Introduction

Besides the main combustion products, carbon dioxide and water, the uncontrolled conditions in wildfires produce large amounts of other combustion products such as polycyclic aromatic hydrocarbons (PAHs), carbon monoxide (CO), nitrogen oxides (NO_x), volatile organic compounds (VOCs), sulphur dioxide (SO₂), and fine particulate matter (PM_2.5, particles with aerodynamic diameter ≤2.5 μm) consisting of soot, organic carbon compounds, and inorganic ash. These air pollutants have half-lives of days in the atmosphere and they can be transported to hundreds of kilometres. Depending on the affected area determined by the prevailing winds and air mass movements, millions of people can be exposed to the harmful combustion products.

Ambient fine particles have been identified as the most harmful pollutant affecting public health in Europe. In the Clean Air for Europe (CAFE) Study, the central estimate presented for yearly excess mortality associated with PM_2.5 air pollution was 348,000 (EC, 2005). Source apportionment studies have shown that long-range transported air pollution forms a significant fraction of levels of fine particles. Koistinen et al. (2004) estimated that 46% of fine particle mass (PM_2.5) in the ambient air and 31% of personal exposures of the working age population in Helsinki consisted of secondary inorganic long-range transported particles. The corresponding observed mass concentrations were 4.7 and 3.3 μg/m³, respectively. Vallius et al. (2003) reported slightly higher contribution (51%) of long-range transport to PM_2.5.

Besides the substantial contribution to the long-term average levels, episodes of elevated PM caused by long-range transported particles occur regularly. Niemi et al. (2006) found 30 episodes during which the PM_2.5 levels were above 25 μg/m³ (i.e., approximately three times higher than the long-term average) in 1999–2005 in Helsinki. Elemental data allowed more careful source analysis of 20 of the episodes between 2001 and 2005. Practically, all these episodes originated from Eastern Europe, including Baltic countries, Russia, Belarus, Ukraine, and Poland and 10 episodes were either directly caused or significantly affected by biomass burning products from wild and agricultural fires. Episodes occurred typically in March–April, less frequently in January–February and in August–October. Both the highest hourly and daily PM_2.5 concentrations during episodes in Helsinki (46 and 81 μg/m³, respectively) were observed in 2002 and were associated with wildfires (Niemi et al., 2006).

Boman et al. (2003) reviewed epidemiological studies specific to biomass combustion smoke. Their PubMed search produced nine relevant studies, including five general population studies and four panel studies. Only one of the studies (Hales et al., 2000) looked at the effects on daily mortality, reporting 1% increase in risk of all-cause mortality and 4% increase for respiratory mortality per 10 μg/m³ increase in the ambient PM₁₀ level during the day before. The exposure–response relationships reported for wood smoke in studies reviewed by Boman et al. (2003) were in most cases equal or higher than those reported for the same pollutants in urban settings without influence from wood combustion. In a recent experimental study of wood smoke exposure in healthy humans (Barregard et al., 2006), this exposure appeared to elicit effects on systemic and pulmonary inflammation; the effects were probably stronger than those shown after experimental diesel exposure at similar particle mass concentrations.

The statistical power of the epidemiological studies on the public health impacts of wildfires tends to be substantially weaker than that of the studies on urban air pollution, in general, mainly due to the limited duration of the exposure periods. For example, Vedal and Dutton (2006) analysed a wildfire episode case in Denver (CO, USA), where a population of 2 million was exposed to 40–50 μg/m³ levels of wildfire smoke for 2 days in June 2002. They were not able to detect mortality effects from the observed data. However, assuming an increase of 1% in daily mortality per 10 μg/m³ for the exposures in the Denver case, with the average daily background mortality of 35 cases, the expected value of additional mortality would be less than 5 × 1% × 35=1.75 cases, which is impossible to observe from 2 days of episode data when the normal daily standard deviation (SD) of mortality is 6.7 cases. For the excess mortality to be clearly visible in such data 2 × SD, or more than 13 daily cases would be needed. This would require approximately 10 times higher toxicity for the smoke PM_2.5 particles than observed in the urban PM epidemiology studies. Studies covering several seasons may be necessary, as illustrated by a recent study in Australia showing an association between bushfires and hospital admissions, when data from three seasons in the 2000s were combined (Johnston et al., 2007).

During some extreme episodes, such as the one in Indonesia in 1997, huge numbers of people have been exposed, but there has been limited information available for exposure assessment as well as scarce data on health outcomes (Brauer and Hisham-Hashim, 1998). Sastry (2002) used PM₁₀ measurements (≥210 μg/m³) and visibility data to identify high air pollution days and observed 19%–22% increase in the next day all-cause mortality. Translated to the traditional risk ratio per 10 μg/m³ increase in PM₁₀ this corresponds approximately 1%. During the same episode, Mott et al. (2005) and Emmanuel (2000) reported significant increases in outpatient attendance for haze-related conditions. Nevertheless, in their recent review, Naeher et al. (2007) concluded that there is insufficient evidence at present to conclude that wood smoke particles are significantly less or more damaging to health than general ambient fine particles of similar size.

Similar position is taken by the World Health Organization. The recently updated Guidelines for Air Quality (WHO, 2006) are given for mass concentrations (μg/m³) based on the implicit assumption that all fine particles are equally harmful and that the guideline value depends only on the particle size distribution (defined in the Guidelines as PM_2.5 or PM₁₀). The exact value of relative risks associated with even the more general urban fine particles contains nevertheless some uncertainty. WHO (2004) reviewed time series and panel studies on fine particles (PM_2.5) for a meta-analysis of dose–response functions and found very few European studies on all-cause or cause-specific mortality. In the West Midlands Study, the relative risk for all-cause mortality was 1.0034 (0.9915, 1.0154) (Anderson et al., 2001). This compares with 1.0057 (0.9980, 1.0136) and 0.9837 (0.9677, 0.9999) from studies in the Czech Republic (Peters et al., 2000) and Erfurt (Wichmann et al., 2000), respectively.

The objectives of the current work are (i) to estimate the population exposure to fine particles from a serious transboundary wildfire episode in Finland; (ii) to estimate additional mortality associated with the episode using published exposure–response relationships of urban fine particles; and (iii) to use observed mortality data to investigate the relative risk specific to the episode particles and compare it with the relative risk estimate chosen for the risk assessment. In the current work, independent results from parallel studies on the atmospheric conditions leading to the episode, source apportionment and chemical characterization of PM, and evaluation of PM toxicity in vitro are used. These are shortly summarized before the Materials and methods section.

Atmospheric conditions of the episode

In August–September 2002, North European weather was dominated by a long period of stagnant anticyclones above the Eastern Europe and a cyclone in the Northern Atlantic. These conditions typically lead to sunny and warm weather in the Eastern Europe: average temperatures were higher and precipitation values much lower than the average values for the summer season (Chubarova et al., 2003). As a consequence of the hot and dry conditions, the summer was an extreme with fires on drained peat soils in Russia (Bannikov et al., 2003; Goldammer, 2003). Between April and October, 35,800 fires were recorded by the Russian Aerial Forest Protection Service Avialesookhrana, and almost 2 million hectares was affected by fires in Russia alone (Davidenko and Eritsov, 2003). A continuous patchwork of fires 2500 km long and for the most part 1000 km wide extends from Ural in the East to the borders of Ukraine and Poland, and Russia and the Baltic countries in the West (Figure 1).

Figure 1.

A map adjusted from the Web Fire Mapper (http://maps.geog.umd.edu) output, showing active fires (grey dots) detected from the MODIS satellite data for the period 26 August–8 September 2002. Each detection represents a 1-km² pixels identified as containing an active fire. NOTE: Figure 1 caption: “grey dots” on black and white print, red in colour.

Full figure and legend (257K)

The significantly degraded visibility and irritating smell of the smoke together with associated symptoms of eyes and airways experienced by the population lead to alarmed calls to the authorities, including the National Public Health Institute and raised the issue into the TV news, radio, and, during the consequent days, the headlines. The current analysis was started in response to these events. Different aspects of the episode have been analysed and published elsewhere, and are shortly summarized to the extent where the results are relevant for the current risk analysis.

Backward Trajectory Analysis for the Origin of the Aerosol

The trajectories shown in Figure 2 as well as the more detailed backward trajectory analyses conducted by Niemi et al. (2005) and Sillanpää et al. (2005) show that the aerosols observed in Finland during 26 August–6 September originated from the wildfire areas in Eastern Europe. In comparison to the fire maps (Figure 1), the trajectories emphasize especially the role of fires in the Baltic countries. The spread of the smoke was also clearly visible in the Terra-satellite images affecting large areas in Poland, Germany, and Denmark (http://www.sat.dundee.ac.uk; data not shown).

Figure 2.

Four-day backward trajectories (HYSPLIT model, NOAA Air Resource Laboratory) for the highest episode peaks observed in (1) Southern Sweden (4.9 at 18 O'clock), (2) Helsinki (5.9. at 20 O'clock) and (3) Kuopio (6.9. and 10 O'clock), highlighting the role emissions in the Baltic countries. Trajectory height indicated in the figure.

Full figure and legend (75K)

Observed Air Quality

Continuous fixed site-monitoring networks are operated in 13 cities in Finland (locations in Figure 3, concentration data for three of the stations in Figure 4). Daily average PM_2.5 (Helsinki) and PM₁₀ (other cities) concentrations were obtained from eight cities within the affected area. Peak concentrations, raising wide public attention, approached 200 μg/m³ in Lappeenranta and in the other cities 100 μg/m³ (hourly averages).

Figure 3.

Locations of continuous air quality monitoring stations (o), provinces (numbered), and relative exposure levels in provinces (darkness of colour only indicative; see actual exposure levels in Table 1) during the forest fire episode. PM₁₀ data were not available from Tampere, and PM_2.5 data were available only from Helsinki, Vaasa, and Oulu.

Full figure and legend (95K)

Figure 4.

Observed mortality data by provinces and ambient air quality data from three cities (PM_2.5 for Helsinki, PM₁₀ for Kuopio, and Lappeenranta; hourly and daily values).

Full figure and legend (294K)

When the worst episode day reached Kuopio area on Friday morning, 6 September 2002, emergency measurements of PM_2.5 particles were started within 2 h time at KTL. Size selective Climet optical particle counter (Climet Inc.) and pDR-1200 (Thermo Electron Inc., MA, USA) and DustTrak (TSI Inc.) optical particle monitors with PM_2.5 cyclones were placed on the roof of the KTL three-storey building. The Climet monitor classifies particles into five size classes between 0.3 and 25 μm based on their aerodynamic diameter and counts continuously the respective number concentrations. The other two monitors estimate the particle mass concentrations based on the optical aerosol properties and were logged on a 1-min basis. Besides the optical measurements, the US EPA reference method, EPA-Wins sampler with PM_2.5 impactor, was used to collect four consecutive samples for analysis of mass concentrations and chemical composition.

Source apportionment and PM toxicity

Two independent studies on the source apportionment and chemical characterization of the episode particles in Helsinki, and consequent in vitro studies on the cellular toxicity of the size-segregated high-volume PM samples are summarized to support the assumptions of the current work. Two shorter parts of the current 2-week episode period (26–29 August and 5–6 September) were analysed by Niemi et al. (2005). Biomass combustion was identified as the main cause for the elevated PM_2.5 levels, but elevated SO₂ and NH₄ levels also indicated potential impacts from emissions of industrial and other sources.

Composition of particles in Helsinki was also investigated by Sillanpää et al. (2005) using nine multiday size-segregated high-volume PM samples collected between 23 August and 23 September 2002, overlapping the current episode. Four of the samples were collected during the target period of the current study. A high correlation of PM_2.5 mass was observed with succinate, malonate, oxalate, K⁺, Ca²⁺, and monosaccharide anhydrides, and a good correlation with NH₄⁺, Br⁻, Al, Si, and P supporting the origin of the particles from wildfires.

These high-volume samples were further used in in vitro toxicological experiments. Mouse macrophage cultures were exposed to particles of four different sizes (<0.2, −1.0, −2.5, and −10 μm) to study their inflammogenic and cytotoxic activities (Jalava et al., 2006). They reported that the episode particles were associated with increased inflammogenic and cytotoxic activities per inhaled cubic metre of air due to the greatly increased particulate mass concentration in the accumulation size range (Jalava et al., 2006).

Materials and methods

Exposure Assessment

PM_2.5 or PM₁₀ data were available from eight stations on an area covering approximately 100,000 km² and 3.4 million inhabitants (Table 1). To estimate the exposure levels, the concentrations from various stations were plotted as time series together, displaying convincing similarities in the profiles. Exposures in the 11 provinces affected by the smoke were estimated using the most representative urban background monitoring data (Table 1) after subtraction of the seasonal background. Seasonal background concentrations were visually estimated using the baseline levels of hourly concentration data and subtracted from the observations using the minimum daily average value.

Table 1 - Observed daily mortality by provinces and additional PM_2.5 population exposure as estimated from the air quality monitoring data during 26 August–8 September 2002 in Finland.

Full table

Owing to the lack of PM_2.5 monitoring stations in all other cities, but Helsinki (Kallio) (and the PM_2.5 emergency measurements in Kuopio that covered the last couple of days of the full period), the PM_2.5 size fraction of the particles originated from the forest fires had to be estimated indirectly. The average PM_2.5 and PM₁₀ concentrations during the 2-week episode in 2002 in Helsinki were compared with the monthly average of those parameters at the same season 1 year after the episode. Assuming that the background levels during the episode would have been at the same level as the monthly average values in September 2003, the particle size fraction originated from the forest fires could be estimated. This analysis showed that 95% of the increase of PM₁₀ was due to the increase of PM_2.5. Comparison of the difference between the minimum and maximum levels of PM_2.5 and PM₁₀ during the episode produced similar results (data not shown). Theoretical considerations as well as measurements conducted by KTL in Kuopio also showed that the part of the episodic haze observed as PM₁₀ at the fixed monitoring stations corresponded quite well to gravimetric PM_2.5 measurements. Thus, the concentrations above the seasonal regional PM₁₀ background level were considered as PM_2.5 particles in the exposure assessment.

Moreover, PM₁₀ and PM_2.5 measurements in Helsinki metropolitan area and PM₁₀ levels in Imatra in the Southeastern part of the country followed roughly the same temporal pattern throughout the whole month (Niemi et al., 2005), indicating a significant, evenly distributed long-range transportation contribution in the PM_2.5 fraction.

Risk Assessment

The additional daily mortality caused by the episode was estimated by using air quality monitoring data in combination with exposure–response values from epidemiological studies as suggested by WHO (Schwela et al., 1999), but with the exposure–response factor of PM_2.5 instead of the value for PM₁₀. The calculation was based on the assumption that the daily mortality is increased by 1% per each increment of 10 μg/m³ (WHO, 2004, 2006). Population and daily mortality for the provinces affected by the episode were provided by Statistics Finland (Table 1, top part of Figure 4). To assess the uncertainty of our analysis, a range of mortality estimates was tested. A lower estimate of 0.5%/10 μg/m³ PM_2.5 was selected on the basis of the lowest estimates for total mortality presented in previous original studies selected into the meta-analysis of WHO (2004) and review by Pope and Dockery (2006).

Epidemiology

Short duration (2 weeks), relatively low additional exposure level (ranging from 10.3 to 26.5 μg/m³ in the various provinces; population-weighted mean 15.7 μg/m³), and limited population affected (3.4 million) made it unreasonable to expect statistically significant associations between PM exposure and mortality in the current work. Similar limitations have haunted previous studies on the effects of wildfire particles. Nevertheless, having access to a reasonably well-built and evaluated population exposure model, it was interesting to test the observations in an epidemiological model.

Poisson's regression was used to evaluate the associations between daily levels of PM and mortality during the 2-week period. The area-specific models included linear term for time trend; smooth term was not considered to avoid overadjusting the short study period. Inclusion of a dummy variable for weekdays did not affect the results. Daily temperature cycle remained in practice unchanged, afternoon temperatures being 19–22°C in the Southern Finland during the study and, consequently, was not included in the model. The pooled effect estimate was calculated as the weighted average of city-specific regression coefficients using the inverse of the squared standard errors of the regression coefficients as weights. The results were calculated for an increase of 10 μg/m³ in PM_2.5 to be comparable to previous studies.

Results

Exposure Assessment

Exposures of the general Finnish population were estimated for the 11 provinces most seriously affected by the wildfire smoke. Increase in the population exposure to PM_2.5 was estimated to be 15.7 μg/m³ for the 2-week period in August–September 2002 (Table 1).

Daily variability in the PM_2.5 and PM₁₀ levels throughout the period at the three selected sites shown in Figure 4 exhibit clear similarities in the levels. In the beginning of the 2-week period, all stations represent approximately 3 days of clearly elevated levels, in Lappeenranta PM₁₀ exceeding 100 μg/m³, PM_2.5 in Helsinki reaching 60 μg/m³, and PM₁₀ similar levels in Kuopio. Towards the end of the period, a shorter and sharper peak with small differences in timing occurs, and the minimum levels in between represent plausible values for seasonal background concentrations.

Risk Assessment

The additional mortality potentially associated with the episode was calculated to be 17 (lower and upper estimates 9 and 34, respectively) cases in 2 weeks (Table 2). Risk in each province is a combination of the size of population and the exposure level listed in Table 1. Highest contribution (3.5 cases) was associated with Uusimaa (no. 1) containing the Metropolitan area of Helsinki, due to the largest population and relatively high exposure, followed by provinces Kymenlaakso (no. 8; 2.5 cases), Pirkanmaa (no. 6, 2.4 cases), and Etelä-Savo (no. 10, 2.2 cases). The result for Pirkanmaa is related to high population, whereas the two others represent high exposures.

Table 2 - Estimated excess daily mortality by provinces caused by the wildfire episode during 26 August–8 September 2002 in Finland.

Full table

Epidemiology

Relative risk for daily mortality (RR) varied between 0.8% and 2.1% per additional 10 μg/m³ of PM_2.5 exposure in the various regression model calculated (Table 3). Lowest values were observed for same day concentration (lag 0) and highest values for lags 2–3 and for 5-day average concentration. All regression coefficients were statistically nonsignificant as could be expected already from the theoretical considerations. Nevertheless, the consistency of the results over the various lags and averaging times investigated suggests an association.

Table 3 - Poisson's estimates for relative daily mortality presented as % change per 10 μg/m³ of PM.

Full table

Discussion

Statistical Power

The epidemiological attempt to estimate the relative mortality risk produced results that were in line with the other studies, but far from statistical significance due to the limited duration of the episode. A much longer single episode or a pooled data from multiple episodes would be needed to detect effects of PM from wildfires more clearly. Standard deviation of the daily mortality in the current study population (3.4 million) was 7.25 for the 2-week period and the estimated daily mortality (central estimate) was 1.2 cases per day (17 cases/14 days). Crude calculation for the needed data to reduce the standard error of the additional daily mean mortality to below 1 cases (i.e., to smaller than the estimated potential mortality of 1.2 cases per day) would require 7.25² or 52 days of similar episodic conditions. Therefore, it seems reasonable to assume that by pooling the episodes from 5 to 10 years should yield sufficient data to obtain statistically significant results on daily mortality.

Intermediate Time Scale Effects for Multiday Episodes

Recent scientific literature has suggested that the traditional split used in epidemiology to chronic (one to several years) and daily effects may not fully characterize the temporal variability in toxic effects of PM. The exposure–response values presented for chronic exposures are one order of magnitude higher than those presented for daily exposures, and it is quite reasonable to assume, that there is a more continuous transition between short- and long-term effects than depicted by these two points (Pope and Dockery, 2006).

If this is the case, a substantially increased PM_2.5 concentration for consecutive days during a period of 2 weeks would be likely to lead to some cumulative effects. According to previous comparisons of daily and intermediate time scale (5–20 days) mortality estimates for PM (Pope and Dockery, 2006), it can be estimated that these cumulative effects could double the daily mortality estimate.

Extent of the Wildfire Problem

The current analysis covers only 2 weeks, but the data shown by Niemi et al. (2005) indicated that substantially longer period was affected by the wildfires. Although the levels were not quite as high throughout the 6 weeks, the estimated mortality would probably more than double if the whole period was accounted for. Niemi et al. (2006) found that in 2001–2005, there have been on the average two wildfire episodes per year affecting Helsinki. In contrast, in August 2002, the devastating air quality and population health problems in Moscow got wide attention and, for example, air traffic was stopped because of the degraded visibility caused by the fires. Poland, Germany, Denmark, and Sweden were affected for variable periods of time as indicated by satellite images (data not shown). The global health risks of the fires are by no means captured by the current analysis. Moreover, it has also been suggested that the climate change may increase the occurrence and extent of forest fires in the future (Westerling et al., 2006).

Conclusion

The current work investigated the population exposures to smoke originating from East European wildfires that affected the air quality in Finland in 2002. During the studied episode, 3.4 million people in Finland (total population 5.3 million) were exposed to elevated PM_2.5 levels. The mean exposure level for 2 weeks was 15.4 μg/m³ above the seasonal background.

Population mortality risk was estimated based on the exposures, background mortality, and the relative risks estimated for general urban fine particles. Central estimate for the associated additional mortality was 17 cases (9–34 cases as lower and upper estimates, respectively).

Observed increase in daily all-cause mortality in the time series analysis was not statistically significant due to the short duration of the episode, but was in line with the effect estimates reported elsewhere for urban PM air pollution in general.

References

1. Anderson H.R., Bremner S.A., Atkinson R.W., Harrison R.M., and Walters S. Particulate matter and daily mortality and hospital admissions in the west midlands conurbation of the United Kingdom: associations with fine and coarse particles, black smoke and sulphate. Occup Environ Med 2001: 58(8): 504–510. | Article | PubMed | ChemPort |
2. Bannikov M.V., Umarova A.B., and Butylkina M.A. Fires on drained peat soils of Russia — causes and effects. Int Forest Fire N 2003: 28: 29–32.
3. Barregard L., Sällsten G., Gustafson P., Andersson L., Johansson L., Basu S., and Stigendal L. Experimental exposure to wood-smoke particles in healthy humans: effects on markers of inflammation, coagulation, and lipid peroxidation. Inhal Toxicol 2006: 18(11): 845–853. | Article | PubMed | ChemPort |
4. Boman B.C., Forsberg A.B., and Järvholm B.G. Adverse health effects from ambient air pollution in relation to residential wood combustion in modern society. Scand J Work Environ Health 2003: 29(4): 251–260. | PubMed | ChemPort |
5. Brauer M., and Hisham-Hashim J. Fires in Indonesia: crisis and reaction. Environ Sci Technol 1998: 32: 404A–407A. | Article | ChemPort |
6. Chubarova N., Prilepsky N., Riebau A., Shoettle A., Musselman R., Uliumdzhieva N., Yurova A., Potter B., Rublev A., Zhmylev P., and Karpuhina E. A natural fire experiment in Central Russia: meteorology, radiative and optical properties of atmosphere and resulting effects in sub-boreal forest plants. 5th Symposium on Fire and Forest Meteorology Joint with 2nd International Wildland Fire Ecology and Fire Management Congress, Orlando, FL, November 16–20, 2003. American Meteorological Society, Boston, MA, 2003 Available at http://www.ncrs.fs.fed.us/pubs/jrnl/2003/nc_2003_chubarova_001.pdf.
7. Davidenko E.P., and Eritsov A. The Wildland Fire Season 2002 in the Russian Federation: report of the Aerial Forest Fire Service Avialesookhrana. Int Forest Fire N 2003: 28: 15–17.
8. EC. Proposal for a directive of the European parliament and of the council on ambient air quality and cleaner air for Europe. Brussels, September 21, 2005, COM (2005) 447 final, 2005/0183 (COD). Available at http://ec.europa.eu/environment/air/cafe/pdf/com_2005_447_en.pdf, accessed on 14 June 2007.
9. Emmanuel S. Impact to lung health of haze from forest fires: the Singapore experience. Respirology 2000: 5(2): 175–182. | Article | PubMed | ChemPort |
10. Goldammer J.G. The Wildland Fire Season 2002 in the Russian Federation: an assessment by the Global Fire Monitoring Center (GFMC). Int Forest Fire N 2003: 28: 2–14.
11. Hales S., Salmond C., Town G., Kjellström T., and Woodward A. Daily mortality in relation to weather and air pollution in Christchurch, New Zealand. Aust Nz J Publ Heal 2000: 24(1): 89–91. | Article | ChemPort |
12. Jalava P., Salonen R.O., Hälinen A., Penttinen P., Pennanen A., Sillanpää M., Sandell E., Hillamo R., and Hirvonen M.-R. In vitro inflammatory and cytotoxic effects of size-segregated particulate samples collected during long-range transport of wildfire smoke to Helsinki. Toxicol Appl Pharmacol 2006: 215: 341–353. | Article | PubMed | ChemPort |
13. Johnston F.H., Baile R.S., Pilotto L.S., and Hanigan I.C. Ambient biomass smoke and cardio-respiratory hospital admissions in Darwin, Australia. BMC Public Health 2007: 7: 240. | Article | PubMed |
14. Koistinen K., Edwards R., Mathys P., Ruuskanen J., Künzli N., and Jantunen M. Sources of PM_2.5 in personal exposures and residential indoor, outdoor and workplace microenvironments in EXPOLIS-Helsinki, Finland. Scand J Work Environ Health 2004: 30(Suppl 2): 36–46. | PubMed | ChemPort |
15. Mott J., Mannino D., Alverson C., Kiyu A., Hashim J., Lee T., Falter K., and Redd S. Cardiorespiratory hospitalizations associated with smoke exposure during the 1997 Southeast Asian forest fires. Int J Hyg Environ Health 2005: 208(1–2): 75–85. | Article | PubMed |
16. Naeher L.P., Brauer M., Lipsett M., Zelikoff J.T., Simpson C.D., Koenig J.Q., and Smith K.R. Woodsmoke health effects: a review. Inhal Toxicol 2007: 19: 67–106. | Article | PubMed | ChemPort |
17. Niemi J.V., Saarikoski S., Aurela M., Tervahattu H., Hillamo R., Luoto T., Aarnio P., Koskentalo T., Makkonen U., Martikainen J., Vehkamäki H., Hussein T., and Kulmala M. Pienhiukkasten kaukokulkeumaepisodit Etelä-Suomessa jaksolla 1999–2005. YTV Pääkaupunkiseudun julkaisusarja PJS B 2006: 18. ISBN 951-798-614-9. Available at http://www.ytv.fi/NR/rdonlyres/864ECC74-E383-4530-B479-12E0071
    7E3FC/0/kaukokulkeuma_netti.pdf(in Finnish; a documentation page in English and in Swedish).
18. Niemi J.V., Tervahattu H., Vehkamaki H., Martikainen J., Laakso L., Kulmala M., Aarnio P., Koskentalo T., Sillanpaa M., and Makkonen U. Characterization of aerosol particle episodes in Finland caused by wildfires in Eastern Europe. Atmos Chem Phys 2005: 5: 2299–2310. | ChemPort |
19. Peters A., Skorkovsky J., Kotesovec F., Brynda J., Spix C., Wichmann H.E., and Heinrich J. Associations between mortality and air pollution in Central Europe. Environ Health Perspect 2000: 108: 283–287. | Article | PubMed | ISI | ChemPort |
20. Pope C.A., and Dockery D. Health effects of fine particulate air pollution: lines that connect. J Air Waste Manag Assoc 2006: 56: 709–742. | PubMed | ChemPort |
21. Sastry N. Forest fires, air pollution, and mortality in Southeast Asia. Demography 2002: 39(1): 1–23. | Article | PubMed |
22. Schwela D.H., Goldammer J.G., Morawska L.H., and Simpson O., (Eds.). Health Guidelines for Vegetation Fire Events. World Health Organization, Geneva, 1999.
23. Sillanpää M., Saarikoski S., Hillamo R., Pennanen A., Makkonen U., Spolnik Z., Van Grieken R., Koskentalo T., and Salonen R.O. Chemical composition, mass size distribution and source analysis of long-range transported wildfire smokes in Helsinki. Sci Tot Environ 2005: 350: 119–135. | Article | ChemPort |
24. Vallius M., Lanki T., Tiittanen P., Koistinen K., Ruuskanen J., and Pekkanen J. Source apportionment of urban ambient PM_2.5 in two successive measurement campaigns in Helsinki, Finland. Atmos Environ 2003: 37(5): 615–623. | Article | ChemPort |
25. Vedal S., and Dutton S. Wildfire air pollution and daily mortality in a large urban area. Environ Res 2006: 102: 29–35. | Article | PubMed | ChemPort |
26. Westerling A.L., Hidalgo H.G., Cayan D.R., and Swetnam T.W. Warming and earlier spring increase Western U.S. forest wildfire activity. Environ Sci Technol 2006: 313: 940–943. | ChemPort |
27. WHO. Meta-analysis of time-series studies and panel studies of particulate matter (PM) and ozone (O₃). Task Group Report, 80pp. World Health Organization, Copenhagen, Denmark, 2004. Available at http://www.euro.who.int/document/e82792.pdf, accessed on 5 February 2008.
28. WHO. World Health Organisation Air Quality Guidelines, Global Update 2005. 484pp. World Health Organization, Copenhagen, 2006 Available at http://www.euro.who.int/Document/E90038.pdf, accessed on 14 June 2007.
29. Wichmann H.-E., Spix C., Tuch T., Wölke G., Peters A., Heinrich J., Kreyling W.G., and Heyder J. Daily mortality and fine and ultrafine particles in Erfurt, Germany. Part I: Role of particle number and particle mass. Health Effects Institute Research Reports, no. 098, 104pp. 2000. Available at http://pubs.healtheffects.org/getfile.php?u=189.

Acknowledgements

We thank the Helsinki Metropolitan Area Council (YTV) and the environmental offices of the cities of Kuopio and Lappeenranta for providing us with the air quality monitoring data, the Finnish Meteorological Institute for air quality data from the other cities and IC analysis of the PM_2.5 filters, and Juhani Tarhanen from the University of Kuopio for data on VOC concentrations. Diane Davies and Minnie Wong from the Department of Geography, University of Maryland provided numerical MODIS fire mapping data combining the West and East European Regions. We are thankful also for Peter Wåhlin from the Danish National Environmental Research Institute (NERI) for providing information on the assessment of the episode in Denmark. This work has been supported by intramural funding in KTL.