Content uploaded by Dmitry Schepaschenko
Author content
All content in this area was uploaded by Dmitry Schepaschenko on Oct 13, 2014
Content may be subject to copyright.
Available via license: CC BY 3.0
Content may be subject to copyright.
Differences in satellite-derived NO
x
emission factors between Eurasian
and North American boreal forest fires
S.F. Schreier
a
,
b
,
*
, A. Richter
a
, D. Schepaschenko
b
, A. Shvidenko
b
, A. Hilboll
a
,
J.P. Burrows
a
a
Institute of Environmental Physics, University of Bremen, Germany
b
International Institute for Applied Systems Analysis, Laxenburg, Austria
highlights
A satellite-based approach to estimate NO
x
EFs for boreal forests is presented.
The results indicate differences between Eurasian and North American boreal forests.
Our EFs are in good agreement with recent reported values.
However, EFs applied in frequently used emission inventories are 3e5 times larger.
article info
Article history:
Received 1 April 2014
Received in revised form
26 August 2014
Accepted 27 August 2014
Available online xxx
Keywords:
Satellite measurements
Tropospheric NO
2
Fire radiative power
NO
x
emission factor
Boreal forest
abstract
Current fire emission inventories apply universal emission factors (EFs) for the calculation of NO
x
emissions over large biomes such as boreal forest. However, recent satellite-based studies over tropical
and subtropical regions have indicated spatio-temporal variations in EFs within specific biomes. In this
study, satellite measurements of tropospheric NO
2
vertical columns (TVC NO
2
) from the GOME-2 in-
strument and fire radiative power (FRP) from MODIS are used for the estimation of fire emission rates
(FERs) of NO
x
over Eurasian and North American boreal forests. The retrieval of TVC NO
2
is based on a
stratospheric correction using simulated stratospheric NO
2
instead of applying the reference sector
method, which was used in a previous study. The model approach is more suitable for boreal latitudes.
TVC NO
2
and FRP are spatially aggregated to a 1
1
horizontal resolution and temporally averaged to
monthly values. The conversion of the satellite-derived tropospheric NO
2
columns into production rates
of NO
x
from fire (P
f
) is based on the NO
2
/NO
x
ratio as obtained from the MACC reanalysis data set and an
assumed lifetime of NO
x
. A global land cover map is used to define boreal forests across these two regions
in order to evaluate the FERs of NO
x
for this biome. The FERs of NO
x
, which are derived from the gradients
of the linear relationship between P
f
and FRP, are more than 30% lower for North American than for
Eurasian boreal forest fires. We speculate that these discrepancies are mainly related to the variable
nitrogen content in plant tissues, which is higher in deciduous forests dominating large parts in Eurasia.
In order to compare the obtained values with EFs found in the literature, the FERs are converted into EFs.
The satellite-based EFs of NO
x
are estimated at 0.83 and 0.61 g kg
1
for Eurasian and North American
boreal forests, respectively, which is in good agreement with the value found in a recent emission factor
compilation. However, recent fire emission inventories are based on EFs of NO
x
that are 3e5 times larger,
which indicates that there are still large uncertainties in estimates of NO
x
from biomass burning,
especially on the regional scale.
©2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/3.0/).
1. Introduction
Boreal forest, commonly referred to as taiga, is one of the largest
terrestrial biomes and covers about 30% of the total global forest
area (Pan et al., 2011). The taiga is dominated by evergreen
(coniferous) and deciduous forests storing large amounts of
*Corresponding author. Institute of Environmental Physics, University of Bre-
men, Otto-Hahn-Allee 1, D-28359 Bremen, Germany.
E-mail address: schreier@iup.physik.uni-bremen.de (S.F. Schreier).
Contents lists available at ScienceDirect
Atmospheric Environment
journal homepage: www.elsevier.com/locate/atmosenv
http://dx.doi.org/10.1016/j.atmosenv.2014.08.071
1352-2310/©2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/3.0/).
Atmospheric Environment xxx (2014) 1e11
Please cite this article in press as: Schreier, S.F., et al., Differences in satellite-derived NO
x
emission factors between Eurasian and North
American boreal forest fires, Atmospheric Environment (2014), http://dx.doi.org/10.1016/j.atmosenv.2014.08.071
nitrogen (N), which is an essential nutrient for all living organisms.
However, wildfires across these ecosystems release large masses of
N in form of nitrogen oxides (NO
x
¼NO þNO
2
) and ammonia (NH
3
)
by flaming and smoldering fires, respectively, and affect the
biosphere and atmosphere.
In general, the total amount of emissions released from boreal
forest fires is relatively low compared to tropical and subtropical
vegetation fires (Lobert et al., 1999). However, during exceptional
climatic years such boreal forest fires can spread over large areas.
Interestingly, Wooster and Zhang (2004) found that the intensity of
fires is generally lower in Russian boreal forests than in North
American boreal forests, and relate this difference to the respective
dominance of surface fires in Russia and crown fires in North
America.
Current model estimates of future fire regimes in the boreal
zone predict a doubling of the number of wildfires by the end of the
21st century, an increase in the occurrence of catastrophic fires and
fires that escape from control, a substantial increase of the burning
severity, and changes in the intensity and gas composition of fire
emissions caused by increased soil burning (Flannigan et al., 2009).
Moreover, Shvidenko et al. (2011) suggested a linkage between
catastrophic fires and large scale atmospheric anomalies for recent
exceptionally large wildfires (e.g. fires in the Russian Far East vs.
flooding in China in 2003; fires in European Russia vs. flooding in
Pakistan and India in 2010). Thawing permafrost and the following
drying up of typical landscapes in higher latitudes could lead to a
dramatic loss of forested areas (Shvidenko and Schepaschenko,
2013). The increasing occurrence of catastrophic boreal wildfires
in the future is likely to have an impact on air pollution and at-
mospheric chemistry. Therefore, accurate estimates of current fire
emissions are needed to better understand the increasing future
role of boreal forest fires. This can only be achieved by reducing
uncertainties in fire emission estimates.
During the burning process, the nitrogen bound in the fuel is
converted in part into oxides and N present in amino acids is
converted to NO. However, NO
x
may also result from the reaction of
molecular nitrogen (N
2
) with O
2
from the atmosphere at very high
temperatures (Andreae and Merlet, 2001). While the higher ther-
mal energy during the flaming stage leads to the break-up of plant
materials into simpler N molecules enabling a full oxidation to NO
x
,
smoldering fires are rather incomplete. Chen et al. (2010) have
examined the combustion efficiency by laboratory-controlled
combustion experiments. Their results show that the fuel mois-
ture content decreases the combustion efficiency and prolongs the
smoldering phase before flames start. Emission factors (EFs) of NO
x
,
which are used in bottom-up emission inventories for the trans-
lation of biomass burned into trace gas emissions, can consequently
change as a function of the fuel moisture content.
The amount of NO
x
emissions being released from forest fires
also depends on the N content in plant tissues. For boreal forest
plants, this content varies substantially and depends on species,
plant part burned, site productivity, geographical location, and
other factors. For instance, in Siberian dark coniferous forests
(dominated by Pinus sibirica,Abies sibirica,Picea excelsa and Picea
obovata) of the southern taiga zone, the N content varies in the
range of 0.1e0.45% (of dry matter mass) in stem wood, 0.5e1.0% in
branches, 0.8e1.8% in needles, 0.3e0.9% in bark, and 0.4e1.2% in
roots (Protopopov, 1975). Similar relative amounts are found for
light coniferous forests (Rodin and Bazilevich, 1965; Bazilevich and
Rodin, 1971). In the taiga of West Siberia, lichens contain 0.5%,
mosses 0.9%, and grasses 1.1e2.0% of N (Bakhnov, 2001). The results
of a study in northern Alberta suggest that deciduous forests
(mainly trembling aspen) have a higher N content than coniferous
(mainly white spruce) and mixed forests (Jerabkova et al., 2006).
Although the differences in total litter mass and total N content in
the canopy are negligible between the three forest types, the input
of leaves and/or needles in terms of mass and N content is higher in
deciduous stands. Moreover, it was shown that the nitrogen frac-
tion in the observed smoke plume differs considerably among
various fuel types burned in laboratory-controlled combustion
experiments. While the concentration ratio of NO
x
to grand total
carbon (including C in CO
2
, CO, and PM
2.5
) is higher than 20% for
litter, it is less than 10% for other fuel types such as leaves and stems
(Chen et al., 2010). The results described above clearly indicate that
the amounts of fuel nitrogen in the vegetation are rather hetero-
geneous throughout the taiga. Consequently, the EF of NO
x
can vary
within the boreal forest due to a changing N content in different
tree species and fuel types that are burned.
There has been much discussion about the leading cause of NO
x
emissions from wildfires. Andreae and Merlet (2001) suggest that
temperatures in typical wildfires are not high enough to produce
large amounts of NO
x
via the oxidation of N
2
. Nevertheless, the
production of NO
x
is more efficient during the flaming combustion.
The fraction of reactive nitrogen in volatilized fuel nitrogen emitted
accounts for 25e50%, with NO
x
and NH
3
being the dominant
reactive nitrogen species during flaming and smoldering combus-
tion, respectively (Goode et al., 2000; Yokelson et al., 2008). In
comparison, up to 50% of the fuel nitrogen might be converted into
N
2
(Lobert et al., 1990). However, the exact mechanism of N
2
for-
mation remains unclear as the detection of N
2
from open fires is
influenced by the large N
2
fraction of the ambient air. For example,
Mebust and Cohen (2013) found a seasonal cycle in NO
x
emissions
from African woody savanna fires, but could only speculate about
the exact mechanisms.
The composition of total reactive nitrogen oxides (NO
y
) in the
continental troposphere mainly consists of nitric acid (HNO
3
) and
peroxyacyl nitrates (PANs), whereas NO
x
constitutes only 15%
(Singh et al., 2007). Smaller NO
x
/NO
y
fractions of 5% (spring) and
10% (summer) were derived from in-situ measurements taken
during the ARCTAS (Arctic Research of the Composition of the
Troposphere from Aircraft and Satellites) airborne campaigns in the
high northern latitudes (Singh et al., 2010). Nevertheless, it was
shown that the relatively small amount of NO
x
can be transported
into remote areas where it leads to a significant increase of O
3
(Singh et al., 2007). This can be explained by the efficient ozone
production at low NO
x
levels (Jacob, 1993). Val Martin et al. (2008)
have shown that boreal wildfire emissions were responsible for
higher levels of NO
x
in remote areas downwind from the boreal
region. They further concluded that the NO
x
background levels
during such fires were increased, and thus, the tropospheric O
3
budget was affected over large parts of the northern hemisphere.
Recent fire emission inventories are based on the translation of
estimated biomass burned into trace gas emissions by applying
uniform EFs for a relatively small number of biomes (e.g. Van Der
Werf et al., 2010; Kaiser et al., 2012). However, more recent
satellite-based studies have indicated substantial spatio-temporal
variations of NO
x
EFs within a specific biome and between
different regions (Mebust and Cohen, 2013; Castellanos et al., 2014;
Schreier et al., 2014). As the bulk of these results are confined to
tropical and subtropical regions, we here expand this research to
higher latitudes and estimate fire emission rates (FERs) and EFs of
NO
x
for boreal forests.
The following Sect. 2gives a description of satellite measure-
ments used in this study and outlines the approach to estimate
FERs and EFs of NO
x
from these measurements. The results are
presented and discussed in Sect. 3. An overview on possible un-
certainties in the approach is given in Sect. 4, followed by a sum-
mary and conclusions in Sect. 5.
S.F. Schreier et al. / Atmospheric Environment xxx (2014) 1e112
Please cite this article in press as: Schreier, S.F., et al., Differences in satellite-derived NO
x
emission factors between Eurasian and North
American boreal forest fires, Atmospheric Environment (2014), http://dx.doi.org/10.1016/j.atmosenv.2014.08.071
2. Data retrieval and data analysis
2.1. Satellite measurements of NO
2
Measurements from the GOME-2 instrument on board the
MetOp-A satellite (Callies et al., 2004) are used for the retrieval of
tropospheric NO
2
. The retrieval is based on the Differential Optical
Absorption Spectroscopy (DOAS) method, which is described in
detail elsewhere (Platt and Stutz, 2008). Here, we follow the
retrieval approach as described in Richter et al. (2011) and Schreier
et al. (2014). Briefly, the slant column densities (SCDs) are retrieved
from the GOME-2 spectral measurements by fitting the absorption
cross section of NO
2
and other trace gases in the spectral window
between 425 and 497 nm. Instead of applying the reference sector
method (Richter and Burrows, 2002), the stratospheric influence on
the total column measurements has been estimated using strato-
spheric NO
2
fields simulated by the Bremen 3d CTM (B3dCTM) and
scaled to observations over the Pacific.
The B3dCTM is a combined model approach based on the
“Bremen transport model”(Sinnhuber et al., 2003a) and the
chemistry code of the “Bremen two-dimensional model of the
stratosphere and mesosphere”(Sinnhuber et al., 2003b; Winkler
et al., 2008), which evolved from SLIMCAT (Chipperfield, 1999).
B3dCTM is driven by ECMWF ERA Interim meteorological rean-
alysis fields (Dee et al., 2011). The detailed model setup is described
in Hilboll et al. (2013b). The simulated stratospheric profiles are
interpolated in space and time for each satellite measurement to
yield stratospheric vertical columns and airmass factors (AMFs)
(Hilboll et al., 2013a).
The impact of clouds on the retrieval of NO
2
is accounted for by
removing measurements with a cloud fraction larger than 20%,
based on the FRESCOþretrieval (Wanget al., 2008). Finally, the SCDs
are converted into vertical column densities (VCDs) by dividing
through an AMF, which corrects for the different sensitivity of the
measurements in different altitudes. A detailed description of the
AMFs used in this study can be found in Nüß (2005) and Richter et al.
(2005). The final retrieved tropospheric NO
2
vertical columns (TVC
NO
2
) are binned to a horizontal resolution of 1
1
in order to
reduce the effect of horizontal transport of NO
2
.
Uncertainties in tropospheric NO
2
slant columns originating
from the stratospheric correction are usually up to
510
14
molec cm
2
, but can be as large as 2.5 10
15
molec cm
2
at
high latitudes in winter (Hilboll et al., 2013a). Due to the absence of
fires in winter, however, these larger uncertainties are not of rele-
vance for this study. The satellite-based retrieval of TVC NO
2
is
affected by uncertainties, which are mainly caused by the conver-
sion of SCDs into VCDs by applying AMFs (Boersma et al., 2004). A
priori information used for the calculation of AMFs introduces the
bulk of these uncertainties into the retrieval of TVC NO
2
. For
instance, inaccurate a priori information on aerosols can have a
substantial influence on the accuracy of the AMFs, and thus, on the
precision of TVC NO
2
. The vertical position of the aerosol layer
relative to the NO
2
layer is of particular interest. While the mea-
surement sensitivity of the instrument is reduced when the aerosol
layer is located above the NO
2
plume, it is increased when aerosols
are located within or below the NO
2
plume (Leit~
ao et al., 2010). A
decrease (increase) in the measurement sensitivity by not correctly
accounting for the aerosol information would lead to an over-
estimation (underestimation) of the AMFs, and thus, to an under-
estimation (overestimation) of TVC NO
2
. Besides the location of the
aerosol layer, uncertainties in aerosol amounts and optical prop-
erties can also deteriorate the accuracy of TVC NO
2
.Leit~
ao et al.
(2010) have shown that the effect of a varying single scattering
albedo (SSA) on the AMF can be larger than 70% for polluted
atmospheres.
2.2. Satellite measurements of fire radiative power
Fire radiative power (FRP) is a measure for the power radiated
from a fire in terms of its temperature, based on the black body
concept described by the StefaneBoltzmann law (Kaufman et al.,
1998). MODIS observations on board the near-polar orbiting sat-
ellites Terra (10:30 LT) and Aqua (13:30 LT) are available since 1999
and 2002, respectively. Amongst other parameters, FRP is retrieved
from these measurements at a 1 km
2
horizontal resolution. Ac-
cording to the equatorial overpass time of the GOME-2 instrument
(9:30 LT), we make use of the MOD14CM FRP product from the
Terra satellite (10:30 LT), provided at a 1
1
horizontal resolution
(ftp://neespi.gsfc.nasa.gov/data/s4pa/Fire/MOD14CM1.005/).
2.3. Global land cover map
The Collection 5 MODIS Global Land Cover Type product (Friedl
et al., 2010), which is available at https://lpdaac.usgs.gov/products/
modis_products_table/mcd12q1, is used for the definition of boreal
forests. According to the horizontal resolution selected for T VC NO
2
and FRP, the land cover product is spatially aggregated to 1
1
by
applying a majority filter. The 14-class University of Maryland
classification (UMD) is used to define boreal forest pixels (Hansen
et al., 2000). The aggregation of 1
1
pixels covered by conif-
erous forests, deciduous forests, or a mixture of both form the basis
of the estimation of FERs and EFs of NO
x
for boreal forests in Eurasia
and North America. In the UMD classification, woody savanna is
defined as a mixture of trees (40e60%) and grassland. As woody
savannas cover considerable areas in Alaska, Russia, and Scandi-
navia, and in order to increase the data points used for the esti-
mation of FERs of NO
x
, we included woody savannas in the
definition of boreal forest pixels. The aggregation of the four land
cover types (evergreen needleleaf forest, deciduous needleleaf
forest, mixed forest, and woody savannas) between 50
N and 80
N
is shown in Fig. 1 and the definition of the regions Eurasia and
North America is highlighted in Table 1.
Fig. 1. Boreal forests in Eurasia and North America as defined by the aggregation of
evergreen needleleaf forest, deciduous needleleaf forest, mixed forest, and woody
savannas on a grid of 11horizontal resolution (see Table 1) using the UMD
classification scheme (Hansen et al., 2000; Friedl et al., 2010).
S.F. Schreier et al. / Atmospheric Environment xxx (2014) 1e11 3
Please cite this article in press as: Schreier, S.F., et al., Differences in satellite-derived NO
x
emission factors between Eurasian and North
American boreal forest fires, Atmospheric Environment (2014), http://dx.doi.org/10.1016/j.atmosenv.2014.08.071
2.4. Satellite measurements of aerosol optical depth
In order to assess the possible influence of aerosols on the
retrieval of TVC NO
2
, and thus, on the magnitude of the estimated
FERs and EFs of NO
x
for boreal forests in Eurasia and North America,
we have analyzed the aerosol optical depth (AOD) and vertical
profiles of wildfire emissions for these regions. The AOD is a
parameter describing the column integrated extinction over the
entire atmosphere and is derived from MODIS spectral measure-
ments between 470 and 2130 nm (Remer et al., 2008). The aerosol
product from MODIS on board Terra at 550 nm is used to evaluate
the possible relative influence of the AOD on the retrieval of TVC
NO
2
between the Eurasian and North American boreal forest pixels.
The AOD product has been downloaded from ftp://ladsweb.
nascom.nasa.gov/allData/51/MOD08_M3/2005/.
2.5. Vertical profiles of wildfire emissions
A 4-dimensional data set of fire smoke with a resolution of
1
1
500 m has been recently calculated by Sofiev et al. (2013).
The estimated vertical profiles are based on a semi-empirical for-
mula for the plume-top height, satellite observations of active fires,
and meteorological conditions derived from a numerical weather
prediction model. The vertical profiles are available for the entire
globe and cover the whole time period analyzed in this study
(2007e2012). These data are used to deduce the possible role of a
variable mean injection height on the retrieval of TVC NO
2
between
the two regions.
2.6. An adapted approach for the boreal region
In general, fire emission inventories (e.g. GFEDv3.1 and
GFASv1.0) estimate the total amount of NO
x
(usually reported as
NO) released by fires. As satellite instruments, such as GOME-2,
measure NO
2
, the conversion of TVC NO
2
into production rates of
NO
x
from fire (P
f
) is a further step to prepare the data for the
analysis.
The aim of this study is the satellite-based estimation of FERs
and EFs of NO
x
for the Eurasian and North American boreal forests.
As recent satellite-based studies have indicated substantial spatio-
temporal variations in EFs for African (Mebust and Cohen, 2013)
and South American (Castellanos et al., 2014) biomes, we hypoth-
esize that EFs could also fluctuate among the large taiga forests.
We build on the approach described in Schreier et al. (2014) to
estimate linear gradients between P
f
and FRP, here referred to as
FERs of NO
x
. The assumption made for the following approach is
that the fire radiative power is mainly related to the amount of fuel
burned and not to the temperature of the individual fire. The
amount of NO
x
emitted in turn is assumed to be linearly related to
the amount of fuel burned.
We start the analysis with evaluating the temporal correlation
between the gridded monthly mean T VC NO
2
and FRP for the boreal
forest pixels. Schreier et al. (2014) have indicated strong correlation
for larger regions located in the tropics and subtropics. However,
they also pointed out the comparatively weak correlation between
the two parameters for higher latitudes. One possible reason for the
weaker correlation is related to the fact that the bulk of boreal
forest fires are generally smaller in size (e.g. Stocks et al., 2003)
when compared to the rather extensive slash and burn activities in
Africa and South America. As a result, the measurement sensitivity
of the satellite instrument might be too low for the detection of NO
2
produced from such smaller fires. Another possible explanation for
the weak correlations found by Schreier et al. (2014) could be the
larger uncertainty in tropospheric NO
2
vertical columns in higher
latitudes, which are introduced by the stratospheric correction
method used in their study. Basically, the applied reference sector
method assumes that there are no tropospheric sources of NO
x
over
a specific and rather remote region over the Pacific and that the
stratospheric NO
2
column varies only with latitude, but not with
longitude. However, these assumptions can introduce negative NO
2
columns by overestimating the stratospheric NO
2
, especially in
higher latitudes mainly during winter and spring. Hilboll et al.
(2013a) suggest that these negative NO
2
columns could be related
to the polar vortex and the resulting zonal inhomogeneity. In order
to improve the quality of TVC NO
2
over the boreal regions, we use
stratospheric NO
2
columns as calculated by the B3dCTM simula-
tions in this study (see Sect. 2.1).
Using the linear relationship between TVC NO
2
and FRP (see Eq.
(1)), the y-intercepts are subtracted from the TVC NO
2
(see Eq. (2)).
This step is performed to isolate the tropospheric NO
2
column
contribution produced by fire (TVC
f
NO
2
), assuming that y-in-
tercepts represent the NO
2
background (TVC
b
NO
2
) and that there is
no or a negligible small seasonal cycle in NO
2
background.
TVC½NO
2
¼slope*FRP þTVC
b
½NO
2
(1)
TVC
f
½NO
2
¼TVC½NO
2
TVC
b
½NO
2
(2)
In a second step, the obtained monthly gridded values of TVC
f
NO
2
(in units of 10
15
molec cm
2
) are then converted into monthly
gridded values of NO
x
production rates P
f
(in units of g NO
x
s
1
) (see
Eq. (3)).
P
f
¼
TVC
f
½NO
2
*M1þ
NO
NO
2
A
p
N
A
*t
;(3)
where TVC
f
NO
2
is the number density of NO
2
molecules produced
by fires integrated over the tropospheric vertical column (in
Table 1
Selected regions and land cover types with their respective share of the total boreal forest area. The land cover types are used to define boreal forests in Eurasia and North
America (see Sect. 2.3).
Region Latitudes Longitudes Land cover types
a
Share of total area
b
Share of total area
c
Eurasia 50
e80
N0
e180
E Evergreen needleleaf forest 7.1% 0.5%
Deciduous needleleaf forest 17.6% 31.2%
Mixed forest 53.7% 39.8%
Woody savannas 21.6% 28.5%
North America 50
e80
N 170
e35
W Evergreen needleleaf forest 57.6% 59.3%
Mixed forest 12.2% 9.3%
Woody savannas 30.2% 31.4%
a
Based on the Collection 5 MODIS Global Land Cover Type product (UMD classification).
b
Percentages represent share of total boreal forest area as defined in this study and shown in Fig. 1.
c
Percentages represent share of total boreal forest area remaining after data filtering (see Fig. 3).
S.F. Schreier et al. / Atmospheric Environment xxx (2014) 1e114
Please cite this article in press as: Schreier, S.F., et al., Differences in satellite-derived NO
x
emission factors between Eurasian and North
American boreal forest fires, Atmospheric Environment (2014), http://dx.doi.org/10.1016/j.atmosenv.2014.08.071
molecules cm
2
) and Mis the molar mass of NO (in g mol
1
). The
term 1 þNO/NO
2
accounts for the NO
2
/NO
x
ratio (without units) as
derived from the MACC reanalysis data set (see below) and A
p
is the
respective pixel area (in cm
2
). N
A
denotes Avogadro's number (in
molecules mol
1
) and
t
is the assumed lifetime of NO
x
(in seconds).
We use a constant value of 6 h for
t
, which is based on the findings
of Beirle et al. (2011) for the megacity Moscow and Takegawa et al.
(2003) for biomass burning plumes over northern Australia. Ac-
cording to the described conversion of TVC NO
2
, the FRP values
have also been multiplied by A
p
. Further details about the conver-
sion are given in Schreier et al. (2014).
A detailed description of the MACC data assimilation system for
chemically reactive gases, which is based on ECMWF IFS and
MOZART-3 CTM simulations, can be found in Inness et al. (2013).
We have calculated daily weighted averages of the NO
2
/NO
x
ratio
for the 8 given UT hours (3, 6, 9, 12, 15,18, 21, and 0 UT) by including
29 hybrid sigma-pressure levels between the surface and ~10 km
altitude to reflect tropospheric values. We have then interpolated
between these UT hours by including daily gridded maps of the
geographical location of the GOME-2 overpass time (UT) to
construct daily values of the NO
2
/NO
x
ratio. Finally, we have
computed monthly means of the NO
2
/NO
x
ratio with a horizontal
Fig. 2. Correlation coefficients (r) of the linear regression (see Eq. (1)) of TVC NO
2
against FRP (upper) and statistical significance (lower) for the period 2007e2012. A p-value
smaller than 0.05 means that the correlation is statistically significant within a 95% confidence level. The retrieval of T VC NO
2
is based on the removal of stratospheric NO
2
applying
the reference sector (a) and (c) as well as using B3dCTM simulations (b) and (d). All colored 11pixels are defined as boreal forest (see Fig. 1 and Table 1).
S.F. Schreier et al. / Atmospheric Environment xxx (2014) 1e11 5
Please cite this article in press as: Schreier, S.F., et al., Differences in satellite-derived NO
x
emission factors between Eurasian and North
American boreal forest fires, Atmospheric Environment (2014), http://dx.doi.org/10.1016/j.atmosenv.2014.08.071
resolution of 1
1
for the entire period (2007e2012). These
values are included in the formula (Eq. (3)) for the conversion of
TVC NO
2
into P
f
, instead of assuming a constant value for the NO
2
/
NO
x
ratio as done in Schreier et al. (2014). Although there is no
systematic difference in the results when using the gridded values
of NO
2
/NO
x
ratio obtained from the MACC reanalysis data instead of
a constant value of 0.75 (not shown), we use these values as they
account for the seasonal variability.
A linear regression model is used for the calculation of gradients
between P
f
and FRP for all boreal forest pixels with a correlation
coefficient (r) between TVC NO
2
and FRP higher than 0.3. As the
major parts of boreal forests examined in this study are located in
rather remote areas, we have tested whether the population den-
sity filter as applied in Schreier et al. (2014) is crucial for this study.
We found that this filter has no significant effect on the obtained
results (not shown) and have therefore not applied such filtering.
In short, the approach to estimate FERs of NO
x
is similar to the
approach used by Schreier et al. (2014), but three parameters have
been modified for this study. Firstly, TVC NO
2
is based on a
stratospheric correction using model simulations to reduce the
uncertainty from stratospheric NO
2
. Secondly, monthly gridded
maps of the NO
2
/NO
x
ratio derived from the MACC reanalysis data
are used to account for seasonality and spatial variation of this ratio.
Thirdly, filtering by population density data has been omitted as it
proved to be not necessary.
3. Results and discussion
3.1. Statistical evaluation of regression coefficients
The correlation coefficients as calculated from the pixel-wise
linear relationship between monthly means of TVC NO
2
and FRP
are shown in Fig. 2 for the entire boreal forest (as defined in Fig. 1).
The correlation coefficients as computed in Schreier et al. (2014) by
applying the reference sector method for the retrieval of TVC NO
2
(a) are compared with the correlation coefficients based on the new
TVC NO
2
as computed by subtracting the B3dCTM simulated
stratospheric NO
2
(b). There is some degree of improvement,
especially in the Far East Russia and parts of Siberia, due to the
improved stratospheric correction of the NO
2
satellite measure-
ments. Negative correlation coefficients indicate the influence of
anthropogenic emissions, which are higher during winter months
despite the absence of fires. As already stated in Schreier et al.
(2014), negative values are not included in the approach to derive
FERs of NO
x
. In order to give evidence about the statistical signifi-
cance, p-values have been computed for the linear relationship
between TVC NO
2
(based on the reference sector method (c) and
simulated stratospheric NO
2
(d)) and FRP (see Eq. (1)). The adapted
approach used in this study for the evaluation of FERs and EFs of
NO
x
in the boreal region clearly increases the number of pixels with
p-values <0.05 (statistical significance within a 95% confidence
level). There are fewer pixels having correlation coefficients >0.3 in
the boreal region when compared to tropical and subtropical re-
gions (Schreier et al., 2014). This might be related to the fact that
there are many smaller fires with NO
2
signals falling below the
detection limit of GOME-2. Although large fires account for over
85% of the total area burned in the Canadian boreal forest, they
account for less than 5% of the fires (Stocks et al., 2003). Therefore,
we anticipate that the estimation of FERs and EFs of NO
x
is based on
rather larger forest fires in this study. The gradients of the linear
regression model used for estimating fire emission rates are
calculated for those 1
1
pixels having p-values <0.05 and cor-
relation coefficients r>0.3. By this data selection, we focus on areas
where there is a clear link between satellite observed fires and
satellite retrieved NO
2
columns. As shown in Fig. 3, the number of
boreal forest pixels used for the estimation of FERs of NO
x
is
reduced by these thresholds to 141 and 63 for Eurasia and North
America, respectively. Although a correlation coefficient of 0.3 is
not high, it is statistically significant in our study, even for the
relatively low amount of data points available in the selected time
series. The use of a higher threshold value (e.g. r>0.4) would lead
to an even smaller data set, which would not be beneficial for the
analysis. Therefore, we feel confident to use the above mentioned
threshold values (r>0.3 and p-values <0.05) for this study.
In Fig. 3, the y-intercepts of the linear regressions (see Eq. (1))
are shown for the selected boreal forest pixels, based on the filter
criteria (p-value <0.05 and r>0.3). These values reveal useful
information about the background level of tropospheric NO
2
in the
selected areas as they reflect levels of tropospheric NO
2
produced
from other NO
x
sources than fire. The y-intercept values are sub-
tracted from the TVC NO
2
grid cells to obtain the tropospheric NO
2
produced from fire (TVC
f
NO
2
), which is used for the conversion of
TVC NO
2
into P
f
(see Eq. (2) and Sect. 2.6). The subtraction of these
values is performed under the assumption that there is no seasonal
cycle in background tropospheric NO
2
.
3.2. Determination of fire emission rates of NO
x
Scatter plots for the relationship between P
f
and FRP are shown
in Fig. 4 for Eurasia (red) and North America (blue). All boreal forest
pixels with a p-value <0.05 and r>0.3 are plotted in the graph. The
total number of data points, i.e., TVC NO
2
averaged over one grid
cell and one month, included in this plot is 6791 and 3298 for
Eurasia and North America, respectively (see Table 2). A maximum
Fig. 3. Y-intercepts of the linear regression of TVC NO
2
against FRP for the six-year
period (2007e2012). The y-intercepts representing the background levels in NO
2
columns are subtracted from the total tropospheric NO
2
columns (see Sect. 2.6). All
pixels with a p-value <0.05 and r>0.3 are shown in the Figure and used as a selection
criterion for the estimation of FERs and EFs of NO
x
.
S.F. Schreier et al. / Atmospheric Environment xxx (2014) 1e116
Please cite this article in press as: Schreier, S.F., et al., Differences in satellite-derived NO
x
emission factors between Eurasian and North
American boreal forest fires, Atmospheric Environment (2014), http://dx.doi.org/10.1016/j.atmosenv.2014.08.071
number of 72 data points would result for a single boreal forest
pixel if the satellite instruments would have detected a signal for
each month of the selected time period (2007e2012) for both TVC
NO
2
and FRP. As boreal wildfires mainly occur in summer and to a
lesser extent in spring and fall, the maximum number of data
points for the individual pixels is reduced due to the absence of fires
in winter. A division of the total number of data points (6791 and
3298) by the number of boreal forest pixels remaining after data
filtering (141 and 63) results in average numbers (or months con-
taining signals of both TVC NO
2
and FRP) of 48 and 52 for an in-
dividual pixel located in Eurasia and North America, respectively.
In general, a large spread of data is expected because of mea-
surement uncertainties, the simplifications made in the conversion
of NO
2
columns to NO
x
production rates and from the horizontal
and vertical changes of N and moisture contents in the vegetation
(Fig. 4). For instance, a dry deciduous forest with increased N
content will release more NO
x
per unit FRP than a rather humid
coniferous forest. There is a clear indication that single fire events
can be more intensive in North America than in Russia. Wooster
and Zhang (2004) suggest that the higher fire intensity in North
American boreal forests is linked to more intensive crown fires.
In order to reduce the spread of data and exclude outliers, data
points are averaged within consecutive FRP-intervals of
15 MW pixel
1
. The application of the binning method moreover
leads to a result that represents spatio-temporally averaged FERs of
NO
x
for the respective region. Due to the larger amount of data
available for Eurasia, the threshold criterion for the binning could
be set to 10 data points available within the interval, whereas for
North America it is set to 5.
After binning, clear linear relationships are visible for both re-
gions, albeit with different slopes. This result shows that there are
differences in FERs of NO
x
between the two selected regions with
higher values observed for the Eurasian boreal forest fires (Fig. 5). In
other words, the emissions of NO
x
per unit of FRP are lower on
average for forest fires in North America. One possible explanation
could be the lower N content in evergreen (coniferous) species
dominating large parts of the North American region (see Table 1
and Jerabkova et al., 2006). Moreover, Chen et al. (2010) have re-
ported that the concentration ratio of NO
x
over the grand total
carbon is highest for litter combustion, which is more likely in
deciduous forests dominating the Eurasian region analyzed in our
study. Wooster and Zhang (2004) found an overall higher-
temperature flaming combustion in North American boreal for-
ests. Assuming that the production of NO
x
via the oxidation of N
2
would be the dominant source of NO
x
from boreal wildfires, higher
FERs of NO
x
would be expected for boreal forest fires in North
America. As this is not the case in our study, we can only speculate
that the observed differences in FERs of NO
x
between Eurasian and
North American boreal forests are rather related to the variable N
content in plant tissues. We argue that the higher FERs of NO
x
derived for the Eurasian boreal forests are likely attributed to the
larger proportion of deciduous stands such as larch forests
(Schepaschenko et al., 2011) and/or to the non-existence of such
forests in North America. Large differences in the N content are
especially found in the canopy litter, with ~30% larger amounts
reported for deciduous forests (Jerabkova et al., 2006). In addition,
the N content in grasses is higher than in other plant tissues
(Bakhnov, 2001). Consequently, the dominance of surface fires in
deciduous forests in Eurasia (see Wooster and Zhang, 2004) could
explain the larger emissions of NO
x
per unit of FRP.
3.3. Conversion into emission factors of NO
x
In order to compare our values with the values found in litera-
ture, the FERs are translated into EFs of NO
x
by applying a
Fig. 4. Estimated productions rates of NO
x
from fire (P
f
) plotted against associated FRP
values over boreal forests Eurasia (red) and North America (blue). All 11pixels
with a p-value <0.05 and r>0.3 are included in the plot (see Sect. 2.6). (For inter-
pretation of the references to color in this figure legend, the reader is referred to the
web version of this article.)
Table 2
Spatio-temporally averaged fire emission rates (FERs) of NO
x
in [g s
1
MW
1
] and
emission factors (EFs) of NO
x
in [g kg
1
] for the Eurasian and North American boreal
forest fires analyzed in this study. FERs and EFs of NO
x
are reported as NO. The
approach to derive the FERs and EFs of NO
x
is described in Sect. 2.6.
Region NFER
a
r
2
EF
b
Eurasia 6791 0.34 ±0.03 0.87 0.83 ±0.07
North America 3298 0.25 ±0.03 0.79 0.61 ±0.07
Uncertainty of FERs and EFs is given as the standard error of the slope as shown in
Fig. 5.
a
Derived by applying a binning procedure as described in the text.
b
Based on the conversion factor of 0.41 kg MJ
1
as suggested by Vermote et al.
(2009).
Fig. 5. Spatio-temporally averaged fire emission rates (FERs) of NO
x
for boreal forest in
Eurasia (red) and North America (blue) derived by applying a binning procedure as
explained in Sect. 3.2. The gradients, here referred to as FERs, are calculated by
applying a linear regression of monthly means of P
f
against monthly means of FRP by
including all pixels with a p-value <0.05 and r>0.3. The error bars show one standard
deviation of P
f
values within the consecutive FRP-intervals. A summary of the statistics
is given in Table 2. (For interpretation of the references to color in this figure legend,
the reader is referred to the web version of this article.)
S.F. Schreier et al. / Atmospheric Environment xxx (2014) 1e11 7
Please cite this article in press as: Schreier, S.F., et al., Differences in satellite-derived NO
x
emission factors between Eurasian and North
American boreal forest fires, Atmospheric Environment (2014), http://dx.doi.org/10.1016/j.atmosenv.2014.08.071
conversion factor of 0.41 kg MJ
1
as suggested by Vermote et al.
(2009). The values for the spatio-temporally averaged FERs and
EFs of NO
x
, the total number of data points included in the analysis
(N), and the coefficient of determination (r
2
) are summarized in
Table 2.
A comparison of the obtained EFs with the EF provided by Akagi
et al. (2011) indicates very good agreement, as their reported value
of 0.9 g kg
1
for the whole boreal forest biome is in good agreement
with the value derived for the Eurasian boreal forest (0.83 g kg
1
)in
our study. However, the EF estimated for the North American boreal
forest (0.61 g kg
1
) is about 30% lower than the average value re-
ported by Akagi et al. (2011).Wiedinmyer et al. (2006) have esti-
mated fire emissions for North America by assigning EFs for each
land cover type in the Global Land Cover (GLC2000) classification.
The EFs of NO
x
used in that study are in the range of 2.1e2.7 g kg
1
for sub-polar needleleaved and broadleaved forests and thus,
around three times larger than the EFs obtained in our study. Van
Der Werf et al. (2010) and Kaiser et al. (2012) applied constant
values of 3.41 and 3.4 g NO
x
kg
1
, respectively, for the entire
extratropical forest biome. When compared to our values, their
values are five times larger. As about 80% of global burned area is
found in the tropics and subtropics (e.g. Van Der Werf et al., 2010),
the given difference in EFs of NO
x
for the boreal forest is less
important for fire emission inventories on a global scale. However,
this difference would affect the magnitude of fire emissions
considerably on the regional level.
4. Possible uncertainties in the approach
In order to make sure that the observed differences in FERs and
EFs of NO
x
between the two regions are related to characteristics of
the vegetation and fire type, we analyze and discuss possible fac-
tors that could affect the retrieval of TVC NO
2
and the conversion
into P
f
. The bulk of uncertainties in the retrieval of TVC NO
2
results
from the conversion of SCDs into VCDs by the use of AMFs (see Sect.
2.2). Here, we focus on the impact of aerosol amounts and prop-
erties as well as on injection profiles of fire smoke that could affect
the magnitude of FERs of NO
x
between Eurasia and North America
differently. Additionally, the lifetime of NO
x
and the NO
2
/NO
x
ratio
are discussed in terms of possible relative influences on the con-
version of TVC NO
2
into P
f
between the two selected regions.
4.1. Impact of aerosol amounts and properties
First, the AOD retrieved from MODIS on board Terra (in accor-
dance to the overpass time of GOME-2 on board MetOp-A) is
analyzed over the Eurasian and North American boreal forests. In
Fig. 6, the AOD is plotted against FRP for boreal forest pixels with a
p-value <0.05 and r>0.3 in Eurasia (red) and North America (blue).
Clearly, the AOD is lower over North American forests when the FRP
value of the respective pixel is lower than 500 MW. This might be
an indication that aerosol amounts are largely dominated by fires in
this region. In comparison, aerosols could be transported from
anthropogenic emission sources (from coal, gas, and oil burning)
into the Eurasian boreal forest pixels, as a certainproportion of data
points indicate higher AOD values in Eurasia. In terms of AMF cal-
culations, an increase in AOD generally results in higher measure-
ment sensitivity when the aerosols are below or mixed with the
NO
2
molecules, and thus, increase the AMF. A decreased AMF
comes along with a reduced sensitivity due to the location of
aerosols above the NO
2
plume (shielding effect). Stohl et al. (2013)
highlighted the important role of black carbon emissions from gas
flaring in the Arctic. According to their findings, black carbon
emissions from gas flaring, especially in Russia, are transported into
boreal forest pixels analyzed in our study. Black carbon is
considered as a fine black fluffy particle with highly absorbing
properties. Highly absorbing aerosols, either located above or
mixed with the NO
2
molecules, can only reduce the measurement
sensitivity, and thus, decrease the AMF. In order to assess the in-
fluence of the increased AOD over the Eurasian boreal forest pixels
on the magnitude of TVC NO
2
, it is important to know the relative
location and the properties of the additional aerosol load in the
troposphere. However, this information is still highly unknown as
no accurate data sets for the selected regions are currently avail-
able. Thus, we can only speculate that the effect of an increased
AOD could lead to an underestimation of the FERs and EFs of NO
x
over Eurasia due to an overestimation of the AMF.
Secondly, the single scattering albedo (SSA) of aerosols and its
possible impact on the retrieval of TVC NO
2
shall be discussed. The
SSA describes the scattering and absorbing properties of aerosols
and is simply defined as the ratio between scattering and extinction
(scattering þabsorption). While highly absorbing aerosols are
characterized by a lower SSA, the SSA of highly scattering aerosols
tends towards one. In general, a less complete combustion (smol-
dering fire) leads to a larger fine mode fraction of aerosols, which
increases scattering, and thus, the SSA (Eck et al., 2009). Giles et al.
(2012) have reported an average SSA of 0.95 at 440 nm for Bonanza
Creek in Alaska. In contrast, the SSA in Western Siberia is estimated
at 0.92e0.93 at 440 nm in an altitude between 500 and 2000 m,
which is in good agreement with a measurement site at Tomsk
(Panchenko et al., 2012). Both locations match with parts of the
boreal forest pixels that are analyzed in this study. The lower SSA in
Russia could be related to the influence of anthropogenic emissions
sources in the vicinity of boreal forest pixels included in the anal-
ysis. However, a larger fraction of grasses burned could also
contribute to the lower value. By assuming that the two reported
SSA values are representative for the two selected regions, we
conclude that the FERs of NO
x
in Eurasia are rather underestimated
than overestimated.
4.2. Impact of injection heights
The mean seasonal daytime injection profiles of the plumes
from fires are shown in Fig. 7 over boreal forest fires in Eurasia (left)
Fig. 6. Scatterplot between AOD and FRP for Eurasia (red) and North America (blue).
The data points represent all boreal forest pixels (2007e2012) with a p-value <0.05
and r>0.3. (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
S.F. Schreier et al. / Atmospheric Environment xxx (2014) 1e118
Please cite this article in press as: Schreier, S.F., et al., Differences in satellite-derived NO
x
emission factors between Eurasian and North
American boreal forest fires, Atmospheric Environment (2014), http://dx.doi.org/10.1016/j.atmosenv.2014.08.071
and North America (right). As for the analysis of AOD, all boreal
forest pixels with a p-value <0.05 and r>0.3 are included in the
computation of injection profiles as presented in Fig. 7. In general,
the seasonal distribution of the injection profiles is similar in both
regions with the highest injection heights observed in April, May,
June, July, and August (see Sofiev et al., 2013), which is the main
forest fire season. As the differences are very small, the relative
effects on the retrieval of TVC NO
2
between the two regions are
assumed to be negligibly small. Val Martin et al. (2010) found a
quantitative link between median injection heights and FRP in
North America. The slightly higher injection heights in North
America could thus be related to the increased FRP values observed
in this region when compared with maximum FRP values in Eurasia
(see Fig. 4). Nevertheless, the higher injection heights observed
over North American boreal forest fires could potentially result in
an overestimation of the AMF. This can be explained by the fact that
the measurement sensitivity, which is smallest close to the Earth's
surface, is overestimated if the NO
2
plume is higher in the atmo-
sphere than assumed (see Leit~
ao et al., 2010). Therefore, FERs and
EFs of NO
x
in North America could potentially be overestimated
relative to those in Eurasia.
4.3. Impact of NO
x
lifetime and NO
2
/NO
x
ratio
With respect to the lifetime of NO
x
and the NO
2
/NO
x
ratio, the
influence of an increased NO
2
layer is twofold. While the lifetime of
NO
x
is slightly increased towards higher altitudes, the NO
2
/NO
x
ratio is decreased as the relative proportion of NO
2
decreases with
increasing altitude. However, the changes are too small to expect a
significant relative influence on the magnitude of FERs of NO
x
be-
tween the two regions.
5. Summary and conclusion
In this study, the fire emission rates (FERs) and emission factors
(EFs) of NO
x
are estimated for boreal forest fires in Eurasia and
North America, based on the empirical relationship between
satellite-derived tropospheric NO
2
vertical columns (TVC NO
2
) and
fire radiative power (FRP). The retrieval of TVC NO
2
is based on a
model-based correction of the stratosphere instead of the previ-
ously used reference sector method, which clearly improves the
empirical relationship between TVC NO
2
and FRP at mid and high
latitudes. As the GOME-2 retrievals provide NO
2
columns, a
simplified formula including the lifetime of NO
x
and the NO
2
/NO
x
ratio is used to convert tropospheric NO
2
column densities into
production rates of NO
x
from fire (P
f
) in terms of mass concentra-
tions. Instead of assuming a constant value of 0.75 for the NO
2
/NO
x
ratio (see Schreier et al., 2014), gridded values obtained from the
MACC reanalysis data set are applied in this study. Although these
monthly means account for the seasonal variability, no improve-
ment was found for the empirical relationship.
The boreal forest pixels are defined according to the collection 5
MODIS global land cover product and confined between 50
and
80
N. The approach used to estimate FERs of NO
x
only includes
boreal forest pixels that exceed a value of 0.3 for the temporal
correlation coefficient between TVC NO
2
and FRP and are statisti-
cally significant within a 95% confidence. This criterion has been
chosen to exclude regions with an even weaker link between
observed NO
2
columns and FRP, which are not beneficial for the
analysis. On the other hand, a higher threshold value applied for the
correlation coefficient (e.g. r>0.4) decreases the available data
points.
The spatio-temporally averaged FERs of NO
x
are estimated at
0.34 and 0.25 g s
1
MW
1
for Eurasian and North American boreal
forest fires, respectively. We speculate that the observed difference
is related to changes in the N content and moisture conditions of
the fuel types burned. Moreover, the type of fire (surface fires vs.
crown fires) and the linked combustion of dead material on the
ground and tops of trees could affect the magnitude of FERs.
For a better comparison with values found in the literature, the
FERs are translated into EFs of NO
x
by simply applying a conversion
factor of 0.41 kg MJ
1
, assuming the findings by Vermote et al.
(2009). The satellite-based values are estimated at 0.83 and
0.61 g kg
1
for Eurasian and North American boreal forests,
respectively. A comparison with the emission factor reported by
Akagi et al. (2011) for the entire boreal forest (0.9 g kg
1
) indicates
good agreement. However, recent fire emission inventories have
used EFs of NO
x
that are 3e5 times larger. Our findings have
possible implications for future estimates of fire emissions, espe-
cially on the regional scale where our results indicate less fire
related NO
x
emissions.
We discussed possible factors that could affect the observed
differences in FERs and EFs of NO
x
between North America and
Eurasia and suggest that considering possible systematic biases in
the NO
2
retrievals, the real differences of NO
x
EFs between the two
regions could even be larger. Therefore, we conclude that the
observed differences of FERs and EFs of NO
x
between Eurasia and
North America are real and should be investigated by other, more
direct methods in the future.
Acknowledgements
Parts of this study were performed while Stefan F. Schreier was
participating in the Young Scientists Summer Program (YSSP) and
working in collaboration with the Ecosystems Services and Man-
agement (ESM) Program at the International Institute for Applied
Fig. 7. Mean seasonal daytime injection profiles (Sofiev et al., 2013) over boreal forest fires in Eurasia (left) and North America (right). All boreal forest pixels (2007e2012) with a p-
value <0.05 and r>0.3 are included in the averaging of injection profiles.
S.F. Schreier et al. / Atmospheric Environment xxx (2014) 1e11 9
Please cite this article in press as: Schreier, S.F., et al., Differences in satellite-derived NO
x
emission factors between Eurasian and North
American boreal forest fires, Atmospheric Environment (2014), http://dx.doi.org/10.1016/j.atmosenv.2014.08.071
Systems Analysis (IIASA). Stefan F. Schreier wishes to acknowledge
financial support provided by IIASA and the Earth System Science
Research School (ESSReS), an initiative of the Helmholtz Associa-
tion of German Research Centres (HGF) at the Alfred Wegener
Institute for Polar and Marine Research (AWI). GOME-2 lv1 data
have been provided by EUMETSAT. We thank NASA for the free use
of the MODIS data and wish to thank M. Sofiev for providing the
injection profiles. We gratefully acknowledge the MACC team and
Anne Blechschmidt for providing the reanalysis data. Finally, we
wish to thank two anonymous reviewers for their useful
comments.
References
Akagi, S.K., Yokelson, R.J., Wiedinmyer, C., Alvarado, M.J., Reid, J.S., Karl, T.,
Crounse, J.D., Wennberg, P.O., 2011. Emission factors for open and domestic
biomass burning for use in atmospheric models. Atmos. Chem. Phys. 11,
4039e4072. http://dx.doi.org/10.5194/acp-11-4039-2011.
Andreae, M.O., Merlet, P., 2001. Emission of trace gases and aerosols from biomass
burning. Glob. Biogeochem. Cycles 15, 955e966.
Bakhnov, V.R., 2001. General scheme of evolution of swampy soils and swamps of
taiga belt. In: Vasiliev, S.V., Titlyanova, A.A., Velichko, A.A. (Eds.), Weat Siberian
Peatlands and Carbon Cycle: Past and Present. Institute of Soil Science and
Agrochemistry, Siberian Branch of RAS, Novosibirsk, pp. 10e11.
Bazilevich, N.I., Rodin, L.E. (Eds.), 1971. Biological Productivity and Mineral Cycling
in the Terrestrial Plant Communities. Nuka Publ., Leningrad, p. 313 (in Russian).
Beirle, S., Boersma, K.F., Platt, U., Lawrence, M.G., Wagner, T., 2011. Megacity emis-
sions and lifetimes of nitrogen oxides probed from space. Science 333,
1737e1739 .
Boersma, K.F., Eskes, H.J., Brinksma, E.J., 2004. Error analysis for tropospheric NO
2
retrieval from space. J. Geophys. Res. D Atmos. 109, D04311e04320.
Callies, J., Corpaccioli, E., Eisinger, M., Lefebvre, A., Munro, R., Perez-Albinana, A.,
Ricciarelli, B., Calamai, L., Gironi, G., Veratti, R., Otter, G., Eschen, M., Van Riel, L.,
2004. GOME-2 ozone instrument on-board the European METOP satellites. In:
Weather and Environmental Satellites, Proceedings of SPIE. The International
Society for Optical Engineering, Denver, CO, pp. 60e70.
Castellanos, P., Boersma, K.F., Van Der Werf, G.R., 2014. Satellite observations indi-
cate substantial spatiotemporal variability in biomass burning NO
x
emission
factors for South America. Atmos. Chem. Phys. 14, 3929e3943. http://
dx.doi.org/10.5194/acp-14-3929-2014.
Chen, L.-W.A., Verburg, P., Shackelford, A., Zhu, D., Susfalk, R., Chow, J.C.,
Watson, J.G., 2010. Moisture effects on carbon and nitrogen emission from
burning of wildland biomass. Atmos. Chem. Phys. 10, 6617e6625. http://
dx.doi.org/10.5194/acp-10-6617-2010.
Chipperfield, M.P., 1999. Multiannual simulations with a three-dimensional chem-
ical transport model. J. Geophys. Res. 104, 1781e1805.
Dee, D.P., Uppala, S.M., Simmons, A.J., Berrisford, P., Poli, P., Kobayashi, S., Andrae, U.,
Balmaseda, M.A., Balsamo, G., Bauer, P., Bechtold, P., Beljaars, A.C.M., van de
Berg, L., Bidlot, J., Bormann, N., Delsol, C., Dragani, R., Fuentes, M., Geer, A.J.,
Haimberger, L., Healy, S.B., Hersbach, H., H
olm, E.V., Isaksen, L., Kållberg, P.,
K€
ohler, M., Matricardi, M., McNally, A.P., Monge-Sanz, B.M., Morcrette, J.-J.,
Park, B.-K., Peubey, C., de Rosnay, P., Tavolato, C., Th
epaut, J.-N., Vitart, F., 2011.
The ERA-Interim reanalysis: configuration and performance of the data
assimilation system. Q. J. Roy. Meteorol. Soc. 137, 553e597. http://dx.doi.org/
10.1002/qj.828.
Eck, T.F., Holben, B.N., Reid, J.S., Sinyuk, A., Hyer, E.J., O'Neill, N.T., Shaw, G.E., Vande
Castle, J.R., Chapin, F.S., Dubovik, O., Smirnov, A., Vermote, E., Schafer, J.S.,
Giles, D., Slutsker, I., Sorokine, M., Newcomb, W.W., 2009. Optical properties of
boreal region biomass burning aerosols in central Alaska and seasonal variation
of aerosol optical depth at an Arctic coastal site. J. Geophys. Res. D Atmos. 114
http://dx.doi.org/10.1029/2008JD010870.
Flannigan, M.D., Krawchuk, M.A., De Groot, W.J., Wotton, B.M., Gowman, L.M., 2009.
Implications of changing climate for global wildland fire. Int. J. Wildland Fire 18,
483e507. http://dx.doi.org/10.1071/WF08187.
Friedl, M.A., Sulla-Menashe, D., Tan, B., Schneider, A., Ramankutty, N., Sibley, A.,
Huang, X., 2010. MODIS collection 5 global land cover: algorithm refinements
and characterization of new datasets. Remote Sens. Environ. 114, 168e182.
http://dx.doi.org/10.1016/j.rse.20 09.08.016.
Giles, D.M., Holben, B.N., Eck, T.F., Sinyuk, A., Smirnov, A., Slutsker, I., Dickerson, R.R.,
Thompson, A.M., Schafer, J.S., 2012. An analysis of AERONET aerosol absorption
properties and classifications representative of aerosol source regions.
J. Geophys. Res. D Atmos. 117 http://dx.doi.org/10.1029/2012JD018127.
Goode, J.G., Yokelson, R.J., Ward, D.E., Susott, R.A., Babbitt, R.E., Davies, M.A.,
Hao, W.M., 2000. Measurements of excess O
3
,CO
2
, CO, CH
4
,C
2
H
4
,C
2
H
2
, HCN,
NO, NH
3
, HCOOH, CH
3
COOH, HCHO, and CH
3
OH in 1997 Alaskan biomass
burning plumes by airborne Fourier transform infrared spectroscopy (AFTIR).
J. Geophys. Res. 105, 22147e22166. http://dx.doi.org/10.1029/2000JD900287.
Hansen, M.C., Defries, R.S., Townshend, J.R.G., Sohlberg, R., 2000. Global land cover
classification at 1 km spatial resolution using a classification tree approach. Int.
J. Remote Sens. 21, 1331e1364.
Hilboll, A., Richter, A., Rozanov, A., Hodnebrog, Ø., Heckel, A., Solberg, S., Stordal, F.,
Burrows, J.P., 2013a. Improvements to the retrieval of tropospheric NO
2
from
satellite estratospheric correction using SCIAMACHY limb/nadir matching and
comparison to Oslo CTM2 simulations. Atmos. Meas. Tech. 6, 565e584. http://
dx.doi.org/10.5194/amt-6-565-2013.
Hilboll, A., Richter, A., Burrows, J.P., 2013b. Long-term changes of tropospheric NO
2
over megacities derived from multiple satellite instruments. Atmos. Chem.
Phys. 13, 4145e4169. http://dx.doi.org/10.5194/acp-13-4145-2013.
Inness, A., Baier, F., Benedetti, A., Bouarar, I., Chabrillat, S., Clark, H., Clerbaux, C.,
Coheur, P., Engelen, R.J., Errera, Q., Flemming, J., George, M., Granier, C., Hadji-
Lazaro, J., Huijnen, V., Hurtmans, D., Jones, L., Kaiser, J.W., Kapsomenakis, J.,
Lefever, K., Leit~
ao, J., Razinger, M., Richter, A., Schultz, M.G., Simmons, A.J.,
Suttie, M., Stein, O., Th
epaut, J.-N., Thouret, V., Vrekoussis, M., Zerefos, C., The
MACC team, 2013. The MACC reanalysis: an 8 yr data set of atmospheric
composition. Atmos. Chem. Phys. 13, 4073e4109. http://dx.doi.org/10.5194/acp-
13-4073-2013.
Jacob, D.J., 1993. Factors regulating ozone over the United States and its export to
the global atmosphere. J. Geophys. Res. 98, 14817e814826.
Jerabkova, L., Prescott, C.E., Kishchuk, B.E., 2006. Nitrogen availability in soil and
forest floor of contrasting types of boreal mixedwood forests. Can. J. For. Res. 36,
112 e122. http://dx.doi.org/10.1139/x05-220.
Kaiser, J.W., Heil, A., Andreae, M.O., Benedetti, A., Chubarova, N., Jones, L.,
Morcrette, J.J., Razinger, M., Schultz, M.G., Suttie, M., Van Der Werf, G.R., 2012.
Biomass burning emissions estimated with a global fire assimilation system
based on observed fire radiative power. Biogeosciences 9, 527e554. http://
dx.doi.org/10.5194/bg-9-527-2012.
Kaufman, Y.J., Justice, C.O., Flynn, L.P., Kendall, J.D., Prins, E.M., Giglio, L., Ward, D.E.,
Menzel, W.P., Setzer, A.W., 1998. Potential global fire monitoring from EOS-
MODIS. J. Geophys. Res. D Atmos. 103, 32215e32238.
Leit~
ao, J., Richter, A., Vrekoussis, M., Kokhanovsky, A., Zhang, Q.J., Beekmann, M.,
Burrows, J.P., 2010. On the improvement of NO
2
satellite retrievals eaerosol
impact on the airmass factors. Atmos. Meas. Tech. 3, 475e493. http://
dx.doi.org/10.5194/amt-3-475-2010.
Lobert, J.M., Scharffe, D.H., Hao, W.M., Crutzen, P.J., 1990. Importance of biomass
burning in the atmospheric budgets of nitrogen-containing gases. Nature 346,
552e554.
Lobert, J.M., Keene, W.C., Logan, J.A., Yevich, R., 1999. Global chlorine emissions from
biomass burning: reactive Chlorine Emissions Inventory. J. Geophys. Res. D
Atmos. 104, 8373e8389.
Mebust, A.K., Cohen, R.C., 2013. Observations of a seasonal cycle in NO
x
emissions
from fires in African woody savannas. Geophys. Res. Lett. 40, 1451e1455. http://
dx.doi.org/10.1002/grl.50343.
Nüß, J.H., 2005. Improvements of the Retrieval of Tropospheric NO
2
from GOME and
SCIAMACHY Data (Ph.D. thesis). University of Bremen.
Pan, Y., Birdsey, R.A., Fang, J., Houghton, R., Kauppi, P.E., Kurz, W.A., Phillips, O.L.,
Shvidenko, A., Lewis, S.L., Canadell, J.G., Ciais, P., Jackson, R.B., Pacala, S.W.,
McGuire, A.D., Piao, S., Rautiainen, A., Sitch, S., Hayes, D., 2011. A large and
persistent carbon sink in the world's forests. Science 333, 988e993.
Panchenko, M.V., Zhuravleva, T.B., Terpugova, S.A., Polkin, V.V., Kozlov, V.S., 2012. An
empirical model of optical and radiative characteristics of the tropospheric
aerosol over West Siberia in summer. Atmos. Meas. Tech. 5, 1513e1527. http://
dx.doi.org/10.5194/amt-5-1513-2012.
Platt, U., Stutz, J., 2008. Differential Optical Absorption Spectroscopy. Physics of
Earth and Space Environments. Springer, Berlin.
Protopopov, V.V., 1975. Sredoobrazuiushchaia Rol' Temnokhvoinogo Lesa.
Novosibirsk.
Remer, L.A., Kleidman, R.G., Levy, R.C., Kaufman, Y.J., Tanr
e, D., Mattoo, S.,
Martins, J.V., Ichoku, C., Koren, I., Yu, H., Holben, B.N., 2008. Global aerosol
climatology from the MODIS satellite sensors. J. Geophys. Res. D Atmos. 113
http://dx.doi.org/10.1029/2007JD009661.
Richter, A., Burrows, J.P., 2002. Tropospheric NO
2
from GOME measurements. Adv.
Space Res. 29, 1673e1683.
Richter, A., Burrows, J.P., Nüß, H., Granier, C., Niemeier, U., 2005. Increase in
tropospheric nitrogen dioxide over China observed from space. Nature 437,
129e132. http://dx.doi.org/10.1038/nature04092.
Richter, A., Begoin, M., Hilboll, A., Burrows, J.P., 2011. An improved NO
2
retrieval for
the GOME-2 satellite instrument. Atmos. Meas. Tech. 4, 1147e1159. http://
dx.doi.org/10.5194/amt-4-1147-2011.
Rodin, L.E., Bazilevich, N.I., 1965. Dynamics of Organic Matter and Biological Turn-
over of Ash Constituents and Nitrogen in Major Types of Vegetation of the
Globe. Academy of Sciences of the USSR, Moscow-Leningrad, p. 253 (in
Russian).
Schepaschenko, D., McCallum, I., Shvidenko, A., Fritz, S., Kraxner, F., Obersteiner, M.,
2011. A new hybrid land cover dataset for Russia: a methodology for integrating
statistics, remote sensing and in situ information. J. Land Use Sci. 6, 245e259.
http://dx.doi.org/10.1080/1747423X.2010.511681.
Schreier, S.F., Richter, A., Kaiser, J.W., Burrows, J.P., 2014. The empirical relationship
between satellite-derived tropospheric NO
2
and fire radiative power and
possible implications for fire emission rates of NO
x
. Atmos. Chem. Phys. 14,
2447e2466. http://dx.doi.org/10.5194/acp-14-2447-2014.
Shvidenko, A.Z., Shchepashchenko, D.G., Vaganov, E.A., Sukhinin, A.I.,
Maksyutov, S.S., McCallum, I., Lakyda, I.P., 2011. Impact of wildfire in Russia
between 1998e2010 on ecosystems and the global carbon budget. Dokl. Earth
Sci. 441, 1678e1682. http://dx.doi.org/10.1134/S1028334X11120075.
S.F. Schreier et al. / Atmospheric Environment xxx (2014) 1e1110
Please cite this article in press as: Schreier, S.F., et al., Differences in satellite-derived NO
x
emission factors between Eurasian and North
American boreal forest fires, Atmospheric Environment (2014), http://dx.doi.org/10.1016/j.atmosenv.2014.08.071
Shvidenko, A.Z., Schepaschenko, D.G., 2013. Climate change and wildfires in Russia.
Lesovedenie 5, 50e61. http://dx.doi.org/10.1134/S199542551307010X.
Singh, H.B., Salas, L., Herlth, D., Kolyer, R., Czech, E., Avery, M., Crawford, J.H.,
Pierce, R.B., Sachse, G.W., Blake, D.R., Cohen, R.C., Bertram, T.H., Perring, A.,
Wooldridge, P.J., Dibb, J., Huey, G., Hudman, R.C., Turquety, S., Emmons, L.K.,
Flocke, F., Tang, Y., Carmichael, G.R., Horowitz, L.W., 2007. Reactive nitrogen
distribution and partitioning in the North American troposphere and lower-
most stratosphere. J. Geophys. Res. D Atmos. 112 http://dx.doi.org/10.1029/
2006JD007664.
Singh, H.B., Anderson, B.E., Brune, W.H., Cai, C., Cohen, R.C., Crawford, J.H.,
Cubison, M.J., Czech, E.P., Emmons, L., Fuelberg, H.E., Huey, G., Jacob, D.J.,
Jimenez, J.L., Kaduwela, A., Kondo, Y., Mao, J., Olson, J.R., Sachse, G.W., Vay, S.A.,
Weinheimer, A., Wennberg, P.O., Wisthaler, A., The ARCTAS Science Team, 2010.
Pollution influences on atmospheric composition and chemistry at high
northern latitudes: boreal and California forest fire emissions. Atmos. Environ.
44, 4553e4564. http://dx.doi.org/10.1016/j.atmosenv.2010.08.026.
Sinnhuber, B.-M., Weber, M., Amankwah, A., Burrows, J.P., 2003a. Total ozone during
the unusual Antarctic winter of 2002. Geophys. Res. Lett. 30, 1580e1584. http://
dx.doi.org/10.1029/2002GL016798.
Sinnhuber, M., Burrows, J.P., Chipperfield, M.P., Jackman, C.H., Kallenrode, M.-B.,
Künzi, K.F., Quack, M., 2003b. A model study of the impact of magnetic field
structure on atmospheric composition during solar proton events. Geophys.
Res. Lett. 30, 1818e1821. http://dx.doi.org/10.1029/2003GL017265.
Sofiev, M., Vankevich, R., Ermakova, T., Hakkarainen, J., 2013. Global mapping of
maximum emission heights and resulting vertical profiles of wildfire emissions.
Atmos. Chem. Phys. 13, 7039e7052. http://dx.doi.org/10.5194/acp-13-7039-
2013.
Stocks, B.J., Mason, J.A., Todd, J.B., Bosch, E.M., Wotton, B.M., Amiro, B.D.,
Flannigan, M.D., Hirsch, K.G., Logan, K.A., Martell, D.L., Skinner, W.R., 2003.
Large forest fires in Canada, 1959e1997. J. Geophys. Res. 108, 8149. http://
dx.doi.org/10.1029/2001JD000484.
Stohl, A., Klimont, Z., Eckhardt, S., Kupiainen, K., Shevchenko, V.P., Kopeikin, V.M.,
Novigatsky, A.N., 2013. Black carbon in the Arctic: the underestimated role of
gas flaring and residential combustion emissions. Atmos. Chem. Phys. 13,
8833e8855. http://dx.doi.org/10.5194/acp-13-8833-2013.
Takegawa, N., Kondo, Y., Koike, M., Ko, M., Kita, K., Blake, D.R., Nishi, N., Hu, W.,
Liley, J.B., Kawakami, S., Shirai, T., Miyazaki, Y., Ikeda, H., Russel-Smith, J.,
Ogawa, T., 2003. Removal of NOx and NOy in biomass burning plumes in the
boundary layer over northern Australia. J. Geophys. Res. Atmos. 108, 4308.
http://dx.doi.org/10.1029/2002jd002505.
Val Martin, M., Honrath, R.E., Owen, R.C., Lapina, K., 20 08. Large-scale impacts of
anthropogenic pollution and boreal wildfires on the nitrogen oxides over the
central North Atlantic region. J. Geophys. Res. D Atmos. 113 http://dx.doi.org/
10.1029/2007JD009689.
Val Martin, M., Logan, J.A., Kahn, R.A., Leung, F.-Y., Nelson, D.L., Diner, D.J., 2010.
Smoke injection heights from fires in North America: analysis of 5 years of
satellite observations. Atmos. Chem. Phys. 10, 1491e1510. http://dx.doi.org/
10.5194/acp-10-1491-2010.
Van Der Werf, G.R., Randerson, J.T., Giglio, L., Collatz, G.J., Mu, M., Kasibhatla, P.S.,
Morton, D.C., Defries, R.S., Jin, Y., Van Leeuwen, T.T., 2010. Global fire emissions
and the contribution of deforestation, savanna, forest, agricultural, and peat
fires (1997e2009). Atmos. Chem. Phys. 10, 11707e11735. http://dx.doi.org/
10.5194/acp-10-11707-2010.
Vermote, E., Ellicott, E., Dubovik, O., Lapyonok, T., Chin, M., Giglio, L., Roberts, G.J.,
2009. An approach to estimate global biomass burning emissions of organic and
black carbon from MODIS fire radiative power. J. Geophys. Res. D Atmos. 114
http://dx.doi.org/10.1029/2008JD011188.
Wang, P., Stammes, P., van der, A.R., Pinardi, G., van Roozendael, M., 2008. FRESCOþ:
an improved O2 A-band cloud retrieval algorithm for tropospheric trace gas
retrievals. Atmos. Chem. Phys. 8, 6565e6576. http://dx.doi.org/10.5194/acp-8-
6565-2008.
Wiedinmyer, C., Quayle, B., Geron, C., Belote, A., McKenzie, D., Zhang, X., O'Neill, S.,
Wynne, K.K., 2006. Estimating emissions from fires in North America for air
quality modeling. Atmos. Environ. 40, 3419e3432. http://dx.doi.org/10.1016/
j.atmosenv.2006.02.010.
Winkler, H., Sinnhuber, M., Notholt, J., Kallenrode, M.-B., Steinhilber, F., Vogt, J.,
Zieger, B., Glassmeier, K.-H., Stadelmann, A., 2008. Modeling impacts of
geomagnetic field variations on middle atmospheric ozone responses to solar
proton events on long timescales. J. Geophys. Res. 113, 11. http://dx.doi.org/
10.1029/2007JD008574. D02302.
Wooster, M.J., Zhang, Y.H., 2004. Boreal forest fires burn less intensely in Russia than
in North America. Geophys. Res. Lett. 31, L20505. http://dx.doi.org/10.1029/
2004GL020805, 20501e20503.
Yokelson, R.J., Christian, T.J., Karl, T.G., Guenther, A., 2008. The tropical forest and
fire emissions experiment: laboratory fire measurements and synthesis of
campaign data. Atmos. Chem. Phys. 8, 3509e3527. http://dx.doi.org/10.5194/
acp-8-3509-2008.
Web references
ftp://neespi.gsfc.nasa.gov/data/s4pa/Fire/MOD14CM1.0 05/ (last access August
2014).
https://lpdaac.usgs.gov/products/modis_products_table/mcd12q1 (last access
August 2014).
ftp://ladsweb.nascom.nasa.gov/allData/51/MOD08_M3/2005/ (last access August
2014).
S.F. Schreier et al. / Atmospheric Environment xxx (2014) 1e11 11
Please cite this article in press as: Schreier, S.F., et al., Differences in satellite-derived NO
x
emission factors between Eurasian and North
American boreal forest fires, Atmospheric Environment (2014), http://dx.doi.org/10.1016/j.atmosenv.2014.08.071
