for four different emission source subcategories in Finland in 2000: (1) domestic wood combustion in residential buildings, (2) domestic wood combustion in recreational buildings, (3) traffic exhaust and wear emissions, and (4) traffic resuspension emissions. The iFs were estimated for both total population and for subpopulations with different gender, age, and educational status. Primary PM_2.5 emissions were based on the Finnish Regional Emission Scenario model and the dispersion of particles was calculated using the Urban Dispersion Modeling system of Finnish Meteorological Institute. Both emissions and dispersion were estimated on a 1 km spatial resolution. The iFs for primary PM_2.5 emissions from (1) residential and (2) recreational buildings were 3.4 and 0.6 per million, respectively. The corresponding iF for (3) traffic exhaust and wear and (4) traffic resuspension emissions were 9.7 and 9.5 per million, respectively. The differences in population-weighted outdoor concentrations were significant between subpopulations with different educational status so that people with higher education were exposed more to traffic-related PM_2.5.

Keywords

iF, Intake fraction, Exposure, Particulate matter, Domestic combustion, Traffic

Introduction

Exposure to air pollution, especially for fine particulate matter (PM2.5; particles with aerodynamic diameter of less than _2.5 μm), have been associated with increases in mortality and hospital admissions due to respiratory and cardiovascular diseases (Dockery et al. 1993; Pope et al. 2002; WHO 2003; Beelen et al. 2008; Brunekreef and Holgate 2002). The Clean Air for Europe program have estimated that PM_2.5 is associated with more than 348,000 premature deaths and to loss of nearly 3.7 Ma of life in Europe in 2000 (Watkiss et al. 2005). The other study, based on measured PM concentrations, estimated that PM_2.5 air pollution causes 492,000 premature deaths and loss of 4.9 Ma of life lost in Europe in 2005 (de Leeuw and Horàlek 2009).

In Finland, an integrated assessment model for primary PM_2.5 air pollution has been developed to estimate emissions, dispersion, and adverse health effects of primary PM_2.5 air pollution (Kukkonen et al. 2008; Kauhaniemi et al. 2008). In a previous assessment of the present authors, dispersion and exposure of primary PM2.5 were evaluated in Finland using regional scale dispersion model system SILAM (Sofiev et al. 2006) on the spatial resolution of approximately 5 km (Tainio et al. 2009a) and the adverse health effects were estimated for the whole population (Tainio et al. 2010). The primary PM_2.5 emissions from Finland were concluded to cause annually approximately 179 premature deaths in Finland (Tainio et al. 2010). Over half of the deaths were associated with domestic wood combustion and traffic emissions from Finland.

One of the main conclusions of Tainio et al. (2009a) was that the emission–exposure calculations for emission sources with low emission height, such as traffic and domestic wood combustion, need to be refined. Traffic emissions occur to considerable extent in densely populated urban environments, while the wood combustion emissions occur both in rural and in urban environments. This has been illustrated in a study where the vicinity of primary PM_2.5 emissions and population was evaluated over Finland (Tainio et al. 2009b).

Karvosenoja et al. (2010) compared population exposure effects of traffic and domestic wood combustion using two different source-receptor dispersion matrix set-ups: (1) identical to this study on the horizontal grid resolution of 1 km, and (2) based on the SILAM atmospheric model identical to Tainio et al. (2009a) on the resolution of 10 km. The study pointed out the benefits of using finer spatial resolution for low-altitude emissions especially in the vicinity of high population densities.

Intake fraction describes the fraction of a pollutant released from source and summed over exposed individuals and divided by the emission (Bennett et al. 2002). It has been used in a substantial number of air pollution studies (e.g., Levy et al. 2002a; Greco et al. 2007). For example, Stevens et al. (2007) have used the iF concept to compare different exposure assessment methods with each other in Mexico City due to PM_2.5 emissions from mobile sources. In that study, the iFs for primary PM_2.5 varied from 23 to 120 per million depending on the method used.

Different subpopulations have different exposure levels. This could be a result of living in different locations (e.g., families with children might live in different areas than the elderly) or due to differences in time activity (e.g., adults may spend more time in traffic than other subpopulations). In previous studies, Levy et al. (2002b) have estimated that people with higher education have lower premature mortality due to air pollution. Greene and Morris (2006) evaluated the health risks of PM_2.5 for different subpopulations and concluded that females and adults had highest lifetime risks for developing new lung cancer cases due to PM_2.5. Beelen et al. (2008) concluded that population with lower education had higher adverse health effects due to traffic air pollution.

In this study, we estimated emission–exposure relationships for primary PM_2.5 emissions due to domestic wood combustion and traffic in Finland in 2000. Although, for simplicity, concepts such as “emission–exposure relationship” are used throughout this article, strictly speaking, the term “exposure” refers here to population-weighted outdoor concentrations (of PM_2.5). The emissions and atmospheric dispersion were estimated on a finer spatial resolution of 1 km. The emission–exposure relationships for both domestic wood combustion and traffic were evaluated for two emission source subcategories. We also estimated emission– exposure relationships and the population-weighted outdoor concentrations for subpopulations with different gender, age, and educational status.

Material and methods

Emissions

Primary PM_2.5 emissions of domestic wood combustion and road traffic were calculated with the Finnish Regional Emission Scenario (FRES) model of the Finnish environment institute (Karvosenoja et al. 2008). The emissions of primary PM_2.5 are calculated by combining (1) activity levels, (2) emission factors, and (3) emission control technologies (for detailed description of these parameters, see Karvosenoja et al. 2008). The basic spatial and temporal domains of the model are Finland and 1 year, respectively, which are then disaggregated to 1 km and 1 h resolutions, respectively. The 1 km and hour resolutions were used in the dispersion computations (see below). The national level emission calculation and data sources for traffic exhaust and domestic wood combustion are presented in Karvosenoja et al. (2008), and for traffic non-exhaust in Kupiainen et al. (2007). The basis for the spatial and temporal allocations of emissions are presented in detail in Karvosenoja et al. (2010). The description of emission source categories and subcategories are presented in Tables 1 and 2.

Domestic wood combustion and traffic emissions are calculated separately for various wood combustion appliance and vehicle types, direct wear products of road, tires and brakes, and particle resuspension from road surfaces. In this paper emissions are presented aggregated to four categories based on their different temporal variations: (1) wood heating in residential buildings, (2) wood heating in recreational buildings (3) traffic exhaust and wear, and (4) resuspension from road surfaces. The temporal disaggregation in time was carried out using typical temporal patterns based on data from wood use and traffic surveys. Wood combustion emissions from residential buildings are highest during winter, whereas for recreational buildings the maximum occurs during non-winter months. Hourly maximum of wood combustion emissions occurs at early evening hours. For exhaust traffic emissions, monthly variation is less pronounced, and for resuspension emissions, the maximum occurs during winter and spring months (Kupiainen 2007). Traffic emissions have their maximum during the morning and afternoon rush hours, and also the lower emissions during weekends are taken into account.

Atmospheric dispersion

The dispersion of primary PM_2.5 was evaluated with the Urban Dispersion Modeling system (UDM-FMI) developed at the Finnish Meteorological Institute (Karppinen et al. 2000a, b). It includes a multiple source Gaussian plume model and a meteorological pre-processor MPP-FMI (Karppinen et al. 1997, 1998). For the selected calculation grid, the system can be used to compute an hourly time series of concentrations. The parameters used in dispersion calculations are presented in Table 2.

The MPP-FMI model uses synoptic weather observations and meteorological soundings, and its output consists of hourly time series of meteorological parameters, including atmospheric turbulence parameters and the mixing height of the atmospheric boundary layer. In order to have input data representative for different areas in Finland, meteorological observations from ten synoptic weather stations were used in the calculations. The stations selected were Helsinki, Turku, Lappeenranta, Jokioinen, Kankaanpää, Jyväskylä, Kuopio, Oulu, Rovaniemi, and Sodankylä. The corresponding sounding observations were taken from the Jokioinen and Sodankylä observatories.

In this study, the UDM-FMI model was used to calculate source-receptor matrices. These matrices describe the dispersion of inert particles from a single emission source of unit strength, and they are combined with the actual emissions to compute the PM_2.5 concentrations. The size of the unit source was set at 1×1 km2, corresponding to the resolution of the emission inventory. The source was assumed to be in the center of a square calculation grid of size 40×40 and 20×20 km² for domestic wood combustion and traffic emissions, respectively. The size of the domain was chosen so that it covers most of the substantially elevated concentrations from both categories of sources. Detailed numerical computations show that the computational domain includes at least 85% and 90–95% of the pollutant mass for wood combustion and traffic emissions, respectively. The diurnal and seasonal variations of the emissions were taken into account, as described above. Removal of particles through deposition was neglected, due to the relatively small particle size and the short time scales considered.

The source-receptor matrices were computed for four different emission source subcategories (Table 1). The differences between the various emission categories were the different seasonal and diurnal variation of the emissions, and varying source heights (Table 2). The emission heights including the estimated initial plume rises were assumed to be 7.5 and 2.0 m for domestic wood combustion and traffic, respectively. The concentrations were calculated at the height of 2.0 m. The dispersion matrices were calculated from hourly concentrations as annual averages for the year 2000, for the ten stations mentioned above, and for the above mentioned four emission categories, i.e., altogether 40 matrices were computed.

There are differences between matrices especially between different stations, due to varying meteorological conditions, e.g., prevailing wind direction and speed. There are also differences between the matrices for traffic and wood combustion emissions, due to varying release heights. Traffic emissions have lower release height, and therefore concentrations in the center of the calculation area are higher than for domestic wood combustion. Different seasonal variation between the emission subcategories also results in some differences in dispersion patterns. Some examples of sourcereceptor matrices are given in Figs. 1 and 2 to illustrate the different dilution of the emission for different cases. The effect of different meteorological conditions between stations is depicted in Fig. 1a–b. An example of differences in dispersion due to different seasonal variation of emissions is given in Fig. 2a–b for domestic wood combustion— emissions from residential buildings are highest in winter, whereas emissions from recreational buildings have their maximum during non-winter months, which results in a different dispersion pattern.

The matrices were combined with the data regarding the spatial distribution of emissions to calculate the concentration levels due to nearby emissions of wood combustion and road traffic for the whole of Finland. Each emission value was multiplied with the appropriate value from the source-receptor matrix, and the resulting concentration fields were summed up. For most cases, the sourcereceptor matrix of the nearest station was used, with two exceptions. The border between Oulu and Rovaniemi, as well as in southern Finland the border between the effect of coastal stations (Helsinki and Turku) and inland stations, were redefined in order to achieve a better compliance with the borders of the climatic zones.

Population

The population data were provided by the Statistics Finland Grid Database (http://www.stat.fi/tup/ruututietokanta/ index_en.html). The dataset contained population numbers for Finland on a resolution of 250×250 m² for 2004. The same population data were used also in Tainio et al. (2009a). In addition to population densities of the total population, the dataset contained population densities for subpopulations with different gender, age, and educational status (Table 3). The population data were combined with the concentration data so that population numbers of each grid were joined to the nearest concentration point. The joining of data was done with the ArcMap version 9.2 (http://www.esri.com/)

Emission–exposure relationship

The emission–exposure relationships for different primary PM2.5 emission source categories and subcategories were estimated using intake fraction (iF) concept (Bennett et al. 2002). The iF is the fraction of an emission of a pollutant that is inhaled by the population (Bennett et al. 2002). The intake fraction iFs were calculated with Eq. 1:

Failed to parse (Missing <code>texvc</code> executable. Please see math/README to configure.): \frac{iF_{s,i,j}=\sum(c_{s,i}*Pop_{i,j}*BR)}{Q_s}

where

Failed to parse (Missing <code>texvc</code> executable. Please see math/README to configure.): Q_s = the total emission from the emission source category or subcategory s

Failed to parse (Missing <code>texvc</code> executable. Please see math/README to configure.): c_{s,i} = the incremental outdoor concentration of PM (grams per cubic meter) in a grid cell Failed to parse (Missing <code>texvc</code> executable. Please see math/README to configure.): i (i=1..n, where Failed to parse (Missing <code>texvc</code> executable. Please see math/README to configure.): n is the total number of grid cells), caused by Failed to parse (Missing <code>texvc</code> executable. Please see math/README to configure.): Q_s

Failed to parse (Missing <code>texvc</code> executable. Please see math/README to configure.): Pop_{i,j} = the population size in the grid cell Failed to parse (Missing <code>texvc</code> executable. Please see math/README to configure.): i for subpopulation Failed to parse (Missing <code>texvc</code> executable. Please see math/README to configure.): j

Failed to parse (Missing <code>texvc</code> executable. Please see math/README to configure.): BR = is the average individual breathing rate. A nominal breathing rate of 20 m³ day⁻¹ person⁻¹ (∼0.00023 m³ s⁻¹person⁻¹) was adopted in this study. We used the same breathing rate for all subpopulations resulting differences in spatial distribution.

The total number of grid cells for the whole Finland (i) was 783,900. The population–weighted outdoor concentration of PM2.5 was used as a proxy for the population exposure while calculating the iFs. The multiplying of population numbers with PM2.5 concentrations (C×Pop in