Prescribed burning of logging slash in the boreal forest of Finland: emissions and effects on meteorological quantities and soil properties

Atmos. Chem. Phys., 14, 4473–4502, 2014 www.atmos-chem-phys.net/14/4473/2014/ doi:10.5194/acp-14-4473-2014 © Author(s) 2014. CC Attribution 3.0 License. Atmospheric Chemistry and Physics Open Access Prescribed burning of logging slash in the boreal forest of Finland: emissions and effects on meteorological quantities and soil properties A. Virkkula1,2, J. Levula 1,3, T. Pohja 3, P.P. Aalto 1, P. Keronen 1, S. Schobesberger 1, C. B. Clements 4, L. Pirjola 5, A.-J. Kieloaho 1, L. Kulmala 6, H. Aaltonen 2,7, J. Patokoski1, J. Pumpanen 7, J. Rinne 1, T. Ruuskanen1, M. Pihlatie 1, H. E. Manninen 1, V. Aaltonen 2, H. Junninen 1, T. Petäjä1, J. Backman 1, M. Dal Maso1, T. Nieminen1, T. Olsson 2, T. Grönholm 8, J. Aalto 3, T. H. Virtanen 2, M. Kajos 1, V.-M. Kerminen 1, D. M. Schultz 1,2,9, J. Kukkonen 2, M. Sofiev 2, G. De Leeuw 1,2, J. Bäck 7, P. Hari 7, and M. Kulmala 1 1Department of Physics, University of Helsinki, 00014, Helsinki, Finland 2Finnish Meteorological Institute, Erik Palménin aukio 1, 00101, Helsinki, Finland 3Hyytiälä Forestry Field Station, University of Helsinki, 35500, Korkeakoski, Finland 4Department of Meteorology and Climate Science, San José State University, San José, CA 95192, USA 5Department of Technology, Metropolia University of Applied Sciences, 00079, Helsinki, Finland 6Finnish Forest Research Institute, P.O. Box 18, Vantaa, Finland 7Department of Forest Sciences, P.O. Box 27, University of Helsinki, 00014, Helsinki, Finland 8Finnish Environment Institute, Joensuu Office, 80101 Joensuu, Finland 9Centre for Atmospheric Science, School of Earth, Atmospheric and Environmental Sciences, University of Manchester, Simon Building, Oxford Road, Manchester, M139PL, UK Correspondence to: A. Virkkula ([email protected]) Received: 20 June 2013 – Published in Atmos. Chem. Phys. Discuss.: 22 August 2013 Revised: 20 January 2014 – Accepted: 3 March 2014 – Published: 7 May 2014 Abstract. A prescribed fire experiment was conducted on 26 June 2009 in Hyytiälä, Finland, to study aerosol and trace gas emissions from prescribed fires of slash fuels and the effects of fire on soil properties in a controlled environment. A 0.8ha forest near the SMEAR II measurement station (Station for Measuring Ecosystem-Atmosphere Relations) was cut clear; some tree trunks, all tree tops and branches were left on the ground and burned. The amount of burned organic material was ∼46.8tons (i.e., ∼60tonsha−1). The flaming phase lasted 2h 15min, the smoldering phase 3h. Measurements were conducted on the ground with both fixed and mobile instrumentation, and in the air from a research aircraft. In the middle of the burning area, CO2concentration peaked around 2000–3000ppm above the baseline, and peak vertical flow velocities were ∼9ms−1, as measured with a 10Hz 3-D sonic anemometer placed within the burn area. In the mobile measurements the peak particle number concentrations were approximately 1–2×106cm−3in the plume at a distance of 100–200m from the burn area. On the ground at the SMEAR II station the geometric mean diameter of the mode with the highest concentration was 80±1nm during the flaming phase and in the middle of the smoldering phase, but, at the end of the smoldering phase, the largest mode was 122nm. In the volume size distributions, geometric mean diameter of the largest volume mode was 153nm during the flaming phase and 300nm during the smoldering phase. The lowest single-scattering albedo of the groundlevel measurements was 0.7 in the flaming-phase plume and ∼0.9 in the smoldering phase. Elevated concentrations of several volatile organic compounds (VOC) (including acetonitrile, a biomass burning marker) were observed in the smoke plume at ground level. Measurements at the forest floor (i.e., a richly organic layer of soil and debris, characteristic of forested land) showed that VOC fluxes were generally low and consisted mainly of monoterpenes, and VOC flux peaked after the burning. After one year, the fluxes had Published by Copernicus Publications on behalf of the European Geosciences Union. 4474 A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland nearly stabilized close to the level before the burning. The clear-cutting and burning of slash increased the total longterm CO2release from the soil, and altered the physical, chemical and biological properties of the soil, such as increased the available nitrogen contents of the soil, which in turn, affected the long-term fluxes of greenhouse gases. 1 Introduction Gaseous and aerosol emissions from wildfires have significant climatic (e.g., Bowman et al., 2009; Grell et al., 2011) and health effects (Naeher et al., 2007; Johnston et al., 2012) on local to hemispheric scales. In the Northern Hemisphere, smoke from wildfires can be transported over long distances from the boreal forest areas in Eurasia and North America to the Arctic (e.g., Radke et al., 1991; Goldammer et al., 1996; Lavoué et al., 2000; Randerson et al., 2006; Stohl, 2006; Law and Stohl, 2007; Shindell et al., 2008; Paris et al., 2009; Hirdman et al., 2010; Lamarque et al., 2010; AMAP, 2011, 2011). Smoke originating from wildfires and agricultural fires in eastern Europe can affect extensive regions in western and central Europe and the Arctic (Law and Stohl, 2007; Saarnio et al., 2010; Klein et al., 2012). Fires directly emit long-lived greenhouse gases (e.g., carbon dioxide (CO2), nitrous oxide (N2O)) and short-lived greenhouse gases (e.g., methane (CH4)), countless volatile organic compounds (VOC) and nitrogen oxides (NOx) that are precursors of ozone (O3), a short-lived greenhouse gas (e.g., Andreae and Merlet, 2001; Akagi et al., 2011; Simpson et al., 2011; Jaffe and Widger, 2012; Yokelson et al., 2013). The particles emitted by fires are short-lived climate forcers that can have either negative or positive forcing effects, depending on their optical and cloud-forming properties and on the albedo of the underlying surface (e.g., Randerson et al., 2006; Quinn et al., 2008; Ramanathan and Carmichael, 2008). Black carbon emitted from wildfires may get transported and deposited on snow or ice, where it has a positive radiative forcing due to the reduction of the albedo of the surface (e.g., Ramanathan and Carmichael, 2008; Bond et al., 2013). Wildfires also change the surface albedo of the forests, which can impact the climate. For instance, Randerson et al. (2006) showed that the warming impact of increasing boreal forest fires may be limited or even result in regional cooling because of the loss of canopy overstory and consequently higher albedo values during winter and spring. Active wildfires and burned areas can be observed from space by using satellite imagery (e.g., Flannigan and Haar, 1986; Lentile et al., 2006; French et al., 2008; Sofiev et al., 2009; Giglio et al., 2010). Satellite images give information on the area that is burning, but not on the amount of fuel consumed or smoke emitted. van der Werf et al. (2010) used a biogeochemical model and satellite-derived estimates of area burned, fire activity, and plant productivity to calculate the total global carbon emissions due to deforestation, savanna, forest, agricultural and peat fires. They estimated that the boreal region accounted for about 9% of total global carbon emissions from fires. To estimate the amount of aerosols and trace gases emitted, emission factors, defined as the amount of emitted aerosol or trace gases per mass unit of burned biomass, are needed. Recent reviews of emission factors include Andreae and Merlet (2001), Reid et al. (2005a, b), Janhäll et al. (2010), Akagi et al. (2011), Simpson et al. (2011) and Yokelson et al. (2013). Detailed measurements of gas and aerosol emissions are difficult to obtain in real wildfires. The fire may be too large and uncontrolled, and placing instrumentation near the fire may be difficult. To address these problems, researchers often choose the more controlled environment of a prescribed burn to study fires. Prescribed fire is used for fire prevention, site preparation and maintaining habitat quality (Bowman et al., 2009, 2011). The total area of prescribed fire in the USA was nearly 1 million hectares during 2011 (National Interagency Fire Center, 2011). The areas of prescribed burns that have been used for research vary by several orders of magnitude. Radke et al. (1991) described measurements of smokes from 17 biomass fuel fires, including 14 prescribed fires and 3 wildfires primarily in the temperate zone of North America. The prescribed fires were in forested lands and logging debris and the areas burned varied from 10 to 700 hectares (ha). The “Smoke, Clouds and Radiation – California”, SCAR– C, experiment was conducted in September 1994 in the Pacific Northwest of the United States (Kaufman et al., 1996; Hobbs et al., 1996; Gassó and Hegg, 1998). In SCAR–C the emissions from clear-cut prescribed burns with areas ranging from 19.4ha to 44.5ha were measured with instruments on an aeroplane (Hobbs et al., 1996). Close to natural wildfires were for instance the Bor Forest Island Fire Experiment in which the forest on a 50ha Siberian island was burned in 1993 (FIRESCAN Science Team, 1996), the International Crown Fire Modelling Experiment (ICFME) that involved 18 experimental high-intensity crown fires ranging from 0.56ha to 2.25ha in size in Canada’s Northwest Territories between 1995 and 2001 (Cofer III et al., 1998; Conny and Slater, 2002; Alexander et al.,2004; Payne et al.,2004; Stocks et al., 2004), the FROSTFIRE experiment in which approximately 365ha of black spruce was burned in Alaska in 1999 (Ferguson et al., 2003; Hinzman et al., 2003; Cahill et al., 2008), and the Fire Effects in the Boreal Eurasian Forests (FIRE BEAR) experiments where 22 plots of approximately 4ha were burned in central Siberia in 2000 and 2001 (McRae et al., 2006). In Finland, there is a long tradition of managed forest burning. The use of burn-beating cultivation to produce corn and root crops existed for several hundred years and ended around 1910 (Heikinheimo, 1915). In the 1920s, prescribed burning of clear-cut areas began (Viro, 1969). The idea of prescribed burning is to burn the logging waste, surface vegetation and the uppermost part of the raw humus layer. This Atmos. Chem. Phys., 14, 4473–4502, 2014 www.atmos-chem-phys.net/14/4473/2014/ A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland 4475 practise promotes the regeneration of the tree stand and is normally followed by the seeding of Scots pine and occasionally silver birch. Prescribed burning was widely used in Finland in the 1950s and 1960s, with over 10000ha typically burned annually. Since then, more effective mechanical soilpreparation methods superseded prescribed burning (Finnish Forest Research Institute, 1991). One reason for the reduction in the areas burned was also the fear of the fire getting out of control. Nowadays 500–1000ha is burned each year (Finnish Forest Research Institute, 2011). The main reason of the present-day burnings is to enhance biodiversity. The use and effects of managed burning of forest are investigated at the University of Helsinki. As part of this work, we conducted a prescribed burning about 300–500m south-southwest from the SMEAR II measurement station (Station for Measuring Ecosystem–Atmosphere Relations; Hari and Kulmala, 2005) in Hyytiälä, Finland (61◦5004700 N, 24◦1704200 E, 181m a.m.s.l.) on 26 June 2009. The experiment was an integral part of two large projects: the European Integrated project on Aerosol Cloud Climate and Air Quality Interactions (EUCAARI; Kulmala et al., 2011) and the Integrated Monitoring and Modelling System for Wildland Fires (IS4FIRES; Saarikoski et al., 2007; Sofiev et al., 2009). A 0.8ha forest area near SMEAR II was cut clear. Some tree trunks, all tree tops and all branches were left on the ground and burned. The prescribed burning therefore represents fires in a forested area containing clear-cut blocks of logging slash. It does not represent fires in a full-grown boreal forest. The burned area is small compared with the experiments mentioned above (e.g., Radke et al., 1991; Hobbs et al., 1996; Hinzman et al., 2003; McRae et al., 2006). During burning, we conducted measurements on the ground and in the air. Ground-based instrumentation included the SMEAR II station and meteorological and ecological measurements on and around the site. We measured ground-level dispersion of particles and trace gases both by using the research van Sniffer and by walking in the forest with portable particle counters at different distances from the burning area. We measured the vertical and horizontal dispersion of the plume with instruments installed in a Cessna 172. Soil temperature, humidity and trace gas efflux were measured within the burn and unburned reference areas. The general goal of the experiment was to collect data for estimating the effect of fires on air quality and climate. The specific goals and objectives were (1) to obtain emission factors of aerosols and gases, (2) to characterize the climatically relevant physical properties of the smoke aerosol (e.g., size, optical properties), (3) to quantify the connections between ground-based smoke observations and satellite remote sensing, (4) to obtain data for testing an improving model of atmospheric dispersion of the fire plume, and (5) to quantify the changes taking place in soil carbon stocks and greenhouse gas (CO2, CH4and N2O) fluxes following clear-cutting and prescribed burning. The experiment fulfils the requirements to be called a prescribed fire: it is the application of prescribed burning in a skilled manner, under exacting weather conditions, in a definite place, and to achieve specific results (Wade and Lunsford, 1989; Alexander and Thomas, 2006). The purpose of this article is to provide an overview of the experiment by describing the preparations for the experiment, estimates of burned biomass, meteorological conditions during the experiment, characterization of the aerosols and gases emitted, and the observed dispersion of aerosols both at ground level and in the airborne measurements. The aim is also to evaluate the performance of the setup for repeating similar experiments. Following the recommendation of Alexander (2010) for the documentation of prescribed fires, we also present photographs from the different phases of the fire. 2 Methods 2.1 The site preparation A suitable burn area was found in summer 2008 approximately 300–500m south-southwest of the measurement buildings of the SMEAR II station. We selected the site to be burned so that prevailing southwesterly wind, specifically, with a wind direction from 180–200◦, would bring the smoke aerosols and gases to the SMEAR II station during the burning. To determine the suitability of the burn area, the 30min averaged wind direction from the SMEAR II mast data over the layer 33.6m to 73m was averaged from every month of June of the years 1996–2008 to get the mean wind direction above the treetops. Based on this climatology, wind directions of 180–200◦occurred 9.6% of the time with no particular preference for a specific time of day. When the wind occurred in this direction, the average wind speed was 3.1±1.3ms−1and 86% of wind speeds were less than the 5ms−1threshold required for a safe burn. In addition to the burn area, we also selected a control site near the burning area (Fig. 1). At the burn site, there was a mature spruce-dominated (Picea abies (L.) H. Karst. known as a Norway Spruce or European Spruce) stand with a stem volume per hectare of about 400m3. The area was cut clear in February 2009. After the clear-cutting most tree trunks were transported away; some of them and all tree tops and all branches were left on the ground in the burn area. An estimation of the biomass was done before and after the burning. 2.2 Estimation of burned organic material The tree stand was measured in July 2008 from 13 relascope plots from which the species, diameter at breast height (DBH), diameter at the height of 6.0m, the living crown length and height (H) were recorded for each tree. Then, biomass models (Repola et al., 2007) were used to calculate the biomass for the different tree components. The merchantable wood was harvested in February 2009, after www.atmos-chem-phys.net/14/4473/2014/ Atmos. Chem. Phys., 14, 4473–4502, 2014 4476 A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland a) b) Fig. 1. Aerial photographs of the study site during the flaming phase. (a) From the south, with the progress of the ignition described by the red, blue, white and light blue arrows, and the average wind direction by the pink arrow. The yellow bars denote the poles (not in scale) with meteorological sensors (MS1–MS4) on top. (b) From the north at approximately 300m above ground level. The majority of the aerosol and gas measurement instrumentation of SMEAR II is located within the dashed white oval. The control area is within the yellow oval. The blue arrow points approximately north. AC: aerosol cottage; REA: relaxed eddy accumulation cottage; AODTWR: Aerosol Optical Depth Tower. which all the non-merchantable trees were also felled. After burning, the amount of unburned wood was sampled from 21 plots of 0.5m2. All the wood was collected from the plots, dried (24h, 105◦C) and weighed. The amount of burned tree biomass was finally calculated as an extraction of the non-merchantable tree biomass (treetops, branches and non-merchantable trees) and unburned wood biomass. The surface vegetation, dominated by feather mosses and dwarf shrubs, was systematically sampled from 13 plots of 0.0625m2in July–August. The vegetation was cut along the surface of the litter layer, collected, dried (24h, 105◦C) and weighed. The organic matter content of the uppermost, organic soil layers (litter layer and humus layer) was systematically sampled both before the clear-cutting in August 2008 and soon after the burning in July 2009. A total of 25 samples was collected on both occasions with a 45mm-diameter soil auger. The samples were dried (24h, 105◦C) and weighed. The mass of burned organic material in the organic soil layer was calculated as the difference of the mass before and after the burning. 2.3 Gas, aerosol and meteorological measurements A list of the measurements conducted during the campaign is presented in Table 2. In short, trace-gas concentrations, aerosol physical properties aerosol chemical composition, and meteorological parameters were measured both at fixed sites and on mobile platforms. 2.3.1 Measurements at fixed positions At the SMEAR II measurement station, both aerosols and gases were measured with the setup described by Hari and Kulmala (2005). Measurements were conducted at five different locations: the main building of the station, the 73mhigh SMEAR II mast, the aerosol cottage, the relaxed eddy accumulation (REA) cottage, and the aerosol optical depth tower (AODTWR) about 100m east of the aerosol cottage. The above measurements are within 300–400m of the burn area (Fig. 1b). The concentrations of CO2, H2O, O3, NO, NOx, SO2and CO were measured alternately at six heights along the 73m mast. The instruments were located in the main building, and sample air was taken through six sample lines consisting of polytetrafluoroethylene (PTFE) tubes, each 100 m long with 14mm inner diameter and 16 mm outer diameter. There was a continuous flow rate of 45Lmin−1in the lines, which resulted in an estimated lag time of 20s. For each gas component, there was one analyzer for measuring the concentrations, except for NO and NOxwhich were measured with one instrument. The response times of the analyzers were about 30s, so when sampling a new height, a flush time of about 30s was needed. The sample line system and instrumentation at the station is designed for measuring accurately the concentration profiles which then degrade the temporal coverage of the results for the separate measurement heights. The combined response and flush time set the signal recording time step to 1min and the overall time spacing of the data per measurement height to 6 min. In general, the averaging times were different for different analyzers. Exact averaging times cannot be given for every analyzer because of the combination of averaging caused by flushing of sample volumes, signal averaging and “sample measurement – reference measurement sequencing” of an analyzer vs. the switching cycle of sample heights in the tower. For the CO2analyzer, each 1min signal is approximately a 30s average; for the CO analyzer, each 1min signal is approximately a 1min average. The signal standard deviations of the CO2and the CO Atmos. Chem. Phys., 14, 4473–4502, 2014 www.atmos-chem-phys.net/14/4473/2014/ A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland 4477 analyzers were 0.04ppm and 6ppb, respectively, when sampling calibration gas. These are the estimates for the precision of any 1min value. VOCs were measured with Proton Transfer Reaction Mass Spectrometers (PTR-MS, Ionicon Analytik, Innsbruck, Austria; e.g., Hewitt et al., 2003) at two locations: one at the SMEAR II main building with an inlet above the roof at about 10m above ground level and the other in the REA cottage with the inlet above canopy at about 16m above ground level at the REA tower. The PTR-MS instrument measures charged VOCs at a given mass that was assigned to the VOCs that likely dominated each signal. The assignment of massto-charge ratios (m/z) to VOCs and the measurement setup are described by Taipale et al. (2008). Usually, m/z =69 is assigned to the biogenic VOCs isoprene and 2-Methyl-3buten-2-ol (MBO), but, in this case, m/z =69 was assigned to furan, which is associated with burning processes (de Gouw and Warneke, 2007). The VOC measurements were sampled every 1min. An Aerodyne Aerosol Mass Spectrometer (AMS) (e.g., Jayne et al., 2000; Jimenez et al., 2003; Drewnick et al., 2005) was used for measuring the concentrations of ammonium (NH+ 4), sulfate (SO2− 4), nitrate (NO− 3), chloride (Cl−) and organics in particles with Dp<600nm. The AMS was located in the SMEAR II main building, and it took its sample from the same inlet as the PTR-MS, above the roof at about 10m above ground level. The AMS measurements were taken every 5min. In the aerosol cottage, particle number size distributions for particles of 3–1000nm in diameter were measured with a custom-made twin differential mobility particle sizer (TDMPS) system (Aalto et al., 2001) and a TSI aerodynamic particle sizer (APS) in the aerodynamic diameter size range of 0.53–20µm. In the overlapping range of the TDMPS and the APS, the number concentrations from the TDMPS were used up to 700nm. Data were collected every 10min. A Neutral cluster and Air Ion Spectrometer (NAIS) was used to measure the mobility and size distributions of atmospheric ions and neutral clusters in the size range of 0.8–47nm (e.g., Manninen et al., 2009; Asmi et al., 2009) for the first time in a wildfire smoke plume. The NAIS measurements were taken every 2min. The aerosol optical measurements at SMEAR II were described by Virkkula et al. (2011). In short, total scattering coefficients (σsp) and backscattering coefficients (σbsp) were measured with a TSI 3λnephelometer, averaged over a 5min period. A Magee Scientific 7λAethalometer (AE-31) was used for measuring light absorption, also at a 5min averaging time. The absorption coefficient (σap) was calculated from the aethalometer and nephelometer data using the algorithm by Arnott et al. (2005). Aerosol optical depth was measured with a Cimel CE-318 sunphotometer in a tower about 100m east of the aerosol cottage (Fig. 1), above the canopy level. The sunphotometer made one instantaneous measurement every 15min. In the same tower, the light absorption coefficient at a wavelength (λ) of 637nm was measured with a Multi-Angle Absorption Photometer (MAAP). The MAAP reports the absorption coefficient as black carbon concentrations using the mass absorption coefficient of 6.6m2g−1. MAAP measurements were available every 1min. In addition to the SMEAR II measurements, meteorological instrumentation was installed on top of poles within and around the area to be burned (hereafter called stations). The poles were prepared by cutting the branches of five trees that were left standing in the slash. Four stations were outside the burning area, and one was within it (Fig. 1a). The distances of the stations 1, 2, 3 and 4 from the perimeter of the burn area were 10, 8, 9 and 6m, respectively. In situ meteorological instrumentation (Vaisala WTX510) was deployed on the burn perimeter, and a sonic anemometer (Applied Technologies, Inc. (ATI) Sx-Probe) and Vaisala GMP-343 CO2sensor were placed within the burn area on the area on the top of a pole at about 12m in height. Total heat flux, Q, including radiative heat flux, was measured at 1m above the surface with a water-cooled Hukseflux SBG01 sensor. Fine-wire thermocouples (Omega, 5SC-TT-E-40-36, Type-E) were mounted along the pole to measure near-surface plume temperatures. 2.3.2 Mobile measurements Ground-level dispersion of aerosols and gases was measured in Sniffer from the Metropolia University of Applied Sciences, Helsinki (Pirjola et al., 2004, 2006). Sniffer was driven along the surrounding forest roads and stopped at several locations for some minutes. Sampling occurred above the windshield of the van at 2.4m altitude. Particle number concentration and size distribution were measured by an Electrical Low Pressure Impactor (ELPI, Dekati Ltd) at a flow rate of 10Lmin−1(Keskinen et al., 1992). ELPI was equipped with a filter stage (Marjamäki et al., 2002) and a stage to enhance the particle size resolution for nanoparticles (YliOjanperä et al., 2010). The ELPI classifies particles in the size range of 7nm–10µm (aerodynamic diameter) into 12 classes with samples every 1s. Sniffer also monitored concentrations of CO, NO, NO2and CO2at 1s intervals. Furthermore, PM2.5and PM10 were recorded by two TSI DustTrak aerosol monitors. The DustTraks measure light scattering, but they were not specifically calibrated for smoke aerosol. A weather station on the roof of Sniffer at 2.9m height provided meteorological parameters (temperature, relative humidity, wind speed and wind direction). A global positioning system (GPS) was used to record the van’s speed and the driving route. In addition to Sniffer, the dispersion at ground level was measured by students walking around the area with three portable TSI model 3007 condensation particle counters (CPCs) and GPS receivers. There were three different routes at three distances from the burning area. The CPCs used in the nearest two routes were equipped with diluters because, www.atmos-chem-phys.net/14/4473/2014/ Atmos. Chem. Phys., 14, 4473–4502, 2014 4478 A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland according to the manual, the model 3007 CPC measures concentrations up to 105cm−3. The diluters were calibrated afterwards and a flow rate of 0.7Lmin−1in the 3007 CPC produced a dilution ratio of about 0.32. Thus, the concentrations from the two nearest routes were divided by 0.32, resulting in the upper limit of the concentration range increasing to about 3×105cm−3. Vertical and horizontal dispersion were measured with instruments installed in a Cessna 172 research aircraft (Schobesberger et al., 2013; Virkkula et al., 2013). There were three CPCs for measuring particle number concentrations at three cutoffs (3, 6 and 10nm). The 3nm cutoff was with a TSI model 3776 CPC. The other two were TSI model 3772 CPCs equipped with 1:10 diluters and set up for cutoff sizes of 6nm and 10nm. In the present article, the discussion of particle number concentrations is based on the model 3776 CPC only. The scattering coefficient (σsp) at λ=545nm was measured with a Radiance Research model 903 nephelometer, and the absorption coefficient (σap) was measured with a Radiance Research 3-λParticle Soot Absorption Photometer (PSAP) at λ=467nm, 530nm and 660nm. A LI-COR LI-840 measured CO2concentrations. The data were saved at 1Hz frequency. The scattering and absorption coefficients will be discussed in the companion paper (Virkkula et al., 2013). 2.3.3 Soil and flux measurements The changes in soil physical, chemical and biological environment were monitored with a long-term perspective, as similar high-frequency instrumentation described above for atmospheric aerosol and trace gas concentrations are not available for the soil parameters. Also, although the soil conditions do change rapidly during and after the fire, many of the biological processes and responses to changing conditions have a time lag and therefore require several years of monitoring. These slowly changing responses were expected, for instance in soil pH and the concentrations of available nitrogen, as well as soil greenhouse gas fluxes. Long-term ecological measurements were begun in the mature forest in 2008 before the clear-cutting and partial burning. The measurements were performed at three sites: (1) in the area that was later clear-cut, (2) in the area that was later clear-cut and also burned and (3) in an area that remained as a mature forest. These measurements comprised automatic soil temperature and moisture measurements in the organic layer and in the A and B mineral soil horizons, manual measurements of the heights of soil organic layers, and the total carbon and nitrogen content, as well as available nitrogen species and pH in organic and mineral soil horizons. The soil horizon is a layer parallel to the soil surface, whose physical characteristics differ from the layers above and beneath. The measurement campaign ended in late 2011, two and a half years after the burn. The long-term effects of the burning of slash on the CO2and CH4fluxes from the soil were quantified by manual chamber measurements from the burned area every two weeks, together with the corresponding measurements from the clear-cut and a control forest (Kulmala et al., 2014). The fluxes were measured every two weeks from early May to the end of November for one year before and for three years after the treatment. The flux measurements were performed by placing a chamber on a collar inserted into the soil to an approximate depth of 5cm. Eleven collars were inserted for CO2measurement and eight collars were inserted for CH4 measurement at each site, and one closure took 4min for CO2and 35min for CH4, as described in detail by Kulmala et al. (2014) and Pihlatie et al. (2013). During 2008–2010, CO2fluxes at each site were also measured using an automatic chamber described in detail by Kulmala et al. (2010). We approximated the cumulative release of CO2at each site after the treatments by interpolating the effluxes from each treatment separately between the days. The emission of forest-floor VOCs was measured at the burn site twice before (23–24 July and 1–2 October 2008) and three times after the burning (14–15 July 2009, 15–16 September 2009 and 24–25 October 2010). The VOC fluxes were measured on five permanently installed collars with a manual steady-state chamber system. The VOC sampling and analysis method is described by Aaltonen et al. (2011). 2.4 Formulas used for data processing By using the measurements in the 12m pole within the burning area, the turbulent sensible heat flux was calculated from the covariance of the vertical velocity and sonic temperature perturbation as Hs=ρcpw0T0,(1) where ρis the air density, assumed to be constant, and cpis the heat capacity of air at constant pressure. The turbulent kinetic energy (TKE) is calculated from the sum of the velocity variances: TKE =1 2ρ2 u+ρ2 v+ρ2 w.(2) In the smoke plume, concentrations of trace gases and aerosols were elevated. The concentration of trace gas X above the background is defined as the excess concentration and denoted as 1X. The concentrations of several trace gases vary also smoothly during the day due to biological processes such as photosynthesis, so the background was not taken as a constant value for the whole day. Instead, at every time step t, all concentrations of a trace gas X measured in the time range of [t–30min, t+30min] were taken into account, and the background was the lowest concentration of X during this time. This was applied both for the data obtained from the mast and for the PTR-MS data. For the trace gases measured from the mast at alternating altitudes, the background Atmos. Chem. Phys., 14, 4473–4502, 2014 www.atmos-chem-phys.net/14/4473/2014/ A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland 4479 was calculated without taking the altitude into account (i.e., by considering the time series as if it had been measured at one altitude only). Trace gas emission from biomass burning can be expressed as an emission factor (EF), which signifies the emitted mass of trace gas X divided by the burned dry biomass, or as emission ratios (ER) that relate the emission of the species X to that of a reference species (e.g., Andreae and Merlet, 2001; Simpson et al., 2011). When CO is used as the reference species, the emission ratio is ERX/CO =1X 1CO.(3) If there are several simultaneous 1X and 1CO values, the emission ratio can also be calculated by fitting the line 1X = ERX/CO1CO with linear regression with the offset forced to zero (e.g., Yokelson et al., 1999; Simpson et al., 2011). The aerosol number size distributions were used for calculating volume size distributions and the integrated mass concentrations were calculated by assuming a density of 1.5gcm−3. Three to five lognormal modes were fitted to the data up to 10µm. The fitting yields the modal parameters (geometric mean diameter (Dg), geometric standard deviation (σg), and number or volume concentration of the mode). The in situ aerosol optical data were analyzed as discussed in Virkkula et al. (2011). Here we calculated three intensive aerosol optical properties: the single-scattering albedo (ω0), the Ångström exponent of scattering (αsp) and the backscatter fraction (b). ω0=σsp σsp +σap (4) is a measure of the darkness of aerosols; for purely scattering aerosols, it equals 1. For freshly generated pure Black Carbon (BC) ω0has been measured to be 0.2±0.1 (e.g., Bond and Bergstrom, 2006; Mikhailov et al., 2006; Cross et al., 2010; Bond et al., 2013). The Ångström exponent of scattering αsp describes the wavelength dependency of scattering, and it was calculated for the nephelometer wavelength range by taking the logarithm of scattering coefficients and the respective wavelengths and fitting the data line to the line ln(σsp)= −αsp ln(λ) +C, (5) where Cis a constant not relevant in this study. In general, large values (αsp>2) indicate the dominance of small particles, and small values (αsp<1) indicate the dominance of large particles. This relationship is not unambiguous, however (e.g., Schuster et al., 2006; Virkkula et al., 2011). The backscatter fraction b=σbsp σsp ,(6) where σbsp is the backscattering coefficient, is a measure related to the angular distribution of light scattered by aerosol particles. From b, it is possible to estimate the average upscatter fraction βand the aerosol asymmetry parameter, which are the key properties controlling the aerosol direct radiative forcing (e.g., Andrews et al., 2006). In general, larger particles scatter less light backwards than small particles so the size relationship of bis qualitatively similar to that of αsp. The radiative forcing efficiency (1F/δ), i.e., aerosol forcing per unit optical depth (δ), was calculated from 1F δ= −DS0T2 at(1−Ac)ω0β(7) (1−Rs)2−2Rs β 1 ω0−1, where Dis the fractional day length, Sois the solar constant, Tat is the atmospheric transmission, Acis the fractional cloud amount, Rsis the surface reflectance, and β is the average upscatter fraction calculated from b. If the non-aerosol-related factors are kept constant and if it is assumed that βhas no zenith angle dependence, this formula can be used for assessing the intrinsic radiative forcing efficiency by aerosols (e.g., Sheridan and Ogren, 1999; Delene and Ogren, 2002). The constants used were D=0.5, So=1370Wm−2,Tat =0.76, Ac=0.6, and Rs=0.15 as suggested by Haywood and Shine (1995), and βwas calculated from β=0.817+1.8495b–2.9682b2(Delene and Ogren, 2002). The value of Rs=0.85 was used to assess the effect of the aerosols above snow surfaces. 3 Results and discussion 3.1 General description of the burning Whenever prescribed fires are conducted, parameters affecting fire weather, such as biomass moisture, temperature, relative humidity, wind speed and rainfall should be measured and documented prior to and during the burning operation, as recommended by Alexander (2006). The measurement setup was ready at the beginning of May 2009, waiting for the proper conditions. For our experiment the required conditions were: (1) wind direction was to be in the range of 175◦– 215◦to blow smoke toour ground-based instrumentation, (2) wind speed had to be less than 5ms−1to keep the fire under control as suggested by the Finnish handbook of prescribed burns (Lemberg and Puttonen, 2002), (3) soil had to be dry enough to burn properly, and (4) the sky was to be clear so that the smoke plume (i.e., the visible column of smoke) could be followed with the aircraft and possibly even from satellites. Fulfilment of requirement 3 was assessed from the forest fire warnings issued by the Finnish Meteorological Institute. In Finland, forest fire danger is assessed by the Forest Fire Index (FFI), which is calculated from soil moisture, so that for very wet conditions FFI <2 and for very dry conditions FFI=6.0 (Vajda et al., 2013). www.atmos-chem-phys.net/14/4473/2014/ Atmos. Chem. Phys., 14, 4473–4502, 2014 4480 A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland Table 1. Estimated amount of burned organic material during the experiment. s.e. = standard error of the estimate. Mass ±s.e. Mass/area ±s.e. Carbon ±s.e. Carbon/area ±s.e. (kg) (kgha−1) (kg) (kgha−1) Tree biomass 30700 38030 15400 19080 Surface vegetation 1850 2300 930 1150 Organic soil layer 14200 17600 7100 8800 Sum 46800 ±10900 58000 ±13500 23400 ±5500 29000 ±6800 The last rain before the burning day was on 20 June, and on 25 June fuel volumetric moisture had decreased below 0.2m3m−3and FFI increased above 4 in the scale from 1 to 6. On the morning of 26 June, a handheld smoke signal was ignited soon after 07:00 Eastern European Time (EET=UTC +2h) in order to make the final decision of whether to start the fire. Wind was blowing from the right direction, the sky was clear, and soil moisture was 0.18kgkg−1, so FFI was 4.2. Relative humidity and temperature at 4.2m above ground was 56% and 19◦C, respectively. The conditions were acceptable, so the area was set on fire at 07:45EET. (All times presented below will be inEET, not in Eastern European Summer Time.) The burning was performed against the wind as a backing burn; first the fire was ignited against the wind and then ignition slowly proceeded in both directions (Fig. 1a). The idea was to slowly burn the edges of the site until a horseshoelike shape was achieved and more than half of the area was burned (Fig. 2a). This phase of our experiment took about 110 min. Then, the edges were rapidly ignited in both directions so that the edges of the site were enclosed with the fire (Fig. 2b). Thereafter, the fire proceeded rapidly downwind, and flaming was over within about 25min. Flame height was not measured, but we estimate that during the flaming phase, it was approximately 1–3m (from the photographs in Fig. 2a–f). Flame lengths varied from about 1m during backing fires along the burn plot edge to ∼5–10m during heading fires. While the rate of fire spread (ROS) was not measured during the experiment, estimates were made in the field. About 50% of the firing operations consisted of backing fires with an estimated ROS ∼0.01ms−1while the remaining ignitions were made using head fires with an average ROS of ∼0.2ms−1. The flaming or active burning was over at 10:00EET, and there was only a little visible smoke at 13:00EET. These times will be shown in the figures below as the indicators of the flaming and smoldering phases of the burning, although the ends of both periods were not well defined. There were flames in some parts of the area while most of it was already smoldering, and smoldering biomass does not always emit visible smoke. After the burning, the amount of burned organic material was estimated as described above (Sect. 2.2). The amount of unburned wood was 30700kg. The burned area was approximately 0.81ha, so the amount of burned wood ~7 m a) b) c) d) e) f) g) h) 09:32 09:24 09:38 09:39 09:39 09:41 11:41 10:52 ~7 m ~7 m ~12 m Fig. 2. Photographs of the burn area during the flaming phase (a–f) and the smoldering phase (g, h). biomass was about 38030kgha−1. All the surface vegetation, 1850kg (2300kgha−1), was burned. The mass of burned organic material in the organic soil layer was 14200kg (17600kgha−1). The total amount of burned organic material (46800kg, 58000kgha−1) was calculated as a sum of burned tree biomass, surface vegetation and organic soil layer (Table 1). The burned biomass was strongly dominated by slash (non-commercial wood), and the roundwood material consisted of treetops, with a diameter of 7cm or less, and branches typically smaller than 5cm in diameter. The mass portion of the needle foliage from the total burned biomass was 37%. Schlesinger (1997) noted that the carbon content of biomass is generally between 45% and 50% (by oven-dry mass). Table 1 also presents an estimated amount of carbon released by multiplying the biomass by 0.5. Atmos. Chem. Phys., 14, 4473–4502, 2014 www.atmos-chem-phys.net/14/4473/2014/ A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland 4481 Table 2. Measurements made during the prescribed fire experiment, the instrument type, location of the instrument, and the responsible institute. Quantity Instrument Location Institute Meteorology Heat flux Hukseflux SBG01 mast within the burn area SJSU (San Jose State University) Temperature Fine wire thermocouples mast within the burn area SJSU Wind 2-D anemometers SMEAR II mast, several altitudes UHEL (University of Helsinki) Vaisala WXT 520 small masts around burn area FMI (Finnish Meteorological Institute) 3-D sonic anemometer, ATI, Sx probe mast within the burn area SJSU Trace gases CO2URAS 4 SMEAR II mast, several altitudes UHEL mast within the burn area SJSU CO Horiba APMA SMEAR II mast, several altitudes UHEL NOxTEI 42CTL SMEAR II mast, several altitudes UHEL O3TEI 49C SMEAR II mast, several altitudes UHEL VOCs Proton Transfer Reaction Mass Spectrometer REA cottage UHEL (PTR-MS) SMEAR II, new part UHEL Aerosol physical properties Size distribution Dp: 3–1000nm differential mobility particle sizer Aerosol cottage UHEL Dp: 0.5 – 10 µm aerodynamic particle sizer Aerosol cottage UHEL Dp: 0.4–40nm Neutral cluster and Air Ion Spectrometer (NAIS) Aerosol cottage UHEL Scattering coefficient 3-λnephelometer Aerosol cottage UHEL Absorption coefficient 7-λaethalometer Aerosol cottage UHEL Multi-Angle Absorption Photometer (MAAP) AODTWR FMI Aerosol optical depth Sunphotometer AODTWR FMI Aerosol chemical composition NO− 3Aerosol Mass Spectrometer (AMS) SMEAR II, new part UHEL SO2− 4AMS SMEAR II, new part UHEL NH+ 4AMS SMEAR II, new part UHEL Cl−AMS SMEAR II, new part UHEL Organics AMS SMEAR II, new part UHEL Mobile measurements On ground Number concentration 3 portable condensation particle counters Walking UHEL Number concentration Electric Low Pressure Impactor (ELPI) Sniffer van MUAS (Metropolia University of Applied Sciences) Size distribution Sniffer van MUAS CO2Sniffer van MUAS CO Sniffer van MUAS NOxSniffer van MUAS Airborne Number concentration 2 TSI Model 3762 and 1 model 3772 CPC Cessna 172 UHEL Scattering coefficient Radiance Research, 1-λnephelometer, 545 nm Cessna 172 UHEL Absorption coefficient Radiance Research, 3-λPSAP Cessna 172 UHEL CO2LI-COR LI-840 Cessna 172 UHEL Soil measurements Temperature iButtons, PT100 Burned and reference area UHEL Moisture ThetaProbe Burned and reference area UHEL pH Burned and reference area UHEL C/N-ratio Burned and reference area UHEL Nitrogen compounds Burned and reference area UHEL CO2efflux Vaisala GMP343 Burned and reference area UHEL CH4flux Agilent Gas Chromatograph model 7890A Burned and reference area UHEL VOC flux Tenax-Carbopack-B +GC-MS Burned area UHEL (Gas Chromatograph-Mass Spectrometer) 3.2 Winds Most of the smoke ascended almost vertically, as seen from the aerial photographs taken during the flaming phase of the experiment (Fig. 1), indicating that wind speed was not high and no strong temperature inversion was present to inhibit the rising smoke. That the wind speed was low is also shown by measurements at the SMEAR II 73m mast. At the ignition time, wind speed was<2ms−1at all altitudes of the tower, but it increased to 2–4ms−1during the morning (Fig. 3a). www.atmos-chem-phys.net/14/4473/2014/ Atmos. Chem. Phys., 14, 4473–4502, 2014 4488 A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland Fig. 8. Relationships of excess concentrations (1X) of organic trace gases to excess carbon monoxide (1CO) concentrations in the smoke plume peaks presented in Fig. 7. mode occurred at 106nm, an accumulation mode occurred at 264nm, and a coarse mode occurred at 3.6µm. The integrated mass concentration was 9.4µgm−3for Dp<10µm. The size distribution at 08:00EET is the clearest one obtained from the smoke plume during the flaming phase. In the number size distribution, there were four modes, the largest of which was at Dg=80nm. The geometric standard deviations (i.e., the widths of the modes) were quite small, ranging from 1.15 to 1.25, so the fitting was done by assuming that instead of the three largest modes there is only one large mode with Dg=81nm, σg=1.58 (the dashed line at 08:00). In the volume size distribution, there were four modes, the highest concentration of which was at Dg=153nm. The integrated mass concentration was 21.6µgm−3, the highest during the flaming phase. At 09:20EET, there was a very clear nucleation mode at Dg=8.8nm, simultaneously with the high positive and negative air ion concentration in the sub-10nm size range (Fig. 9). At this time, the number concentrations in the Aitken and accumulation modes were lower than in the smoke plume size distribution at 08:00EET and not very different from those in the background size distribution at 07:50EET; the mass concentration of 8.9µgm−3was actually lower than at 07:50EET. Therefore, it is reasonable to interpret this size distribution as representing natural new particle formation that is frequently observed at SMEAR II during sunny days (Dal Maso et al., 2005). The size distribution at 12:40EET was measured from the thickest smoke plume arriving at the aerosol cottage during the smoldering phase. It is worth noting that already an hour earlier at 11:41 (Fig. 2h) the visual smoke emissions from the burn were clearly smaller than during the flaming phase (Fig. 2a–f). At 12:40, CO also reached the maximum concentration (Fig. 3e), so the timing of these maxima suggests that this part of the otherwise very patchy plume was wide. In this size distribution, the integrated mass concentration of 28.6µgm−3was the largest observed in the aerosol cottage during the experiment. In this number size distribution, the largest mode was at Dg=79nm, essentially the same size as in the flaming-phase-plume size distribution at 08:00EET, but the accumulation mode Dg=244nm was larger than that in the flaming phase size distribution. The volume size distribution at 12:40EET was clearly different from that during the flaming phase at 08:00EET. First, the mode with the largest concentration was at Dg=318nm whereas at 08:00EET, it was at Dg=153nm. Second, in the smoldering-phase-plume volume size distribution, the contribution of the coarse-mode particles was much higher than in the flaming-phase-plume size distribution. Actually, the broad shape of the supermicron size distribution and the high σg=2.4 suggest there were even more modes in the coarse sizes. At 13:40EET, another smoke plume was observed at the aerosol cottage, again simultaneously with a CO peak at the mast. This was at a time when little or no visible smoke was observed at the burned site. The number size distribution was narrower with the largest mode at Dg=122nm. The volume size distribution also had two clear accumulation modes and Atmos. Chem. Phys., 14, 4473–4502, 2014 www.atmos-chem-phys.net/14/4473/2014/ A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland 4489 a) b) c) d) e) f) Fig. 9. Data of selected ground-based aerosol measurements at SMEAR II on 26 June 2009. (a) Particle number size distributions measured with the APS and the DMPS, and the positive and negative air ion size distributions measured with the NAIS. (b) Total particle number concentrations measured with a CPC and integrated from the DMPS. (c–e) Air ion number concentrations from the NAIS data in three different size ranges. (f) Concentrations of organics and the sum of all compounds measured with an AMS, mass concentration of concentration of particles smaller than 10µm calculated from the size distributions measured with the DMPS and the APS using the density of 1.5gcm−3, and black carbon concentration at two locations. www.atmos-chem-phys.net/14/4473/2014/ Atmos. Chem. Phys., 14, 4473–4502, 2014 4490 A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland Fig. 10. Selected particle size distributions measured with DMPS and APS in the aerosol cottage. The left column: number size distributions; right column: volume size distributions. In each plot, the grey line represents the measured size distribution and the associated numbers: the number concentration in cm−3(left column), the mass concentration integrated to 10µm inµgm−3calculated assuming a density of 1.5gcm−3(right column), and the geometric mean diameter (Dg) of the whole size distribution. The modal parameters are the geometric mean diameter, the geometric standard deviation, and the number or volume concentration of the mode. The grey shaded band in the smoke size distributions shows the range of count median diameters (CMD) and volume median diameters (VMD) of smoke aerosol from temperateforest prescribed fires and wildfires in the review by Reid et al. (2005a). a broad coarse particle size distribution. The fast passage of the smoke plume creates uncertainty regarding the modal parameters since the smoke plume passages were shorter than the time used for scanning one size distribution. Nevertheless, in both of the size distributions that were measured during the smoldering phase, the mass size distribution had much larger modes than during the flaming phase. The mode diameters of the size distributions observed during the flaming phase were clearly smaller than those measured by Radke et al. (1991) and those presented in the papers reviewed by Reid et al. (2005a). In the smoldering phase of our experiment, the accumulation mode of the particle size distributions was in the same range as those in Reid et al. (2005a). In contrast to our observations, Hobbs et al. (1996) found that the mode diameter of the size Atmos. Chem. Phys., 14, 4473–4502, 2014 www.atmos-chem-phys.net/14/4473/2014/ A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland 4491 a) b) c) d) Fig. 11. Optical properties of aerosol observed in the aerosol cottage: (a) scattering and absorption coefficients at λ=550nm; (b) single-scattering albedo at λ=550nm; (c) Ångström exponent of scattering over 450nm–700nm (αsp) and backscatter fraction bat λ=550nm; (d) intrinsic aerosol forcing efficiency for two surface reflectances (Rs). distribution was smaller in the plume during the smoldering phase than during the flaming phase. 3.4.2 Aerosol optical characterization In the first smoke plume observed during the flaming phase, the light scattering coefficient (σsp) at λ=550nm was 127Mm−1(Fig. 11). Because the mass concentrations obtained from the combined DMPS+APS data are available every 10min, scattering data, which are available at 5min intervals, were averaged over 10min for comparison. The peak σsp in the first smoke plume passage was 93.8Mm−1, whereas the mass concentration in the size range Dp<10µm was 21.6µgm−3(Fig. 10, volume size distribution at 08:00EET), which yields a mass scattering efficiency of 4.3m2g−1. The highest 5min-averaged σsp = 137Mm−1was observed in the smoldering phase at the time the mass concentration reached the maximum value of 28.6µgm−3(Fig. 10). The corresponding 10min-averaged σsp =116.5Mm−1resulted in a mass scattering efficiency of 4.1m2g−1. These mass scattering efficiencies are somewhat higher than the value of 3.1±0.9m2g−1that was obtained from the 3yr time series at SMEAR II (Virkkula et al., 2011) and than the median value for the whole burning day that was also 3.1m2g−1, but in good agreement with other published values (e.g., Reid et al., 2005b; Malm and Hand, 2007). Light scattering increased when the smoke plume passed the aerosol cottage, but the absorption coefficient increased only during two short periods in the flaming phase (Fig. 9a),in agreement with laboratory studies showing that Fig. 12. Backscatter fraction bat λ=550nm vs. Ångström exponent of scattering (αsp) in the aerosol cottage before the flaming phase and after the last plume was observed (crosses), during the flaming phase (red circles), and during the smoldering phase (grey circles). The regression line was fitted with the smoldering-phase data. BC is produced in flaming combustion but less in smoldering combustion (e.g., McMeeking et al., 2009). The aerosol was not very dark: the single-scattering albedo ω0is about 0.2±0.1 for pure BC, but during the experiment the lowest ω0was about 0.7 and in the strongest plume during the flaming phase 0.82. These values are in line with the average ω0=0.83±0.11 of the fires studied by Radke et al. (1991). During the smoldering phase, ω0was ≈0.9 and did not deviate from the background values during the smoke plumes (Fig. 9b). That ω0was larger during the smoldering phase is also in agreement with earlier studies (e.g., Hobbs et al., 1996; Reid et al., 2005b). In general, the backscatter fraction bof larger particles is smaller than that of smaller particles, so the size relationship of the backscatter fraction bis qualitatively similar to that of the Ångström exponent of scattering, αsp. This relationship was also observed in the smoke plumes. There were clear differences in αsp and bbetween the flaming and smoldering phases; both parameters were clearly lower in the smoke plumes observed during the latter phase (Fig. 8). In the plumes during the flaming phase, the average αsp and bwere 2.25±0.01 and 0.171±0.001, respectively, and in the smoldering phase 1.56±0.07 and 0.134±0.001, respectively. These observations and the higher contribution of coarse-mode particles in the smoldering phase (Fig. 7) than in the flaming phase are in line with the general picture of the size relationships of both αsp and b. The two parameters were especially well correlated during the smoldering phase (Fig. 9). www.atmos-chem-phys.net/14/4473/2014/ Atmos. Chem. Phys., 14, 4473–4502, 2014 4492 A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland Fig. 13. Vertical profiles of virtual potential temperature (θv), excess CO2and total particle number concentrations during the experiment day. For flight 3, no excess CO2data are shown because they did not deviate from zero. The θvdata are from meteorological soundings from two sites, the Jokioinen Observatory and the Tikkakoski Airport in Jyväskylä. The single-scattering albedo and the backscatter fraction were used for estimating the radiative forcing efficiency 1F/δ from Eq. (7). 1F/δ is negativefor dark surfaces(Rs= 0.15), both during the flaming and smoldering phases, even for the darkest aerosol during the flaming phase (Fig. 8d). For Rs, the value of 0.85 was used to assess the effect of the aerosols above snow surfaces. There, the observed aerosol would have a positive radiative forcing (Fig. 8d). The flaming-phase aerosol would heat the atmosphere much more strongly than the smoldering phase aerosol. To estimate the direct radiative forcing, 1F/δshould be multiplied by the aerosol optical depth δ. However, we do not have any measurement data on the smoke plume optical depth. The sunphotometer that was in the tower east of the aerosol cottage did not detect the smoke at all even though the MAAP that was at the same location did. The main reason is that most of the smoke plume did not flow between the sunphotometer and the Sun. Another reason is that the sunphotometer made one instantaneous measurement every 15min according to AERONET (AErosol RObotic NETwork) settings and the smoke plume passed by the mast only during short 1–2min periods (Fig. 6). Satellitedatawere also studied for possible signs of the fire plume. The Moderate Resolution Imaging Spectroradiometer (MODIS) onboard the Terra and Aqua platforms made two overpasses of the site shortly after the fire at 11:40EET (Terra) and 11:55EET (Aqua). The 250m resolution visible wavelength (MODIS bands 1 and 2) and 1km resolution thermal infrared channels data were inspected, but no signs of the fire plume were observed. The 10×10km MODIS Aerosol Optical Depth (AOD) product is not elevated near the site (AOD<0.05). The Multi-angle Imaging SpectroRadiometer (MISR) 275m resolution data were also checked for the same Terra overpass, with no sign of the plume. The satellite overpasses missed the strongest flaming phase, and the resolution is not high enough to catch the remaining plume. 3.5 Observations on mobile platforms There were three different types of mobile measurements: the research aircraft, Sniffer and the portable particle counters. Here we discuss observations that were aimed at studying the horizontal and vertical dispersion of particles. Three research flights were conducted during the day. The plan was to fly through the smoke plume at several altitudes up to about 3000m above ground level. The first flight was flown during the flaming phase, the second flight was flown during the clear smoldering phase, and the last flight was flown when no smoke was observed on the ground. During flight 1 (07:40–10:15EET), elevated particle number and CO2concentrations indicated the smoke plume up to an altitude of about 1500m a.m.s.l. (Fig. 13). This altitude was 200m lower than the stable layer (i.e., where the virtual potential temperature θvincreased dramatically with height) in the routine meteorological sounding at the Tikkakoski Airport in Jyväskylä, about 90km northeast from Hyytiälä, at 08:00EET at the beginning of the flaming phase. Therefore, the smoke rose up to about the top of the boundary layer, but not above it. We flew through the smoke plume at several altitudes and analyzed the 25 most obvious plume crossings and determined the plume width from the particle number concentration data. With a flight speed of 38±2ms−1, at about 100m above ground level, the plume was crossed in about 3– 4s, yielding a plume width of about 120m. At 1200m above ground level, the plume width was 800±100m. This analysis is found in the companion paper by Virkkula et al. (2013). Atmos. Chem. Phys., 14, 4473–4502, 2014 www.atmos-chem-phys.net/14/4473/2014/ A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland 4493 a) b) Fig. 14. Horizontal spread of the smoke plume (a) at ground level as observed with the portable CPCs and (b) as observed in the aircraft during flight 1. In (a) both the diameter and the symbol color are scaled according to the concentration. In (b) the color scale is for the concentration and the circle size is scaled according to the diameter of the plume. The concentrations in (a) are all concentrations larger than 104cm−3and in (b) the maximum concentrations in the plume crossings. During flight 2 (11:05–13:40EET), the boundary layer depth had increased to about 2300m a.m.s.l. according to the sounding at the Jokioinen Observatory at 14:00EET (Fig. 13). We observed elevated number concentrations at 2000m a.m.s.l., but CO2concentrations did not exceed background concentrations at any level. During flight 3 (15:50– 17:55EET), particle number concentrations exceeded background concentrations up to about 1000m a.m.s.l., but CO2 concentration did not rise above the background concentration. The horizontal dispersion of the smoke plume is visualized by plotting the concentrations measured with the portable CPCs on the ground and with the CPC in the aircraft as a function of the latitude and longitude. For the groundlevel measurements, Fig. 14a shows the paths the pedestrians walked and the concentrations at those locations where the concentration was larger than 10000 particlescm−3. For the airborne measurements, Fig. 13b shows the locations, the maximum concentrations and the widths of the plumes in the 27 plume passages mentioned above. Both at ground level and aloft, the plume was transported in the direction of the average wind direction at 73m in elevation measured at the SMEAR II mast. The location and concentration data from ground-based and airborne measurements were used to estimate the decreaseof the concentrations as a function of the distance from the center of the burn area (Fig. 15). The three-dimensional distance was calculated from the center of the burn area to the point location of the measurement. The pedestrian data were arranged in 100m distance bins, and the maximum of each of these was used for the calculations. The three data points from Sniffer are the average concentrations over 3–5min at distances of 120m, 180m and 250m downwind and 250m downwind from the edge of the burned area. For the airborne data, the maxima of each plume crossing were used. Fig. 15. Particle number concentration as a function of 3dimensional distance from the center of the burning area measured in the Cessna, with the portable CPCs on the ground (walking, NW), in the aeroplane (Cessna, NC), and on the ground in the van (Sniffer,NS). The walking data points are the maximum concentrations of each 100m distance bin, the aircraft data points are the maxima in each plume crossing and the Sniffer data points the 3–5min average concentrations measured at three standing locations of the van. The SMEAR II data point denotes the maximum 1min particle number concentration in the aerosol cottage during the experiment. The solid lines denote fittings of the exponential function N0e−kx and the dashed line the fitting of a power law N0x−b, where xis the distance from the center of the burn area. The highest particle number concentration, 1.6×106cm−3, was measured with the ELPI in Sniffer at 120m from the burn area. The highest particle number concentration from the research aircraft, 0.94×106cm−3, was measured at a height of 118m above ground. The concentration was probably higher, because the particle counter used in the study saturates at 1×106cm−3. The portable CPC model 3007 used by the pedestrians may also have been saturated, with saturation at 1×105cm−3, and, www.atmos-chem-phys.net/14/4473/2014/ Atmos. Chem. Phys., 14, 4473–4502, 2014 4494 A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland with the diluters, the maximum concentration measured was ∼3×105cm−3. The maximum 1min particle number concentration in the aerosol cottage during the experiment was 3.3×104cm−3, an order of magnitude lower than the maxima measured by the pedestrians at some hundreds of meters further away from the burning area. To get a quantitative estimate of the decrease of the number concentrations, an exponential function N=N0e−kx was fitted to the data, where N0is the background concentration, xis the distance from the center of the burn area, and kis the reciprocal of the e-folding distance k−1. The fittings yielded e-folding distances of 98m, 417m and 833m for the Sniffer, walker and aircraft data, respectively. The background particle number concentration at SMEAR II on that day was in the range of 1000–3000cm−3(Fig. 9b). This concentration was reached at ∼2km from the burn area at ground level and at ∼5km from the burn area in the airborne measurements. However, within approximately 1km distance from the burn area (Fig. 15), the exponential form did not fit the aircraft data quite as well as a power law N0x−b. This suggests that a Gaussian formula is not necessarily the best option for describing the dilution of smoke in the immediate vicinity of forest fires. 3.6 Changes in soil properties, greenhouse gas and VOC fluxes The clear-cutting and burning of slash increased soil temperature and moisture, soil pH, and NH4-N and NO3-N concentrations (Kulmala et al., 2014). The increase in the topsoil soil pH and mineral nitrogen concentrations (NH4-N and NO3-N) were rapid, response times ranging from days to a few months, whereas the changes in the deep-soil pH and nitrogen contents were much smaller and were observed with a delay of one to two years. In contrast, the total available nitrogen concentration did not increase after the clear-cutting and burning, but the proportion of mineral nitrogen (NH4-N and NO3-N) of the total increased dramatically (Kulmala et al., 2014). The rates of soil CO2efflux at the sites were similar before the clear-cut and burning of slash. After the treatment, the efflux at the burned site was approximately only half of the flux at the control site (Kulmala et al., 2014). Two years later, the difference between the burned clear-cut and the mature control forest had decreased. The cumulative soil CO2emissions during 2009–2011, interpolated from the chamber measurements and excluding winter months, were the highest at the clear-cut but not at the burned site. Nevertheless, taking into account the rapid CO2release during the burning, the burned clear-cut site had the highest CO2emissions over the 3yr period (Kulmala et al., 2013) Before and after the clear-cut and burning, all three sites acted as CH4sinks. Similar to the CO2exchange, the soil uptake of CH4decreased significantly soon after the burning. Burning did not seem to have a long-term effect on soil CH4uptake as the differences between the three sites disappeared during the following years (Kulmala et al., 2013). The decrease in soil CH4uptake after clear-cutting and burning may be related to changes in soil temperature and moisture and to the increased soil NH4-N content, as mineral nitrogen in the soil may inhibit CH4oxidation (Saari et al. 1997, 2004; Maljanen et al., 2006). The soil VOC emissions were generally low compared to the total VOC emissions from similar forest ecosystems. This result is also in line with the general finding that the forest floor makes up a maximum of ∼10% of the total ecosystem VOC emissions in coniferous forest (Hayward et al., 2001; Hakola et al., 2006; Taipale et al., 2008, 2009). The VOC fluxes between the chambers differed greatly at the burned site, which is a phenomenon often observed with forest-floor VOC flux measurements (Aaltonen et al., 2011). After one year, the emissions of VOCs had nearly stabilized close to the level before the burning. 4 Summary and conclusions The general goal of the prescribed fire experiment on 26 June 2009 in Hyytiälä, Finland, was to collect data for estimating the effect of prescribed fires burning slash fuels on air quality and climate. The experiment was designed from the beginning to be multidisciplinary, and it had several more detailed goals: to obtain emission factors of aerosols and gases from boreal wildfires, to characterize the climatically relevant physical properties of the smoke aerosol, to quantify the connections between ground-based smoke observations and satellite remote sensing, to obtain data for testing an improved model of atmospheric dispersion of the fire plume, and to quantify the changes taking place in soil carbon stocks and greenhouse gas fluxes following clear-cutting and prescribed burning. In the campaign, a 0.8ha region of forest near the SMEAR II was cut clear, and some tree trunks, all treetops and all branches were left on the ground and burned. The amount of burned organic material was estimated to be about 46.8 tons (i.e., about 60tonsha−1), of which 64% consisted of the cut tree material, 32% of organic litter and humus layer and 4% of surface vegetation. During the burning, various measurements were conducted on the ground with both fixed and mobile instrumentation, and from a research aircraft. Most of the time the smoke was not transported to the SMEAR II station due to the low wind speed or calm meteorological conditions that were associated with substantial, sudden variations of the wind direction. The low wind speeds in combination with the substantial buoyancy of the fire plumes resulted in an almost vertical rise of a substantial fraction of the effluents. The fire was started when the wind was from the right direction in terms of the main measuring stations, but the wind direction soon turned. The ideal wind direction to bring smoke to Atmos. Chem. Phys., 14, 4473–4502, 2014 www.atmos-chem-phys.net/14/4473/2014/ A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland 4495 SMEAR II would have been 190±10◦, whereas the average wind direction was 134–140◦at all levels of the SMEAR II 73m mast. Therefore, the smoke plumes were located west of the station, and the smoke reached the sample line inlets of the instruments at SMEAR II only during short periods. In a related article (Kukkonen et al., 2014), we have presented an overview of a mathematical model, BUOYANT, that was designed for the evaluation of the dispersion of buoyant plumes. We have compared model predictions with the data of (i) the SCAR–C experiment in Quinault (US) in 1994 (e.g., Kaufman et al., 1996; Hobbs et al., 1996; Gassó and Hegg, 1998), and (ii) the prescribed burning experiment addressed in this article. In the case of the latter experiment, the model predictions were compared with plume elevations and diameters, determined based on particulate matter number concentration measurements on board an aeroplane. The agreement of modeled and measured results was good for both of these experiments, provided that in the case of the Hyytiälä prescribed burning, one assumes the measured maximum convective heat fluxes as input data for the model. The results demonstrated that in both cases there were substantial uncertainties in estimating (i) the source terms for the atmospheric dispersion computations, and (ii) the relevant vertical meteorological profiles. The burnt area in the Quinault experiment was substantially larger, approximately 20 ha, than the 0.8ha in our experiment. Correspondingly, the maximum convective heat flux in the Quinault experiment was clearly higher than that in the Hyytiälä experiment (Kukkonen et al., 2014). The meteorological conditions were also substantially different in these two experiments; there was an elevated inversion in the case of the Quinault experiment. The plume in the Hyytiälä experiment ascended to higher altitudes compared with that in the Quinault experiment, according to both the measurements and the model predictions (Kukkonen et al., 2014). This was mainly caused by the different vertical structure of the atmosphere, especially the temperature inversion in the Quinault case. Despite the changing of the wind direction and the intensive plume rise, concentrations originating from the fire were detected both from within the burn area and part of the time also at the ground-based stations. In the middle of the burning area, CO2concentration peaks were around 2000– 3000ppm above the baseline, and peak vertical flow velocities were 6±3ms−1. The meteorological stations placed near the perimeter of the burn area produced data for the analysis of fire dynamics. A strong rise in moisture was observed in the plume, which has been suggested as possibly modifying plume dynamics. The concentrations of the trace gases O3, NOx, SO2, CO and CO2, which are routinely measured from six different altitudes in the mast, should be elevated in a biomass-burning plume. The most distinct exceedances above the background values were for CO, NOxand SO2, but no obvious smokeplume-related variations were observed for O3and CO2. The lack of a signal in the CO2measurements may indicate that the sensitivity or the response time of the CO2monitor was not sufficient. Even though the CO2concentrations did not rise, there were several other indicators of smoke arriving from the burning biomass: elevated particle number concentrations, higher scattering coefficients, elevated CO concentrations, and elevated concentrations of many VOCs that are known to be emitted during biomass burning. They were detected almost simultaneously with the elevated CO concentrations, so linear regressions were calculated between excess concentrations of VOCs and CO. For most of the compounds, the emission ratios (defined as 1X/1CO) in the present study are larger than those in other studies presented on boreal forest fires. The values for formaldehyde were smaller, and values for monoterpenes were close to other published ratios. The reason for the larger emission ratios may either be a true difference or artefact due to the non-collocated measurements of the VOCs and CO. If it were a true difference, it would mean that the biomass burned is different from that burned by Akagi et al. (2011) and Simpson et al. (2011). Further studies are needed to confirm either of these hypotheses. Peak particle number concentrations were approximately 1–2×106cm−3in the plume at the distance of 100–200m from the burn area on the ground and in the research aircraft. At SMEAR II, the total particle number concentrations increased from about 1000–2000cm−3before the smoke arrived at the instrumentation to about 30000cm−3within the plume. The air ion measurements showed that cluster-mode and intermediate-mode ions were depleted in the strongest smoke plume passages, suggesting that the ions were attached to the larger aerosols in the plume. The maximum particle mass concentrations in the smoke plume observed in the aerosol cottage were 21.6µgm−3and 28.6µgm−3during the flaming and smoldering phases, respectively. It is worth noting that the latter peak was observed simultaneously with the highest CO peak and at time when the visual smoke emissions from the burn were clearly smaller than during the flaming phase. In the number size distribution, there were 3–4 modes. The geometric mean diameter of the mode with the highest concentration was 80±1nm during the flaming phase and in the middle of the smoldering phase. The mode diameters of the size distributions observed during the flaming phase were clearly smaller than those presented from similar experiments. A probable explanation is that the fires in the compared experiments were considerably larger than ours. For instance, Radke et al. (1991) measured size distributions in an aircraft in smoke plumes from prescribed fires where the burned areas were in the range of 10–700ha (i.e., one to three orders of magnitude larger than in our experiment), and Hobbs et al. (1996) measured within fires of 20–40ha. In those fires, the amount of burned biomass was much larger than in our experiment, so the smoke definitely also contained more condensable material to grow particles. At the end of the smoldering phase, the largest mode was 122nm, larger than during the flaming phase. In the volume size www.atmos-chem-phys.net/14/4473/2014/ Atmos. Chem. Phys., 14, 4473–4502, 2014 4496 A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland distributions, the geometric mean diameter of the largest volume mode was 153nm during the flaming phase and 300nm during the smoldering phase. In contrast to our observations, Hobbs et al. (1996), during the SCAR–C project, and Payne et al. (2004), during the ICFME, found that the mode diameter of the size distribution was smaller in the plume during the smoldering phase than during the flaming phase. Both of these observations were made in airborne measurements. One possible explanation to this very obvious discrepancy is that during the flaming phase the particles in the plume that was detected in the aerosol cottage were smaller than those in the main plume that ascended with the hot air. Unfortunately, it is not possible to prove or disprove this hypothesis because in our aircraft we did not measure particle size distributions. The aerosol optical properties were in agreement with other published data from wildfire smoke aerosols. The particles were not dark; the lowest single-scattering albedo of the ground-level measurements was 0.7 in the flaming-phase plume and ∼0.9 in the smoldering phase. The mass scattering efficiency of 4.1–4.3m2g−1at λ=550nm is also in agreement with other published wildfire aerosol data. There were changes in soil physical and chemical properties, which influenced the soil greenhouse gas (CO2, CH4) fluxes for several years after the burning. The VOC fluxes were generally low and consisted mainly of monoterpenes, but a clear peak was observed after the burning. One year after the burning, the fluxes had nearly stabilized close to the level before the burning. The discussion above shows that some of the goals of the experiment were reached, some were not. The emission factors of aerosols and gases could not be obtained because most of the smoke did not reach the instrumentation. The climatically relevant physical properties of the smoke aerosol were partially obtained, but the lack of aerosol optical depth data and hygroscopic growth factors prevent an estimate of the smoke-aerosol–climate-forcing effect. The quantification of the connections between ground-based smoke observations and satellite remote sensing failed because the smoke was not detected in any satellite images. The MODIS onboard the Terra and Aqua platforms made two overpasses of the site shortly after the fire but no signs of the forest fire plume were observed. This is not a unique case: the global number of fires so small that they cannot be observed with MODIS is not well known but their contribution to biomass burning emissions has lately been estimated to be considerable (Randerson et al., 2012). The plan was successful in that we obtained data for testing and improved models of atmospheric dispersion of the fire plume, data on the recovery of the forest after burning, and data of the changes taking place in soil-carbon stocks and greenhouse gas fluxes following clear-cutting and prescribed burning. The airborne measurements have already been used for evaluating and refining of a plume-rise model (Kukkonen et al., 2014). The data are available also for the evaluation of other corresponding models. The experiment taught us several lessons. First, the selection of the meteorological conditions is critical for the success of the experiment. In particular, a higher wind speed that would result in a more inclined plume would be more suitable with regard to the ground-based measurements. However, we had to set an upper limit of 5m/s for the wind speed at 10m height, for safety reasons. There were also several other requirements, especially regarding the wind direction, that limited the choice of a suitable date. We selected the burning site so that the measurement stations were almost in the direction of the prevailing winds (southwesterly). According to the climatological survey conducted before the experiment, favorable conditions would most likely occur several times within the selected two-month period (May and June of 2009). However, unluckily, the acceptable conditions did not occur until 26 June. Second, we would recommend more mobile platforms to ensure good measurements of the fire plumes, even in situations in which the wind direction unexpectedly changes. Airborne measurements of particles and trace gases are especially useful for measuring the dispersion of strongly buoyant, almost vertically rising smoke plumes released from wildfires and prescribed fires. In large natural wildfires, a wide range of aircraft with instruments for measuring aerosol optical properties, size distributions, chemical composition and trace gases can be used for such measurements. However, in a small prescribed fire, such as the present experiment, the plume was so narrow that even a small aircraft flew through it in about 3s at the lowest altitude levels and in about 30 seconds even at the widest plume crossings. The aircraft data obtained was useful; however, the short measurement time and the small aircraft size also had limitations. For instance, the above-mentioned time intervals are too short for measuring some advanced properties, for example the hygroscopic growth of particles. Using a helicopter would enable us to fly more slowly through the smoke plume; however, a helicopter also needs to have a sufficient forward speed to avoid the rotor downwash effects (e.g., Cofer III et al., 1998). Third, in fire experiments the response time of all measurements should be set as fast as possible. In our experiment some trace gas and aerosol instruments were not operating at the fastest response time and this compromised some of our results. Fourth, visual documentation of the phases of prescribed fires is important for the interpretation of the experiment. Even standard photographs proved to be valuable but more information could have been obtained with video surveillance at both visible and infrared wavelengths. In our experiment it was not arranged but it would have given better information for instance on flame height, fire spread, and temperature. Atmos. Chem. Phys., 14, 4473–4502, 2014 www.atmos-chem-phys.net/14/4473/2014/ A. Virkkula et al.: Prescribed burning of logging slash in the boreal forest of Finland 4497 Acknowledgements. The financial support by the Academy of Finland as part or the Centre of Excellence program (project no. 1118615) and as part of the IS4FIRES project (decision no. 122870) is gratefully acknowledged. The experiment was also supported by the European Commission 6th Framework Program Project (EUCAARI), contract 036833-2; the European Research Council; the University of Helsinki; the Finnish Meteorological Institute; the Academy of Finland; the TEKES (project KASTU, decision no. 40208/08 and KASTU-2, decision no. 40393/10); San José State University, CA, USA; and the Institute for Tropospheric Research, Leipzig, Germany. J. Pumpanen was funded by Academy of Finland project number 218094. Partial funding for D. M. Schultz came from Vaisala Oyj. We gratefully acknowledge the help given by the personnel of SMEAR II. Edited by: S. M. Noe References Aalto, P., Hämeri, K., Becker, E., Weber, R., Salm, J., Mäkelä, J. M., Hoell, C., O’Dowd, C.D., Karlsson, H., Hansson, H.-C., Väkevä, M., Koponen, I. K., Buzorius, G., and Kulmala, M.: Physical characterization of aerosol particles during nucleation events, Tellus, 53B, 344–358, 2001. Aaltonen, H., Pumpanen, J., Pihlatie, M., Hakola, H., Hellén, H., Kulmala, L., Vesala, T., and Bäck, J.: Boreal pine forest floor biogenic volatile organic compound emissions peak in early summer and autumn, Agr. Forest Meteorol., 151, 682–691, 2011. Achtemeier, G. L.: Measurements of moisture in smouldering smoke and implications for fog, Int. J. Wildl. Fire, 15, 517–525, 2006. Akagi, S. K., Yokelson, R. J., Wiedinmyer, C., Alvarado, M. J., Reid, J. S., Karl, T., Crounse, J. D., and Wennberg, P. O.: Emission factors for open and domestic biomass burning for use in atmospheric models, Atmos. Chem. Phys., 11, 4039–4072, doi:10.5194/acp-11-4039-2011, 2011. Alexander, M. E.: Prescribed burn documentation and fire danger ratings: a case study, edited by: Viegas, D. X., Proceedings of the 5th International Conference on Forest Fire Research, 27– 30 November 2006, Figueira da Foz, Portugal, Elsevier, Amsterdam, 12 pp., 2006. Alexander, M. E.: Fire behavior in moderately heavy logging slash: documenting the past with photographs, edited by: D. X. Viegas, Proceedings of the 6th International Conference on Forest Fire Research, 15–18 November 2010, Coimbra, Portugal. University of Coimbra, Coimbra, Portugal. 12 pp., 2010. Alexander, M. E. and Thomas, D. A.: Prescribed fire case studies, decision aids, and planning guides, Fire Management Today, 66, 5–20, 2006. Alexander, M. E., Stefner, C. N., Mason, J. A., Stocks, B. J., Hartley, G. R., Maffey, M. E., Wotton, B. M., Taylor, S. W., Lavoie, N., and Dalrymple, G. N.: Characterizing the jack pine – black spruce fuel complex of the International Crown Fire Modelling Experiment (ICFME). Canadian Forest Service Northern Forestry Centre Information Report NOR-X-393, 2004. AMAP2011:TheImpactofBlack CarbononArctic Climate,edited by: Quinn, P. K., Stohl, A., Arneth, A., Berntsen, T., Burkhart, J. F., Christensen, J., Flanner, M., Kupiainen, K., Lihavainen, H., Shepherd, M., Shevchenko, V., Skov, H., and Vestreng, V., Arctic Monitoring and Assessment Programme (AMAP), Oslo, 72 pp., 2011. Andreae, M. and Merlet, P.: Emission of trace gases and aerosols from biomass burning, Global Biogeochem. Cy., 15, 955–966, 2001. Andrews, E., Sheridan, P. J., Fiebig, M., McComiskey, A., Ogren, J. A., Arnott, P., Covert, D., Elleman, R., Gasparini, R., Collins, D., Jonsson, H., Schmid, B., and Wang, J.: Comparison of methods for deriving aerosol asymmetry parameter. J. Geophys. Res., 111, D05S04, doi:10.1029/2004JD005734, 2006. Arnott, W. P., Hamasha, K., Moosmüller, H., Sheridan, P. J., Ogren, J. A.: Towards aerosol light-absorption measurements with a 7wavelength aethalometer: evaluation with a photoacoustic instrument and 3-wavelength nephelometer, Aerosol Sci. Technol. 39, 17–29, 2005. Asmi, E., Sipilä, M., Manninen, H. E., Vanhanen, J., Lehtipalo, K., Gagné, S., Neitola, K., Mirme, A., Mirme, S., Tamm, E., Uin, J., Komsaare, K., Attoui, M., and Kulmala, M.: Results of the first air ion spectrometer calibration and intercomparison workshop, Atmos. Chem. Phys., 9, 141–154, doi:10.5194/acp-9-141-2009, 2009. Bond, T. C. and Bergstrom, R. W.: Light absorption by carbonaceous particles: An investigative review, Aerosol Sci. Technol., 40, 27–67, 2006. Bond, T. C., Doherty, S. J., Fahey, D. W., Forster, P. M., Berntsen, T., DeAngelo, B. J., Flanner, M. G., Ghan, S., Kärcher, B., Koch, D., Kinne, S., Kondo, Y., Quinn, P. K., Sarofim, M. C., Schultz, M. G., Schulz, M., Venkataraman, C., Zhang, H., Zhang, S., Bellouin, N., Guttikunda, S. K., Hopke, P. K., Jacobson, M. Z., Kaiser, J. W., Klimont, Z., Lohmann, U., Schwarz, J. P., Shindell, D., Storelvmo, T., Warren, S. G., and Zender, C. S.: Bounding the role of black carbon in the climate system: A scientific assessment, J. Geophys. Res., 118, 5380–5552, doi:10.1002/jgrd.50171, 2013. Bowman, D. M. J. S., Balch, J. K., Artaxo, P., Bond, W. J., Carlson, J. M., Cochrane, M. A., D’Antonio, C. M., DeFries, R. S., Doyle, J. C., Harrison, S. P., Johnston, F. H., Keeley, J. E., Krawchuk, M. A., Kull, C. A., Marston, J. B., Moritz, M. A., Prentice, I. C., Roos, C. I., Scott, A. C., Swetnam, T. W., van der Werf, G. R., and Pyne, S. J.: Fire in the Earth System, Science 324, 481–484, 2009. Bowman, D., Balch, J., Artaxo, P., Bond, W., Cochrane, M., D’Antonio, C., DeFries, R., Johnston F., Keeley, J., Krawchuk, M., Kull, C., Mack, M., Moritz, M., Pyne, S., Roos, C., Scott, A., Sodhi, N., and Swetnam, T.: The human dimension of fire regimes on Earth, J. Biogeogr., 38, 2223–2236, 2011. Cahill, C. F., Cahill, T. A., and Perry, K. D.: The sizeand timeresolved composition of aerosols from a sub-Arctic boreal forest prescribed burn, Atmos. Environ., 42, 7553–7559, 2008. Clements, C. B., Potter, B. E., and Zhong, S.: In situ Measurements of Water Vapor, Heat and CO2Fluxes within a prescribed Grass Fire. International Journal of Wildland Fire, 15, 299–306, 2006. Clements, C. B., Zhong, S., Goodrick, S., Li, J., Bian, X., Potter, B. E., Heilman, W. E., Charney, J. J., Perna, R., Jang, M., Lee, D., Patel, M., Street, S., and Aumann, G.: Observing the dynamics of wildland grass fires: FireFlux – a field validation experiment, B. Am. Meteorol. Soc., 88, 1369–1382, 2007. www.atmos-chem-phys.net/14/4473/2014/ Atmos. Chem. Phys., 14, 4473–4502, 2014