Impact of fuel moisture on chaparral emissions under flaming and smoldering conditions

Abstract

Fourier-transform infrared (FTIR) spectroscopy was used to quantify gas and particulate emissions from flaming and smoldering combustion of four dominant chaparral species (eucalyptus, coastal live oak, manzanita, and toyon) with different fuel moisture content from California, USA under controlled laboratory conditions. Results show that the higher fuel moisture content enhanced the production of incomplete combustion products, mostly CO and hydrocarbons. Smoldering combustion consistently produced 10–70 times higher particulate matter emission factors compared to flaming across all species, and this amplification ratio varied substantially depending on both fuel type and moisture content. Live eucalyptus exhibited the highest hydrocarbon emissions under smoldering conditions at 37.18 ± 0.65 g/kg.

Cross-ecosystem comparisons revealed that chaparral emission profiles were broadly similar to western shrubland and grassland fuels, with notable differences primarily in nitrogen-containing species, highlighting the influence of combustion scale and mixing conditions on emission profiles. These findings provide important emission factor data on chaparral fuels, demonstrating the impact of fuel moisture, combustion mode, and fuel-specific properties on emissions, contributing to wildfire emission inventories and air quality modeling.

1. Introduction

In California’s Mediterranean-type shrublands, chaparral is the most widespread vegetation type, comprising approximately 73% of the total shrubland area and covering about 6% of the state’s land surface [1, 2]. These landscapes are dominated by sclerophyllous (hard-leaved, drought-tolerant) shrubs such as chamise (Adenostoma fasciculatum), manzanita (Arctostaphylos glandulosa), and toyon (Heteromeles arbutifolia), many of which are highly adapted to fire, possessing underground burls for resprouting and accumulating resin-rich biomass that enhances their flammability [1, 3, 4]. Wildfires in chaparral are often particularly intense due to the dense, woody structure and the presence of high concentrations of ether extractives, such as waxes, oils, and terpene [1, 5]. The high fuel loading, large surface area to volume ratio, and abundance of volatile compounds in these species contribute to rapid fire spread, intense crown fires, and a complex mixture of chemical emissions during combustion [6-8]. In recent decades, wildfires in chaparral fuels have increased in frequency and intensity, primarily due to prolonged drought and decreased precipitation. Meanwhile, continued population growth and expansion of wildland–urban interface (WUI) regions have increased human-caused ignitions and can promote fuel accumulation through long-term fire suppression, thereby increasing the likelihood of large and high-intensity wildfire events [9-11]. Chaparral fires generate smoke plumes abundant in trace gases and particulate matter (PM), significantly impacting regional air quality and facilitating the formation of ozone and secondary organic aerosols (SOA) in downwind areas [1, 8, 12]. Thus, quantification of chaparral emissions is crucial as it directly impacts health assessments and exposure risks [13, 14]. Modern advances in instrumentation, such as high-resolution open-path/airborne Fourier-transform infrared (FTIR) spectroscopy and aerosol mass spectrometry (HR-AMS), have made it possible to concurrently measure various major and trace gases released during wildland fires with improved accuracy [8, 12, 15-17].

Besides different fuel types, fuel moisture content (FMC) is another critical factor that influences combustion dynamics and emission chemistry in chaparral fires [18-23]. While higher FMC can have a positive or negative impact on PM emissions depending on specific conditions, it tends to suppress flaming combustion and promote smoldering or incomplete oxidation, which increases emissions of carbon monoxide and volatile organic compounds (VOCs) [18-20, 23]. Recent studies have shown that the effects of FMC on combustion emissions can vary by fuel species, with a significantly greater impact observed during flaming combustion compared to smoldering [21, 22]. However, these findings are predominantly based on coniferous fuels, and further research is needed as there is limited quantitative data on how FMC affects combustion efficiency and emissions in chaparral species under controlled conditions.

The purpose of this study is to quantify gaseous and particulate emissions from four dominant chaparral species commonly found in California, USA under controlled laboratory conditions. Using a high-resolution FTIR spectrometer and a custom linear tube heater setup, we varied FMC and combustion conditions to assess their effects on emission profiles. Through detailed emission factor (EF) calculations and analysis, this research explores the relationships between various fuel properties, combustion modes, and emission chemistry, which are discussed in the following sections.

2. Material and Methods

2.1. Experimental Setup

The experiments were carried out on a custom linear tube heater apparatus as shown in Figure 1. The system was adapted from the DIN 53436 tube furnace smoke generation method, which has been evaluated for simulating fire-relevant flaming combustion conditions [24]. Similar steady-state tube furnace combustion systems have been widely used for fire toxicity, smoke characterization, and vegetative combustion emission studies under controlled laboratory conditions [21, 25, 26]. This setup consisted of an 182 cm quartz tube (3.5 cm ID) and a cylindrical ceramic heater (7.6 cm ID, 15.2 cm width), mounted on a linear actuator. Further specifications on the setup can be found in previous work [21, 22].

Fig. 1.

The schematic diagram of the custom linear tube-heater smoke generating and sampling apparatus, not to scale.

Gaseous emissions were collected using a Thermo Scientific Nicolet iG50 FTIR spectrometer (2 m gas cell, MCT detector, 0.5 cm⁻¹ resolution, 0.025 Hz sampling frequency). The FTIR system was maintained at 100°C throughout the test day to prevent gas condensation. All components of the FTIR sampling stream, including heated transfer lines, filter housing, and the gas cell, were maintained at this temperature. The FTIR system operated at 650–700 Torr, which lowers the effective dew point of water vapor and prevents condensation of HO and semi-volatile species within the sampling system. Before each test, nitrogen purging was used to remove emission residues. Real-time concentrations of 25 gas species that are commonly reported in combustion studies were measured based on the gas calibration and built-in least squares algorithm of the FTIR [8, 12, 16, 27, 28]: carbon dioxide (CO₂), carbon monoxide (CO), sulfur dioxide (SO₂), nitric oxide (NO), nitrogen dioxide (NO₂), ammonia (NH₃), methanol (CH₃OH), hydrogen cyanide (HCN), hydrogen chloride (HCl), hydrogen fluoride (HF), hydrogen bromide (HBr), nitrous acid (HNO₂), acetic acid (CH₃COOH), formic acid (CH₂O₂), formaldehyde (CH₂O), methane (CH₄), water (H₂O), acetylene (C₂H₂), ethane (C₂H₆), propane (C₃H₈), butane (C₄H₁₀), isobutene (C₄H₈), ethene (C₂H₄), propene (C₃H₆), and isoprene (C₅H₈). Concentrations were determined using factory calibration spectra at sampling conditions of 100°C and 650 torr, with detection limits above 0.01 ppm.

Time-averaged PM was obtained through gravimetric analysis using a stainless-steel sampling tube (9.24 mm ID) with 2 mm holes across the exhaust duct [21, 29]. PM was drawn from both ends through a 6.35 mm tubing to a 37 mm filter cassette with a glass fiber filter, using a pump operating at a fixed flow rate of 2.5 L min⁻¹. Particulate matter was sampled from a well-mixed turbulent exhaust, as combustion emissions are dominated by ultrafine particles that are less sensitive to isokinetic sampling bias. Filters were immediately weighed post-sampling using a Sartorius balance (resolution 0.01 mg). Carbon mass balance validation showed an average fuel consumption of 96.0 ± 2.2% across all tests, confirming efficient fuel consumption and reliable emission factor quantification [8, 30].

2.2. Fuel Preparation and Moisture Content

Leaves of four different chaparral species were tested in this study: eucalyptus (Eucalyptus globulus), coastal live oak (Quercus agrifolia), toyon (Heteromeles arbutifolia), and manzanita (Arctostaphylos spp.). These species were selected as representative fuels for California’s chaparral ecosystems, which dominate middle-elevation Mediterranean zones and are frequently involved in wildfires throughout the state [1, 3, 31]. Fresh samples were collected during the pre-fire season (March-May) in Berkeley, California, sealed in plastics bags, and stored at 4°C for no more than 48 hours before processing to prevent decomposition. All fuels were trimmed into 0.5 cm segments to ensure uniformity. The elemental composition of each species was analyzed using a Thermo Scientific Flash Smart CHNS Elemental Analyzer, which determines the levels of carbon (C), hydrogen (H), nitrogen (N) and Sulfur (S) through combustion-based gas analysis. Samples were tested under three distinct moisture conditions: freshly collected (“live”), partially dried (“wet”, oven-dried at 70°C for 3 hours), and completely dried (“dry”, oven-dried at 70°C for 72 hours). The drying temperature of 70°C was selected to remove moisture while minimizing loss of volatile organic compounds (VOCs) and to maintain consistency with our previous biomass combustion experiments [29]. FMC was determined as the percentage of moisture present in the fuel sample relative to its wet mass [21]:

$$FMC=\frac{M_W-M_D}{M_W}\times 100$$ (1)

where $M_W$ is the wet mass of the sample (g) and $M_D$ is the dry mass of the sample (g). The dry mass was determined by drying the fuel samples in an oven for either 3 hours (for wet conditions) or 72 hours (for completely dry conditions). Moisture content was continuously monitored using an A&D MF-50 moisture analyzer to ensure accuracy. FMC values and elemental composition results for each fuel were recorded in Table 1.

Table 1.

FMC and CHNS Result. Sulfur contents were not included since they were below the instrumentation detection limits (< 0.01%). Errors represent one standard deviation (shown in parentheses).

Fuels	FMC (%)			CHN (%, mean (SD))
	Live	Wet	Dry	C	H	N
Eucalyptus	46.98	15.15	3.30	53.27 (0.16)	6.16 (0.01)	0.79 (0.09)
Live Oak	46.75	19.38	4.70	46.97 (0.30)	6.27 (0.01)	2.23 (0.06)
Manzanita	44.35	22.88	6.45	53.50 (0.86)	6.46 (0.05)	0.99 (0.02)
Toyon	-	-	5.12	48.52 (0.35)	6.35 (0.09)	1.05 (0.03)

2.3. Test procedure

During the experiments, three parameters were varied: FMC (recent live, wet, and dry), fuel type (eucalyptus, coastal live oak, toyon, and manzanita), and external heat flux. Smoldering and flaming tests were performed at 25 kW/m² and 50 kW/m², respectively. Based on prior calibration, these conditions correspond to furnace surface temperature of 450°C and 650°C. For each test, fuel samples (2.5 g for smoldering and 5 g for flaming tests) were uniformly spread over the first half of an 80 cm quartz boat and then placed inside the quartz tube. Then, oxidizer flow was applied with a prescribed oxygen concentration at 3 L/min. The primary airflow is provided in the opposite direction of the heater travel direction to prevent downstream fuel preheating [21]. The heater traveled at a constant traverse rate of 30 mm/min for flaming and 20 mm/min for smoldering along the quartz tube for 15 min. This duration ensured complete heater passage through the sample region, and the mass of fuel and heater speed were optimized to achieve steady and consistent fuel combustion. FTIR sampling began one minute post-ignition after concentrations stabilized and was conducted with a time resolution of 32 seconds per scan. Detailed post-run cleaning and maintenance protocols for the FTIR system have been described in previous studies. [22, 29]. Gaseous and PM sampling for all experiments was repeated 3 times to ensure robust calibration and measurement accuracy. Experiments were conducted under controlled ambient conditions at 25 °C, 50 % relative humidity, and 1 atm atmospheric pressure.

2.4. Emission Factor and Modified Combustion Efficiency

Emission factors (EFs) were calculated using the carbon balance approach, which assumes that all carbon from the fuel is conserved in the carbon-containing emissions [8, 30, 32]. For a given species $i$, EF can be determined as Eq. 2 [30]:

$$E{F}_i(g∕kg)=F_c\times 1000\times \frac{M{M}_i}{12}\times \frac{E{R}_i}{C_T}$$ (2)

where $F_c$ is the carbon content (%) of the fuel, $M{M}_i$ is the molar mass of the species $i$ (g/mol), 12 is the molar mass of carbon, and $E{R}_i$ is the emission ratio of $i$ to $C{O}_2$. $C_T$ was calculated as:

$$C_T=\sum _{j=1}^n{N}_j\times \frac{\Delta C_j}{\Delta C{O}_2}$$ (3)

where $n$ is the number of carbon-containing species, $N_j$ is the number of carbon atoms in species $j$, and $\Delta C_j$ is the excess concentration of species $j$ relative to $\Delta C{O}_2$.

The modified combustion efficiency (MCE) is a widely used metric to characterize combustion conditions, distinguishing between flaming and smoldering combustion phases [30, 32, 33]:

$$MCE=\frac{\Delta C{O}_2}{\Delta C{O}_2+\Delta CO}\times 100$$ (4)

where $\Delta C{O}_2$ and $\Delta CO$ represent the excess mixing ratios of carbon dioxide and carbon monoxide, respectively. Higher MCE values indicate more complete combustion, and decreasing MCE suggests less efficient burning.

3. Results and discussion

3.1. CO, CO₂, TPM and other major emissions

EF and MCE for all tested chaparral fuel types under varying moisture conditions and combustion phases are presented in Tables 2 and 3. These summaries include total particulate matter (TPM), CO, CO₂, total hydrocarbons (HC), SO₂, and NO_x, and reveal clear patterns influenced by both fuel moisture and fuel-specific combustion characteristics. To statistically assess the significance of observed differences, two-sample t-tests were also performed to determine whether EFs varied significantly among fuel types, FMC, and combustion modes. Toyon was evaluated only under dry conditions, as live and wet Toyon failed to maintain stable emission profiles during combustion, particularly under flaming conditions, exhibiting unstable ignition, frequent self-extinguishment, and highly variable emissions. These observations provide useful qualitative context for Toyon combustion behavior. This behavior is likely attributable to toyon’s inherently lower flammability relative to other chaparral species [34].

Table 2.

Emission Factor (g/kg) for flaming conditions. Errors shown in parentheses represent one standard deviation between tests. *Toyon was only tested under dry conditions.

Fuel Type	FMC	TPM	CO	CO2	HC	SO2	NOx	MCE (%)
Eucalyptus	Dry	1.12 (0.37)	8.42 (1.51)	1937.40 (2.39)	0.68 (0.18)	0.40 (0.17)	3.62 (0.30)	99.36 (0.23)
	Wet	1.79 (0.31)	14.11 (0.45)	1923.51 (0.16)	1.71 (0.33)	0.04 (0.04)	3.38 (0.03)	98.86 (0.16)
	Live	1.41 (0.20)	77.63 (3.42)	1825.21 (6.66)	1.22 (0.21)	0.49 (0.29)	3.66 (0.08)	93.83 (2.40)
Live Oak	Dry	1.79 (0.56)	18.44 (4.52)	1686.07 (4.74)	0.87 (0.35)	0.45 (0.30)	3.10 (0.12)	98.25 (0.96)
	Wet	0.91 (0.40)	23.97 (7.37)	1681.40 (11.01)	1.65 (0.76)	0.45 (0.45)	3.35 (0.38)	97.24 (1.56)
	Live	0.50 (0.17)	72.11 (2.41)	1601.25 (3.16)	1.85 (0.11)	-	4.89 (0.12)	93.32 (2.28)
Manzanita	Dry	0.96 (0.32)	4.86 (0.29)	1949.91 (0.73)	0.29 (0.04)	-	3.23 (0.19)	99.61 (0.09)
	Wet	1.50 (0.16)	32.52 (10.16)	1909.23 (17.08)	0.28 (0.13)	-	2.64 (0.31)	98.40 (0.01)
	Live	0.86 (0.37)	107.13 (8.50)	1782.75 (13.82)	1.99 (0.22)	-	3.15 (0.01)	90.56 (1.84)
Toyon	Dry	1.00 (0.37)	19.00 (6.27)	1728.30 (17.04)	0.33 (0.10)	-	3.30 (0.39)	97.32 (3.37)

Table 3.

Emission Factor (g/kg) for smoldering conditions. Errors shown in parentheses represent one standard deviation between tests. *Toyon was only tested under dry conditions.

Fuel Type	FMC	TPM	CO	CO2	HC	SO2	NOx	MCE (%)
Eucalyptus	Dry	37.50 (2.39)	176.66 (17.01)	1554.09 (39.90)	31.66 (2.44)	1.69 (0.86)	1.56 (0.34)	84.94 (1.95)
	Wet	45.80 (3.25)	188.93 (12.91)	1531.24 (26.68)	32.88 (0.74)	0.15 (0.15)	1.00 (0.14)	83.65 (3.56)
	Live	41.09 (3.56)	183.40 (4.17)	1512.15 (9.07)	37.18 (0.65)	1.72 (0.35)	1.51 (0.28)	84.01 (0.89)
Live Oak	Dry	24.99 (2.38)	203.99 (2.37)	1329.87 (0.57)	16.82 (1.44)	0.87 (0.62)	1.72 (0.35)	80.59 (1.37)
	Wet	31.05 (2.65)	193.80 (6.50)	1344.98 (19.72)	18.47 (1.01)	1.60 (0.58)	1.33 (0.22)	80.68 (1.97)
	Live	35.44 (2.55)	192.11 (1.80)	1325.26 (6.70)	22.65 (1.34)	0.54 (0.54)	2.24 (0.07)	81.23 (1.15)
Manzanita	Dry	18.32 (3.87)	186.59 (2.63)	1538.17 (3.48)	31.79 (0.54)	-	1.46 (0.05)	84.00 (0.49)
	Wet	20.52 (2.99)	166.81 (1.80)	1610.72 (4.84)	20.79 (0.64)	-	1.83 (0.69)	86.17 (0.79)
	Live	24.73 (2.82)	135.00 (3.20)	1658.24 (8.79)	18.27 (0.38)	-	3.81 (0.25)	88.37 (0.86)
Toyon	Dry	19.30 (2.28)	147.10 (8.21)	1457.11 (21.24)	23.06 (1.08)	-	0.68 (0.16)	86.06 (4.54)

Across all species, live and wet fuels produced substantially higher emissions of incomplete combustion products, such as CO, HC, methanol, and formaldehyde, along with elevated TPM. This suggests that higher fuel moisture inhibits complete oxidation and promotes the formation of these compounds. Live eucalyptus, in particular, exhibited the highest total HC emissions under smoldering condition compared to other fuels. Two-sample t-tests revealed that species identity and moisture content served as primary determinants of emission factor variability, with significant differences (P < 0.05) observed between multiple species comparisons under identical moisture conditions and between dry and live fuel states within species groups. Notably, the distributions of MCE for some species and moisture conditions overlapped, particularly between live and wet fuels, indicating that MCE alone may not fully capture differences in combustion completeness under varying fuel and combustion conditions.

Emission data revealed distinct contrasts between flaming and smoldering combustion conditions as shown in Figure 2. It is readily apparent that CO emissions demonstrated clear FMC-dependent patterns, increasing from dry to live conditions during flaming combustion for eucalyptus, coastal live oak, and manzanita. This trend aligns with Chen et al. [35], who found that higher FMC increases CO emissions from wildland biomass combustion, even at high ignition temperatures, due to delayed ignition and pre-flame smoldering. These findings are further supported by May et al. [36], who reported similar increases in CO emissions with higher FMC in live fuels under laboratory burning conditions. During smoldering conditions, CO emissions became relatively stable across different FMC conditions compared to those under flaming, although notable fuel-specific differences emerged. CO EF for coastal live oak and manzanita decreased with increasing FMC from dry to live conditions. In contrast, eucalyptus exhibited consistently high CO EF across all moisture levels, with an average value of 183.0 ± 12.6 g/kg, which closely aligns with the CO EF of approximately 180 g/kg reported by Wang et al. [37] for dried eucalyptus leaf. Under flaming conditions, manzanita exhibited the broadest range of CO EF from a minimum of 4.86 ± 0.29 g/kg (dry fuel) to a maximum of 107.13 ± 8.50 g/kg (live fuel). Under smoldering conditions, coastal live oak had the highest CO EF values for all FMC conditions, while live manzanita exhibited the lowest CO EF. Comparing flaming and smoldering conditions, all fuels showed significantly higher CO EF under smoldering conditions, consistent with recent laboratory results [22, 38].

Fig. 2.

Average EFs of major gaseous and particulate emissions for (a) eucalyptus, (c) coastal live oak, (e) manzanita, and (g) toyon under flaming conditions; (b) eucalyptus, (d) coastal live oak, (f) manzanita, and (h) toyon under smoldering conditions

CO₂ EF, as expected, decreased modestly from dry to live conditions during flaming combustion, as increased FMC reduces combustion efficiency and results in lower CO₂ emissions correlated with reduced MCE [35]. Statistical analysis showed that the differences in CO₂ EF between most fuel types under flaming conditions were significant (p < 0.05), except for live eucalyptus and manzanita. Under dry flaming conditions, manzanita had the highest CO₂ EF, while coastal live oak exhibited the lowest CO₂ EF. Under smoldering conditions, CO₂ EFs were lower and remained consistent across moisture levels. The consistent reduction in CO₂ emissions from flaming to smoldering reflects the reduced combustion completeness and lower efficiency inherent in smoldering combustion phases.

Across all chaparral species, TPM EFs were substantially higher under smoldering than flaming by 10 to 70 times due to the inherently lower combustion efficiency, consistent with findings by Garg et al. [22]. Notably, this amplification ratio depended both on fuel type and FMC. For example, TPM EF for dry eucalyptus rose from 1.12 ± 0.37 g/kg in flaming to 37.50 ± 2.39 g/kg under smoldering conditions, maintaining a relatively stable amplification range of 25-34 times across different FMC levels. In contrast, coastal live oak showed a dramatic range from 14 to 71 times amplification from dry to live fuel conditions. For manzanita, the amplification ratio was seen to decrease from 21 to 12 times with increasing FMC, suggesting that species respond very differently to FMC. Examining FMC’s effects alone, TPM EFs under smoldering typically increased from dry to live conditions for both coastal live oak and manzanita, while eucalyptus exhibited its highest TPM under wet conditions. Under flaming conditions, TPM EFs peaked under wet conditions for eucalyptus (1.79 ± 0.31 g/kg) and manzanita (1.50 ± 0.16 g/kg), but decreased from dry to live conditions for coastal live oak from 1.79 to 0.50 g/kg, aligning with its declining MCE.

Generally, NO_x emissions remained higher during flaming conditions compared to smoldering. Across different fuel types and FMCs, wet manzanita produced the lowest NO_x EF (2.64 ± 0.31 g/kg) while live coastal live oak generated the highest (4.89 ± 0.12 g/kg) under flaming conditions. SO₂ emissions were minimal and varied among fuel types and conditions [39]. Live eucalyptus consistently produced the highest SO₂ EF (0.49 ± 0.29 g/kg in flaming, increasing to 1.72 ± 0.35 g/kg in smoldering), while manzanita and toyon generated negligible SO₂ across both combustion phases.

Total HC emissions showed strong dependence on both fuel moisture and combustion phase, with clear fuel-specific differences. Across all fuel types, HC EF consistently increased from flaming to smoldering with decreasing MCE. In flaming conditions, HC EF was highest for live manzanita and lowest for wet manzanita, suggesting a substantial influence of moisture within the same species. Under smoldering conditions, HC emissions increased significantly for all fuels, with live eucalyptus reaching the highest HC EF of 37.18 ± 0.65 g/kg, while dry live oak exhibited the lowest at 16.82 ± 1.44 g/kg. More detailed analysis is discussed in the following section.

3.2. Hydrocarbons

A total of nine HCs were measured in this study, including CH₄, C₄H₁₀, C₃H₆, C₂H₆, C₂H₄, C₅H₈, C₂H₂, C₄H₈, and C₃H₈. While measured C₃H₈ was negligible for both smoldering and flaming conditions. A consistent observation across all experimental scenarios was the dominance of CH₄ and C₄H₁₀ with respective average values of 8.11 ± 0.26 g/kg and 8.61 ± 0.11 g/kg during smoldering as primary HC contributors, particularly under high FMC. This same trend of CH₄ and C₄H₁₀ dominance across moisture conditions was seen in previous papers focused on combustion of pine needles [38]. Across all fuel types, smoldering combustion produced significantly higher overall HC EF compared to flaming combustion.

Under flaming combustion, seen in Figure 3 (a), different fuel types exhibited individualized HC emission profiles. Live manzanita yielded the highest total HC EF at 1.99 ± 0.22 g/kg, with the lowest flaming MCE of 90.56 ± 1.84 %. In contrast, wet manzanita consistently produced the lowest HC EF at 0.28 ± 0.13 g/kg. These maximum and minimum EFs fall within the standard deviations of previous literature considering flaming emissions from California chaparral [20]. CH₄ EFs were highest under live conditions, particularly for live manzanita, which measured 1.17 ± 0.01 g/kg. Coastal live oak exhibited a dominance of C₅H₈ within its flaming HC emissions, with the highest EF recorded under live conditions at 1.22 ± 0.09 g/kg. This greater EF under live conditions could be explained by isoprene being measured mainly through evaporation during combustion [40]. Wet eucalyptus under flaming combustion showed an elevated C₃H₆ EF at 0.92 ± 0.29 g/kg, while wet manzanita and coastal live oak measured negligible levels. These findings are similar to that of previous literature [19], the EFs of interest increasing exponentially with FMC until leveling off prior to the live condition. For most flaming conditions, lower FMC correlated with the lowest total HC EFs, while higher FMC led to decreasing MCE. C₂H₂ EFs during flaming were substantially observed only under live conditions with high FMC at 0.07 ± 0.01 g/kg for eucalyptus and 0.22 ± 0.03 g/kg for manzanita. This suggests that C₂H₂ generation is related to specific fuel characteristics and moisture levels during the flaming combustion. A similar trend can be found in prior work [18], which also found that the C₂H₂ EF increases directly with FMC.

Fig. 3.

Average EFs of various hydrocarbons for different fuels under (a) flaming conditions and (b) smoldering conditions

During smoldering conditions depicted in Figure 3 (b), live eucalyptus showed the highest overall HC EF at 37.18 ± 0.65 g/kg. Meanwhile, dry coastal live oak exhibited the lowest HC EF of 16.82 ± 1.44 g/kg, and the lowest smoldering MCE at 80.59 ± 1.37 %, suggesting significant roles for moisture content and fuel type in smoldering emissions. A difference between flaming and smoldering was found in the behavior of C₂H₆ : while its presence was negligible under flaming, relatively uniform C₂H₆ EFs were consistently observed between smoldering samples, with an average EF of 1.79 ± 0.05 g/kg. CH₄ smoldering emissions were highest for wet manzanita at 10.70 ± 0.39 g/kg, and lowest for live eucalyptus at 5.78 ± 0.03 g/kg. The maximum CH₄ smoldering EF measured agrees with that seen in literature of smoldering biomass under wet conditions observed in the field [41]. With FMC increase, eucalyptus and coastal live oak displayed increasing HC EFs, for eucalyptus: 31.66 ± 2.44 g/kg when dry became 37.18 ± 0.65 g/kg when live, and for coastal live oak: 16.82 ± 1.44 g/kg when dry became 22.65 ± 1.34 g/kg live. In contrast, manzanita demonstrated its highest HC EF under dry smoldering conditions, 31.79 ± 0.54 g/kg compared with 18.27 ± 0.38 g/kg when live, suggesting different fuel-specific sensitivities to FMC. Live manzanita also experienced the highest MCE for smoldering conditions at 88.37 ± 0.86 %. While C₄H₁₀ was dominant for most fuels, the FMC during maximum measurement varied by fuel type, also observed in past literature correlating combustion phase and FMC [22]. For the cases of manzanita and coastal live oak, increases in FMC correlated to slight increases in MCE, a direct contrast to their MCE under flaming. C₂H₂ behavior under smoldering was more uniform regardless of FMC compared to its flaming emissions, measuring an average of 0.79 ± 0.02 g/kg.

3.3. Oxygenated volatile organic compounds (OVOCs)

Beyond the primary combustion products, the analysis of chaparral fuel emissions consistently revealed the presence of oxygenated volatile organic compounds (OVOCs) across every observed fuel condition. One finding was that the EFs for both methanol and formaldehyde were consistently higher during smoldering combustion, with respective averages of 1.72 ± 0.12 g/kg and 1.14 ± 0.12 g/kg compared to flaming with respective averages of 0.17 ± 0.03 g/kg and 0.17 ± 0.02 g/kg. The dominance of OVOCs during smoldering persisted regardless of the fuel types or FMC, which emphasizes the critical role of combustion modes in emitting these specific OVOCs.

Under flaming conditions shown in Figure 4 (a), (c), (e), and (g), distinct patterns in VOC emissions were observed. Wet eucalyptus exhibited the highest comparable EF for methanol at 1.10 ± 0.11 g/kg, suggesting that higher FMC in this fuel type can lead to increased methanol production during flaming [18]. Although methanol is primarily associated with smoldering emissions, the small methanol signals observed here are generally an order of magnitude lower than corresponding smoldering conditions (Figure 4 (b), (d), (f) and (h)). These lower-level signals likely reflect release during early devolatilization followed by incomplete oxidation due to short residence times and localized quenching near the fuel surface. The increase with FMC is consistent with reduced combustion completeness. Dry toyon and all conditions of coastal live oak consistently demonstrated little to no measured methanol under flaming conditions. Meanwhile, wet coastal live oak showed the lowest formaldehyde emissions under flaming conditions. Further examination of flaming trends showed that eucalyptus and manzanita had similar formaldehyde trends with respect to FMC, with average CH₂O EFs of 0.08 ± 0.01 g/kg and 0.36 ± 0.04 g/kg for dry and live conditions, respectively.

Fig. 4.

Average EFs of methanol and formaldehyde for (a) eucalyptus, (c) coastal live oak, (e) manzanita, and (g) toyon under flaming conditions; (b) eucalyptus, (d) coastal live oak, (f) manzanita, and (h) toyon under smoldering conditions

For smoldering conditions shown in Figure 4 (b), (d), (f), and (h), eucalyptus still emitted the highest methanol and formaldehyde, with wet eucalyptus leading for methanol at 2.55 ± 0.44 g/kg and live eucalyptus for formaldehyde at 2.82 ± 0.93 g/kg. In contrast, dry toyon consistently measured nearly negligible formaldehyde while wet manzanita showed the lowest methanol emission at 0.79 ± 0.18 g/kg. EF profiles of methanol showed similar patterns for eucalyptus and coastal live oak, with an average value of 2.01 ± 0.13 g/kg across different FMCs. These measured EF ranges of methanol were consistent with those reported in the literature for smoldering leaf fuels [42]. Formaldehyde, on the other hand, was more variable and distinct across different fuel and moisture conditions (p <0.001). The lowest EFs were measured under wet fuel conditions with an average of 0.51 ± 0.11 g/kg, while the highest EFs were observed under live fuel conditions with an average of 1.81 ± 0.47 g/kg.

3.4. Comparison with previous field and laboratory measurements

As seen in Table 4, we compared our small laboratory EF results for manzanita and eucalyptus against prior laboratory and field measurements. For manzanita, flaming combustion in our bench scale tests shows MCE values ranging from 0.91 to 1.00, closely matching the large-lab measured MCE of 96 ± 0.01% [16]. CO₂ EF also aligns well, ranging from 1783 to 1950 g/kg fuel in our lab compared to 1698 ± 16 g/kg in the large lab [16]. CO EFs in our tests are systematically lower (4.9–32.5 g/kg) than the large lab 40.6 ± 0.7 g/kg [16], which is expected because the results of the large laboratory and the field reflect combined emissions from heterogeneous fuels and mixed combustion phases, while our laboratory tests could isolate these phases by controlling combustion parameters. Additional differences may also arise from combustion scale effects, reduced flame ventilation, and partial smoldering in soil, which are more representative of field-scale conditions. For eucalyptus, our leaf-only smoldering tests show MCE between 0.84 and 0.85. Corresponding CO2 EFs range from 1512 to 1554 g/kg, which is slightly higher than small-scale burns (1000–1500 g/kg) [37] but bracketing field burns of sclerophyll vegetation complex reported at a mean of 1620 ± 160 g/kg [27]. CO emissions from our eucalyptus smoldering tests (167–189 g/kg) are comparable to values from small scale studies (180–450 g/kg) [37] and exceed those found in field measurements (120 ± 20 g/kg) [27]. These comparisons demonstrate how combustion efficiency, fuel chemistry, and fuel structure influence emissions during different combustion phases.

Table 4.

Cross-scale comparison of emission factors (EF, g/kg) and combustion efficiency (MCE, unitless) for manzanita and eucalyptus. Literature entries are means with one standard deviation in parentheses.

Study (scale)	This study (small lab)	Selimovic et al. (large lab)	This study (small lab)	Wang et al. (small lab)	Paton-Walsh et al. (field)
Fuel, mode	M, F	M, mixed	E, S	E*, S	SVC**, mixed
MCE	0.91–1.00	0.96 (0.01)	0.84–0.85	—	—
CO2	1782.75–1949.91	1698.45 (15.79)	1512.15–1554.09	1000.00–1500.00	1620.00 (160.00)
CO	4.86–32.52	40.62 (0.72)	166.66–189.33	180.00–450.00	120.00 (20.00)
CH4	0.00–1.02	1.14 (0.07)	5.78–9.61	9.00–16.00	3.60 (1.10)
NH3	0.01–0.28	0.52 (0.03)	0.28–0.67	—	1.60 (0.60)
C2H2	0.01–0.22	0.22 (0.09)	0.31–0.70	—	—
C3H6	0.11–0.22	0.17 (0.05)	0.56–0.65	—	—

M = manzanita; E = eucalyptus; F = flaming; S = smoldering; E* = eucalyptus bark, leaf and twig; SVC** = Sclerophyll vegetation complex (eucalypt forest, banksia/hakea heath, shrubland, woodland).

This study uses live chaparral flaming averages to provide a more realistic representation of gaseous emissions, as live fuels better reflect actual field conditions compared to dead or cured fuels. Chaparral, western shrubland, and grassland share comparable emission profiles that correlate with their similar volatile fuel chemistry and fine fuel structure [8, 28, 43]. As shown in Table 5, the measured MCE for chaparral averaged 0.926 ± 0.018, similar to shrubland at 0.935 ± 0.01 and grassland at 0.947 ± 0.018 [43, 44]. Emission factors for CO and CO₂ among these fuel types are also comparable, with chaparral producing CO EF at 86 ± 19 g/kg, slightly higher than shrubland (74 ± 18 g/kg) and grassland (61 ± 21 g/kg) [43, 44]. By contrast, CH₄ in chaparral was strongly suppressed at 0.59 ± 0.51 g/kg compared to shrubland (3.69 ± 1.36 g/kg) and grassland (1.95 ± 1.05 g/kg), suggesting subtle differences in combustion pathways or fuel chemistry. Where chaparral diverges most strongly is in its nitrogen-containing emissions. Flaming chaparral produced NO_x EF of 3.90 ± 0.89 g/kg, nearly double the values reported for the other ecosystems. At the same time, chaparral’s NH₃ EF was very low at 0.027 ± 0.015 g/kg compared to shrubland and grassland. This inverse relationship between NOx and reduced nitrogen species likely reflected the dominance of the pure flaming phase in our experiments, where fuel nitrogen is efficiently oxidized into NO_x. In contrast, large-scale field and prescribed fire studies incorporated more smoldering combustion that increased NH₃ and depressed NO_x, resulting in lower NO_x/NH₃ ratios overall [43, 44]. Southwestern conifer fuels differ more substantially due to their woody, needle-dominated structure and distinct resin chemistry. The conifer MCE was slightly lower at 0.924 ± 0.015, with the lowest CO₂ of 1653 ± 34 g/kg, while its CH₄ and NH₃ values sat between those of chaparral and shrubland/grassland ecosystems.

Table 5.

Cross-ecosystem comparison of emission factors (g/kg) and combustion efficiency (MCE, unitless). Errors shown in parentheses represent one standard deviation. Chaparral column uses live-flaming averages from this study (toyon excluded).

Study	This study	Yokelson et al.	Urbanski et al., Yokelson et al.	Urbanski et al., Yokelson et al.
Ecosystem, mode	Chaparral, flaming	Western shrubland, mixed	Grassland, mixed	SW conifer, mixed
MCE	0.926 (0.018)	0.935 (0.017)	0.947 (0.018)	0.924 (0.015)
CO2	1736 (119)	1674 (38)	1705 (44)	1653 (34)
CO	86 (19)	74 (18)	61 (21)	87 (18)
CH4	0.59 (0.51)	3.69 (1.36)	1.95 (1.05)	3.15 (0.91)
NH3	0.027 (0.015)	1.50 (1.43)	1.50 (1.43)	0.50 (0.69)
NOx	3.90 (0.89)	2.18 (0.78)	2.18 (0.78)	1.88 (1.03)

Chaparral live-flaming averages are computed from Eucalyptus, Live Oak, and Manzanita only. SW conifer represents the prescribed fire in southwestern conifer forest.

Conclusions

Our results demonstrate that chaparral fuels exhibit both similarities and unique distinctions compared to other vegetation or ecosystem types. Consistent with earlier chaparral studies [4, 8], flaming combustion produces higher NO_x and CO₂, while smoldering favors elevated emissions of CO, hydrocarbons, and OVOCs. Notably, the magnitude of these emission differences in chaparral exceeds those reported for conifer fuels under comparable laboratory conditions [22, 38]. For instance, live eucalyptus released HCs up to 37 g/kg under smoldering conditions, compared to typical pine needle values below 20 g/kg [38]. Apart from these fuel-specific chemical influences, FMC was also identified as a dominant control on emissions and combustion efficiency. Elevated FMC shifted combustion to lower MCE, increasing CO and hydrocarbon emissions while reducing CO₂ production. These trends align with prior observations in ponderosa pine and eucalyptus litter burns [19, 37, 38]. In chaparral, in particular, this transition was more pronounced. For example, CO EF of manzanita under flaming conditions increased more than 20 times from dry to live conditions in our study, highlighting strong FMC sensitivity. TPM emissions demonstrated strong combustion phase dependencies, with smoldering producing substantially higher TPM EF than flaming across all chaparral species. This amplification ratio showed strong fuel-specific responses to FMC, indicating distinct fuel-specific moisture sensitivities that should be considered in future emission modeling.

Further comparisons were made across ecosystems, revealing that while chaparral shares similar emission profiles with other fine fuel types, nitrogen-containing gaseous emissions differ primarily due to experimental scale effects. Our laboratory-controlled flaming conditions produced different nitrogen speciation patterns compared to field studies that incorporate mixed combustion conditions, even under similar MCEs. Field-scale emissions may also reflect combustion scale, reduced flame ventilation, and partial soil smoldering not fully captured in laboratory experiments. These findings emphasize the critical need to incorporate phase- and FMC-resolved chaparral emission factors into regional and national fire emission inventories to improve accuracy, especially for Mediterranean-type ecosystems like southern California where chaparral dominates.

Highlights:

• CO and hydrocarbon emissions increased with FMC for most species

• Smoldering produced 10-70x higher TPM than flaming, varying by species/FMC

• Manzanita shows the highest CO variability under flaming conditions

• Eucalyptus exhibited the highest hydrocarbon emissions during smoldering