This work
aims at quantifying the relative contribution of secondary organic aerosol
(SOA) precursors emitted by wildfires to organic aerosol (OA) formation
during summer of 2007 over the Euro-Mediterranean region, where intense
wildfires occurred. A new SOA formation mechanism,
Most of the particle OA concentrations are formed from
I/S/L-VOCs. On average during the summer of 2007 and over the Euro-Mediterranean
domain, they are about 10 times higher than the OA concentrations formed from
VOCs. However, locally, the OA concentrations formed from VOCs can represent
up to 30 % of the OA concentrations from biomass burning. Amongst the VOCs,
the main contributors to SOA formation are phenol, benzene and catechol (CAT; 47 %); USC
Estimating the gaseous I/S/L-VOC emissions from POA or from NMOG has a high
impact on local surface PM
Atmospheric particulate matter (PM) has a strong impact on human health
Organic aerosols (OA) are classified either as primary organic aerosols (POA) or as secondary
organic aerosols (SOA). POA are directly emitted into the atmosphere, whereas SOA are
formed by gas–particle conversion of oxidation products of precursors. OA can
be classified based on their saturation concentrations (
POA
Wildfires are one of the largest sources of primary carbonaceous aerosols
globally. They are also an important source of trace gases including organic
vapors, which themselves can serve as precursors of SOA
SOA formation mechanisms may rely not only on smog chamber experiments but
also on explicit chemical mechanisms when experimental data are not
available. Examples of such mechanisms are the master chemical mechanisms
Recent studies take into account not only the oxidation of selected VOCs but
also gaseous intermediate volatility, semi-volatile and low-volatility organic compounds (I/S/L-VOCs) emitted by biomass burning to model SOA formation
Although primary gaseous I/S/L-VOCs are not considered to be or classified as
unspeciated NMOG in emission inventories, their contribution to the SOA
budget may be substantial, despite being a small fraction of the overall
organic gas emissions
The objective of this work is to quantify the contribution of recently
identified SOA precursors from wildfires (guaiacol, syringol, benzene,
phenol, catechol, cresol, furan, naphthalene, methylnaphthalene and USC
This study aims also to quantify the relative contribution of VOCs and I/S/L-VOCs to OA formation. The OA concentrations are simulated using the chemistry transport model (CTM) Polair3D of the Polyphemus modeling air-quality platform.
This study focuses on two severe fire events that occurred during the summer
of 2007 over the Euro-Mediterranean area.
This paper is structured as follows. Section 2 details the SOA formation
mechanisms from VOCs and I/S/L-VOCs. Then, Sect. 3 describes the model and
the simulation setup during summer of 2007. The main OA
This section presents a new SOA formation mechanism
Laboratory chamber studies provide the fundamental data that are used to
parameterize the atmospheric SOA formation under low- or high-
For each VOC, precursor of SOA and chamber experiment, the SOA mass
yield (
The chamber experimental results are analyzed according to the absorption
gas–particle partitioning model developed by
Under low-
In their studies, and in agreement with the explicit chemical mechanism MCM
version 3.3.1 (MCM.v3.3.1), CAT is the dominant product of the first
oxidation step of phenol. Therefore, catechol is assumed to be the main
intermediary leading to SOA formation from OH oxidation of phenol following
Reaction (
The one-product model with a stoichiometric coefficient
SOA yield from smog chamber experiments under low-
Finally, the oxidation of catechol is modeled following Reaction (
As detailed in the chemical mechanism MCM.v3.3.1, the OH oxidation of cresol
(CRESp) leads to the formation of methylcatechol (MCAT), which is the
dominant product of the first oxidation step of cresol, presented in
Reaction (
The oxidation of methylcatechol by OH leads to the formation of SOA,
following a chemical mechanism detailed in
SOA yield data from smog chamber under low-
Finally, the oxidation of methylcatechol is modeled following
Reaction (
Several studies focus also on the oxidation of cresol by
According to MCM.v3.3.1, benzene (BENZ) reacts with OH to form phenol, as
presented in Reaction (
According to MCM.v3.3.1, furan (FUR) reacts with OH to form an unsaturated
1,4-dicarbonyl product (butendial – ButDial), following Reaction (
According to MCM.v3.3.1, ButDial reacts with OH to form highly volatile
products (not detailed here because they may not form SOA) and a radical
(RADButenalCOO), as presented in Reaction (
Under high-
Under low-
ButenalCOOH is mostly in the gas phase (
Note that the oxidation mechanism of furan presented in this section probably
overestimates the SOA concentrations from the OH-oxidation route because
several reactions such as ozonolysis and photolysis of both ButenalCOOH and
Butenal(COOH)
Furthermore, other routes may be more efficient at forming SOA from furan.
According to
The parameterization is developed for syringol and guaiacol by considering
low-
SOA experimental and modeled yield data from smog chamber for
syringol under low-
The second reaction step for SOA formation is then represented with the
following Reactions (
Similarly, for guaiacol, the two
SOA experimental and modeled yield data from smog chamber for
guaiacol under low-
The surrogate compound chosen to represent SOA formation in both conditions
is
The second part of the OH-oxidation mechanism for guaiacol follows Reactions (
As detailed in
It is
not easy to design a chemical mechanism for the structurally assigned and
unassigned compounds with at least six carbon atoms per molecule (USC
Table
Different parameterizations may be used to describe the formation of SOA from the gaseous I/S/L-VOCs emitted from wildfires, with or without an ageing scheme: a one-step oxidation scheme (no ageing) and multi-generational oxidation scheme.
In the one-step oxidation scheme, used, for example, in
Compared to the primary products, the volatility of the secondary products is
reduced by a factor of 100, and their molecular weight is increased by 40 %
For the multi-generational scheme, the VBS approach based on the hybrid
VBS
The impact of wildfires on PM concentrations and optical depths in the
Euro-Mediterranean during the summer of 2007 was studied by
Here, the CTM Polair3D or Polyphemus
Two domains are considered in this study (Fig.
Simulation domains,
including one large domain (with a 0.5
The CB05 gas-phase chemical mechanism is used in conjunction with the
chemical mechanism
As in
Dry deposition of gaseous I/S/L-VOCs from wildfires is parameterized based on
The reference simulation uses the same setup as
To assess the relative influence of emissions of VOCs and I/S/L-VOCs from
wildfires on OA concentrations, six sensitivity simulations are performed.
The setup of the different simulations is summarized in Table
The reference simulation OnestepISLVOC uses the default setup, i.e., the setup used in the
previous study
To assess the impact of VOCs on SOA formation, the Simulation Multstep-withVOC
uses the same setup as the simulation MultstepISLVOC, but all the VOCs, which are SOA precursors, are added to the
model, as detailed in Sect.
The sensitivity of two parameters involved in the modeling of the ageing of
these VOCs is also assessed: the enthalpy of vaporization (
Several studies consider
Summary of the sensitivity simulations performed by Polyphemus (n/a: not applicable).
To better understand the contribution of OA
Daily fire emissions of
toluene, xylene, phenol, benzene and furan are estimated by the APIFLAME fire
emission model
However, cresol, catechol, syringol, guaiacol, naphthalene and methylnaphthalene
emission factors are missing from the
For two types of vegetation
Considering only these two types of vegetation (crop residue and chaparral)
for which emission ratios are available may lead to an underestimation of
the emission factors and therefore the emissions of cresol, catechol,
guaiacol, syringol, naphthalene and methylnaphthalene emissions. Indeed,
Fig.
Percentage of the different vegetation types in the burned area detected over the subregion during the summer of 2007.
According to
Emissions of the OA
Relative contribution of VOCs to gaseous precursors (VOCs plus gaseous I/S/L-VOCs; %) emitted by wildfires over the Mediterranean area during the summer of 2007.
Number of burned area detections for temperate forest on 25 July 2007.
The gaseous I/S/L-VOC emissions from wildfires are estimated either from the
POA emissions released from wildfires, by multiplying them by a constant
ratio of
Fig. 7a presents the emissions of total
(gas plus particle) OA
The spatial distribution of the relative contribution of VOCs to gaseous
precursors emissions (I/S/L-VOCs from NMOG plus VOCs) is assessed in
Fig.
Fig. 7b shows the distribution of VOCs
between the different compounds emitted over the subregion during the summer of 2007. USC
Mean surface OA
Daily mean surface OA concentrations from wildfires
The influence of VOCs and I/S/L-VOCs on OA and OA
Distribution of the OA concentrations formed from the different VOCs emitted by wildfires over the subregion during the summer of 2007 (simulation Multstep-withVOC).
Sensitivity of surface PM
Daily mean POA
Figure
The emissions of VOCs are lower than those of gaseous I/S/L-VOCs estimated
from NMOG (or POA) emissions by almost a factor of about 2.5. This
preponderance of I/S/L-VOCs is observed not only for emissions but also for
concentrations. The primary and secondary OA concentrations from gaseous
I/S/L-VOCs (estimated from NMOG emissions and from POA emissions) are about
10 times higher than the OA concentrations from VOCs. Most of the OA and
OA
Across our cases, 28 % to 42 % of the OA concentrations from I/S/L-VOC emissions are primary. The amount of POA from I/S/L-VOC emissions in simulation OnestepISLVOC (28 %) is lower than the one in the simulation MultstepISLVOC (42 %) because of the differences in the volatility properties of the species in the two ageing schemes.
The OA concentrations simulated with the one-step and the multi-generational
schemes are nearly similar (about 5 % difference). However, the primary and
secondary OA
A large part of OA
Using the SOA formation mechanism of naphthalene rather than the SOA
formation mechanism of phenol affects the OA
Figure
Figure
To assess the sensitivity of PM
Figure
Concerning the influence of the gaseous I/S/L-VOC ageing scheme, the relative
differences between the simulations OnestepISLVOC and MultstepISLVOC are low
(below 5 %). The differences can be positive or negative because the
one-step oxidation scheme and the multi-step oxidation schemes lead to SOA of
different volatilities. The sign of the differences depends on the SOA
volatilities and on the partitioning between the gas and the particle phases
of I/S/L-VOCs, which itself depends on PM
The emissions of the added VOCs (namely benzene, phenol, cresol, catechol,
furan, guaiacol, syringol, naphthalene, methylnaphthalene, and the
structurally assigned and unassigned compounds with at least six carbon atoms
per molecule (USC
Estimating the gaseous I/S/L-VOC emissions from POA rather than from NMOG
results in higher local PM
This study quantified the relative contribution of OA
Data can be requested from the corresponding author (marwa.majdi@enpc.fr).
The VOCs that are SOA precursors and their emission
factors (EFs) and SOA yields. EFs from
Properties of the compounds added to the model.
Chemical structure of SOA compounds considered in this study.
Reactions leading to SOA formation added to CB05.
Ageing mechanism of I/S/L-VOCs using Couvidat approach
Properties of primary and secondary I/S/L-VOCs.
Ageing mechanism of I/S/L-VOCs using Ciarelli approach
Properties of the VBS species (primary and secondary I/S/L-VOCs).
Summary of the parameters used to compute the dry-deposition velocities of the gaseous I/S/L-VOCs.
MM, KS, GL and FC developed the chemical mechanisms. ST and MM prepared VOC emissions from fires. MM performed the simulations, with help from MC and KS for the post-processing. MM, KS and GL prepared the paper with contributions from all co-authors.
The authors declare that they have no conflict of interest.
CEREA is a member of the Institut Pierre-Simon Laplace (IPSL). A PhD grant from École des Ponts ParisTech partially funded this research.
This paper was edited by Alma Hodzic and reviewed by two anonymous referees.