Residential wood combustion, particulate matter 2.5 sampling project, Whitehorse: chapter 3

• 3.1 Meteorology
• 3.2 PM_2.5 Results
  • 3.2.1 Measured PM_2.5
  • 3.2.2 Modelled PM_2.5
• 3.3 Source Apportionment of PM_2.5
  • 3.3.1 Wood Smoke Markers
  • 3.3.2 Aethalometer Black Carbon
• 3.4 Episode Analysis

3. Results and Discussion

3.1 Meteorology

Meteorological data from the study period is shown in Figures 3 and 4(a, b, c). Temperatures remained below freezing for most of the sampling campaign, with an average of -12°C at the two sites. Average temperatures at the airport were -18.2, -16.0 and -9.9 for January, February and March, respectively, making it a slightly colder winter than the 30-year period from 1971 to 2000 (-17.7°C, -13.7°C and -6.6°C for January, February and March, respectively).

Wind roses show similar distribution of wind directions at the two sampling sites, with the winds generally following the topography of the Yukon River Valley. The Downtown site had a higher frequency of stronger winds than Riverdale, which is more sheltered by the escarpment. Downtown, having a primarily south, and to lesser extent south- southeast, wind contribution, could be impacted by emissions in Riverdale. Similarly, Riverdale, having a significant northwest and north-northwest wind contribution could be impacted by emissions from the downtown core. Whitehorse Airport shows a similar wind vector distribution to the Downtown site but with a higher predominance of winds from the south. This site, being atop the escarpment, is not as affected by the local topography.

(a)

(b)

(c)

3.2 PM_2.5 Results

3.2.1 Measured PM_2.5

All PM_2.5 and OC/EC concentration results are shown in Table 1. Several sampling events were missed or had to be discarded due to inclement weather and sampler problems. Average/median concentrations were 7.7/6.4 µg/m³ (n=18) and 10.2/7.1 µg/m³ (n=12) at the Downtown and Riverdale sites, respectively. Higher concentrations at Riverdale are likely the result of lower wind speeds and presumably closer proximity to wood smoke sources, although such a detailed emission inventory was not undertaken as part of this study. A partial PM_2.5 mass reconstruction at the Riverdale site shows that, on average (n=12), carbon made up the bulk of the PM_2.5 mass, with OC contributing 55% and EC contributing 7%. The remaining 38% would be made up of atoms associated with OC as well as inorganic species.

Overall, PM_2.5 concentrations ranged from 1.6 µg/m³ to a maximum of 31.3 µg/m³ (Riverdale on February 11). The City of Whitehorse and Yukon Territory do not have air quality standards or objectives in place for PM_2.5. The Canada-wide Standard for 24-hour average PM_2.5 is 30 µg/m³, based on the 98th percentile, averaged over 3 consecutive years, while British Columbia has established a 24-hour average PM_2.5objective of 25 µg/m³ and an annual average objective of 8 µg/m³. For comparison, average January to March PM_2.5 concentrations for a selection of interior B.C. cities and towns are shown in Table 2.

PM_2.5 results from this study are higher than typically found at the Whitehorse NAPS site for the January to March period (see Figure 1). A comparison between PM_2.5 measurements taken Downtown (filter-based) with the NAPS site (TEOM) was made for the nine sampling days where there were concurrent data available (Figure 5). The correlation is poor, with the filter-based sampler generally reading higher. It is difficult to make conclusions based on a small data set; however, the different locations are likely a factor (see Figure 2). Although both sites are located in the downtown core, the NAPS station is adjacent to the river and may sample cleaner river-channeled winds relative to the Downtown site, which was located approximately 500 m away. An additional causative factor could be that TEOMs read lower than filter-based samplers in cold winter locations due to the loss of volatile particulate matter (Allen et al., 1997; Dann et al., 2006). Coincidentally, the NAPS station will be moved to a new downtown site in 2011, one block from the Wood Street Annex.

3.2.2 Modelled PM_2.5

To estimate the 24-hour average concentrations of PM_2.5 throughout the sampling period, a simple linear regression model was used to relate 24-hour average black carbon data, measured by the aethalometer at Riverdale, to daily average PM_2.5 concentrations measured at the two sites (Figures 6a, 6b). One outlier point (where the standard residual was greater than 2σ) was removed for each regression. The regressions show a good fit, with r² values of 0.93 and very low statistically significant p values for both Downtown and Riverdale.

Assuming a consistent linear relationship between Riverdale black carbon and PM_2.5 at both sites throughout the study period and within the range of black carbon measured, daily average PM_2.5 concentrations can be predicted from the linear regression, from January 27 to March 31. Predicted values and associated 95% confidence intervals along with the observed values are shown in Figure 7. It can be seen that there are five days at Riverdale where PM_2.5 is predicted to have exceeded the value of the Canada-wide Standard, all occurring during a nine-day period in February, with a maximum predicted PM_2.5 concentration of 41.7 µg/m³ (95% CI = 33.1-47.4) on February 13. At the Downtown site, the maximum was predicted to be 25.4 µg/m³ (95% CI = 22.5-28.3), also on February 13.

3.3 Source Apportionment of PM_2.5

3.3.1 Wood Smoke Markers

All levoglucosan and ¹⁴C results are shown in Table 3. Average concentrations of levoglucosan were 453 and 754 ng/m³ at Downtown and Riverdale, respectively, but varied widely from sample to sample, as the concentration of PM_2.5 varied substantially over the course of the study. Ratios (% w/w) of levoglucosan to PM_2.5were 4.7 ± 1.6 and 6.0 ± 2.4 (avg. ± s.d.) at Downtown and Riverdale, respectively, indicating a slightly higher wood smoke contribution at Riverdale. Note that to allow for more data points in calculating the ratios, damaged filters were included, as it was assumed that the ratio between PM_2.5 and levoglucosan mass would not be affected. Correlations of levoglucosan to PM_2.5 mass at both sites are shown in Figure 8. A stronger correlation for the Riverdale site (r² = 0.96) indicates that there is less variability in source contribution here compared to the Downtown site (r² = 0.86), which is expected to have more contribution from traffic and heating from commercial buildings.

(b)

The fraction of modern carbon (f_m) results as determined by the University of Arizona Accelerator Mass Spectrometry Facility are shown in Table 3. Due to high uncertainties at low filter loadings, sample filters having less than 100 µg of carbon were not included. While there can be several sources of modern carbon particulate in addition to wood smoke, such as cooking, cigarette smoke and secondary biogenic aerosols, it can be assumed that the former two are negligible compared with wood smoke in Whitehorse. The average OC:EC ratio was 8.9 in this study, indicating that secondary organics were present (Ward et al., 2006); however, it is likely a reasonable assumption that at temperatures typical of the sampling period, the biogenic contribution to secondary aerosol formation is low. One uncertainty associated with using the ¹⁴C method for source apportionment is the need to determine the theoretical fraction of modern carbon that should be present in wood typically burned in Whitehorse. ¹⁴C levels in the atmosphere have varied over time, peaking in 1965 due to nuclear testing and declining since then (Ward et al., 2006). The fraction of modern carbon found in trees, which incorporate ¹⁴C through respiration of CO₂, therefore will be dependent on the age of the wood. In calculating the contribution of wood smoke shown (Table 3), the laboratory used annual atmospheric ¹⁴C levels over the last 130 years. The minimum and maximum % wood smoke were determined from calculations using average atmospheric values of ¹⁴C for the last 5, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, 120 and 130 years. Given that the range of age of the dead wood harvested in the southern lakes area for fuel wood in Whitehorse is 60 to 120 years (K. Price, pers. comm., 15 Jan. 2010), this calculation should provide a reasonable estimate of wood smoke contribution for this study. Wood smoke contribution ranged from 61% to 93%, with an average range of 70-84%. It is expected that in the wintertime in Whitehorse, the dominant source of wood smoke is RWC. The only other source of wood combustion is industrial/commercial wood boilers; however, these units are relatively few in number and are not expected to be a significant source of wood smoke relative to RWC, particularly in Riverdale.

The results of this study are in relatively close agreement with earlier studies in Riverdale in which chemical mass balance modelling showed that RWC constituted 90% of particulate mass, results that were confirmed with ¹⁴C analysis, showing that nearly all aerosol carbon was from modern sources (McCandless, 1984). Results are also in close agreement with the most recent emission inventory for Whitehorse, which estimated that 82% of PM_2.5 is attributable to wood combustion for heating (Senes, 2008), although this number is projected to be higher for the winter season.

Table 4 shows levoglucosan concentration and source apportionment results for some recent studies in western North America. Results from the current study are comparable to results from a study in Golden, B.C., which determined the wood burning contribution to PM_2.5 using positive matrix factorization modelling. It is expected that wood burning conditions in Golden would be similar to those in Whitehorse. In addition, both the Golden and Prince George studies used the same laboratory for levoglucosan analysis as the current study, making results directly comparable.

3.3.2 Aethalometer Black Carbon

Five-minute average aethalometer black carbon and UV minus black carbon data from Riverdale for the duration of the study are shown in Figure 9. Concentrations of black carbon ranged from 0 to 8.6 µg/m³, with an average of 0.9 µg/m³ and a median of 0.2 µg/m³. Black carbon results correlated well with the co-located filter-based EC results (r² = 0.94). It can be seen that UV-black carbon is greater than 0 much of the time, particularly during black carbon peaks, indicating that the driving factor of high black carbon concentrations is wood smoke. Diurnal and weekly concentration patterns are also indicative of wood smoke source (Figures 10 and 11). Both the diurnal black carbon and UV-black carbon show a pattern that is typical of home heating with concentrations rising in the early morning as residents start up their wood stoves, falling off in the late morning through the afternoon and rising again in the early evening. The daily black carbon shows lower concentrations on Friday and Saturday, presumably due to residents being out of the house more on these evenings. If there was a significant traffic impact at this site, one would expect Saturday and Sunday concentrations to be lower.

(b)

3.4 Episode analysis

Meteorological data from the five days having the highest PM_2.5 as predicted by linear regression modelling with black carbon data (Section 3.2.2) at Riverdale are summarized in Table 5. These days are characterized by having average to well-below-average daily average temperatures and low daily average wind speeds. The mixing heights and ventilation index, calculated using data generated from Environment Canada’s GEM-Regional forecast model at the time of daily maximum temperature, are also low on these five days, creating ideal conditions for accumulation of wood smoke in the Whitehorse area (note: Ventilation Index values from 0 to 33 are categorized as “Poor,” 34 to 54 as “Fair” and 55 to 100 as “Good” ventilation). RWC contribution at Riverdale determined on two of the days is higher than average, indicating again that this source can be an important factor in causing higher PM_2.5 concentrations.

Over the course of this study, there were 14 days having similar meteorological conditions, which appear to be conducive to elevated PM_2.5 (daily average temperature at YXY <10°C, daily average wind speed at YXY <1.5 m/s and poor ventilation at YXY) Riverdale black carbon concentrations on these 14 days ranged from a minimum of 0.73 µg/m³ to a maximum of 3.11 µg/m³, corresponding to predicted PM_2.5 concentrations of 10.3 µg/m³ (95% CI = 7.0-10.8) to 41.7 µg/m³ (95% CI = 33.1-47.4) at Riverdale and 6.9 µg/m³ (95% CI = 6.0-7.8) to 25.4 µg/m³ (95% CI = 22.5-28.3) at Downtown. For the months of January, February and March, there were 24 days where these meteorological conditions were met, compared with 12 days over the same period in 2008 and 7 days over the same period in 2010.

Figures 12a, b, c, d and e show the surface weather maps over northwestern Canada and the temperature soundings from Whitehorse Airport (YXY) at 4 p.m. local time for the five days listed in Table 5. Soundings show the existence of surface-level temperature inversions on all five days, leading to the stable conditions observed that would further explain the high predicted PM_2.5 values. Observations from February 11, 12 and 13 show the presence of a strong stable ridge of cold arctic air and a particularly strong surface- level inversion. February 9 and 17 show a different synoptic pattern; however, strong winds from the southwest and west at 500 mb height on these days could be causing subsidence as air flows over the coastal mountain range, which would lead the observed surface-level inversions, and stable conditions.

Figure 12 (a, b, c, d and e): Environment Canada Surface Weather Maps and Temperature Sounding Data

(a) February 9 at 4 p.m.

(b) February 11 at 4 p.m.

(c) February 12 at 4 p.m.

(d) February 13 at 4 p.m.

(e) February 17 at 4 p.m.