Long-range transport of Siberian biomass burning emissions and impact on surface ozone in western North America



[1] During the summer of 2003, biomass fires burned a large area of Siberia, the largest in at least 10 years. We used the NRL Aerosol Analysis and Prediction System (NAAPS) model to forecast the transport of the smoke from these fires. Transport of these airmasses to North America was confirmed by aircraft and surface observations. The fires resulted in enhancements in summer background CO and O3 of 23–37 and 5–9 ppbv, respectively, at 10 sites in Alaska, Canada and the Pacific Northwest. From the area burned, we estimate that the Siberian fires generated 68 Tg of CO and 0.82 Tg of NOx (as N). In addition, we show that the background O3 enhancement contributed to an exceedance of the ozone air quality standard in the Pacific Northwest. These results show that regional air quality and health are linked to global processes, including climate, forest fires and long-range transport of pollutants.

1. Introduction

[2] Previous studies have identified trans-Pacific transport of atmospheric pollutants during spring [Jaffe et al., 2003a, 2003b; Fiore et al., 2002]. Occasionally these pollutants are transported downward and result in substantial concentrations at the surface [Jaffe et al., 2003a]. However we know little about how this phenomenon contributes to regional air quality. During summer, large-scale biomass burning in boreal forests can contribute to regional and hemispheric air pollution [McKeen et al., 2002; Wotawa et al., 2001].

[3] During the summer of 2003 we conduced vertical profiles near the coast of Washington state to obtain information on the background levels of CO, O3 and aerosol light scattering between 0 and 6 km above sea level (asl). The methods used in the summer of 2003, were nearly identical to our previous measurements conducted in the previous few years [Bertschi et al., 2004, and references therein]. The results of our summer 2003 aircraft measurements are described in a separate paper (I. Bertschi and D. Jaffe, Aircraft observations of plumes from Asian boreal fires during 2003, submitted to Journal of Geophysical Research, 2004, hereinafter referred to as Bertschi and Jaffe, submitted manuscript, 2004). In the present paper, we show how large biomass fires in Siberia during 2003 impacted both CO and O3 over a wide region of western North America and contributed to an exceedance of the O3 air quality standard in the Pacific Northwest.

2. Experimental

[4] We used results from the Naval Research Laboratory Aerosol Analysis and Prediction System (NAAPS, http://www.nrlmry.navy.mil/aerosol/) to identify possible episodes of long-range transport. The NAAPS model has recently been modified to incorporate real-time observations of biomass burning based on the joint Navy/NASA/NOAA Fire Locating and Modeling of Burning Emissions system (FLAMBE, http://www.nrlmry.navy.mil/flambe/) [Reid et al., 2004]. For smoke fluxes in Siberia, twice daily MODIS fire products at ∼10:30 and 14:00 local time were used, along with the methods described by Reid et al. [2004].

[5] We used O3 and CO data from 10 rural and background sites in the Northwestern U.S., British Columbia and Alaska1. We also used the GEOS-CHEM model, which is a global 3-D chemical transport model driven by assimilated meteorological data from Goddard Earth Observing System (GEOS) of the NASA Global Modeling and Assimilation Office. Tropospheric ozone-NOx-hydrocarbon-sulfur chemistry in the model includes 31 transported tracers, 120 chemical species, and over 200 reactions, including treatment of aerosols [Bey et al., 2001; Martin et al., 2003]. The simulations presented here use GEOS-CHEM model version 5.03 with a horizontal resolution of 4 degrees latitude by 5 degrees longitude and 30 vertical layers.

[6] For this work, we conducted a reference simulation using GEOS-CHEM, with a climatological inventory of biomass burned [Lobert et al., 1999; Duncan et al., 2003] with emission factors described by Staudt et al. [2003], yielding CO and NOx emissions over Asiatic Russia (40–90N; 60–180E) of 17 Tg CO and 0.21 Tg NOx (as N). A comparisons of this reference simulation to surface CO observations throughout the North Pacific for 2001 showed no significant bias [Liang et al., 2004]. We also conducted a simulation with enhanced biomass burning emissions for the 2003 fire season using data from the Global Fire Monitoring Center (GFMC, http://www.fire.uni-freiburg.de/). Using the GFMC estimate of 18.9 million hectares burned in 2003, average fuel consumption of 30 tons/hectare and emission factors of 120 g CO/kg fuel burned and 1.4 g N/kg fuel burned, we calculate CO and NOx emissions from the fires of 68 Tg and 0.82 Tg (as N), respectively, or 3.9 times the climatological emissions. While there is some uncertainty in these values, the values we have used are consistent with other reports for high latitude biomass fires [e.g., Kasischke and Bruhwiler, 2002; Cofer et al., 1998; Goode et al., 2000, and references therein].

[7] To allocate the emissions, we used the spatial distribution of monthly fires detected by MODIS/TERRA between May and September 2003 and vertically distribute them between 1.5 and 4.5 km altitude [Lavoue et al., 2000]. Because assimilated meteorological fields for 2003 were not available at the time of this work, we used meteorological fields for the spring-summer of 2001, which was climatologically similar to 2003. While this prevents us from making a direct day-by-day comparison with the measurements, it is unlikely to influence the mean summer results shown below.

3. Results

[8] Satellite images and model forecasts from NAAPS for May–August 2003 suggested substantial emissions and trans-Pacific transport of smoke from the large fires burning in Siberia at that time. Figure 1 shows that the NAAPS model predicted aerosol optical depths of 0.1–0.4 over the U.S. Pacific Northwest (PNW) and north-eastern Pacific on June 2, 2003 due to smoke from the Siberian fires. On the same day we measured substantial concentrations of CO, O3 and aerosols during our aircraft vertical profiles, in westerly flow off the coast of Washington state. This confirmed the arrival of the Siberian smoke to the North American lower troposphere. Back-trajectories, calculated with the NOAA-Hysplit model, corroborate that this airmass passed over Siberia, in the region where the fires were burning. Nonetheless, because this airmass passed over industrial as well as some desert regions during transport, it is likely that industrial pollutants and mineral dust were also present. Transport of smoke from the Siberia fires was identified on several other occasions during the summer of 2003, from both the aircraft observations and the NAAPS model (Bertschi and Jaffe, submitted manuscript, 2004).

Figure 1.

NAAPS model aerosol optical depth (AOD) for June 2, 2003, 12 GMT for the smoke component. The enhanced AOD over western North America originated from the Siberia fires located near 50°N 110°E.

[9] The 2003 Russian area burned, 18.9 million hectares, was more than twice the annual average for Russian fires between 1996 and 2003. Based on MODIS fire hotspot data, the Russian forest fires near Lake Baikal in May were the largest and most intense episodes seen in Siberia since MODIS began collecting such data in 2000. So while there is some uncertainty in these burned areas, the estimated area burned in 2003 was comparable to, or more likely larger than, the 14.4 million ha, that was reported for Asian boreal fires in 1987 [Cahoon et al., 1991] and the 17.9 million ha that was reported for all boreal regions in 1998 [Kasischke and Bruhwiler, 2002]. However it should be noted that the area burned may not directly correlate with emissions, since the fires can vary from flaming to smoldering and burn over a variety of ecosystem types.

[10] In Figure 2 we show the annual area of Russian biomass fires since 1996 (GFMC data), along with summer-mean O3 and CO concentrations from 8 sites in western North America. Data from all sites, along with a map showing the locations, is given in the supplementary electronic section. The annual areas burned, shown in Figure 2, have an uncertainty of approximately 20% (E. Kasischke, personal communication, 2003). Because of the large number of locations considered, Figure 2 shows the sites grouped by region, but the pattern is identical if the data from individual sites is considered. Two sites with shorter data records (6 years or less) were not included in Figure 2, but are included in the supplementary data. At every site examined except one, the summer of 2003 had the highest mean O3 and CO mixing ratios. At Barrow the summer of 2003 had the highest mean CO mixing ratio ever seen, but the third highest O3 value out of 31 years of observations, only 1987 and 1991 had higher mean summer O3 values. Regressions of the seasonal mean O3 and CO mixing ratios with Russian area burned are statistically significant at all sites (p = 0.01–0.10 depending on site). These regressions indicate that the area of Russian biomass burning explains between 42–86% of the interannual variations in seasonal mean CO and O3 for the sites considered.

Figure 2.

Total Russian area burned (GFMC data) and mean summer mixing ratios (June, July, August) vs year for sites in Alaska, British Columbia and Washington. (top): Burned area and Alaskan CO (Barrow, Cold Bay and Shemya). (bottom): Alaskan O3 (Denali N.P. and Barrow), British Columbia O3 (Saturna Island) and Washington O3 (Enumclaw, Mt. Rainier and North Cascades National Park).

[11] For the May–September 2003 period, the largest number of fires occurred in May (45%), centered near 54°N 115°E, followed by August (24%), centered near 64°N 152°E. Data from Rishiri Island, Japan (provided in the electronic supplement), clearly show the influence from the large fires burning in early summer. The May–June average CO and O3 concentrations at Rishiri Island in 2003 were the highest in the data record, being enhanced by 26 and 10%, respectively, over the 6-year mean. In July, the Rishiri site was largely missed by these emissions, which were now further north and west.

[12] Figure 3 shows the CO and O3 enhancements calculated by the GEOS-CHEM model for a simulation with the 2003 Russian biomass burning emissions compared to a simulation with climatological biomass burning emissions. The model simulation calculates enhancements in CO of 20–35 ppbv for sites in Alaska, very similar to what is observed (23–37 ppbv). For O3, the model calculates enhancements of 2–6 ppbv, again similar to the observations (5–9 ppbv). The GEOS-CHEM results confirm that the Russian fires had a significant impact on CO and O3 levels in western North America, consistent with the observations.

Figure 3.

GEOS-CHEM model results showing the mean enhancements in surface concentrations for June 1–August 31 from the 2003 Russian fires, relative to a reference simulation with climatological biomass burning emissions for CO (top) and O3 (bottom). Red dots show the locations of observations given in Figure 2 and in the electronic supplement.

[13] Because O3 is an important summertime air pollutant, we wish to evaluate whether the 2003 Russian fires, had an influence on urban air quality in North America. The June 2, 2003 episode of long-range transport arrived to the Pacific Northwest at a relatively low altitude. Between May 27th and June 9th, surface monitoring sites in Washington state and British Columbia, including Tahoma Woods, Jackson Visitor's Center, Saturna Island and the North Cascades National Park, all had elevated O3, compared to the long-term average for May and June. The average enhancement at these sites for June 1–6th, 2003 was between 9 and 17 ppbv, depending on the site. While aircraft data are only available for June 2nd, the surface monitoring data suggest that an influence from the Siberian fires was present throughout this time period.

[14] On June 6th, 2003, the O3 monitoring site at Enumclaw, Washington, a rural location approximately 50 km south-southeast of Seattle, reported an 8-hour average O3 mixing ratio of 96 ppbv, which exceeds the U.S. air quality standard. While the primary cause for this high O3 was local emissions and photochemistry, combined with poor dispersion conditions, emissions from the Russian fires may have also contributed. To examine this hypothesis, we used temperature as a surrogate indicator of local O3 production [National Research Council (NRC), 1991]. Figure 4 shows the daily maximum 8-hour O3 mixing ratio for the Enumclaw site vs. temperature for late-spring-early fall data from 1996–2003. The positive correlation between O3 and temperature reflects multiple factors including increased reaction rates, increased solar insolation and decreased ventilation [NRC, 1991]. For the Enumclaw data, a statistically significant relationship is found between temperature and O3 above 20°C. Superimposed on Figure 4 are the observations from June 1–6, 2003. These data have higher O3 than indicated by the temperature relationship, and therefore suggest additional O3 from a non-local source. From the O3-temperature regression, shown in Figure 4, we estimate the Siberian fires contributed approximately 15 ppbv to the 96 ppbv observed on June 6th, based on the difference between the regression line and the June 6th observations. Using the same data, but taking a different approach, we select out only those days with a maximum temperature greater than 31°C. This yields 21 points (days) with a mean maximum temperature of 32.3±0.5°C (95% confidence interval) and a mean 8-hour O3 mixing ratio of 79±7 ppbv (95% confidence interval). So while the June 6th, 2003 maximum temperature of 32.2°C is close to the mean of this distribution, the 8-hour O3 average of 96 ppbv is a significant outlier. While these estimates of the enhancement due to long-range transport for June 6th must be considered as approximate, they are consistent with the enhancements of 9–17 ppbv seen at other sites in Washington and B.C. for the June 1–6th period.

Figure 4.

Plot of Enumclaw, Washington daily maximum 8 hour mean O3 mixing ratio vs maximum daily temperature from the Seattle-Tacoma International Airport. Data are from May–September for the years 1996 through 2003. Data from June 1–6, 2003 are shown with red points and the numbers 1–6. The regression line is statistically significant starting at 20°C.

[15] Fiore et al. [2002] used a global model to calculate the influence from long-range transport on background O3 in the U.S. They found that anthropogenic sources in Europe and Asia contribute 4–7 ppbv of O3 at surface sites in the U.S. during summer. However not included in this is the contribution from large-scale episodic boreal fires. Our results show that the contribution from Siberian fires, in some years, adds significantly to these background levels.

[16] Large-scale influence from biomass fires has been reported previously [Wotawa et al., 2001; Tanimoto et al., 2000; McKeen et al., 2002]. Our results show that Russian biomass burning emissions can contribute to elevated O3 over western North America and that the burned area helps explain interannual variations in background O3 mixing ratios. The summer of 2003 saw the largest amount of Siberian biomass burning in the recent past and this was likely a contributor to the elevated O3 seen at numerous sites in western North America. Additionally, for the first time, we show that long-range transport of these emissions contributed to an exceedance of the air quality standard. Given the likelihood of increasing boreal fires in a warmer world [Kasischke et al., 1995; Wotton et al., 2003], these results suggest a linkage between climate change, boreal forest fires, long-range transport of air pollutants and human health. Thus changes in the boreal regions could have important implications for changes in background O3 along the west coast of North America [e.g., Jaffe et al., 2003c].


[17] This work was supported by the National Science Foundation, the Office of Naval Research (Code 322, N0001404WR20188) and the IDS program from NASA ESE. The GEOS-CHEM model is managed by the Atmospheric Chemistry Modeling group at Harvard University with support from NASA.