A subsiding regional forest fire aerosol layer at Whistler , BC : implications for interpretation of mountaintop chemistry observations

Introduction Conclusions References


Introduction
Biomass burning plays an important role in the climate system and is the second largest source of anthropogenic aerosols (IPCC, 2007a).Aerosols produced from the burning of forests, grasslands and crops scatter and absorb solar radiation (direct effect) while also influencing cloud processes by acting as cloud condensation nuclei (indirect effect).Biomass burning also is a source of air pollution and may seriously degrade visibility regionally (e.g.Pahlow et al., 2005).Regional forest fires were a dominant feature of the summers of 2008 and 2009 in western North America and are predicted to increase in frequency and severity under global warming scenarios (IPCC, 2007b) Physical, chemical and optical properties of biomass burning plumes have been studied extensively in a variety of global settings.This work is exhaustively reviewed by Reid et al. (2005 a, b).In summary, approximately 80-90% of smoke aerosol volume is typically in the accumulation mode (particle diameter<1 µm) with smoke particles primarily composed of organic carbon (50-60%) and black carbon (5-10%).However, the properties of smoke vary between fires, and are dependent on such variables as fuel type and moisture, combustion phase, and wind conditions.Furthermore, the physical, chemical, and optical properties of biomass-burning aerosols can change rapidly as they disperse.Typically, aged smoke particles are larger and more spherical than their fresh counterparts (Reid et al., 2005b).For example, aged particles typically have mean diameters ranging from 0.12-0.23 µm, with a mean around 0.18 µm, whereas fresh particles typically have a mean diameter of 0.12 µm.
As with other sources of transported aerosol, such as volcanic and desert aerosols, smoke plumes are subject to meteorological processes that tend to concentrate plumes into layer structures (McKendry and Lundgren, 2000).Previous studies in southwestern Canada indicate that crustal dust plumes from both Asia and Sahara are subject to large scale subsidence (McKendry et al., 2007(McKendry et al., , 2008) ) and mountain wave activity (Hacker et al., 2001), whilst regional smoke plumes from California are trapped by mid-tropospheric inversions (McKendry et al., 2010) .
Development of pollutant layers and their vertical movement has important implications for the interpretation of chemistry data from mountaintop observatories.At such stations, which are frequently assumed to be in the free troposphere (FT), considerable effort is directed at identifying boundary layer (BL) influences.This is often based on the interpretation of diurnal variability in various entities deemed to be BL tracers such as water vapour, or short-lived chemical species (e.g.Obrist et al., 2008;Weiss-Penzias et al., 2006;Baltensperger et al., 1997).Herein, a multi-day smoke episode at Whistler, BC is documented on the basis of observations of mountaintop chemistry at Whistler Mountain, novel deployment of a mountainside mini-lidar (Vaisala CL31 Ceilometer) and ancilliary remote sensing data (Lidar and sunphotometer) from Vancouver and Saturna Island.Primary objectives of this combined meteorological and chemical analysis are to: -Show the critical importance of vertical transport processes (subsidence) in interpretation of mountain-top chemistry observations.
-Characterize the physico-chemical and optical characteristics of regional scale plumes in southwestern BC using lidar and single particle mass spectrometry.
-Demonstrate the utility of remote vertical profiling methodologies to the interpretation of mountaintop chemistry observations.This study is part of a wider project designed to better understand atmospheric aerosol in this region, to better understand trans-Pacific transport of pollution from Asia to Western Canada and to improve understanding of the chemical properties of aerosols in the lower FT.

Ceilometer
The Vaisala CL31 Ceilometer is described in detail in M ünkel et al. (2007).Briefly, the eye-safe instrument has a 5 to 10 m resolution, 0-7500 m range and produces a light source of wavelength 910 nm from a laser diode.The CL31 uses a single-lens design in which the central portion of the lens is used for collimating the outgoing laser beam, while the outer part of the lens is used for focusing the backscattered light 20306 Introduction

Conclusions References
Tables Figures

Back Close
Full onto the receiver.For this study, the ceilometer was operated at the maximum vertical resolution of 5 m with a measurement interval of 3 s and a reporting interval of 60 s.This temporal resolution was deemed sufficient for analyzing the development of BL structure and lower FT features.
Although originally designed to determine cloud heights, in recent years ceilometers have been used for determining BL structure by taking the backscatter intensity of the returned laser light as a proxy for ground-sourced aerosol concentrations (Rsnen et al., 2000;Emeis et al., 2004;Zephoris et al., 2005;van der Kamp et al., 2008;McKendry et al., 2009McKendry et al., , 2010)).Previous studies in the Vancouver Region (see Figure 1) have demonstrated the capability of the CL31 for identifying elevated pollutant layers and smoke plumes (McKendry et al., 2009).For this study the ceilometer was deployed at an elevation of 810 m (a.m.s.l.) on the lower slopes of Whistler Mountain (at the Timing Flats).The ceilometer was installed and operated by Environment Canada as part of a data collection effort related to meteorological services for the 2010 Olympic Winter Games.

Lidar
Additional regional information on the smoke event was derived from 1064/532 nm Lidar observations from the University of British Columbia (part of the Canadian Operational Research Lidar Network (CORALNet) -http://www.coralnet.ca/).The lidar systems deployed for CORALNet use identical transmitter and receiver assemblies and where necessary are housed in a modified cargo trailer.The trailer modifications include a roof hatch assembly, basic meteorological tower, radar interlock system, climate control system and leveling stabilizers.The unit can be operated via an internet link and requires an external power source.A precipitation sensor is used to operate the roof hatch and three pan/tilt webcams capture sky conditions and monitor the lidar system's condition.A remote control interface is used to control all vital components of the system, including the ability to provide hard resets of the laser electronics.Introduction

Conclusions References
Tables Figures

Back Close
Full The laser is the new Continuum Inlite III (small footprint) laser operating at 1064/532 nm with a pulse repetition rate of 10 Hz.The energy output is approximately 150 mJ @ 532 nm and 130 mJ @ 1064 nm.The system emits two wavelengths and measures a return signal at three channels (1064 nm, 532 both polarizations).

Single Particle Mass Spectrometer (SPMS)
An ion trap based single particle MS was built for deployment at the top of Whistler Peak.This instrument follows broadly the design principles of the UT-ATOFMS described by Su et al. (2004).As in that instrument, it incorporates a) an aerosol lens as a vacuum interface, particle concentrator and focusing device, b) dual aerosol scattering devices operating at 532 nm for aerosol sizing and mass spectrometer triggering and c) a pulsed dual polarity mass spectrometer incorporating an ablation/ionization laser operating at 266 nm for ionization of the particles.While the details of the instrument are discussed in a separate publication (Campuzano-Jost and Bertram, in preparation), the main differences compared to the ATOFMS relevant to this study are briefly listed here: -Aerosol sizing module is fully independent of mass spectrometer operation, allowing the acquisition of full particle distributions to compare with standard aerosol instrumentation (Simpson et al., 2009).
-A real-time triggering solution for the ionization laser that ensures consistent beam quality and hence greatly reduces shot to shot fluctuations - (Hanna et al., 2009).
-An ionization region about four times larger than in the Su et al. (2009) design, which should improve MS collection efficiency in some cases.
-A 3-D ion trap for mass analysis.This results in a smaller footprint MS module with a robust mass calibration, high sensitivity, and reduced space charge effects.Introduction

Conclusions References
Tables Figures

Back Close
Full It also allows for tandem mass spectrometry (Simpson et al., 2009), which was not used in this study.
While the full mass range of the ion trap MS is roughly 10-200 m/z, for the data acquired for this paper only a reduced mass range of 10-100 m/z was used.This limits somewhat the analysis of more complex smoke markers.

Whistler mountaintop chemistry
Measurements of particles and trace gases are made at a high elevation site in Whistler, BC, approximately 100 km north of Vancouver (Fig. 1).The site is located at the top of a ski resort at Whistler Mountain, at a height of 2182 m (a.m.s.l.) and is operated by Environment Canada.There are no continuous combustion sources at the peak and occasional influences from maintenance vehicles operating on the mountain have been identified and removed from the data set.Condensation nuclei (CN) are observed using a TSI 3025 Ultrafine Condensation Particle Counter, which measures number concentration of particles larger than approximately 3 nm diameter.The TSI 3025 uses butanol as a condensing fluid to enlarge the particles to a size that can be detected by an optical sensor.Particle size distributions were measured using a MSP Corporation Wide Range Aerosol Spectrometer, which combines a differential mobility analyzer and an optical particle counter to measure from about 0.01 µm to 20 µm diameter.Here, we use only the number concentrations of particles greater than about 0.4 µm diameter.

Saturna Island sunphotometer
AEROCAN CIMELs (AEROCAN is the Canadian sub-network of AERONET) have been important in tracing the transport and characteristics of aerosol layers over British Columbia and, in general, over AEROCAN/AERONET sites across North America (e.g.Thulasiraman et al., 2002).The AEROCAN site on Saturna Island site (Figure 1) is approximately 50 km southwest of Vancouver.The CIMEL sunphotometer instruments Introduction

Conclusions References
Tables Figures

Back Close
Full acquire solar radiances which are transformed into three processing levels of Aerosol Optical Thickness (AOT) (1.0 -non-cloud screened, 1.5 -cloud screened and 2.0cloud screened and quality assured) across eight spectral channels (340, 380, 440, 500, 670, 870, 1020 and 1640 nm).In a separate operational mode, almucanter sky radiances are collected across four channels (440, 670, 870, and 1020 nm) at a nominal sampling resolution which is about 1/20 of the nominal AOT sampling resolution (an hour versus 3 min).The sky radiances, along with AOT estimates at the same four channels, are used to perform inversions for particle size distribution and refractive index (Dubovik et al., 2000).

Ancilliary data
To complement aerosol observations from the summit of Whistler Mountain, hourly PM 2.5 data were obtained from the BC Ministry of Environment's air quality station in the Whistler Valley (Whistler Meadow Park, 641 m a.m.s.l.).Weather maps for the relevant time periods were created from NCEP pressure level reanalysis datasets, available from NOAA's Earth System Research Laboratory (http://www.esrl.noaa.gov/psd/data/gridded/reanalysis).

Background meteorology and fire history
During summer 2009, persistent large-scale weather patterns in North America were responsible for the unusually warm and dry season experienced in southern British Columbia.For most of the late spring and summer a high pressure ridge persisted aloft over the region, deflecting Pacific frontal systems to the north.Precipitation for the June-August period was well below average: 43% of normal for Vancouver and 41% of normal for Whistler (Environment Canada statistics).Record high temperatures were Introduction

Conclusions References
Tables Figures

Back Close
Full observed in the region in late July, including 34.4 • C at Vancouver International Airport (an all-time record high).A significant consequence of this weather pattern was the very active forest fire season that occurred in BC and the northwestern US.
In southwestern British Columbia, in addition to several minor fires, the largest fires (see Fig. 1) were the Kelly Creek Fire near Clinton (reported at 13 572 ha on 28 August) and the Lava Canyon Fire near Williams Lake (reported at 33 815 ha on 26 August).In addition to numerous large fires east of the Coast Mountains, two small fires were reported on Blackcomb Mountain within a few kilometres of observing sites on Whistler Mountain.Thus, degraded air quality from smoke was a common occurrence throughout the summer in southern BC.At times, easterly to northerly wind flows aloft advected smoke through the Whistler area and into Metro Vancouver.
Synoptic weather patterns during the final three days of August resulted in advection of forest fire smoke from southern British Columbia fires (including Kelly Creek, Notch Hill and Lava Creek) westward to Whistler Mountain.This westward transport was associated with a cyclonic circulation situated over the northwestern United States that was evident in both 500 hPa (not shown here) and 700 hPa patterns (Fig. 2) on 29 and 30 August.On 31 August, the low continued inland as low pressure in the northeast Pacific approached the coast, changing the flow over Whistler to southerly.
Between the inland and offshore cyclonic circulations, a ridge of high pressure was apparent at both 500 hPa and 700 hPa .On 30 August, this ridge was oriented southwest-northeast directly over Whistler (Fig. 2b).A maximum in dynamic subsidence (positive ω, expressed in Pa s −1 ) is found downstream of the ridge axis (Figure 3a) and centered over southwestern BC.Ceilometer observations of the smoke plume (Fig. 5b) show a corresponding subsidence rate of ∼1.7 cm s −1 at Whistler over the morning period of 30 August.At the surface, the synoptic scale forcing was rather weak, with a high pressure ridge just offshore of British Columbia (Fig. 3b).This synoptic pattern explains both the timing of the smoke plume in the Whistler area, as well as the vertical motion of the plume that was observed by the Whistler ceilometer.On 29 August, east-northeast winds aloft carried smoke from the Introduction

Conclusions References
Tables Figures

Back Close
Full Clinton-area fires to the Whistler area.On the 30th, easterly winds continued to advect the smoke plume westward as dynamic subsidence forced the plume to descend at the same time.On 31 August, the flow aloft turned southerly and the smoke gradually retreated to the north.Satellite derived smoke detection products from the National Environmental Satellite Data and Information Service (NESDIS found online at -http://map.ngdc.noaa.gov/website/firedetects/viewer.htm)show the regional smoke distribution in the period 29-31 August with a wide swath of smoke extending westward to Vancouver in the south and northward into central British Columbia (Fig. 4).

Lidar and ceilometer
The combination of CORALNet-UBC (Vancouver) lidar imagery, Whistler mountainside ceilometer imagery and Whistler mountaintop chemistry measurements provide a novel opportunity to examine both the regional and local scale processes associated with the subsidence of smoke layers.In Fig. 5a, CORALNet-UBC 1064 nm imagery shows a pattern broadly consistent with that observed by ceilometer at Whistler (Fig. 5b).A dense smoke layer capped below 3000 m a.s.l. was evident over Vancouver on 30 August 2009.Rates of subsidence (∼1.7 cm s −1 ) and midday BL mixing patterns are also similar to those observed by the Whistler ceilometer and suggest that the evolution of the smoke layer at Whistler was characteristic of a broad regional pattern, albeit modified by local topographic effects.Whistler mountaintop CN particle concen- gradient and clear skies allowed thermally induced mesoscale flows to dominate the wind pattern in Whistler Valley.Diurnally-reversing slope and valley winds are common in mountainous terrain under such conditions (Whiteman, 2000).Observations from surface weather stations in the area indicate that diurnal winds occurred during the three days of this study (for brevity, these observations are not presented here).At the valley floor observing location, calm winds were observed in the morning hours, followed by southerly (up-valley) winds in the afternoon, which increased to ∼5 ms with a westerly (upslope) component were dominant during the daytime.Two effects of this wind pattern were likely important to the observed impact of forest fire smoke in the lower elevations of the valley: 1) the slope winds facilitated vertical mixing that was otherwise limited by stable stratification in the lower troposphere; 2) the afternoon up-valley breeze effectively removed much of the smoke from the valley.Evidence for both of these effects can be found in the ceilometer data (Fig. 5b).Enhanced returns in the lowest 500 m around midday are likely due to PM that was mixed downward out of the main smoke plume.The much lower returns found in the lowest 800 m after 17:00 LST indicate cleaner air arriving from up-valley flow.Wind observations in the area are limited, but it is likely that, on days with a well-developed valley wind, marine air reaches Whistler from the head of Howe Sound (approximately 35 km to the south).However, pollutants can be returned to the Whistler Valley during the overnight hours by down-valley winds and ground-level PM concentrations can then increase as the shallow nocturnal boundary layer limits mixing.

Optical characteristics and physico-chemical composition of smoke
Given the regional scope of the smoke event, Saturna Island (Fig. 1) AERONET inversion data provide an appropriate characterization of the optical and physical properties of the 30 August smoke event.These values are presented in Table 1 where they 20313 Introduction

Conclusions References
Tables Figures

Back Close
Full are compared with typical North American biomass burning values as summarized by Reid et al. (2005b).Note, the final four rows describe a bimodal size distribution and show the dominance of the accumulation mode as expected for a forest fire plume.
Other optical parameters are remarkably consistent with those reported by Reid et al. (2005b) based on AERONET inversions for North American boreal forest biomass burning aerosol.
The SPMS was operating at Whistler Peak between 22 August 2009 and 6 September 2009.During this time period, it was only operated during hours when an operator could access it.Therefore the SPMS dataset is limited to daily periods of about 4-6 h between roughly 11:00 a.m. and 06:00 p.m. PDT.During the smoke period, roughly 1 million particles were sampled and sized over the three day period (∼20 h).Total particle concentrations acquired by the SPMS show a steep increase in particle concentration with the plumes' arrival.They were consistent with data acquired by the OPC operating at Whistler Peak, once the detection limit/efficiency of the SPMS was taken into account (Fig. 6, top panel), giving us confidence that the SPMS sampled a representative aerosol distribution.
Mass spectra (161 000 for the smoke period) were background corrected and integrated with unit mass resolution.They were then classified as valid aerosol spectra if at least one particular mass showed an ion current at least 4x above the standard deviation of the noise.Based on this criterion, about 75 000 (47% of acquired) valid MS spectra were included in the final analysis.
The smoke plume was characterized with a variety of well-established markers that were computed on a particle by particle basis for the fraction of valid mass spectra.In order to account for fluctuations in both ionization efficiency and detector gain, all markers were normalized to the total ion current derived from the particle.Finally, for clarity, daily averages were computed.
As for the selection of appropriate markers, according to Reid et al. (2005a)  (1,6-anhydro-β-D -glucopyranose) is a product of cellulose combustion and is well known and stable tracer of biomass burning (Qin et al., 2006 as well as references therein).Aged biomass is also often associated with elevated levels of sulphate (negative m/z 81 and 97), nitrate (negative m/z 46 and 62) and organics (Murphy et al., 2006).Finally, sodium (Na, positive m/z 23) has been sometimes identified in biomass burning plumes, although it seems to be highly dependent on the source of the biomass (Hudson et al., 2004).Figure 6b-f show the daily average ratios of these markers in the smoke plume as well as before and after it with the exception of the organics and black carbon marker.The organics marker (positive m/z 27, 29, 37, see Silva et al., 2001) was discarded due to the aluminum isobar (m/z 27) which was likely present in some dust and pollution events before and after the plume.The black carbon marker was discarded due to its poor expression throughout the sampling period.This is likely due to the overlap of other organic fragments (Murphy et al., 2006).
Figure 6 shows that the aerosol chemistry at Whistler Peak did indeed change during the 29-31 August period with large increases in the relative amounts of potassium and levoglucosan.This is consistent with previous observations of biomass plumes (Qin et al., 2006;Hudson et al., 2004).A significant increase in nitrate is also observed, as would be expected for an aerosol from a source about 100 km away.On the other hand, another aging marker, sulphate, decreases.This probably simply reflects the high levels of sulphate that were present in the background aerosol both before and after the plume.Lastly, no increase is observed for sodium (the increase of this marker around 4 September is due to an influx of Pacific air at the site).As stated before, this might be just a function of the specific composition of BC wood and soil.

Discussion and conclusions
Based on the mountaintop observations of chemical constituents alone, the natural interpretation of the 30 August fire plume would be that the smoke episode was brief.However, lidar and ceilometer data shows that smoke was present at elevations both Figures above and below the mountaintop station and persisted over multiple days.Interpretation of time series of chemical constituents at mountaintop sites therefore requires careful meteorological analysis of vertical motion in relation to layer structures as well as boundary layer growth.Lidar has been used in previous boundary layer studies in mountainous terrain, but in most cases the data have been collected for only short time periods via either aircraft or temporary ground installations (e.g.Reuten et al., 2005;de Wekker et al., 2004).This study confirms that permanent lidar installations in mountainous location can add value to other atmospheric observations from the same area and are crucial to the understanding of vertical transport processes and the analysis of chemical time series.
This study also highlights the importance of synoptic scale weather patterns on the character of air masses sampled at a mountaintop observatory.Studies using a synoptic classification scheme for the European Alps have demonstrated the effects of different circulation patterns and wind directions on aerosol and trace gas measurements at the Jungfraujoch observatory in Switzerland (e.g.Lugauer et al., 1998;Forrer et al., 2000).For the Whistler site, this case is a good example of an easterly wind flow (not common at this location) transporting regionally produced aerosol loads to the site.While the Whistler observatory is well-situated to sample air masses coming off the Pacific Ocean in prevailing westerlies, it is important to recognize times when the synoptic pattern produces trajectories with relatively long residence times over the North American continent.Thus, synoptic scale flow and the more local boundary layer influence should both be considered when characterizing the air mass sampled at a mountain observatory for a particular time period.
In addition to chemically confirming the presence of smoke at Whistler, deployment of a SPMS instrument at Whistler peak represents a significant addition to the suite of instruments at the Whistler mountaintop facility.In this example, the instrument is able to not only confirm the source of local aerosol (biomass burning) but is able to provide the detailed physico-chemical signature of a regional smoke event.When combined with other aerosol measurements in Southwestern BC (e.g.AERONET and CORALNet) the addition of this instrument to the Whistler suite of observations promises to significantly increase understanding of the provenance and chemical properties of aerosols in the lower FT in western Canada.On the basis of this and similar cases, future efforts will be directed toward permanent installation of a CORALNet lidar at Whistler to support the investigation of aerosol transport from hemispheric to local scales.Introduction

Conclusions References
Tables Figures

Back Close
Full  20322 Discussion Paper | Discussion Paper | Discussion Paper | Furthermore, calculation of FT horizontal pollutant fluxes from mountaintop chemistry data assumes that once BL influences are excluded, horizontal advection dominates.Horizontal fluxes of pollutants might therefore be reasonably estimated on the basis of horizontal wind speeds Discussion Paper | Discussion Paper | Discussion Paper | stratified by trajectory (i.e.pollutant concentration × windspeed represents a flux with units gm −2 s −1 ).
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | trations and nearby valley floor PM 2.5 observations (Fig. 5c) were negatively correlated over the 30-31 August period during which the smoke layer was observed to subside below Whistler Peak.As the layer intercepted the peak at ∼03:00 LST on 30 August, CN values spiked and then declined rapidly as the layer moved below the elevation of Whistler Peak.On both 30 and 31 August valley floor PM 2.5 concentrations reached a maximum around midday of ∼30 µg m −3 .Variations in PM concentrations evident in Fig. 5c are likely due to local scale variations in wind and stability.During this period, the weak synoptic scale pressure 20312 Discussion Paper | Discussion Paper | Discussion Paper | afternoon.Reports from several of Environment Canada's Olympic Autostation Network (OAN) sites on the slopes of Whistler Mountain also indicate diurnal flows: winds with an easterly (downslope) component occurred overnight, and winds Discussion Paper | Discussion Paper | Discussion Paper | the main tracers of biomass burning are black carbon (positive m/z 12*n), potassium (K, positive m/z 39 and 41) and levoglucosan (negative m/z 45, 59, 71).Levoglucosan Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Figure 1 :
Figure 1: Map showing places mentioned in the text.

Fig. 1 .
Fig. 1.Map showing places mentioned in the text.