Emissions from prescribed burning of timber slash piles in Oregon

Abstract

Emissions from burning piles of post-harvest timber slash (Douglas fir) in Grande Ronde, Oregon were sampled using an instrument platform lofted into the plume using a tether- controlled aerostat or balloon. Emissions of carbon monoxide, carbon dioxide, methane, particulate matter (PM_2.5), black carbon, ultraviolet absorbing PM, elemental/organic carbon, filter-based metals, polycyclic aromatic hydrocarbons (PAHs), polychlorinated dibenzodioxins/dibenzofurans (PCDD/PCDF), and volatile organic compounds (VOCs) were sampled to determine emission factors, the amount of pollutant formed per amount of biomass burned. The effect on emissions from covering the piles with polyethylene (PE) sheets to prevent fuel wetting versus uncovered piles was also determined. Results showed that the uncovered (“wet”) piles burned with lower combustion efficiency and higher emission factors for VOCs, PM_2.5, PCDD/PCDF, and PAHs. Removal of the PE prior to ignition, variation of PE size, and changing PE thickness resulted in no statistical distinction between emissions. Results suggest that dry piles, whether covered with PE or not, exhibited statistically significant lower emissions than wet piles due to better combustion efficiency.

INTRODUCTION

To reduce wildfire risk and to improve timber forest productivity and health, woody biomass fuels from selective thinning and timber harvests are mechanically treated and piled for burning^{1, 2} This practice is becoming more prevalent as prescribed fire complexity and risk associated with elevated fuel levels (proximity to the wildland/urban interface, smoke effects on air quality and respiratory health) limit the use of broadcast prescribed burning³. Pile burning mitigates concerns about fire safety and air quality as it allows managers to burn under optimal weather conditions and with reduced staffing levels³. Biomass pile burns are often the most economical way to dispose or utilize the biomass due to collection, transportation, and end-product processing costs⁴. Depending on the season and rainfall history, burn piles can smolder for days after they are lit resulting in significant quantities of air pollution⁴. To promote pile combustion, the biomass is preferably dry, resulting in faster, hotter, and more efficient burns, presumably with less pollutants. Common practice involves covering these large piles with polyethylene (PE) film until burn conditions are optimal to prevent moisture saturation during the rainy season. This has raised some questions about emissions from the burning plastic film. The Oregon Department of Forestry (ODF) has used small amounts of PE film sheeting (9.3 m²) per pile through administrative rulemaking (OAR 629–048-0210)⁵. Often this is not enough to keep piles dry for efficient consumption after significant rainfall. Because of this limitation, ODF is seeking data to determine whether or not larger and thicker coverings of PE have deleterious effects on burn emissions.

Only a few studies⁶ have investigated pile burn emissions in the field and often the number of pollutants characterized was limited ^{6, 7} Laboratory burns of pinus ponderosa slash (twigs, needles, and small branches) by Yokelson et al.⁸ characterized emissions from burn piles (1 m x 2 m) using FTIR analysis. Their work determined emission factors for smoldering/flaming phase as partitioned by modified combustion efficiency. Other work⁹ examined emissions from 2 kg mixtures of manzanita stick wood (Arctostaphylos sp.) with 0, 5, and 50 g of shredded low density PE but found no statistical effect of increase PE content on over 190 compounds.

To complement the laboratory scale work previously done on assessing potential contribution of PE to biomass emissions, this work aimed to characterize and compare emissions from burning woody biomass piles, including dried PE-covered piles and wetted piles, in a large-Scale Field Application

METHODS

Biomass piles

Tests were conducted during mid-October in western Oregon, on a timber-harvested Douglas fir (Pseudotsuga menziesii) stand (45° 0’ 44.14” N, −123° 41’ 6.49” W) located about 8 km southwest of Grand Ronde, Oregon and 30 km east of the Pacific coast. The site is at 880 m elevation on a ridge top with an about 10 m change in elevation in the test area. After timber harvesting, the piled material was primarily small branches and limbs of size less than 20 cm in diameter.

Biomass piles approximately 2.5 m high and 5 m in diameter and spaced at least 15 m apart were constructed by the landowner (Figure 1). Three pile types were tested nominally: Dry, Wet, and Dry Polyethylene (PE) covered. Polyethylene sheeting covered eight of the piles throughout the summer to comprise the Dry and PE-covered test piles for the October tests. The PE was removed from four piles prior to testing and were designated Dry piles. The remaining four covered piles were left with the PE in place and were designated Dry PE piles. PE-covered piles had two film thicknesses, 0.10 mm (4 mil) and 0.15 mm (6 mil), and two area sizes, 3.0 m by 3.0 m (10 ft by 10 ft), and 6.1 m by 6.1 m (20 ft by 20 ft) (Table 1). The remaining four piles were uncovered throughout the summer and designated as Wet piles.

Figure 1.

Typical burn pile, uncovered.

Table 1.

Test order and type.

Test Day	Test Order, Type, PE Sizea (if applicable)
Day 1	Burn 1: WET 01
	Burn 2: DRY, PE 6.1×6.1 m, 0.15 mm
Day 2	Burn 3: WET 02
	Burn 4: DRY, uncovered
	Burn 5: DRY, PE 3×3 m, 0.15 mm
Day 3	Burn 6: WET 03
	Burn 7: DRY, uncovered
	Burn 8: DRY, PE 3×3 m, 0.10 mm
	Burn 9: DRY, uncovered
Day 4
	Burn 10: DRY, PE 6.1×6.1 m, 0.15 mm
	Burn 11: DRY, PE 3×3 m, 0.15 mm Ambient background

^aPE = Polyethylene, area in m x m, thickness in mm

Terrain constraints to pile access, a desire to prevent the emissions from upwind smoldering fires from impinging on new burn piles, and effects of week-long meteorological conditions prohibited true random pile testing. The resultant “ordered” testing affects randomness and may have introduced bias into the measurements as a result of dynamic meteorological variables (conditions present at the end of the testing may be different than those at the beginning) confounding the comparisons. Four days of sampling were conducted in later October. Meteorological data for these dates are reported in Supporting Information (SI). The order and notation for the tests are presented in Table 1.

Sampling method

Fires were initiated by drip torch immediately after which emissions were sampled using an aerostat-lofted sampler system (Figure 2) detailed more fully elsewhere^{10, 11}. Briefly, the system consists of a 5 m diameter, helium-filled aerostat, connected with two tethers to all-terrain vehicle (ATV)-mounted winches, upon which is mounted a sampler/sensor system termed the “Flyer”. The Flyer was maneuvered into the burn pile plume by controlling tether length and the location of the ATV-mounted tether winches. Sampling periods consisted of both active flaming and smoldering emissions

Figure 2.

Aerostat with Flyer (Left) and Flyer close up (Right).

Instrumentation on the Flyer

Emission samples were analyzed for carbon monoxide (CO), methane (CH₄), carbon dioxide (CO₂), particulate matter equal to or less than 2.5 μm (PM_2.5), black carbon (BC), ultraviolet absorbing (UVPM), elemental/organic/total carbon (EC, OC, TC), polyaromatic hydrocarbons (PAHs), polychlorinated dibenzodioxins/dibenzofurans (PCDDs/PCDFs), filter-based metals, and volatile organic compounds (VOCs). Targeted emission constituents and their sampling methods are listed in Table 2.

Table 2.

Target pollutants and sampling methods.

Analyte	Method/Instrument	Frequency	Method Reference
CO2	NDIR LICOR-820a	Continuous 1 Hz	U.S. EPA Method 10A19
CO	Electrochemical cell e2V EC4–500-COb	Continuous 1 Hz	U.S. EPA Method 10A19
PM2.5	SKC Impactor, 47 mm filter 2 μm pore size/gravimetric	Batch - 10 L/minc constant flow	40 CFR 50, Appendix J20
PM2.5	DustTrak 8520d	Continuous 1 Hz	Laser optical, factory calibration
PCDD/PCDF/PAHs	Quartz filter/PUF/	Batch - 650 L/min	Modified U.S. EPA
	XAD/PUFe	nominal flowf	Method TO-9A21
VOCs	6 L SUMMA canister	30–60 min integrated sample	U.S. EPA Method TO-1522
CO, CO2, CH4	6 L SUMMA canister	30–60 min integrated sample	Modified U.S. EPA Method 25C23
Black carbon	Aethalometer, AE51g/AE52g	Continuous 1 Hz/0.1 Hz	880 nm by light absorption, factory calibration
UVPM	Aethalometer, AE52g	Continuous 0.1 Hz	370 nm by light absorption, factory calibration
Elemental, organic	SKC Impactor, 47 mm quartz	Batch - 10 L/minc	Modified NIOSH Method
and Total carbon	filter	constant flow	504024

^aLI-COR Biosciences, USA.

^bSGX Sensortech, United Kingdom.

^cLeland Legacy sample pump, SKC Inc., USA.

^dTSI Inc., USA.

^eFilter size 20.3×25.4 cm, Polyurethane foam (PUF) size 7.6×3.8 cm.

^fWindjammer brushless direct current blower AMETEK Inc., USA.

^gAethLabs, USA.

The flyer was equipped with a data acquisition and control program allowing emission samplers to be turned on and off at CO₂ levels above ambient levels (trigger settings: 800 ppm for VOCs and 450 ppm CO₂ for all other emission samplers). The control program was also transmitted to the ground permitting the operator full control of the emission samplers.

The CO₂ analyzer and the CO sensor was calibrated daily in accordance with EPA Method 3A ¹² A precision gas divider Model 821S (Signal Instrument Co. Ltd., England) was used to dilute the high-level span gases for acquiring the mid-point concentrations for CO₂ analyzer and CO sensor calibration curves. The precision gas divider was evaluated in the field as specified in U.S. EPA Method 205 ¹³. The PM_2.5 and EC/OC/TC sample pumps as well as the AE51/AE52 were calibrated with a Gilibrator Air Flow Calibration System (Sensidyne LP, USA) before and after the field campaign. SUMMA canisters were equipped with a manual valve, metal filter (frit), pressure gauge, pressure transducer, and an electronic solenoid valve which enabled the SUMMA to be opened remotely by the ground-based software to maximize desired sample collection and minimize sampling of ambient air.

PCDD/PCDF samples were cleaned and analyzed using an isotope dilution method based on U.S. EPA Method 23¹⁴. Concentrations were determined using high resolution gas chromatography/high resolution mass spectrometry (HRGC/HRMS) with a Hewlett-Packard gas chromatograph 6890 Series coupled to a Micromass Premier mass spectrometer (Waters Corp., Milford, MA, USA) with an RTX-Dioxin 2, 60 m x 0.25 mm x 0.25μm film thickness column (Restek Corp., Bellefonte, PA, USA). For analysis of tetra- through octa-CDDs/Fs, Method 8290A ¹⁵ was followed. The standard used for PCDD/PCDF identification and quantification is a mixture of standards containing tetra-to octa-PCDD/F native and 13C-labeled congeners designed for modified U.S. EPA Method 23 ¹⁴ Not all of the seventeen PCDD/PCDF toxic equivalent factor (TEF) weighted congeners were detected in all samples. The congeners that were not detected (ND) were set to zero in the text, however SI Tables S6-S9 show values both ND = 0 and ND = limit of detection (LOD). The PCDD/PCDF toxic equivalent (TEQ) emission factors were determined using the World Health Organization (WHO) 2005 toxic equivalent factors (TEF) ¹⁶. Only four PCDD/PCDF congeners were detected in all samples (1,2,3,4,6,7,8 - HpCDD, 1,2,3,4,6,7,8,9 - OCDD, 2,3,7,8 - TCDF, 1,2,3,4,6,7, 8- HpCDF) these emission factors were used for intercomparison purposes. These emission factors represent the low end of the absolute emission factor but are the most reliable for intercomparison.

A portion of the methylene chloride extract from the PCDD/PCDF/PAH sample was used for the PAH analysis using a modified EPA Method 8270D ¹⁷ Labeled standards for PAHs were added to the XAD-2 trap before the sample was collected and internal standards were added before mass analysis. The PAHs TEQ emission factors were determined using TEFs by Larsen and Larsen ¹⁸.

Ambient air background samples were collected for each of the target pollutants. Only the VOC emissions were background corrected. PCDD/PCDF, PAH and PM burn samples had over 20, 170, and 200 times higher concentrations than the ambient air background sample, respectively.

Calculations

Emission factors, expressed as mass of pollutant per mass of biomass burned, were based on the carbon balance method²⁵. This method concurrently measures the target analyte along with the amount of fuel burned, the latter assumed to be determined by the ΔCO + ΔCO₂ measurements and assuming a 50% carbon concentration in the biomass fuel. The minor carbon mass emitted as hydrocarbons and PM is ignored without significant effect on the emission factor. The resultant emission factors are expressed as mass of pollutant per mass of biomass consumed (B_c).

The modified combustion efficiency (MCE), ΔCO₂/(ΔCO₂+ΔCO+ΔCH₄) (with CH₄ included in VOC samples only), was calculated for each of the emission samples. Custom photometric calibration factors were derived for each burn conducted for the DustTrak 8520 by simultaneous collection of PM_2.5 mass on a filter (averaged continuous PM_2.5 concentration divided by PM_2.5 by filter mass).

Single factor one-way analysis of variance (ANOVA) with a level of significance a = 0.05 was used to determine any differences in air pollution emissions between PE covered and uncovered biomass piles. To establish significant difference the ANOVA-returned p value (significant value) has to be less than level of significance (0.05) and the F-distribution value (F/F_crit) has to be greater than 1.0.

RESULTS AND DISCUSSION

Eleven pile burns were sampled over a five day period with emission factor results summarized in Table 3. The plumes were sampled with the aerostat/Flyer in close proximity to the fires to maximize the sample collection mass without placing operators or instruments at risk. Typical aerostat heights above the pile burn were 20–70 m. Pile emission sampling averaged 45 minutes. Ambient temperatures ranged from 2–13°C, winds 0–32 km/h, and humidity ranged from 100% for the first two days of testing to 35–40% on the last two days. Additional meteorological data are presented in the Supporting Information.

Table 3.

Results.^a

Pollutant	Unit	WETb	DRYb uncovered	DRY PEc 6.1×6.1 m 0.15 mm	DRY PEc 3×3 m 0.15 mm	DRY PEd 3×3 m 0.10 mm
CO2e	g/kg Bc	1,689 (36%)	1,785 (1.5%)c	1,758d	1,795d	1,756d
COe	g/kg Bc	82 (20%)	29 (56%)c	43d	22d	46d
CH4e	g/kg Bc	5.7 (2.1%)	1.1 (67%)c	2.6d	1.5d	2.0d
PM2.5	g/kg Bc	18 (58%)	4.5 (9.5%)	6.0 (39%)	5.2 (35%)	3.4
BC	g/kg Bc	0.47 (6.2%c)	0.24 (5.7%)	0.27 (38%)	0.28 (14%)	0.28
UVPM	g/kg Bc	0.50d	0.24 (3.5%c)	NS	0.30d	NS
EC	g/kg Bc	0.18 (4.1%c)	0.12 (18%)	0.10 (6.0%)	0.14 (7.9%)	0.13
OC	g/kg Bc	8.2 (2.9%c)	2.5 (22%)	3.5 (56%)	2.5 (38%)	1.8
TC	g/kg Bc	8.4 (2.9%c)	2.6 (21%)	3.6 (55%)	2.7 (37%)	1.9
OC/EC	Ratio	45 (6%)	21 (32%)	34 (52%)	17 (31%)	14
BC/PM2.5	Ratio	0.033 (30%c)	0.052 (9.4%)	0.045 (0.6%)	0.066 (47%)	0.081
EC/PM2.5	Ratio	0.013 (19%c)	0.027 (22%)	0.021 (34%)	0.030 (28%)	0.029
Σ VOCsf	mg/kg Bc	4,106 (50%)	612 (48%)c	1,266	1,036	1,255
Σ PAH16	mg/kg Bc	88 (10%)	15 (27%)	26 (59%)	24 (54%)	14
Σ PAH - TEQ	mg B[a]Peq/kg Bc	2.7 (11%)	0.27 (32%)	0.48 (62%)	0.55 (50%)	0.24
Σ PCDD/PCDF	ng/kg Bc	15 (37%)	5.8 (7.2%)	8.0 (69%)	7.6 (73%)	5.1
Σ PCDD/PCDF TEQg	ng TEQ/kg Bc	0.18 (11%)	0.077 (59%)	0.14 (96%)	0.066 (95%)	0.057
Σ 4 PCDD/PCDF congenersh	ng TEQ/kg Bc	0.015 (19%)	0.0079 (19%)	0.010 (41%)	0.10 (65%)	0.0077

^aUnits in mass of pollutant per mass of biomass consumed (B_c). NS = No sample. Relative standard deviation (RSD) and relative percent difference (RPD) within parentheses.

^bRSD within parentheses.

^cRPD within parentheses.

^dSingle sample.

^eDerived from SUMMA Canisters.

^fSum of 74 VOCs analyzed via U.S. EPA Method TO-15 ²².

^gNot detected congeners set to zero, results for each congener and homologue is presented in SI Table S5-S10.

^hFor intercomparison purpose only, PCDD/PCDF congeners detected in all samples: 1,2,3,4,6,7,8 - HpCDD, 1,2,3,4,6,7,8,9 - OCDD, 2,3,7,8 - TCDF, 1,2,3,4,6,7,8 - HpCDF.

The potential effect of day-of-testing on the results due to, for example meteorological condition changes through the week, were examined by the chronological examination of the emission factors for all targeted pollutants. This analysis is of limited utility due to the non-random order in which the tests were run. Nonetheless, no effects related to testing date, or time of day were found on the Wet/Dry PM_2.5, PAH, and PCDD/PCDF emission factors were found including the Dry PE PCDD/PCDF results. However, an effect of the testing date was found for Dry PE on the PM_2.5 emission factors and was inconclusive on the PAH results.

CO, CH₄, and CO₂

Typical concentration results throughout the duration of a Dry and Wet burns are shown in Figure 3. Fluctuations in the concentrations are typical and reflect wind shifts moving the Flyer in an out of the plume. The CO and CH₄ emission factors were almost twice as high for the wet piles as the dry (Table 3). Hardy ⁶ estimated 1.64 and 5.52 g/kg for CH₄ from flaming and smoldering, respectively. Our work resulted in whole-burn values of 1.1 g/kg (DRY) to 5.7 g/kg (WET). The CO₂, CO and CH₄ emission factors in this study str also in the same range as found in the literature of open burning of Douglas fir 1,601–1,772 g/kg, 74–138 g/kg, 0.3–7.9 g/kg^{26, 27}, respectively.

Figure 3.

Typical concentration traces of CO₂, CO, BC, PM_2.5 and modified combustion efficiency (MCE) for Dry and Wet burns. Traces displayed in 60 seconds moving average.

PM_2.5

The PM_2.5 results show a statistically significant (F = 2.7, p= 0.004) increase in the Wet (18±11 g/kg B_c) versus the Dry uncovered + Dry PE covered (4.9±1.8 g/kg) emission factor (Figure 4 Inset). Individual emission factors (Figure 4) show no distinction between the Dry uncovered and Dry PE covered piles. The PM_2.5 emission factors compare with a value of 6.75 g/kg consumed estimated from hand-pile biomass burns by Wright et al.²⁸. The Wet emission factor (18±11 g/kg B_c) derived at a MCE of 0.839±0.057 is in the same range as found in the literature of open burning of Douglas fir 15.7±5.2 g/kg dry fuel consumed²⁷ at a MCE of 0.916±0.016.

Figure 4.

PM_2.5 results. Inset chart shows Wet versus DRY (PE-covered and uncovered). Error bars represents 1 standard deviation if nothing else stated.

Examination of the relationship between PM_2.5 and the MCE showed that lower combustion efficiencies were correlated with higher PM_2.5 loads. Figure 5 shows that comparison of same-day WET and DRY samples (Day 2 and Day 3) continue to verify the distinction with the passage of time, suggesting that the non-random testing did not affect the conclusions. The distinction in the PM_2.5 emission factors occurs in the initial half of the burns. Figure 6 shows that the early portion of the WET pile burns when the fire is getting started is responsible for the high PM_2.5 emissions. This distinction with the DRY burns persists until the second half of the burn when smoldering was more prevalent.

Figure 5.

The relationship between PM_2.5 emission factor and combustion quality (modified combustion efficiency, MCE).

Figure 6.

Comparison of PM_2.5 emission factors at 4 min intervals throughout the burn durations, comparing the combined WET and combined DRY results.

Black Carbon, UVPM, Elemental/Organic Carbon

BC, EC, OC, and TC values were all higher for the WET burns as compared to all of the DRY and PE burns (Figure 7). No statistical distinctions in these values were observed for the varying sizes and thicknesses of PE. BC showed approximately a factor of two higher values than EC and they did not correlate strongly with each other (R2 of 0.49, SI Figure S1) which may be due to the low number of data points. The EC emission factor, 0.10–0.18 g/kg B_c, is in the same range as found in the literature, 0.19±0.41 g/kg dry fuel, from laboratory burns of Douglas fir²⁶. The relationship between EC and BC emission factors and MCE is shown in Figure 8.

Figure 7.

BC, EC, UVPM, OC and TC results. Inset chart shows Wet versus DRY (PE- covered and uncovered). Error bars represents relative difference if nothing else stated.

Figure 8.

BC and EC in relationship to modified combustion efficiency (MCE).

The OC/EC values, a surrogate for comparison of optical reflectance/warming properties, indicates values ranging between 14 and 45, the latter being the WET burns (Table 3). Values greater than unity are common and anticipated for biomass burns. These values are the opposite of what is observed with, for example, crude oil combustion ²⁹, where the OC/EC ratio is about 1/15.

Volatile Organic Compounds (VOCs)

VOC results for the most concentrated species are shown in Table 4. The full set of VOC emission factors are summarized in Supporting Information, Tables S11-S13. ANOVA analysis (Figure 9) of acrolein, benzene, styrene and 1,3-butadiene showed statistical differences between WET and DRY piles, (Benzene F = 1.6, p = 0.0208; Acrolein F = 3.3, p = 0.004; Styrene F = 1.9, p = 0.015; 1,3-Butadiene F = 1.4, p = 0.026). Benzene is a common VOC associated with incomplete combustion. Acrolein is a toxic, irritant, 3-C carbonyl and is not listed as a carcinogen on EPA or international lists. 1,3-butadiene is listed as a human carcinogen. Styrene is “reasonably anticipated to be a human carcinogen” ³⁰. The relationship between emission factors for these select VOCs and MCE is shown in Figure 10.

Table 4.

VOC result.

	WETa	DRY uncoveredb	DRY PE 3×3 m 0.10 mm	DRY PE 3×3 m 0.15 mm	DRY PE 6.1×6.1 0.15 mm
Compound	mg/kg biomass consumed
Benzenec	757±416	115 (37%)	216	289	222
Propene	682±373	119 (53%)	252	199	250
Acetone	668±280	32	163	78	ND
Acroleinc	463±168	97 (50%)	134	99	180
Vinyl Acetatec	309±133	52 (58%)	78	51	134
Toluenec	297±172	52 (55%)	100	98	116
1,3-Butadiene	231±136	31 (50%)	78	71	74
2-Butanone (MEK)	156±76	27 (69%)	49	21	72
Styrenec	111±59	16 (52%)	25	33	35
Acetonitrile	69±40	17 (60%)	34	12	38
m,p-Xylenesc	68±41	13 (68%)	22	15	27
Ethylbenzene	43±26	7.5 (53%)	14	12	15
alpha-Pinene	41±31	8.7 (60%)	17	17	14
d-Limonene	31±21	6.7 (3.9%)	8.7	12	13
Acrylonitrilec	27±14	6 (25%)	12	7.0	11
o-Xylenec	23±14	4.4 (73%)	8.0	4.5	9.1
1,2,4-Trimethylbenzene	12±5.8	2.4 (1.7%)	3.8	1.9	4.2
1,3,5-T rimethylbenzene	3.5±1.6	1.2	1.2	0.49	1.2

^aRange of data equal one standard deviation.

^bRange of data equals relative percent difference.

^cOn U.S. EPA’s list of hazardous air pollutants³¹. The VOCs shown here were selected based on the number of samples detectable above three times the detection limit and their relevance to the EPA’s list of hazardous air pollutants list and their role as greenhouse gas/ozone precursors. Full list of the 74 analyzed VOCs and their emission factors are presented in SI Tables S11-S12.

Figure 9.

VOC results. Error bars represent one standard deviation for WET burns and DRY combined burns, and relative difference for DRY uncovered burns. * = On U.S EPA’s list of hazardous air pollutants.

Figure 10.

The effect of modified combustion efficiency (MCE) on select VOC emission factors.

PCDD/PCDF

Results for PCDD/PCDF emission factors for Dry, Wet, and PE are summarized in Table 3. Figure 11 presents data for four of the 17 congeners that comprise the PCDD/PCDF TEQ value ¹⁶ that were present in all 11 samples (complete data are shown in SI Tables S5-S10). As such, these emission factors represent the low end of the absolute emission factor but are the most reliable in terms of intercomparisons. Wet PCDD/PCDF values are higher than Dry uncovered piles [F = 2.0, p = 0.017]. Dry and PE values show no statistical difference between them [F = 0.01, p = 0.814]. Within the PE grouping, no distinction was observed between the PE sheet size and thickness, although the limited number of tests limits the statistical power of this test. Figure 12 examines the effect of combustion quality as measured by MCE on the PCDD/PCDF emission factors. Three distinct groupings of emission factors for Dry, Wet, and PE are indicated. While Wet results show no apparent trend with MCE, PE results suggest that PCDD/PCDF emission factors decline with increased MCE (r² = 0.93). This is similar to observations for both PM_2.5 and select VOCs. Evaluation of the whole data set shows an r² = 0.82 with declining emission factor and MCE. Additional data are necessary to verify these MCE indications, although this trend is consistent with historical observations that equate improved combustion conditions with decreased PCDD/PCDF emissions.

Figure 11.

PCDD/PCDF emission factors in ng TEQ/kg biomass consumed. Error bars represent 1 standard deviation if nothing else stated.

Figure 12.

PCDD/PCDF emission factors in ng TEQ/kg biomass consumed by group versus MCE.

These PCDD/PCDF emission factors are approximately ten times lower than literature values of 0.11–0.22 ng TEQ/kg B_c from open burning of pine savannas^{10, 32}

PAHs

Individual PAH emission factors (for the 16 EPA PAHs) are shown in Table 5 and Sum of the 16 EPA PAHs are shown in Figure 13. Similar to observations of PM_2.5, select VOCs, and PCDD/PCDF, Wet piles resulted in greater emissions (statistically significant, F = 14.3, p < 0.0001), by a factor of 4–5. No distinction was observed, however, between any of the Dry (cover and uncovered) PAH emission factors. These emission factors compared to a value of 28 mg/kg burning Douglas fir in a laboratory setting³³.

Table 5.

PAH emission factors.

	WETa	DRYa uncovered	DRY PEb 6.1×6.1, 6 mm	DRY PEb 3×3, 6 mm	DRY PEc 3×3, 4 mm
PAHs	mg/kg biomass consumed
Naphthalene	17 (3.4%)	4.4 (37%)	8.1 (50%)	7.4 (54%)	5.0
Acenaphthylene	16 (14%)	2.5 (24%)	4.6 (65%)	4.1 (53%)	2.3
Acenaphthene	1.6 (21%)	0.34 (24%)	0.60 (67%)	0.46 (59%)	0.27
Fluorene	6.4 (35%)	0.97 (27%)	1.7 (66%)	1.5 (61%)	0.75
Phenanthrene	19 (20%)	3.3 (26%)	4.8 (64%)	4.5 (57%)	2.5
Anthracene	4.1 (15%)	0.65 (28%)	1.0 (63%)	0.98 (56%)	0.50
Fluoranthene	6.9 (3.4%)	0.90 (30%)	1.4 (59%)	1.6 (54%)	0.76
Pyrene	6.2 (10%)	0.78 (31%)	1.3 (59%)	1.5 (51%)	0.68
Benzo(a)anthracene	2.1 (10%)	0.24 (28%)	0.43 (64%)	0.44 (54%)	0.20
Chrysene	2.5 (10%)	0.38 (24%)	0.61 (62%)	0.58 (55%)	0.30
Benzo(b)fluoranthene	1.3 (14%)	0.13 (28%)	0.24 (61%)	0.25 (51%)	0.11
Benzo(k)fluoranthene	1.7 (6.9%)	0.16 (35%)	0.29 (61%)	0.34 (47%)	0.15
Benzo(a)pyrene	1.7 (12%)	0.16 (33%)	0.29 (62%)	0.34 (49%)	0.14
Indeno(1,2,3-cd)pyrene	0.84 (12%)	0.073 (38%)	0.13 (60%)	0.17 (47%)	0.067
Dibenz(a,h)anthracene	0.20 (14%)	0.021 (28%)	0.037 (63%)	0.041 (51%)	0.022
Benzo(ghi)perylene	0.98 (14%)	0.086 (38%)	0.15 (58%)	0.21 (45%)	0.079
SUM 16-EPA PAH	88 (11%)	15 (27%)	26 (59%)	24 (54%)	13.8

^aRange of data within parentheses equals relative standard deviation.

^bRange of data within parentheses equals relative percent difference.

^cSingle sample.

Figure 13.

Average PAH emission factors for each category.

The PAH measurements reflect both gas phase and particle-bound PAH compounds. The relationship between the emission factors for PM_2.5 and PAHs were examined in Figure 14. Predictably higher PM_2.5 is associated with higher PAHs.

Figure 14.

Comparison of PAH emission factors and PM_2.5 emission factors.

The relationship between PAHs and combustion quality (MCE) is shown in Figure 15. As with previous emissions, lower combustion quality (MCE) is associated with higher PAH emissions. All of the Wet results have the lowest MCE and highest PAH levels.

Figure 15.

Comparison of PAH emission factors with modified combustion efficiency (MCE).

COMPARISON WITH OTHERS’ DATA

Comparison of our results with previously compiled data on open pile burning of woody biomass from twelve sources⁴ places our data within the range of reported results. Literature values for PM (total) ranged from 3–22 kg/kg dry biomass burned whereas our results were 3–18 kg/kg B_c (these units are similar but derived differently). Likewise, reported CO emission factors were 17–164 g/kg in comparison to our results of 22–82 g/kg B_c. CH₄ values were reported at 0.9–11 g/kg versus ours at 1–6 g/kg B_c. Few other pollutants for field pile burns are characterized in the literature

CONCLUSION

Field sampling of eleven biomass pile burns determined emission factors for a wide range of pollutants. Comparison of piles that were naturally wetted versus those that were dry showed statistically higher emission factors for PM_2.5, PAHs, VOCs, and PCDD/PCDF for the wet piles. Emission levels were negatively correlated with combustion quality as represented by MCE. Variation of PE cover size and thickness showed no statistically significant difference in emission factor for any of the pollutants suggesting that the PE was not contributing significantly to any of the measured pollutants. Time-resolved PM_2.5 emissions were highest at the beginning of the burns; for the Dry pile tests, this startup period lasted for less than 4 min; for the Wet pile tests, it was four times longer, about 16 min. For the Wet pile tests, PM_2.5 emission factors were higher than those of the Dry pile tests for at least half of the burn durations, after which they were similar. These tests suggest that use of PE as a biomass pile cover results in lower emission factors than those from piles exposed to moisture, reducing pollutant levels during slash pile burns. These emission factors, together with estimates of burn pile numbers, size, and fuel consumption, can be used by management and regulatory communities to minimize smoke impacts while limiting the potential hazard of biomass fuel loading.