Access to this full-text is provided by Copernicus Publications on behalf of European Geosciences Union.
Content available from Geoscientific Model Development
This content is subject to copyright.
Geosci. Model Dev., 17, 5545–5571, 2024
https://doi.org/10.5194/gmd-17-5545-2024
© Author(s) 2024. This work is distributed under
the Creative Commons Attribution 4.0 License.
Model evaluation paper
Evaluating CHASER V4.0 global formaldehyde (HCHO)
simulations using satellite, aircraft, and ground-based
remote-sensing observations
Hossain Mohammed Syedul Hoque1, Kengo Sudo1,2, Hitoshi Irie3, Yanfeng He1, and Md Firoz Khan4
1Graduate School of Environmental Studies, Nagoya University, Nagoya, 4640064, Japan
2Japan Agency for Marine-Earth Science and Technology (JAMSTEC), Kanagawa, 2370061, Japan
3Center for Environmental Remote Sensing (CEReS), Chiba University, Chiba, 2638522, Japan
4Department of Environmental Science and Management, North South University, Dhaka, Bangladesh
Correspondence: Hossain Mohammed Syedul Hoque (hoquesyedul@gmail.com,
hoque.hossain.mohammed.syedul.u6@f.mail.nagoya-u.ac.jp)
Received: 12 March 2024 – Discussion started: 25 March 2024
Revised: 27 May 2024 – Accepted: 9 June 2024 – Published: 24 July 2024
Abstract. Formaldehyde (HCHO), a precursor to tropo-
spheric ozone, is an important tracer of volatile organic
compounds (VOCs) in the atmosphere. Two years (2019–
2020) of HCHO simulations obtained from the global chem-
istry transport model CHASER at a horizontal resolution
of 2.8° ×2.8° have been evaluated using the Tropospheric
Monitoring Instrument (TROPOMI) and multi-axis differ-
ential optical absorption spectroscopy (MAX-DOAS) obser-
vations. In situ measurements from the Atmospheric To-
mography Mission (ATom) in 2018 were used to evalu-
ate the HCHO simulations for 2018. CHASER reproduced
the TROPOMI-observed global HCHO spatial distribution
with a spatial correlation (r) of 0.93 and a negative bias
of 7 %. The model showed a good capability to reproduce
the observed magnitude of the HCHO seasonality in differ-
ent regions, including the background conditions. The dis-
crepancies between the model and satellite in the Asian re-
gions were related mainly to the underestimated and miss-
ing anthropogenic emission inventories. The maximum dif-
ference between two HCHO simulations based on two dif-
ferent nitrogen oxide (NOx) emission inventories was 20%.
TROPOMI’s finer spatial resolution than that of the Ozone
Monitoring Instrument (OMI) sensor reduced the global
model–satellite root-mean-square error (RMSE) by 20 %.
The OMI- and TROPOMI-observed seasonal variations in
HCHO abundances were consistent. The simulated season-
ality showed better agreement with TROPOMI in most re-
gions. The simulated HCHO and isoprene profiles correlated
strongly (R=0.81) with the ATom observations. However,
CHASER overestimated HCHO mixing ratios over dense
vegetation areas in South America and the remote Pacific
region (background condition), mainly within the planetary
boundary layer (<2 km). The simulated seasonal variations
in the HCHO columns showed good agreement (R > 0.70)
with the MAX-DOAS observations and agreed within the 1σ
standard deviation of the observed values. However, the tem-
poral correlation (R∼0.40) was moderate on a daily scale.
CHASER underestimated the HCHO levels at all sites, and
the peak occurrences in the observed and simulated HCHO
seasonality differed. The coarseness of the model’s resolu-
tion could potentially lead to such discrepancies. Sensitiv-
ity studies showed that anthropogenic emissions were the
highest contributor (up to ∼35 %) to the wintertime regional
HCHO levels.
1 Introduction
Formaldehyde (HCHO), the most abundant carbonyl com-
pound in the atmosphere, is a high-yield oxidation product of
all primary biogenic and anthropogenic non-methane volatile
organic compounds (NMVOCs). Methane (CH4) oxidization
produces background HCHO concentrations of 0.2–1.0 ppbv
(Burkert et al., 2001; Singh et al., 2004; Sinreich et al., 2005;
Published by Copernicus Publications on behalf of the European Geosciences Union.
5546 H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations
Weller et al., 2000). Along with secondary sources (i.e., the
oxidization of NMVOCs), biomass burning, industrial pro-
cesses, and fossil fuel combustion are the primary HCHO
emission sources (Fu et al., 2008; Hak et al., 2005; Lee et
al., 1997). However, the oxidization of NMVOCs drives the
spatial variability of HCHO on a global scale (Franco et
al., 2015). HCHO removal mechanisms include photolysis
at wavelengths below 400nm, oxidization by hydroxyl rad-
icals (OH), and wet deposition. The atmospheric lifetime of
HCHO is around a few hours (Arlander et al., 1995). There-
fore, HCHO observations can help elucidate chemical pro-
cesses in the atmosphere. A few examples are the following:
(1) the ozone (O3) production regime can be determined from
the HCHO to nitrogen dioxide (NO2) ratio (Duncan et al.,
2010; Hoque et al., 2022; Martin et al., 2004); (2) midday
OH levels can be quantified from the oxidation of isoprene
into HCHO (Kaiser et al., 2015); and (3) HCHO, being an
intermediate product in the oxidation chain of NMVOCs, en-
genders the formation of carbon monoxide (CO) and carbon
dioxide (CO2). Consequently, CO chemical production from
NMVOCs and CH4can be quantified from HCHO measure-
ments (De Smedt et al., 2021).
Given its importance, global HCHO observations started
in 1995 with the launch of a nadir-viewing ultraviolet (UV)
sensor, the Global Ozone Monitoring Experiment (GOME;
Burrows et al., 1999). Since then, there have been numerous
sensors: the SCanning Imaging Absorption SpectroMeter for
Atmospheric CHartographY (SCIAMACHY; De Smedt et
al., 2008, 2010; Wittrock et al., 2006) on board the ENVISAT
satellite, the Ozone Monitoring Instrument (OMI) (Levelt
et al., 2018), the Global Ozone Monitoring Experiment-2
(GOME-2) (Munro et al., 2016), and the Ozone Mapping
and Profiler Suite (González Abad et al., 2016, new refer-
ence). The HCHO observations from these sensors have been
used extensively to evaluate models, air quality, and climate
change (Chutia et al., 2019; De Smedt et al., 2008, 2010,
2015; Hoque et al., 2022). The Tropospheric Monitoring In-
strument (TROPOMI) (De Smedt et al., 2021; Veefkind et
al., 2012), launched on the European Copernicus Sentinel-
5 Precursor (S5P) satellite on 13 October 2017, is a recent
addition to the series of nadir-viewing UV sensors providing
HCHO data. The unprecedented original spatial resolution of
3.5 ×7 km2(across track ×along track), which was refined
to 3.5 ×5.5 km2on 6 August 2019, is the crucial feature of
TROPOMI. Such spatial resolution is almost 16 times finer
than its predecessor, OMI (De Smedt et al., 2021). Such high-
resolution observations will likely reduce uncertainties in the
HCHO products for multiple research purposes.
Several studies that used the TROPOMI HCHO product
have been reported in the literature. De Smedt et al. (2021)
and Vigouroux et al. (2020) have validated TROPOMI
HCHO comprehensively against the MAX-DOAS and FTIR
networks. Both studies have concluded that TROPOMI
HCHO products have achieved the pre-launch accuracy re-
quirement of <40 %–80 %. Ryan et al. (2020) and Chan
et al. (2020) reported good agreement (temporal correlation
R > 0.70) between TROPOMI and MAX-DOAS in Mel-
bourne and Munich. In addition to validation studies, HCHO
products have been used to infer changes in the global HCHO
levels during the shutdown prompted by the COVID-19 pan-
demic (Levelt et al., 2022; Souri et al., 2021; Sun et al.,
2021), demonstrating the role of anthropogenic emissions in
global HCHO variability.
Among the multitude of applications of TROPOMI HCHO
observations, few efforts have specifically evaluated HCHO
simulations from global chemistry transport models (CTMs).
This work evaluates the HCHO spatiotemporal distribu-
tion simulated by the global Chemical Atmospheric General
Circulation Model for the Study of Atmospheric Environ-
ment and Radiative Forcing (CHASER) (Sekiya and Sudo,
2014; Sudo et al., 2002; Sudo and Akimoto, 2007) against
TROPOMI HCHO observations. In addition, airborne and
ground-based observations are used to validate the simulated
HCHO profiles and surface mixing ratios in a few regions.
CHASER simulations of NO2, OH, and O3have been eval-
uated against satellite and ground-based observations (e.g.,
Sekiya and Sudo, 2014; Sekiya et al., 2018). Moreover,
CHASER is a forward model in the chemical reanalysis sys-
tem (TCR) developed by Miyazaki et al. (2017, 2020). The
model simulations are performed at a horizontal resolution
of 2.8° ×2.8° (T42). Although the model can run at higher
resolutions, T42 is the most commonly used framework for
CHASER applications. Therefore, it is used for this study.
Hoque et al. (2022) validated CHASER-simulated NO2
and HCHO against OMI and MAX-DOAS observations
for 2017. CHASER showed good skills in reproducing the
HCHO abundances observed by OMI (spatial correlation
r=0.74) and MAX-DOAS (temporal correlation R > 0.80).
The study found that biomass burning contributes ∼50 % to
the HCHO levels observed at the site in Thailand. However,
the limitations of the study were that (1) the simulated HCHO
partial column and profile were evaluated against MAX-
DOAS observations on a seasonal scale only, (2) the model
sensitivity studies were site specific and thus did not provide
global statistics on the emission contribution, and (3) satellite
observations were used as supporting datasets, so the model–
satellite comparison was not comprehensive. This study uti-
lizes multi-satellite (TROPOMI and OMI) HCHO observa-
tions, different NOxemission inventories, aircraft measure-
ments, and daily and diurnal MAX-DOAS data to provide ro-
bust and comprehensive statistics on the model HCHO sim-
ulations.
2 Model, observations, and methods
2.1 CHASER
CHASER V4.0 (version 4) is a global CTM that studies the
atmospheric environment and radiative forcing. It is cou-
Geosci. Model Dev., 17, 5545–5571, 2024 https://doi.org/10.5194/gmd-17-5545-2024
H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations 5547
pled online with the MIROC atmospheric general circulation
model (AGCM) and the SPRINTAS aerosol transport model
(Takemura et al., 2005, 2009). The latest version of CHASER
(Ha et al., 2023; He et al., 2022) entails several updates, in-
cluding those related to the formation of aerosol species and
related chemistry, radiation, and cloud processes.
Through 263 multi-phase (gaseous, aqueous, and hetero-
geneous) chemical reactions, CHASER calculates the con-
centrations of 92 species by considering the chemical cycle
involving O3, NOx(nitrogen oxides), HOx(hydrogen ox-
ides), and CH4-CO along with the oxidation of NMVOCs
(Ha et al., 2023; He et al., 2022; Hoque et al., 2022; Miyazaki
et al., 2017; Sekiya et al., 2023). The chemical mechanism
adopted comes mainly from the master chemical mechanism
(MCM) (Jenkin et al., 2015). The stratospheric O3chem-
istry simulations are based on the Chapman mechanism, the
catalytic reaction of halogen oxides, and polar stratospheric
clouds. The dry and wet depositions are calculated based
on resistance-based parameterization (Wesley et al., 2007),
cumulus convection, and large-scale condensation parame-
terization. Advective trace transport is calculated using the
piecewise parabolic method (Colella and Woodward, 1984)
and flux-form semi-Lagrangian schemes. Tracer transport
is simulated on a sub-grid scale in the framework of the
prognostic Arakawa–Schubert cumulus convection scheme
(Emori et al., 2001) and a vertical diffusion scheme (Mel-
lor and Yamada, 1974). The simulations were performed at
a horizontal resolution of 2.8° ×2.8°, employing 36 verti-
cal layers from the surface to approx. 50 km altitude and a
20 min time step. At every time step, meteorological fields
obtained from the MIROC AGCM were nudged toward the
6-hourly NCEP FNL reanalysis data.
CHASER incorporates emissions from biomass burn-
ing, anthropogenic sources, lightning, and soil. Anthro-
pogenic NOxemissions for 2018 are obtained from the
HTAP_v3 inventory (Crippa et al., 2023). Other anthro-
pogenic emissions are taken from HTAP_v2.2 for 2008, and
the biomass burning emissions are from MACC-GFAS (In-
ness et al., 2013). The monthly soil NOxemissions, de-
rived from Yienger and Levy (1995), are constant across
years. Biogenic emissions of VOCs are obtained from a
process-based biogeochemical model: the Vegetation Inte-
grative SImulator for Trace gases (VISIT) (Ito and Inatomi,
2012). VISIT is part of the CHASER modeling frame-
work and incorporates the biogenic flux estimate scheme
of Guenther (1997) (Ito et al., 2012). The global iso-
prene emissions in VISIT and the CAMS global bio-
genic emission inventory (Sinderolova et al., 2022; based
on MEGANv2.1) are 400 and 450 Tg C yr−1, respectively.
Lightning NOxproduction estimates are based on the pa-
rameterization of Price and Rind (1992) and linked to the
convection scheme of the AGCM. Global NOxemissions in
CHASER are set to 43.80 Tg N yr−1, with industrial produc-
tion (23.10 Tg N yr−1), biomass burning (9.65 Tg N yr−1),
soil (5.50 Tg N yr−1), lightning (5 Tg N yr−1), and aircraft
(0.55 Tg N yr−1) considered as significant emission sources.
Annual monoterpene, acetone, and other non-methane
volatile organic compound (ONMV) emissions are 102, 20,
and 60 Tg C yr−1, respectively. Direct emissions of HCHO
from anthropogenic sources and biomass burning are not
considered in CHASER. However, secondary production of
HCHO from VOCs (C2H6, C3H8, C2H4, C3H6, CH3COCH3,
and ONMV) emitted directly from anthropogenic and pyro-
genic sources is considered.
Sekiya et al. (2018) comprehensively assessed CHASER-
simulated NO2abundances using OMI observations.
CHASER reproduced the ATom-observed OH spatiotempo-
ral variation well (Sekiya et al., 2018). The quality of O3sim-
ulations has been explained in the work of Sudo et al. (2007).
Ha et al. (2023) and He et al. (2022) updated the heteroge-
neous chemistry and lightning NOxschemes, respectively.
These updates have not been considered in the current study.
The effect of these updates on the HCHO simulations will
be addressed in a separate study. Multiple simulations with
varying emission inputs were performed for the study. They
are presented in Table 1.
To account for the altitude dependence of TROPOMI ob-
servations, averaging kernel (AK) information obtained from
the level (L2) files was applied to all simulations, follow-
ing the method of Sekiya et al. (2018). First, the simulated
HCHO profiles were sampled closest to the TROPOMI over-
pass of 13:30 LT (local time). Secondly, AKs averaged on a
2.8° bin grid were applied to the sampled profiles. Then, the
total column was calculated. Thirdly, the AK-applied model
columns on the available measurement days were selected.
2.2 TROPOMI
The TROPOMI operational L2 offline (OFFL) HCHO ver-
tical column density (VCD) (ver. 1.1.5.7) data from 2019
to 2020 have been used for this study. The S5P TROPOMI
HCHO L2 product user manual (Veefkind et al., 2012) pro-
vides a detailed product description. The TROPOMI HCHO
retrieval algorithm is based on the DOAS technique and was
adapted directly from the OMI QA4ECV product retrieval
algorithm (De Smedt et al., 2017). The three-step retrieval al-
gorithm was explained explicitly by De Smedt et al. (2018).
Slant columns were retrieved from the UV part of the spectra
(channel 3) in the 328.5–358 nm fitting window. The HCHO
cross-section data reported by Meller and Moortgart (2000)
were used to fit the spectra. All the cross-sections were con-
volved with the instrument’s slit function (adjusted after the
launch) for every row separately. Spectra averaged over the
tropical Pacific region from the prior day were used as ref-
erence spectra for the DOAS fit (De Smedt et al., 2021;
Vigouroux et al., 2020). The slant columns, therefore, ex-
ceed the average Pacific background HCHO levels because
they were derived from the differences between the local and
reference spectra. The slant columns were converted to tro-
pospheric columns (Nv) using a lookup table of vertically
https://doi.org/10.5194/gmd-17-5545-2024 Geosci. Model Dev., 17, 5545–5571, 2024
5548 H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations
Table 1. Combinations of emission inventories used in the different simulations used in this study.
Simulation name NOxemissions Biogenic emissions Anthropogenic VOC
emissions
Biomass burning
Standard
ANIa
OLNEb
Biogenic_off
Anthropogenic_off
Biomass_off
HTAP_v3
HTAP_v3
HTAP_v2.2
HTAP_v3
HTAP_ v3
HTAP_v3
ON
ON
ON
OFF
ON
ON
ON
Increased 3-fold
ON
ON
OFF
ON
ON
ON
ON
ON
ON
OFF
aAnthropogenic VOC emissions were increased 3-fold in ANI. bThe OLNE simulation used old NOxemissions.
resolved air mass factors (M) at 340 nm calculated with the
radiative transfer model VILDORT v2.6 (Spurr, 2008). The
value of Mdepends on the observation geometry, surface
albedo, cloud properties, and a priori profiles of HCHO.
The surface albedo at a spatial resolution of 1° ×1° was ex-
tracted from the monthly OMI albedo climatology (Kleipool
et al., 2008). Daily HCHO profiles were obtained a priori
from the TM5-MP CTM at a similar spatial resolution. The
independent-pixel approximation (Boersma et al., 2004) ap-
proach was applied to pixels with cloud fractions greater than
0.1. Background correction was performed based on HCHO
slant columns over the Pacific Ocean from the 5 days prior
to account for any remaining global offsets and stripes (De
Smedt et al., 2021). The background HCHO contribution
from CH4oxidation in the reference region was calculated
with TM5-MP. The resulting HCHO tropospheric column
was calculated using Eq. (1):
Nv=Ns−Ns,o
M+Mo
M·NCTM
v,0,(1)
where Mois the air mass factor of the reference sector. Fol-
lowing De Smedt et al. (2021), the following filters ensured
the data quality: (1) a cloud fraction of less than 0.3, (2) qual-
ity assurance values of greater than 0.5, (3) retrievals with a
solar zenith angle (SZA) of less than 70°, (4) a surface albedo
of less than 0.1, and (5) an air mass factor of greater than 0.1.
The total uncertainty in the reprocessed TROPOMI HCHO
columns was estimated as ≥90 % for the fire-free region
(Zhao et al., 2022, and references therein). The uncertainties
in the air mass factors, slant column fitting, and background
HCHO, respectively, account for 75 %, 25%, and 40% of the
total uncertainty. The estimated uncertainty in the retrievals
in regions with strong fires is ∼35 %. The filtering criteria
for the TROPOMI datasets are as follows: quality assurance
value (QA) >0.6, solar zenith angle <70°, cloud fraction
<0.3, air mass factor (AMF) >0.1, and surface reflectivity
<0.2.
TROPOMI observations are averaged spatially and tempo-
rally on the CHASER grid (T42), leading to horizontal rep-
resentativeness errors. However, the random horizontal rep-
resentativeness errors are on the order of 5 %–10 %, which is
lower than the individual retrieval error of the satellite obser-
vations (Boersma et al., 2016). If the model’s horizontal res-
olution is increased by 50 % (i.e., simulated at a horizontal
resolution of 1.4° ×1.4°), the change in HCHO abundances
is less than 6 % (Fig. S1 and Table S1 in the Supplement).
The vertical sensitivity of the satellite retrievals is the most
relevant source of representativeness error (Boersma et al.,
2016). The current study utilizes the TROPOMI AK infor-
mation to minimize the representativeness error. Therefore,
the horizontal representativeness error will likely affect the
results less than other error sources, such as uncertainties in
satellite retrieval, emission inventories, and model chemical
mechanisms.
2.3 OMI
The comparison study used HCHO retrievals from OMI,
a nadir-viewing spectrometer on board the Aura satellite,
which measures backscattering solar radiation in the spec-
tral range of 270–500 nm (Levelt et al., 2018). OMI crosses
the Equator at 13:40 LT (Zara et al., 2018) and provides
daily global coverage of trace gases, including HCHO, at
a spatial resolution of 13 ×24 km2. HCHO columns from
2019 to 2020 retrieved using BIRA-IASBv14 (De Smedt
et al., 2021) were obtained from the Aeronomie website
(i.e., https://www.temis.nl/qa4ecv/hcho/hcho_omi.php, last
access: 1 July 2023) for use in this study. The data-filtering
criteria were cloud fraction <0.3, SZA <70°, quality flag
=0, and cross-track quality flag =0. Like the TROPOMI
data, the OMI data were also averaged spatially and tempo-
rally on the model grid (T42).
2.4 ATom-4 aircraft campaign
The NASA Atmospheric Tomography (ATom) mission used
a DC-8 aircraft to study the remote atmosphere over the Pa-
cific and Atlantic oceans from ∼80° N to ∼65° S (Wofsy
et al., 2018). Repeated flights measured the vertical profiles
from 0.15 to 12 km to provide information related to green-
house gases, reactive and tracer species, and aerosol compo-
sition and size distribution (Kupc et al., 2018). Over 2 years
and four phases, sampling was conducted in one of the four
Geosci. Model Dev., 17, 5545–5571, 2024 https://doi.org/10.5194/gmd-17-5545-2024
H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations 5549
seasons in each stage (Zhao et al., 2022). Here, the 1 min
averaged measurements of HCHO and isoprene during the
ATom-4 flight (Fig. S2) in 2018 are used for the model eval-
uation. The NASA In Situ Airborne Formaldehyde (ISAF)
instrument (Cazorla et al., 2015) performed HCHO sampling
based on the laser-induced fluorescence technique. Isoprene
was measured using two instruments: (a) The University of
Irvine Whole Air Sampler (WAS) and (b) the National Center
for Atmospheric Research (NCAR) Trace Organic Gas An-
alyzer (TOGA). WAS sampled the air every 3–5 min, with
subsequent analyses performed in the laboratory using gas
chromatography (Simpson et al., 2020). TOGO sampling
was conducted every 2 min with a 35 s integrated sampling
time (Apel et al., 2021). The uncertainties in the WAS and
TOGA isoprene observations are, respectively, ±10 % and
15 %. The measurement uncertainty in HCHO was reported
as 10 %. The simulations have been interpolated to the ob-
served spatial and temporal resolutions following the method
of He et al. (2022). The observed and interpolated HCHO and
isoprene vertical profiles were averaged over a 300 m bin.
The Atom campaign took place between 2016 and 2018.
2.5 MAX-DOAS observations
HCHO columns and the volume mixing ratio (vmr) were
retrieved from 2 years of (2019–2020) MAX-DOAS obser-
vations at Phimai (15.18° N, 102.46° E, 212 m a.s.l.), Chiba
(35.62° N, 140.10° E, 21 m a.s.l.), and Kasuga (33.52° N,
130.47° E, 28 m a.s.l.). The MAX-DOAS observations were
conducted within the framework of the International Air
Quality and Sky Research Remote Sensing (A-SKY) net-
work (Irie, 2021). The sites were selected because contin-
uous measurements from 2019 to 2020 were available for
these sites. Phimai is a rural site in Thailand and experiences
a biomass burning influence from January to April. The cli-
mate is divided into two seasons: (1) the dry season (January
to May) and (2) the wet season (June to December). Chiba
and Kasuga are urban sites in central and southern Japan,
respectively. The seasonal classification at these sites is as
follows: spring is from March to May, summer is from June
to August, autumn is from September to November, and win-
ter is from December to February. The observations at these
sites are described elsewhere (i.e., Hoque et al., 2018a; Irie
et al., 2011, 2015).
The A-SKY MAX-DOAS system, including the instru-
ment and algorithm, participated in the Cabauw Intercom-
parison campaign for Nitrogen Dioxide measuring Instru-
ments (CINDI) and CINDI-2 (Kreher et al., 2020; Roscoe
et al., 2010). The instrumentation has been described ex-
plicitly by Irie et al. (2008, 2011, 2015). A UV spec-
trometer (Maya2000Pro; Ocean Insight, Inc.) recorded high-
resolution spectra from 310–515 nm at six elevation angles
(ELs) of 2, 3, 4, 6, 8, and 70°, which were recorded every
15 min. The reference spectra were recorded at an EL of 70°
instead of 90° to avoid saturation intensity. Spectra measured
at all ELs were considered in the retrieved vertical profile and
total columns. Consequently, the choice of reference ELs has
no appreciable effect on the retrieval. The systematic error in
the oxygen collision complex (O4) was reduced by limiting
the off-axis ELs to less than 10° (Irie et al., 2015). How-
ever, this limitation reduces sensitivity above the planetary
boundary layer (PBL), maintaining high sensitivity in the
lower layers of the retrieved profiles. The high-resolution so-
lar spectrum measured by Kurucz et al. (1984) was used for
daily wavelength calibration. The spectral resolution is ap-
proximately 0.4 nm at 357 and 476 nm (Hoque et al., 2022).
Aerosol and trace-gas columns and profiles were retrieved
using the Japanese vertical-profile retrieval algorithm JM2
(ver. 2) (Irie et al., 2011, 2015). Three-step profile and col-
umn retrievals by JM2 are explained explicitly in earlier re-
ports (e.g., Hoque et al., 2018a; Irie et al., 2011, 2015). The
partial VCD values are converted to the volume mixing ratio
(vmr) by scaling the US Standard Atmosphere temperature
and pressure data to the surface measurements at the respec-
tive site. The estimated total error (random and systematic)
in the HCHO product is 30 % (Hoque et al., 2022). Follow-
ing Irie et al. (2011) and Hoque et al. (2018a, 2022), cloud
screening was performed to ensure data quality.
3 Results and discussion
3.1 Comparison of CHASER HCHO with TROPOMI
observations
Figure 1 presents a comparison of the global distributions of
annual mean HCHO columns obtained from TROPOMI re-
trievals and standard CHASER simulations at the TROPOMI
overpass time (13:30 LT). Differences between the observa-
tions and model simulations in the respective years are also
depicted. The statistics related to the comparison are pre-
sented in Table 2. The simulation results agree well with the
TROPOMI observations, with a global spatial correlation (r)
of 0.93, a mean bias error (MBE) (CHASER −TROPOMI)
of −0.20 ×1015 molec. cm−2, and a root-mean-square er-
ror (RMSE) of 0.75 ×1015 molec. cm−2. The r, MBE, and
RMSE values between 60° S and 60° N were, respectively,
0.92, 0.13×1015 molec. cm−2, and 0.82 ×1015 molec. cm−2.
CHASER HCHO columns are negatively biased relative
to the TROPOMI retrievals. Table S2 shows the MBE
and RMSE values obtained for the individual months.
No seasonal variation in the systematic differences be-
tween CHASER and TROPOMI was observed. Biases can
originate from uncertainties in the retrieval and model
assumptions. TROPOMI HCHO retrievals greater than
8×1015 molec. cm−2were negatively biased by 25% rel-
ative to the ground-based MAX-DOAS observations (De
Smedt et al., 2021), whereas direct emissions of HCHO were
not considered in CHASER.
https://doi.org/10.5194/gmd-17-5545-2024 Geosci. Model Dev., 17, 5545–5571, 2024
5550 H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations
Table 2. Comparison of annual mean HCHO (×1016 molec. cm−2)
columns from TROPOMI retrievals and CHASER simulations for
2019 and 2020. MBE and RMSE are the abbreviations for mean
bias error and root-mean-square error, respectively. The units of
MBE and RMSE are ×1015 molec. cm−2. “Correlation” signifies
the spatial correlation between the datasets.
Year Correlation MBE RMSE
2019 0.93 −0.20 0.75
2020 0.93 −0.19 0.75
TROPOMI and CHASER show high HCHO con-
centrations over South America, central Africa, In-
dia, eastern China, and Southeast Asia. Simulated
HCHO magnitudes in the hotspot regions were 0.8–
1.4 ×1016 molec. cm−2, slightly higher than the observed
range of 0.8–1 ×1016 molec. cm−2. The dataset’s greatest
differences (∼4×1015 molec. cm−2) were observed over
Brazil and Southeast Asia. The datasets show a strong
congruence in the high-latitude regions. The simulated
and observed HCHO columns over Europe, the Middle
East, Japan, and Russia were 0.3–0.6 ×1016 molec. cm−2.
Simulated HCHO columns (∼3×1015 molec. cm−2) over
the remote Pacific region were consistent with the observa-
tions, too. The remote Pacific region represents background
conditions that are strongly linked to CH4oxidation. The
congruence with observations in this region suggests that the
simulated CH4estimates in remote areas are reasonable.
Figure 2 compares the observed and simulated seasonal-
ity in HCHO columns (×1016 molec. cm−2) in different re-
gions. Datasets for 2019 and 2020 were used to calculate
the observed and simulated monthly mean values. The MBE
(×1015 molec. cm−2) between the TROPOMI and CHASER
HCHO columns in each region is shown in blue. The compar-
ison statistics are given in Table 3. The regional boundaries
are shown on the global distribution map in Fig. S3. Tem-
poral correlations derived from daily values over 2 years are
provided in Table S2.
3.1.1 Eastern China
Over eastern China (E-China; Fig. 2a), the datasets are
moderately correlated spatially (r=0.44), with MBE and
RMSE values of −0.9 and 1.62 ×1015 molec. cm−2, respec-
tively. The simulated seasonality correlates strongly with
the observations (R=0.97), with a consistent peak (1 ×
1016 molec. cm−2) in the HCHO variability in July. The
HCHO column peaks are compatible with the peak in iso-
prene concentrations (Fig. S4), indicating a strong biogenic
contribution during summer. CHASER mostly underesti-
mated the wintertime HCHO columns in this region. Liu
et al. (2021) reported vehicular exhaust, solvent usage, and
combustion-related regional transport as the primary VOC
emission sources during winter in Shanghai, a megacity in
eastern China. NMVOC emissions from these sources (i.e.,
vehicular exhaust, solvent usage, and transport) are consid-
ered in the HTAP_v2.2 inventory (Crippa et al., 2023). Al-
though CHASER considered HCHO production from the
degradation of anthropogenic VOCs, it is likely underesti-
mated, resulting in a lower simulated wintertime HCHO col-
umn in this region.
3.1.2 Eastern USA, western USA, and Europe
CHASER has reproduced the HCHO spatial variability in the
eastern USA (E-USA; Fig. 2b; r=0.97) and western USA
(W-USA; Fig. 2c; r=0.85) well. The peaks in the HCHO
variability coincide with the isoprene peak in these regions
(Fig. S4). The simulated amplitudes of the HCHO sea-
sonal modulation in E-USA and W-USA are 74% and 62%,
whereas the observed seasonal amplitudes are 74 % and
65 %, respectively. The peak in the HCHO seasonality in E-
USA is similar in both datasets (∼1.2 ×1016 molec. cm−2).
The RMSE value in the W-USA region is 15 % higher than
that in E-USA. Although the spatial correlation in Europe
(Fig. 2d) is moderate (r=0.73), the temporal correlation is
strong (R=0.95). The simulated and observed HCHO sea-
sonal modulations in Europe are 60 % and 62 %, respectively.
The model–satellite discrepancies are prominent in Europe
and W-USA during summer and autumn. In both regions
(i.e., Europe and W-USA), the biogenic and anthropogenic
contributions to the total HCHO level are equivalent during
summer. In autumn, the anthropogenic emission contribu-
tions are higher (Sect. 3.8). This indicates a potential model
underestimation of biogenic HCHO levels in these regions,
linked to the uncertainties in the biogenic emission inven-
tory and isoprene mechanism. However, the model–satellite
agreement is strong during the winter in these regions. Dur-
ing winter, anthropogenic VOC emissions drive the HCHO
variability in these regions (Luecken et al., 2018; Pozzani et
al., 2002). Therefore, the simulated contribution of anthro-
pogenic sources to the HCHO abundances during winter in
these regions is reasonable.
3.1.3 Central, northern, and southern Africa
Over the African regions (Fig. 2e–g, the spatial correlation is
higher than 0.80. The African continent is the single largest
biomass burning emission source (Roberts et al., 2009). The
observed and simulated amplitudes of the HCHO season-
ality in central Africa (C-Africa; Fig. 2e) are, respectively,
45 % and 21 %. The mean simulated and observed HCHO
abundances in North Africa (N-Africa; Fig. 2f)) during the
biomass burning season are ∼1.06 ×1016 molec. cm−2, con-
sistent with the GOME-2 and SCIAMACHY observations
(De Smedt et al., 2008). Figure S5 shows the seasonal fire
radiative power (FRP) cycle over the three African regions.
FRP, a measure of outgoing radiant heat from fires, is a tracer
of changes in atmospheric trace constituents related to py-
Geosci. Model Dev., 17, 5545–5571, 2024 https://doi.org/10.5194/gmd-17-5545-2024
H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations 5551
Figure 1. Annual mean HCHO columns (×1016 molec. cm−2) in 2019 and 2020 obtained from TROPOMI retrievals (first column) and a
standard CHASER simulation (second column). The differences between the model and observations in the respective years are shown in
the third column. The unit of difference is ×1015 molec. cm−2.
Table 3. Comparison of monthly mean tropospheric HCHO (×1016 molec. cm−2) columns obtained from TROPOMI retrievals and stan-
dard CHASER simulations. Coincident dates in 2019 and 2020 are used to calculate the statistics. The units of MBE and RMSE are
×1015 molec. cm−2. The temporal correlations are derived from the seasonal means.
Region MBE (model −
TROPOMI)
RMSE
(model −
TROPOMI)
Spatial correla-
tion (rvalue)
Temporal
correlation (R
value)
E-China
E-USA
W-USA
Europe
C-Africa
N-Africa
S-Africa
S-America
India
IGP
E-India
S-India
SE-Asia
Remote Pacific
−0.91
0.40
−1.25
−0.74
1.13
1.10
−1.45
2.34
−1.20
−1.60
0.24
−0.36
−0.77
0.002
1.62
0.43
1.29
0.92
1.52
1.26
1.64
2.85
1.77
1.99
1.08
0.52
1.22
0.13
0.44
0.97
0.85
0.73
0.93
0.87
0.89
0.56
0.84
0.91
0.86
0.96
0.71
0.86
0.97
0.97
0.95
0.93
0.74
0.83
0.59
0.97
0.18
0.44
0.72
0.34
0.87
0.76
rogenic emissions (Hoque et al., 2018a). The observed and
simulated enhanced HCHO columns in N-Africa are congru-
ent with the high FRP values, indicating the contribution of
biomass burning to the HCHO abundances. CHASER could
not replicate the observed HCHO seasonality over C-Africa.
However, simulations show a decrease in the HCHO abun-
dances in C-Africa from January to March, consistent with
the changes in the coincident FRP values.
Over southern Africa (S-Africa; Fig. 2g), elevated
TROPOMI HCHO columns are consistent with GOME-2
and SCIAMACHY observations (De Smedt et al., 2008). The
observed peaks in HCHO columns and FRP values (Fig. S5)
are consistent and thus can be attributed to biomass burning.
https://doi.org/10.5194/gmd-17-5545-2024 Geosci. Model Dev., 17, 5545–5571, 2024
5552 H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations
Figure 2. Seasonal variation in HCHO columns (×1016 molec. cm−2) in (a) eastern China (E-China; 30–40° N, 110–123°E), (b) the east-
ern United States (E-USA; 32–43° N, 95–71° W), (c) the western United States (W-USA; 32–43° N, 125–100° W), (d) Europe (35–60° N,
−10° W–30° E), (e) central Africa (C-Africa; 4° S–5°N, 10–40°E), (f) northern Africa (N-Africa; 5–15° N, 10° W–30° E), (g) southern Africa
(S-Africa; 5–15° S, 10–30° E), (h) South America (S-America; 20° S–0° N, 50–70°W), (i) India (7.5–35° N, 68–89° E), (j) the Indo-Gangetic
Plain (IGP; 21–33° N, 72–89° E), (k) east India (E-India; 15–25° N, 80–90° E), (l) south India (S-India; 0–15° N, 63–80° E), (m) Southeast
Asia (SE-Asia, 10–20° N, 96–105°E), and (n) the remote Pacific region (28° S–32° N, 117–177° W), as inferred from CHASER simulations
(blue) and TROPOMI observations (red). Blue numbers denote MBE between the TROPOMI and CHASER HCHO columns. The observed
and simulated mean values represent the average of 2019 and 2020.
Pyrogenic emissions contribute ∼36 % of the HCHO in the
high-HCHO columns in this region (Sect. 3.8). TROPOMI
and CHASER have captured the shift in biomass burning
seasons from northern to southern Africa, which agrees well
with earlier observations (i.e., GOME-2 and SCIAMACHY).
The observed amplitude of the HCHO seasonal cycle in
southern and northern Africa is 46 %, signifying an almost
2-fold increase in HCHO abundances during the biomass-
burning season. Earlier studies (e.g., De Smedt et al., 2008;
Müller et al., 2008) found that such a feature (an increase
of a factor of 2) exists only in the southern African region.
This likely indicates an increase in fire intensity in northern
Africa.
3.1.4 South America
CHASER showed moderate skill in reproducing the observed
HCHO spatial distribution in South America (S-America;
Fig. 2h; r=0.56). However, the seasonal variation in the
HCHO columns is strongly correlated (R=0.97). The MBE
and RMSE in the South American continent are 2.34 ×1015
and 2.385 ×1015 molec. cm−2, respectively. The enhanced
HCHO columns during the South American biomass burn-
ing season are well reflected in the datasets. They show
a distinctive seasonal cycle. The observed and simulated
mean HCHO columns from August through October are
∼1.5 ×1016 molec. cm−2. CHASER estimated 46 % sea-
sonal modulation in the HCHO abundances, whereas the
observed modulation is 59 %. The model overestimates the
HCHO columns in S-America, just as it does in C-Africa and
N-Africa, probably because of the uncertainties in biogenic
emission inventories and the isoprene oxidation scheme.
3.1.5 India
CHASER reproduced the observed HCHO spatial distribu-
tion in India well (Fig. 2i; r=0.84), with an MBE and
RMSE of −1.20 ×1015 and 1.775 ×1015 molec. cm−2, re-
spectively. However, the temporal correlation (R=0.18) be-
tween the datasets is low. The observed seasonal modulation
of ∼30 % indicates a less-prominent seasonality in HCHO
abundances in India. The correlation between temperature
variations and isoprene emissions in India is inhomogeneous
(Starvakou et al., 2014). India has a diverse landscape, in-
cluding major forests over the east, northeast, and southwest
regions and deserts in northwestern India (Surl et al., 2018).
Geosci. Model Dev., 17, 5545–5571, 2024 https://doi.org/10.5194/gmd-17-5545-2024
H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations 5553
The Indo-Gangetic Plain (IGP) stretches from Eastern Pak-
istan to Bangladesh and is a major agricultural region in In-
dia (Kuttippurath et al., 2022). Thus, averaging the HCHO
columns over a diverse landscape can lead to less-prominent
seasonality. Moreover, biomass burning compromises 23%
of India’s total NMVOC emissions (13 Tg yr−1; Stewart et
al., 2021). Sensitivity analysis (Sect. 3.8) estimates show that
the biomass burning contribution to the HCHO levels in India
is ∼2 %, indicating that the modeled biomass burning emis-
sions for India are underestimated. Considering the diverse
Indian landscape, model–satellite comparisons for three re-
gions in India (the IGP, east India, and south India) are shown
in Fig. 2j–l.
The model has shown good skill in reproducing the ob-
served HCHO spatial variation in the IGP (Fig. 2j) region
(r=0.91). However, the temporal correlation is moderate
(R=0.44). Several field studies (e.g., Hoque et al., 2018b)
have reported biomass burning influences during spring and
autumn in the IGP, explaining the observed elevated HCHO
columns. The HCHO seasonal variation during January–
June is consistent in both datasets, with an Rvalue of
0.78. The mean observed and modeled HCHO abundances
during spring in the IGP are, respectively, 1.19 ×1016 and
8.72 ×1015 molec. cm−2. However, the model could not re-
produce the biomass burning events in autumn, which re-
duced the overall Rvalue in the IGP region. CHASER under-
estimates winter HCHO columns in the IGP region. Liquid
petroleum gas (LPG) usage, evaporative fuels, and garbage
burning contribute significantly to winter NMVOC levels in
Delhi and Mohali (Kumar et al., 2021). Although NMVOC
emissions from these sources are considered in the simula-
tions, they are likely underestimated for the IGP region.
Over east India (Fig. 2k), both the spatial agreement
(r=0.86) and the temporal agreement (R=0.72) between
TROPOMI and CHASER HCHO are strong. The observed
and modeled amplitudes of the HCHO seasonal cycle are
40 %. Both datasets show enhanced HCHO levels during
spring, consistent with high isoprene concentrations (Fig. S4)
Biogenic emissions are the main driver of the HCHO levels
in east India; however, emissions from mines are also poten-
tial sources of NOxand VOCs (Kuttippurath et al., 2022).
Similarly, CHASER shows a strong capability to repro-
duce the HCHO spatial distribution (r=0.96) in south In-
dia (S-India; Fig. 2l). However, the temporal correlation is
low. The mean observed and simulated HCHO abundances
are, respectively, 4.68 ×1015 and 5.03 ×1015 molec. cm−2.
The HCHO seasonality in S-India is less prominent than in
the other two regions. The coordinate bounds defined for S-
India in this study compromise a large portion of the southern
coastal region, which experiences a tropical maritime climate
with limited seasonal variations in temperature (Surl et al.,
2018). Such a feature can potentially lead to a less prominent
HCHO seasonality in S-India.
3.1.6 Southeast Asia
In Southeast Asia (SE-Asia; Fig. 2m), the rvalue is 0.71.
The MBE and RMSE are, respectively, −0.77 ×1015 and
1.2 ×1015 molec. cm−2. During the dry season (January–
April), prominent biomass burning occurs in this region in
many countries (e.g., Thailand, Malaysia, Indonesia, and
Cambodia). Such fire events degrade local air quality and
cause transboundary pollution (Hoque et al., 2018a, b;
Khan et al., 2016). TROPOMI and CHASER have cap-
tured the enhanced HCHO levels caused by pyrogenic
emissions well. The simulated and observed mean dry-
season HCHO columns are, respectively, 1.07 ×1016 and
1.35 ×1016 molec. cm−2. The observed and simulated am-
plitudes of the seasonal cycle are, respectively, 48 % and
60 %. The CHASER-reproduced columns for the dry season
are underestimated. Potential reasons for such discrepancies
are discussed in Sect. 3.3.
3.1.7 Remote Pacific region
The datasets correlate strongly over the remote Pacific
region (Fig. 2n), which represent the background con-
dition. No prominent seasonal variation is observed in
this region, which CHASER has simulated well. The
simulated and observed background HCHO columns are
2.86 ×1015 molec. cm−2.
3.2 Comparisons for countries with large forested
areas
Figure 3 shows the observed and simulated HCHO columns
over countries where large forested areas are located. The
definition of these countries used in the work of Opacka et
al. (2021) was adopted. The statistics presented in Table 4
include regions with high and low biogenic activities. This
section compares the overall biogenic emissions in the de-
fined regions with literature values and assesses their impact
on model performance.
Over China, CHASER correlates strongly with TROPOMI
(r=0.92), with an MBE of −3×1015 molec. cm−2. The
lowest differences between the datasets are observed pri-
marily in the southeastern and western parts of China. The
megacities of Shanghai, Nanjing, and Guangzhou are located
in southeastern China. Consequently, CHASER has demon-
strated good skills in the areas encompassed by multiple
megacities. The annual isoprene emission for China from
CHASER is 34 Tg C yr−1, which is higher than that found
by Opacka et al. (2021) (9.5–23 Tg C yr−1).
CHASER has shown excellent skill in reproducing
TROPOMI observations over the US. Along with high rval-
ues, the simulated magnitudes of the HCHO columns are
consistent with observations throughout the whole region.
Consequently, the bias between the datasets for the US is
2 %. In CHASER, annual isoprene emissions in the US and
https://doi.org/10.5194/gmd-17-5545-2024 Geosci. Model Dev., 17, 5545–5571, 2024
5554 H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations
Figure 3. Two-year (2019 and 2020) mean CHASER (first column) and TROPOMI (second column) HCHO columns (×1016 molec. cm−2
cm−2) in China (18.19–53.45° N, 73.67–135.02° E), the United States (18.91–45° N, 66–171° W), Indonesia (10° S–6° N, 95–142° E), and
Brazil (33° S–5.24° N, 34–73° W). The differences between the datasets are presented in the third column. Only the coincident dates among
the datasets are used to calculate the annual mean data.
Table 4. Comparison of the 2-year mean HCHO
(×1015 molec. cm−2) columns from TROPOMI and CHASER
over countries with large forested areas. The coordinate bounds of
the regions are adapted from Opacka et al. (2020). “Correlation”
signifies the spatial agreement between CHASER and TROPOMI
calculated from the annual mean data. The unit of MBE is
×1015 molec. cm−2.
Region Spatial correlation MBE
(model vs. TROPOMI) (model −TROPOMI)
China 0.92 −0.84
US 0.93 −0.05
Indonesia 0.81 1.05
Brazil 0.84 1.06
the southeastern US are 22 and 7.8 Tg C yr−1, respectively.
Such values are within the ranges reported by Stavrakou et
al. (2015) and Opacka et al. (2021).
The MBE between TROPOMI and CHASER in Indone-
sia is 1.05 ×1015 molec. cm−2. The rvalue is 0.81. In-
donesia’s annual mean TROPOMI and CHASER HCHO
abundances are 5.06 ×1015 and 6.15 ×1015 molec. cm−2,
respectively. The most significant differences between the
datasets (4 ×1015 molec. cm−2) are observed for the islands
of Sumatra, Borneo, and Sulawesi. Annual isoprene emis-
sions in Indonesia used in the CHASER simulations are
42 Tg C yr−1. Indonesian isoprene emissions vary between
25.5 and 32 Tg C yr−1, depending on the land-use change
(Opacka et al., 2021). Top-down estimates based on OMI and
GOME-2 observations are ∼11 Tg C yr−1(Stavrakou et al.,
2015). However, the value of 11 Tg C yr−1is half of the top-
down estimate based on SCIAMACHY observations. Conse-
quently, isoprene emissions in Indonesia remain very uncer-
tain. However, CHASER estimates using VISIT emissions
are higher than those reported in the literature, likely leading
to overestimation by the model for Indonesia.
CHASER overestimates the HCHO columns over Ama-
zonia (mostly in northern Brazil). Figure S6 shows the ob-
served and simulated seasonal HCHO variation over Brazil.
Although the model reproduced the temporal variability
well, the magnitude was overestimated. This indicates that
emission uncertainties are more prominent than uncertain-
ties related to the chemical mechanism for this region. In
CHASER, annual isoprene emissions over Amazonia are
67 Tg yr−1, consistent with the OMI-based top-down esti-
mate of 70 Tg yr−1obtained using a priori emissions from
MEGAN (Stavrakou et al., 2015). However, deforestation
Geosci. Model Dev., 17, 5545–5571, 2024 https://doi.org/10.5194/gmd-17-5545-2024
H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations 5555
affects the VOC emissions in the Amazon (Yáñez-Serrano
et al., 2020). Massive deforestation occurred in the Ama-
zon between 1985 and 2020, changing 11 % of the Amazo-
nian biome (Cabarello et al., 2022). Depending on the land
use and land cover change (LULCC), isoprene emissions in
Brazil can vary between 79 and 106.5 Tgyr−1(Opacka et al.,
2021). Moreover, although biogenic VOC modeling for the
Amazon has improved, VOC dynamics in the changing Ama-
zonian biome are poorly understood (Salazar et al., 2018;
Taylor et al., 2018). Therefore, updated biogenic VOC and
LULCC inventories can potentially improve the model per-
formance for Brazil.
In addition, CHASER isoprene emission estimates for Eu-
rope and Russia are, respectively, 17 and 15 Tg Cyr−1, which
are comparable to values reported in the literature (e.g.,
Guenther et al., 2006; Sinderolova et al., 2022).
This discussion focuses on isoprene emissions because
isoprene is the dominant biogenic VOC (BVOC). Although
not included in the current discussion, the chemical yield of
HCHO from the oxidation of other BVOCs might also be a
source of model uncertainty.
3.3 Uncertainties related to anthropogenic VOC
emissions
Uncertainties in anthropogenic VOC emissions can also be
crucially important. Sensitivity simulations are performed
by perturbing the anthropogenic VOC emissions. Perturba-
tion effects are relevant when the anthropogenic VOC emis-
sions are increased by 3-fold or more. We select the lowest
perturbed simulation (i.e., a 3-fold increase; hereafter ANI).
The better agreement between ANI and TROPOMI HCHO
columns is attributed to underestimated anthropogenic VOC
emissions in the standard simulation. Figure 4 compares the
TROPOMI HCHO columns and ANI simulations for 2019.
Standard simulation estimates for 2019 are also shown. The
comparison statistics are provided in Table 5.
Over E-China (Fig. 4a) and India (Fig. 4i), ANI shows bet-
ter agreement with TROPOMI than the standard simulation
during winter. In India and China, the contribution of anthro-
pogenic emissions to the NMVOC levels is more significant
during the winter (Kumar et al., 2021; Liu et al., 2021). Thus,
the ANI simulations improve the contribution of the winter-
time anthropogenic VOCs in these regions. The ANI MBE
and RMSE values over E-China are higher than those of the
standard simulation. This indicates that the anthropogenic
VOC estimates in E-China during the other seasons are rea-
sonable. In contrast, the ANI simulations reduce the MBE
values in India, indicating a greater underestimation of an-
thropogenic VOC emissions in this region than in E-China.
Similar to E-China, the ANI MBE and RMSE values are
higher in C-Africa, N-Africa, S-Africa, South America, and
E-USA. Over Europe (Fig. 4d) and W-USA (Fig. 4c), ANI
RMSE values are lower than those of the standard simulation.
The ANI simulations replicated the observed HCHO column
magnitudes in both regions from October to December, re-
sulting in lower RMSE values.
ANI estimates during the dry season in SE Asia (Fig. 4m)
are similar to the standard simulation values, indicating a
small effect of anthropogenic-emission uncertainties. The
dry-season columns are overestimated when the anthro-
pogenic VOC emissions are increased 5-fold (Fig. S7).
Space-based observations have provided substantial evidence
of increasing anthropogenic VOC emissions in Asian cities
(Bauwens et al., 2022). Therefore, the anthropogenic-VOC-
emission inventory should be updated to reduce the discrep-
ancy between CHASER and TROPOMI over SE-Asia.
3.4 Impacts of NOxemission uncertainties on HCHO
simulations
Uncertainties in the NOxemissions can affect the HCHO
abundances through the NOx–HOx–VOC cycle. Such ef-
fects are assessed by comparing simulations that use differ-
ent NOxinventories with the TROPOMI observations. The
CHASER standard, OLNE, and TROPOMI HCHO columns
are depicted in Fig. 5. The HTAP_v3 NOxemission in-
ventory was replaced with the HTAP_v2.2 inventory in the
OLNE simulations without altering the remaining emission
inventories. The differences between the two NOxinven-
tories are that (1) the HTAP_v3 inventory considers the
changes in NOxemissions from 2000 to 2018, whereas the
temporal coverage of HTAP_v2.2 is from 2008 to 2010, and
(2) the emissions in HTAP_v3 have higher sectoral disaggre-
gation (Crippa et al., 2023). The comparison-related statistics
are given in Table S3. NOxemissions from both inventories
are shown in Fig. S8.
On a global scale, HCHO column estimates are mostly un-
affected by the changes in the NOxemission inventories, as
indicated by the MBE values (Table 6). However, the RMSE
is 8 % lower in the case of the standard simulation. OLNE
estimates for higher latitudes (>=50° N) are 5% lower than
those from the standard simulations. Such differences do not
affect the model–satellite agreement in these regions.
The standard HCHO columns in India, China, and South-
east Asia are approximately 10 %–20 % lower than the
OLNE estimates (Fig. 5c). In fact, those differences are con-
sistent with changes in the regional OH estimates (Fig. 5d).
This finding implies that the changes in the NOxemission
estimates have affected the OH and HCHO abundances in
these regions. Satellite data assimilation results reported by
Miyazaki et al. (2017, 2020) indicate that NOxemissions in
India have increased by 30 % since 2008, whereas NOxemis-
sions in China have declined since 2011 (Liu et al., 2016).
Over E-China (Fig. 5a and b), the standard simulations re-
duce the absolute annual mean difference between OLNE
and TROPOMI of 3 ×1015 to 1 ×1015 molec. cm−2, which
is consistent with the lower NOxemissions in this region in
the updated inventory (Fig. S8). Over India and SE-Asia, the
standard OH concentrations are ∼40 % lower (Fig. 5d) than
https://doi.org/10.5194/gmd-17-5545-2024 Geosci. Model Dev., 17, 5545–5571, 2024
5556 H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations
Figure 4. Seasonal variation of HCHO (×1016 molec. cm−2) in selected regions, as inferred from standard simulations (blue), TROPOMI
observations (red), and ANI estimates (green). Anthropogenic VOC emissions are increased 3-fold in the ANI simulations. The blue numbers
denote the MBE between the TROPOMI and CHASER HCHO columns. The MBE between the ANI and TROPOMI columns is shown in
green. The coordinate bounds of the regions are similar to those used in Fig. 2. Simulations and observations in 2019 were used to calculate
the monthly mean values.
Table 5. Comparison of regional mean tropospheric HCHO (×1016 molec. cm−2) columns inferred from TROPOMI observations, the
standard simulation, and ANI estimates. The units of MBE1, MBE2, RMSE1, and RMSE2 are ×1015 molec. cm−2. The simulations and
observations for 2019 were used to calculate the statistics.
Region MBE1
(Standard −
TROPOMI)
MBE2
(ANI −
TROPOMI)
RMSE1
(Standard −
TROPOMI)
RMSE2
(ANI −
TROPOMI)
E-China
E-USA
W-USA
Europe
C-Africa
N-Africa
S-Africa
S-America
India
IGP
E-India
S-India
SE-Asia
−0.84
0.53
−0.72
−0.78
1.19
1.46
−0.99
2.99
−1.05
−1.22
0.26
−0.59
−0.76
1.54
2.22
0.17
0.29
3.32
2.19
0.87
3.92
0.39
0.29
1.64
0.48
0.59
1.40
0.58
0.80
0.92
1.57
1.61
1.32
3.41
1.57
1.69
1.22
0.69
1.16
1.74
2.25
0.43
0.67
3.60
2.30
1.39
4.28
1.50
2.02
2.11
0.58
0.78
Geosci. Model Dev., 17, 5545–5571, 2024 https://doi.org/10.5194/gmd-17-5545-2024
H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations 5557
the OLNE estimates, resulting in lower HCHO columns. The
lower standard HCHO columns can be linked to the increas-
ing NOxemissions in these regions (Fig. S8); however, the
magnitude of the change in the NOxemissions for these re-
gions in the updated inventory is likely overestimated.
In E-USA and W-USA (Table S3), the standard simula-
tion reduces the MBE by 26 % and 12 %, respectively. The
reductions in MBE and RMSE values in Africa and South
America are less than 10 %. Therefore, NOxemission uncer-
tainties mainly affect the HCHO simulations in India and SE
Asia.
3.5 Comparison with OMI HCHO observations
TROPOMI was able to achieve improved precision of HCHO
columns for shorter timescales (De Smedt et al., 2021). The
effect of such a feature on the comparison results is eval-
uated in this section. The method of De Smedt et al. (2021)
has been adopted to minimize the effect of the use of different
cloud retrieval algorithms for OMI and TROPOMI retrievals.
Figure S9 shows the global distribution of the mean HCHO
columns obtained from TROPOMI and OMI retrievals and
CHASER simulations in 2019 during the TROPOMI over-
pass time (13:30 LT). Only the coincident dates among the
three datasets are shown. Global and regional comparison
statistics are presented in Table 6.
The spatial correlation between OMI and CHASER is
0.89 (Table 6). OMI retrievals are positively biased by 7 %
compared to CHASER results. A similar bias is also ob-
served between TROPOMI and CHASER. Despite their sim-
ilar MBE values, TROPOMI reduces the global RMSE by
20 %. Monthly MBE and RMSE values between OMI and
CHASER are higher than those of TROPOMI and exhibit
no seasonality (Table S3). The highest absolute differences
between the model and OMI retrievals are observed in Ama-
zonia in Brazil, C-Africa, and SE-Asia (Fig. S9). The mag-
nitudes of the differences between the model and observa-
tions in these regions are similar for both sensors. Despite
the improved resolution, TROPOMI and OMI show equiva-
lent biases in regions with high HCHO levels (De Smedt et
al., 2021). A regional comparison among the three datasets
is portrayed in Fig. 6. The red (TROPOMI −CHASER) and
green (OMI −CHASER) numbers are the respective MBE
values.
Over E-China (Fig. 6a), the monthly mean TROPOMI
columns are ∼22 % lower than those of OMI, which reduces
the RMSE by 53 %. The simulated spatial distribution shows
better congruence with the new observations. TROPOMI
improved the summer model–satellite agreement consider-
ably. The magnitude of the seasonal modulation in the three
datasets is 50 %. Both sensors show that winter HCHO levels
in E-China are ∼8×1015 molec. cm−2.
Over E-USA (Fig. 6b), the rvalue between CHASER and
OMI is 0.86. CHASER columns are underestimated com-
pared to OMI, with MBE and RMSE values of −1.0×
1015 and 1.1 ×1015 molec. cm−2, respectively. TROPOMI
reduced the model–satellite RMSE by 50 % and improved
the rvalue by 6 %. The most significant improvements were
observed during the summer and autumn.
Over W-USA (Fig. 6c), TROPOMI retrievals are 26 %
lower than OMI observations, which reduces the model–
satellite RMSE by 63 %. The spatial correlation be-
tween OMI and CHASER is moderate. The simulated and
TROPOMI wintertime columns are ∼30 % lower than OMI.
However, the observed peak in HCHO seasonality in July is
consistent with the observational datasets.
OMI and TROPOMI HCHO observations over Europe
(Fig. 6d) are consistent. The seasonal cycle amplitude in-
ferred from both sensors is 60 %. The simulated spatial dis-
tribution shows better agreement with the TROPOMI obser-
vations, demonstrating the effects of improved resolution.
Over C-Africa (Fig. 6e), the RMSE value between
CHASER and OMI is ∼18 % lower than that of TROPOMI.
TROPOMI’s values are biased by 18 % on the low side com-
pared to OMI.
Over N-Africa (Fig. 6f), OMI retrievals are moder-
ately correlated with CHASER. The amplitudes of sea-
sonal modulation inferred from CHASER, TROPOMI, and
OMI are 48 %, 62 %, and 66 %, respectively. The RMSE
and MBE between OMI and CHASER are 1.41 ×1015 and
1.59 ×1015 molec. cm−2, respectively. OMI retrievals are
approximately 13 % higher than TROPOMI retrievals. Simu-
lated North African HCHO columns show better consistency
with the observations during the biomass burning season.
Over S-Africa (Fig. 6g), OMI HCHO columns are bi-
ased by 32 % and 25 % on the high side compared to
TROPOMI and CHASER. The simulated seasonal variabil-
ities and spatial distribution of HCHO show more relevance
to TROPOMI than to OMI.
Over S-America (Fig. 6h), the simulated peak
(1.6 ×1016 molec. cm−2) in the HCHO seasonality
shows strong congruence with the OMI observations.
Despite such consistency, simulated values are higher
than OMI retrievals, with MBE and RMSE values of
∼2×1015 molec. cm−2. Observations and simulations
show that the peak HCHO abundances can vary between
1.0 ×1016–1.8 ×1016 molec. cm−2in September. Although
the rvalue between OMI and CHASER is higher than that of
TROPOMI, the model’s capability to replicate the observed
spatial distribution was limited. OMI HCHO columns are
positively biased by 30 % compared to TROPOMI, thereby
reducing the model–satellite RMSE by 23 %.
Over India (Fig. 6i), CHASER HCHO columns are neg-
atively biased by 23 % compared to OMI observations. Al-
though TROPOMI minimized the model–satellite bias, sea-
sonal discrepancies between the model and observations pre-
vail. Over the IGP region, OMI HCHO retrievals are biased
https://doi.org/10.5194/gmd-17-5545-2024 Geosci. Model Dev., 17, 5545–5571, 2024
5558 H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations
Figure 5. Annual mean HCHO columns (×1016 molec. cm−2) in 2019, obtained from the (a) standard and (b) OLNE simulations. The
HTAP-2008 NOxemission inventory was used instead of the HTAP-2018 inventory for the OLNE simulations (Table 1). The remaining
emission inventories were similar in both simulations. (c) Global relative differences between the two HCHO simulations (OLNE−Standard).
(d) Relative differences (global) between two OH (OLNE−Standard) simulations. The standard and OLNE OH simulation settings are similar
to those described for Table 1. The OH and HCHO simulations were obtained simultaneously.
Table 6. Comparison of global and regional mean HCHO columns between satellite observations (TROPOMI and OMI) and standard
CHASER simulations. The units of MBE and RMSE are ×1016 molec. cm−2. The rvalue signifies the spatial correlation. The statistics are
based on simulations and observations for 2019.
Region MBE1
(Standard −
TROPOMI)
MBE2
(Standard −
OMI)
RMSE1
(Standard −
TROPOMI)
RMSE2
(Standard −
OMI)
rvalue
(CHASER
vs. TROPOMI)
rvalue
(CHASER vs.
OMI)
Global
E-China
E-USA
W-USA
Europe
C-Africa
N-Africa
S-Africa
S-America
India
IGP
E-India
S-India
SE-Asia
−0.23
−0.84
0.53
−0.72
−0.78
1.19
1.46
−0.99
2.99
−1.05
−1.22
0.26
−0.59
−0.76
−0.24
−2.54
−1.02
−2.09
−1.31
0.94
1.42
−2.59
2..02
−1.19
−2.85
−0.05
−0.16
−0.83
0.77
1.40
0.58
0.80
0.92
1.57
1.61
1.32
3.41
1.57
1.69
1.22
0.69
1.16
0.99
3.03
1.12
2.17
1.60
1.28
1.59
2.75
2.61
2.66
3.19
1.34
0.41
1.14
0.93
0.56
0.92
0.83
0.77
0.93
0.81
0.86
0.47
0.85
0.91
0.82
0.96
0.78
0.89
0.17
0.86
0.64
0.67
0.93
0.79
0.84
0.56
0.66
0.84
0.76
0.97
0.86
Geosci. Model Dev., 17, 5545–5571, 2024 https://doi.org/10.5194/gmd-17-5545-2024
H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations 5559
Figure 6. Seasonal variation in HCHO (×1016 molec. cm−2) inferred from TROPOMI (red curve) and OMI (orange curve) retrievals and
standard CHASER (blue curves) simulations. The region definitions are shown in Fig. S2. The blue numbers signify the MBE between
TROPOMI and CHASER, whereas the green numbers represent the MBE between CHASER and OMI. Coincident dates in 2019 among the
datasets are used to calculate the monthly mean data.
by 24 % and 36 % on the high side, respectively, compared
to TROPOMI and CHASER. Both sensors captured similar
HCHO seasonality in the IGP, with a modulation of 49 %.
Although CHASER could not reproduce the seasonality, the
simulated modulation is 48 %. The bias between the model
and observations (OMI and TROPOMI) is ∼4% in E-India
and S-India. The simulated HCHO spatial variation strongly
correlates with the observation datasets (rvalue of ∼0.85).
The amplitude of the seasonal modulation in E-India inferred
from OMI is ∼40 %.
Over Southeast Asia (Fig. 6m), CHASER columns are
negatively biased by 19% compared to the OMI columns.
Despite lower biases, both datasets have similar model–
satellite discrepancies during the dry season. A few reasons
for the underestimation by CHASER in SE Asia during the
dry season have been discussed in Sect. 3.2. In addition, as-
sumptions used and uncertainties in the retrieval could also
potentially engender this model–satellite discrepancy. Fig-
ure S10 compares CHASER and OMI SOA (González Abad
et al., 2016) products. The data selection criterion is simi-
lar to the description presented in Sect. 2. The most relevant
differences between the OMI BIRA and SAO products are
related to the underlying CTMs that simulate the a priori pro-
files and the reference sector correction (Zhu et al., 2016). A
comprehensive list of the differences between the two prod-
ucts is available from Zhu et al. (2016). The comparison
statistics are given in Table S5. CHASER columns during
the dry seasons in SE-Asia show excellent agreement with
the OMI SOA retrievals (Fig. S10m). OMI SOA values dur-
ing the dry season are negatively biased by 7% compared to
TROPOMI observations. The MBE between CHASER and
the SOA product is 0.04 ×1015 molec. cm−2. Based on the
comparison with OMI SOA products, the model performance
during the dry season can be considered excellent. The emis-
sion estimates for SE-Asia in CHASER can be regarded as
reasonable, too.
Similarly, in E-China (Fig. S10a), the OMI SOA prod-
uct reduces the bias between the model and observations
by 11 %. The simulated wintertime columns are consistent
with the SOA estimates but underestimated compared to
TROPOMI. The ANI estimates (Fig. 4a) for this region are
higher than the SOA product, indicating that the anthro-
pogenic emissions in CHASER for this region are rational.
Therefore, uncertainties related to the retrieval procedure can
also significantly affect the comparison results on a regional
scale.
A comparison between CHASER and OMI BIRA HCHO
products shows differences from the results of Hoque et
al. (2022), where the simulation and observations for 2017
were used. The simulations in both studies are similar. How-
ever, the OMI data in the earlier study are systematically
higher, which is the main cause of the statistically significant
https://doi.org/10.5194/gmd-17-5545-2024 Geosci. Model Dev., 17, 5545–5571, 2024
5560 H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations
differences between the study results. A detailed investiga-
tion of the reasons will be addressed in a separate work.
3.6 Validation using MAX-DOAS observations
3.6.1 Seasonal variation
CHASER columns are compared with ground-based MAX-
DOAS observations in Phimai, Chiba, and Kasuga in Fig. 7.
Coincident TROPOMI observations over the sites are used
for comparative discussion. The standard and OLNE simu-
lations are used. MAX-DOAS observations between 12:00
and 15:00 LT were averaged to estimate the monthly mean
columns. Only dates common to the three datasets were com-
pared. De Smedt et al. (2021) compared the TROPOMI and
A-SKY MAX-DOAS datasets in Phimai and Chiba. Because
the comparison between the model and ground-based obser-
vations is the primary focus of this comparison effort, we
do not consider the differences in the vertical sensitivities of
TROPOMI and MAX-DOAS. Thus, the statistics will differ
from those in De Smedt et al. (2021).
In Phimai, the standard CHASER HCHO seasonality cor-
relates strongly (R=0.71) with the MAX-DOAS observa-
tions; it is underestimated by 39 %. However, the bias be-
tween the standard model estimates and TROPOMI obser-
vations is 4 %. Despite a strong correlation, TROPOMI ob-
servations are negatively biased by 37 % compared to MAX-
DOAS (R=0.84). Such underestimation might be related to
the coarse binning of the satellite data. Upon using finer bin-
ning, De Smedt et al. (2021) reported a negative bias of 23%
in Phimai.
Enhancements driven by biomass burning during the dry
season (January–April) are well reflected in the simula-
tions. During the wet season, MAX-DOAS, TROPOMI,
and standard CHASER HCHO columns are mostly lower
than 1 ×1016 molec. cm−2. The simulated standard HCHO
peak in March is consistent with the satellite observations,
whereas the MAX-DOAS observations show a peak dur-
ing February. During the dry seasons of 2015 and 2016,
the HCHO peak was observed in March (e.g., Hoque et al.,
2018a). Consequently, such a shift in the HCHO peak might
be related to fire numbers and fire radiative power changes
(Hoque et al., 2022).
The bias between OLNE and MAX-DOAS observations
is 27 %. OLNE estimates agree better with the TROPOMI
observations during the dry season. However, the overall bias
(13 %) between the model and satellite observations is higher
in the case of OLNE simulations.
At Chiba, the simulated HCHO seasonality correlates
strongly with the MAX-DOAS retrievals (R=0.81) and is
negatively biased by ∼31%. The amplitudes of seasonal-
ity inferred from the simulations, MAX-DOAS observations,
and TROPOMI retrievals are, respectively, 59 %, 60 %, and
34 %. The MAX-DOAS, TROPOMI, and CHASER HCHO
columns, respectively, reach their peaks in September, July,
and June. Similar to Phimai, the HCHO peaks in satellite and
ground-based observations differ. One reason might be the
differences in spatial representativity. The TROPOMI data
used for comparison are spatially averaged over 200km cen-
tered on the Chiba site, whereas the spatial representativity of
the MAX-DOAS is approx. 10 km. Moreover, MAX-DOAS
observations are most sensitive to altitudes near the surface,
whereas satellite sensitivity decreases near the surface. Con-
sequently, the air masses sampled by the instruments at the
same local time might differ, leading to inconsistent observa-
tion peaks.
At Kasuga, the simulated HCHO levels are strongly cor-
related with the TROPOMI observations (R=0.75) and
are negatively biased by 35%. Although the correlation be-
tween the model and MAX-DOAS retrievals is moderate, the
bias between CHASER and MAX-DOAS retrievals is 14 %.
Therefore, CHASER shows better agreement with MAX-
DOAS than with TROPOMI. MAX-DOAS observations ex-
hibit seasonality similar to that at Chiba, with the peak
HCHO column occurring during August. Similar to Chiba,
the satellite-observed and CHASER peaks are observed dur-
ing July and June, respectively. The Chiba and Kasuga sites
are located near the ocean and exhibit similar HCHO vari-
ability, which has been captured well in the simulations.
Although the bias between OLNE and standard simula-
tions for Chiba and Kasuga is ∼4 %, the absolute difference
is ∼1×1015 molec. cm−2. NOxemissions in Japan have not
changed markedly since 2005 (Miyazaki et al., 2017). Differ-
ences between the simulations are observed during the sum-
mer, when isoprene emissions are expected to peak (Hoque
et al., 2018a). Because the OH estimates over Japan are sim-
ilar for both simulations (Fig. 5d), the differences are likely
related to the interaction between the isoprene and NOxin-
ventories.
3.6.2 Diurnal and daily variations
Figure 8 compares the observed and simulated daily and diur-
nal variations in the surface HCHO vmr. The error bars rep-
resent the 1σstandard deviation of the observed mean val-
ues. The daily variation comparison entails only the standard
simulations.
In Phimai, the daily datasets correlate well, with an R
value of 0.67. The slope of the fitted line is 0.35. The ob-
served and simulated daily mean HCHO vmr is ∼4 ppbv.
CHASER daily mean values are negatively biased by 19 %
and 11 %, respectively, during the dry and wet seasons. The
standard diurnal variations at Phimai are also well corre-
lated with the observations (R=0.64). The simulated val-
ues lie within the standard deviation of the observations.
HCHO mixing ratios show a peak (∼6 ppbv) at 08:00 LT in
both datasets. The noontime (12:00 LT) vmr is approximately
4 ppbv, and hourly HCHO levels vary between 2 and 6ppbv.
The OLNE diurnal values are 20 % higher than the standard
Geosci. Model Dev., 17, 5545–5571, 2024 https://doi.org/10.5194/gmd-17-5545-2024
H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations 5561
Figure 7. Seasonal variations of HCHO (×1016 molec. cm−2) columns inferred from satellite retrievals (red), model simulations (blue
and black), and ground-based MAX-DOAS observations (green) in Phimai (Thailand), Chiba (Japan), and Kasuga (Japan). MAX-DOAS
observations and CHASER simulations during 12:00–15:00 LT were selected for comparison. Dates common to the datasets are used to
calculate the monthly mean statistics. The blue and black curves, respectively, signify the standard and OLNE simulations. TROPOMI AKs
have been applied to both simulations. The simulation settings are provided in Table 1.
values. However, the mean absolute difference between the
two simulations is 1 ppbv.
The standard simulation reproduced the observed diur-
nal variations at Chiba with a temporal correlation of 0.79,
higher than at Phimai. Both simulations are biased by 10 %
on the low side compared to the observations. No distinc-
tive peak is observed in the diurnal variations. The increas-
ing daytime HCHO levels in Chiba are well reflected in the
model runs. The simulated daily mean values in Chiba are
negatively biased by 18%, with a temporal correlation of
0.40. The slope of the line fitted to the daily mean concen-
trations is 0.27, which is lower than at Phimai, suggesting
greater underestimation, similar to the total columns (Fig. 7).
In Kasuga, modeled daily variations correlate moderately
(R=0.41) with the observations. The effect of the NOxin-
ventories on the simulated diurnal variations in Kasuga is not
significant. The simulated daily mean values are negatively
biased by 20 %, and the slope of the fit is 0.29. Although
Chiba and Kasuga are similar sites, their observed diurnal
variations are slightly different. However, the simulated val-
ues in both cases agree with the observed standard deviation.
In addition, CHASER HCHO columns are also compared
with MAX-DOAS observations reported in the literature in
Fig. 9. The observed values are obtained from Oomen et
al. (2024). The observed mean values represent the aver-
ages of MAX-DOAS observations made between 12:00 and
15:00 LT from May to September 2019. A similar tempo-
ral filter was applied to the CHASER simulations for 2019.
The coincident TROPOMI HCHO columns are also plotted.
TROPOMI AKs are applied to the CHASER values. The er-
ror bars signify the 1σstandard deviation of the TROPOMI
mean values.
Like the Asian sites, CHASER underestimates the HCHO
columns at the European sites. All three datasets mostly
agree within the 1σvariability range of the satellite obser-
vations. The CHASER and TROPOMI HCHO columns are
lower than the MAX-DOAS observations except in Athens.
CHASER shows better agreement with the MAX-DOAS ob-
servations in Athens. De Smedt et al. (2021) reported the bi-
ases between TROPOMI and MAX-DOAS observations at
these sites, estimated from a daily timescale. As the simu-
lated HCHO magnitude is consistent with the TROPOMI val-
ues, biases between the CHASER and MAX-DOAS HCHO
columns at these sites will likely be equivalent.
3.7 Comparison with ATom-4 flight observations
Comparisons between simulated and observed HCHO and
isoprene profiles along the ATom-4 flight path (Fig. S2) are
depicted in Fig. 10a and c. Only the coincident dates have
been included in the comparison.
https://doi.org/10.5194/gmd-17-5545-2024 Geosci. Model Dev., 17, 5545–5571, 2024
5562 H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations
Figure 8. (a, c, e) Scatter plots showing the correlation between the daily mean observed (MAX-DOAS) and simulated HCHO surface
mixing ratios at the three sites. The standard simulations are used in the scatter plots. The linear fitted lines are shown in red. (b, d, f) Diurnal
variations in the HCHO mixing ratios at the three sites are inferred from the MAX-DOAS observations and standard (blue) and OLNE
(green) simulations. The error bars represent the 1σstandard deviation of the mean values estimated from the observations. Observations
and simulations at coincident dates and times (local) are selected for comparison.
Figure 9. Scatter plot comparing CHASER (red), MAX-
DOAS (green), and TROPOMI (red) HCHO columns
(×1016 molec. cm−2) at a few European sites. The MAX-DOAS
observed values are taken from the work of Oomen et al. (2024).
These values represent the mean HCHO column from May to
September in 2019. Observations made from 12:00–15:00 LT were
used to calculate the mean values. Using a similar temporal filter,
the modeled mean values were calculated from the simulations
for 2019. TROPOMI data for 2019 were filtered as described in
Sect. 2.2. The error bars signify the 1σstandard deviation of the
TROPOMI mean HCHO columns.
The simulated HCHO and isoprene profiles agree well
with the observations, with an Rvalue of 0.95. Above and
below 4 km, CHASER HCHO profiles are positively biased
by 29 % and 62 %, respectively, compared to ATom-4 HCHO
levels. The absolute difference in the isoprene profiles around
1 km is 14 pptv, which strongly correlates with the differ-
ence in the HCHO profile below 2 km. This finding signifies
that overestimated CHASER isoprene mixing ratios induce a
positive bias in the HCHO estimates. Despite non-significant
isoprene mixing ratios at altitudes greater than 2 km, both
datasets show considerable HCHO levels above 2 km. Zhao
et al. (2022) reported a similar finding and attributed the
HCHO mixing ratios above 2 km to enhanced CH4oxidation.
At higher altitudes, HCHO is produced through the CH3O2
(methyl peroxy radical) +CH3O2pathway initiated by CH4
oxidation (i.e., CH4+OH). HCHO production through this
pathway is considered in CHASER. Therefore, despite the
differences in magnitude, CHASER has shown good skills in
reproducing the VOC profiles.
The potential reason for the higher simulated HCHO val-
ues below 2 km could be CHASER’s overestimation of the
HCHO mixing ratios over South America, mainly the Ama-
zon (Fig. 2c). Figure 10e and f depict the observed and sim-
ulated HCHO profiles over the Amazon (10–40° W, 10° S–
Geosci. Model Dev., 17, 5545–5571, 2024 https://doi.org/10.5194/gmd-17-5545-2024
H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations 5563
Figure 10. The top panel shows comparisons between ATom-observed (red) and CHASER-simulated (blue) HCHO (a) and isoprene (c)
profiles along the ATom-4 flight path in 2018. The ATom-4 flight path is depicted in Fig. S2. Standard simulations are used for comparison.
Simulations for the times of the ATom observations were selected. Both datasets were averaged using 0.3km bins. The relative differ-
ences between the observed and simulated (c) HCHO and (d) isoprene profiles are also shown. In the bottom panel, Atom-4-observed and
CHASER-simulated HCHO profiles over (e) Amazonia and (f) the remote Pacific region are compared. Amazonia (10–40°W, 10°S–10° N)
represents a densely vegetated region, whereas the remote Pacific region (160–180°W, 20° S–20°N) represents the background HCHO con-
ditions. The units of the HCHO and isoprene mixing ratios are, respectively, parts per billion by volume (ppbv) and parts per trillion by
volume (pptv).
10° N) and the remote Pacific region (160–180° W, 20° S–
20° N), respectively. The HCHO profiles over the remote
Pacific region represent the background HCHO mixing ra-
tio. The CHASER and ATom background HCHO mixing
ratios within the boundary layer are 0.4 and 0.3 ppbv, re-
spectively. The mean relative differences between the two
datasets within the boundary layer over Amazonia and the
remote Pacific region are ∼60 % and ∼22 %, respectively,
indicating that the uncertainty in the contributions from
the isoprene emissions to the total HCHO uncertainties is
higher. Above 5 km, CHASER underestimates the back-
ground HCHO mixing ratios. However, the simulated and
TROPOMI HCHO columns over the remote Pacific region
showed consistency when gridded over the same horizon-
tal grid (Fig. 1). Consequently, differences in the horizontal
resolution can cause discrepancies between the simulations
and ATom observations over the remote region. Over South
America, the model overestimates the observed (TROPOMI
and ATom) HCHO abundances, irrespective of the horizon-
tal resolution. Therefore, the biogenic emission estimates for
South America in CHASER should be reviewed to reduce the
model–observation biases.
3.8 Contribution estimates
The contributions of different VOC emission sources to the
regional HCHO abundances are presented in Fig. 11. The
contribution estimates are presented in Table 8. A stacked-
bar plot of the annual contributions of the emission sources
is portrayed in Fig. S11.
Over E-China (Fig. 11a), biomass burning has a non-
significant effect on the regional HCHO columns. During
summer, the biogenic and anthropogenic VOC emission con-
tributions are 44 % and 17 %, respectively. In contrast, the
anthropogenic and biogenic contributions to the regional
HCHO level during winter are 35 % and 13 %, respectively.
Non-significant biomass burning effects on the HCHO
columns can be observed over E-USA (Fig. 11b), W-USA
(Fig. 11c), and Europe (Fig. 11d). Biogenic emissions con-
tribute more than 20 % (35 % in E-USA) in these regions.
In these regions, the annual anthropogenic contributions are
higher than the biogenic contribution. Although the simu-
lated winter columns in these regions are consistent with
TROPOMI (Fig. 2), the model values are lower during sum-
mer and autumn. Moreover, the sensitivity results show a
non-significant biogenic contribution during winter and au-
https://doi.org/10.5194/gmd-17-5545-2024 Geosci. Model Dev., 17, 5545–5571, 2024
5564 H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations
tumn, which likely reduces the annual biogenic contribution
estimates.
In C-Africa (Fig. 11e), biogenic emissions (48 %) are
the most significant contributor, followed by anthropogenic
emissions (13 %). Although the biogenic emission contribu-
tions in N-Africa (Fig. 11f; 48 %) and S-Africa (Fig. 11b;
43 %) are equivalent, the pyrogenic contributions are twice
as high in the latter region. Consequently, despite similar
HCHO abundances and modulations in these regions, the
source contributions differ.
Biogenic emissions over South America (Fig. 11h) con-
tribute 61 % of the regional HCHO abundances. The py-
rogenic contribution during the biomass burning period is
12 %, whereas the annual contribution is 7 %.
In SE-Asia (Fig. 11m), the annual anthropogenic contribu-
tions are ∼20 %. During the dry season, the anthropogenic,
pyrogenic, and biogenic contributions are 7 %, 12 %, and
48 %, respectively. Biogenic production comprises 43 % of
the HCHO columns from July to December, whereas anthro-
pogenic emissions account for 9 %.
In India (Fig. 11i), annual pyrogenic emissions contribute
∼2 % of the HCHO levels. A similar source contribution to
the HCHO levels in the IGP (Fig. 11j) is also observed. The
model’s capability to reproduce the observed HCHO season-
ality in India and the IGP region was limited. Consequently,
robust source contribution estimates for these regions cannot
be derived from the current analysis.
Over E-India (Fig. 11k), 44 % of the HCHO levels origi-
nate from biogenic sources, followed by anthropogenic VOC
emissions (36 %). Similar source contributions of biogenic
(30 %) and anthropogenic (29 %) emissions are observed in
S-India (Fig. 11l). Over both regions, the pyrogenic source
contribution is ∼2 %.
3.9 Uncertainties in the chemical mechanism
Uncertainties in the chemical mechanisms affect the HCHO
simulations. The representation of isoprene chemistry can
vary among the gas-phase chemistry mechanisms used in the
CTMs. The most commonly used isoprene schemes underes-
timate the observed HCHO by at least 15 % (Marvin et al.,
2017). Such underestimations are also strongly linked with
the errors in the NOxemission inventories (Anderson et al.,
2017). In addition, potential errors in the acetaldehyde emis-
sions and chemistry can also lead to the underestimation of
the HCHO vmr by up to 75 pptv in the lower troposphere
(Anderson et al., 2017).
4 Conclusions
CHASER-simulated global HCHO spatiotemporal distribu-
tions at a horizontal resolution of 2.8° ×2.8° were evaluated
against multi-platform observations. First, 2 years of sim-
ulation results (2019–2020) were compared with the latest
HCHO satellite observations from TROPOMI. The model–
satellite agreement was excellent, with a global rvalue of
0.93 and an RMSE of 0.75 ×1015 molec. cm−2. The model
showed a good capability to reproduce the HCHO columns in
hotspot and background regions. CHASER HCHO columns
over large forested areas showed good consistency with the
observations, demonstrating that the biogenic emission es-
timates in the model are reasonable. The simulated HCHO
seasonality in a few selected regions was consistent with
the observations. The model was able to reproduce the ob-
served wintertime HCHO columns in E-USA, W-USA, and
Europe, in addition to summer peaks. Disagreement between
TROPOMI and CHASER was observed primarily in India,
China, Amazonia, and SE-Asia. Uncertainties in background
HCHO columns, anthropogenic-VOC-emission inventories,
the chemical mechanisms adopted in the model, and retrieval
algorithms were the potential contributors to these discrepan-
cies. However, those uncertainties did not affect the model–
satellite agreement in Africa and South America. A com-
parison between OMI, TROPOMI, and CHASER HCHO
columns demonstrated that the effect of TROPOMI’s im-
proved spatial resolution was limited globally. However, in
most regions, the simulated HCHO seasonality showed bet-
ter agreement with TROPOMI than with OMI, reducing the
RMSE by up to 63 %. TROPOMI retrievals were, on average,
30 % lower than those of OMI.
Second, CHASER simulations were compared with
2 years of MAX-DOAS observations of HCHO at Phimai,
Chiba, and Kasuga. Daily CHASER HCHO mixing ratios
showed consistency with the observations at the three sites,
with Rvalues of 0.39–0.67. The slopes from linear fitting
were lower for Chiba (0.29) and Kasuga (0.29) than for Phi-
mai (0.37), implying that there is less underestimation by the
model for the latter site. The diurnal variations at the sites
were consistent with the observations. Changing the NOx
emission inventories did not affect the simulated diurnal vari-
ations.
Third, simulated HCHO and isoprene profiles for 2018
were compared with ATom-4 flight observations. Despite
consistent profile shapes, the model overestimated VOC mix-
ing ratios (mainly within the PBL). Uncertainties related to
VOC emission inventories, background HCHO levels, and
the model resolution were potential reasons for the model–
flight discrepancies.
Lastly, sensitivity studies were conducted to estimate the
contributions of the different emissions sources to the to-
tal HCHO columns in different regions. Biogenic emissions
were the most significant contributor in most of the regions.
In a few cases, biogenic and anthropogenic emission contri-
butions were equivalent. In some regions, only summertime
biogenic estimates were found to be reasonable.
Geosci. Model Dev., 17, 5545–5571, 2024 https://doi.org/10.5194/gmd-17-5545-2024
H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations 5565
Figure 11. Seasonal variation of HCHO (×1016 molec. cm−2) inferred from different simulations. The settings of the standard simulation
are presented in Table 1. The model estimates shown in red, green, and blue are simulated by switching off the biomass burning, biogenic,
and anthropogenic emissions. The satellite AKs have been applied to all the simulations. The coordinate bounds of the regions are similar to
those in Fig. 2.
Table 7. Contributions (%) of different emission sources to HCHO abundances in selected regions. The respective emissions were switched
off to estimate the contribution to the total HCHO abundances. The contributions have been calculated with respect to the standard simula-
tions. The satellite AKs were applied to all simulations.
Region Biomass burning
contribution
Biogenic
contribution
Anthropogenic
contribution
E-China
E-USA
W-USA
Europe
C-Africa
N-Africa
S-Africa
S-America
India
IGP
E-India
S-India
SE-Asia
1.4 %
1.7 %
1.8 %
1.2 %
8 %
6 %
15 %
7 %
1.4 %
1.1 %
1.5 %
2.1 %
6 %
32 %
35 %
23 %
20 %
48 %
48 %
43 %
61 %
37 %
39 %
44 %
30 %
45 %
37 %
38 %
39 %
45 %
13 %
17 %
12 %
10 %
34 %
37 %
36 %
29 %
24 %
https://doi.org/10.5194/gmd-17-5545-2024 Geosci. Model Dev., 17, 5545–5571, 2024
5566 H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations
Code availability. The CHASER source code needed to reproduce
the simulations in this work is available from the repository at
https://doi.org/10.5281/zenodo.10892945 (Sudo, 2024).
Data availability. The processed model output and observational
datasets needed to reproduce the results are available from the
repository at https://doi.org/10.5281/zenodo.10052384 (Hoque et
al., 2024). The MAX-DOAS profile and column data provided by
Hitoshi Irie can be accessed from the repository (i.e., Hoque et
al., 2024). TROPOMI (https://scihub.copernicus.eu/dhus/#/home,
last access: 1 July 2023; De Smedt et al., 2021), OMI-BIRA-
product (https://www.temis.nl/qa4ecv/hcho/hcho_omi.php, last ac-
cess: 1 July 2023; De Smedt et al., 2021), and ATom (https://daac.
ornl.gov/ATOM/guides/ATom_nav.html, last access: 1 July 2023;
https://doi.org/10.3334/ORNLDAAC/1581, Wofsy et al., 2018)
data were obtained from the respective websites.
Author contributions. HMSH conceptualized the study, conducted
the model simulations, analyzed the datasets, and drafted the
manuscript. YH helped with the data processing. HI developed the
JM2 code and maintained the A-SKY network. KS developed the
CHASER model and supervised the study. MFK used his exper-
tise to explain the results. All the authors commented and provided
feedback on the final results and manuscript.
Competing interests. The contact author has declared that none of
the authors has any competing interests.
Disclaimer. Publisher’s note: Copernicus Publications remains
neutral with regard to jurisdictional claims made in the text, pub-
lished maps, institutional affiliations, or any other geographical rep-
resentation in this paper. While Copernicus Publications makes ev-
ery effort to include appropriate place names, the final responsibility
lies with the authors.
Acknowledgements. We are grateful to the TROPOMI, OMI, and
ATom scientific teams for making the respective observational
datasets available for public usage. The CHASER model simula-
tions were partly performed with the supercomputer (NEC SX-
Aurora TSUBASA) at the National Institute for Environmental
Studies (NIES), Tsukuba, Japan. The corresponding author ac-
knowledges the valuable advice of Kazuyakai Miyazaki (Jet Propul-
sion Lab., NASA) and Takashi Sekiya (JAMSTEC, Japan).
Financial support. This research has been supported by the Min-
istry of the Environment, Government of Japan (grant nos.
Global Research Fund S-12 and Global Research Fund S-
20), the Japan Society for the Promotion of Science (grant
nos. JP20H04320, JP19H05669, JP19HO4235, JP23H04971,
JP21K12227, JP22H03727, and JP22H05004), the Environ-
mental Restoration and Conservation Agency (grant no. JP-
MEERF20215005), and the Japan Aerospace Exploration Agency
(grant no. 19RT000351).
Review statement. This paper was edited by Jason Williams and re-
viewed by Narendra Ojha and three anonymous referees.
References
Anderson, D. C., Nicely, J. M., Wolfe, G. M., Hanisco, T. F., Salaw-
itch, R. J., Canty, T. P.,Dickerson, R. R., Apel, E. C., Baidar, S.,
Bannan, T. J., Blake, N. J., Chen, D., Dix, B., Fernandez, R. P.,
Hall, S. R., Hornbrook, R. S., Huey, L. G., Josse, B., Jockel, P.,
Kinnison, D. E., Koenig, T. K., Le Breton, M., Marecal, V., Mor-
genstern, O., Oman, L. D., Pan, L. L., Percival, C., Plummer,
D., Revell, L. E., Rozanov, E., Saiz-Lopez, A., Stenke, A., Sudo,
K.,Tilmes, S., Ullman, K., Volkamer, R., Weinheimer, A. J., and
Zeng, G.: Formaldehyde in the tropical western Pacific: Chem-
ical sources and sinks, convective transport, and representation
in CAM-Chem and the CCMI models, J. Geophys. Res., 122,
201–211, https://doi.org/10.1002/2016JD026121, 2017.
Apel, E. C., Asher, E. C., Hills, A. J., and Hornbrook, R. S.: ATom:
Volatile Organic Compounds (VOCs) from the TOGA instru-
ment, Version 2, ORNL DAAC [data set], Oak Ridge, Tennessee,
USA, https://doi.org/10.3334/ORNLDAAC/1936, 2021.
Arlander, D., Brüning, D., Schmidt, U., and Ehhalt, D.: The tropo-
spheric distribution of formaldehyde during TROPOZ II, J. At-
mos. Chem., 22, 251–269, https://doi.org/10.1007/BF00696637,
1995.
Bauwens, M., Verreyken, B., Stavrakou, T., Müller, J., and De
Smedt, I.: Spaceborne evidence for significant anthropogenic
VOC trends in Asian cities over 2005–2019, Environ. Res. Lett.,
17, 015008, https://doi.org/10.1088/1748-9326/ac46eb/, 2022.
Boersma, K. F., Eskes, H. J., and Brinksma, E. J. : Error analysis for
tropospheric NO2retrieval from space. J. Geophys. Res., 109,
D04311, https://doi.org/10.1029/2003JD003962, 2004.
Boersma, K. F., Vinken, G. C. M., and Eskes, H. J.:
Representativeness errors in comparing chemistry transport
and chemistry climate models with satellite UV–Vis tropo-
spheric column retrievals, Geosci. Model Dev., 9, 875–898,
https://doi.org/10.5194/gmd-9-875-2016, 2016.
Burkert, J., Andrés-Hernández, M. D., Stöbener, D., Burrows,
J. P., Weissenmayer, M., and Kraus, A.: Peroxy radical
and related trace gas measurements in the boundary layer
above the Atlantic Ocean, J. Geophys. Res., 106, 5457–5477,
https://doi.org/10.1029/2000JD900613, 2001.
Burrows, J. P., Weber, M., Buchwitz, M., Rozanov, V.,
Ladstätter-Weißenmayer, A., Richter, A., DeBeek, R.,
Hoogen, R., Bramstedt, K., Eichmann, K.-U., Eisinger, M.,
and Perner, D.: The Global Ozone Monitoring Experiment
(GOME): Mission Concept and First Scientific Results,
J. Atmos. Sci., 56, 151–175, https://doi.org/10.1175/1520-
0469(1999)056<0151:TGOMEG>2.0.CO;2, 1999.
Caballero, C. B., Ruhoff, A., and Biggs, T.: Land use and land cover
changes and their impacts on surface-atmosphere interactions in
Brazil: A systematic review, Sci. Total Environ., 808, 152134,
https://doi.org/10.1016/j.scitotenv.2021.152134, 2022.
Geosci. Model Dev., 17, 5545–5571, 2024 https://doi.org/10.5194/gmd-17-5545-2024
H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations 5567
Cazorla, M., Wolfe, G. M., Bailey, S. A., Swanson, A. K., Arkinson,
H. L., and Hanisco, T. F.: A new airborne laser-induced fluores-
cence instrument for in situ detection of formaldehyde through-
out the troposphere and lower stratosphere, Atmos. Meas. Tech.,
8, 541–552, https://doi.org/10.5194/amt-8-541-2015, 2015.
Chan, K. L., Wiegner, M., van Geffen, J., De Smedt, I., Alberti,
C., Cheng, Z., Ye, S., and Wenig, M.: MAX-DOAS measure-
ments of tropospheric NO2and HCHO in Munich and the
comparison to OMI and TROPOMI satellite observations, At-
mos. Meas. Tech., 13, 4499–4520, https://doi.org/10.5194/amt-
13-4499-2020, 2020.
Chutia, L., Ojha, N., Girach, I. A., Sahu, L. K., Alvarado, M. A.
L., Burrows, J. P., Pathak, B., and Bhuyan, P. K.: Distribution of
volatile organic compounds over Indian subcontinent during win-
ter: WRF-chem simulation versus observations, Environ. Pol.,
252, 256–269, https://doi.org/10.1016/j.envpol.2019.05.097,
2019.
Colella, P. and Woodward, P. R.: The piecewise parabolic method
(PPM) for gas-dynamical simulations, J. Comput. Phys., 54,
174–201, https://doi.org/10.1016/0021-9991(84)90143-8, 1984.
Crippa, M., Guizzardi, D., Butler, T., Keating, T., Wu, R., Kamin-
ski, J., Kuenen, J., Kurokawa, J., Chatani, S., Morikawa, T.,
Pouliot, G., Racine, J., Moran, M. D., Klimont, Z., Manseau, P.
M., Mashayekhi, R., Henderson, B. H., Smith, S. J., Suchyta, H.,
Muntean, M., Solazzo, E., Banja, M., Schaaf, E., Pagani, F., Woo,
J.-H., Kim, J., Monforti-Ferrario, F., Pisoni, E., Zhang, J., Niemi,
D., Sassi, M., Ansari, T., and Foley, K.: The HTAP_v3 emission
mosaic: merging regional and global monthly emissions (2000–
2018) to support air quality modelling and policies, Earth Syst.
Sci. Data, 15, 2667–2694, https://doi.org/10.5194/essd-15-2667-
2023, 2023.
De Smedt, I., Müller, J.-F., Stavrakou, T., van der A, R., Eskes,
H., and Van Roozendael, M.: Twelve years of global obser-
vations of formaldehyde in the troposphere using GOME and
SCIAMACHY sensors, Atmos. Chem. Phys., 8, 4947–4963,
https://doi.org/10.5194/acp-8-4947-2008, 2008.
De Smedt, I., Stavrakou, T., Müller, J.-F., van der A, R. J., and
Van Roozendael, M.: Trend detection in satellite observations
of formaldehyde tropospheric columns, Geophys. Res. Lett., 37,
L18808, https://doi.org/10.1029/2010GL044245, 2010.
De Smedt, I., Stavrakou, T., Hendrick, F., Danckaert, T., Vlem-
mix, T., Pinardi, G., Theys, N., Lerot, C., Gielen, C., Vigouroux,
C., Hermans, C., Fayt, C., Veefkind, P., Müller, J.-F., and Van
Roozendael, M.: Diurnal, seasonal and long-term variations of
global formaldehyde columns inferred from combined OMI and
GOME-2 observations, Atmos. Chem. Phys., 15, 12519–12545,
https://doi.org/10.5194/acp-15-12519-2015, 2015.
De Smedt, I., Yu, H., Richter, A., Beirle, S., Eskes, H., Boersma, K.
F., Van Roozendael, M., Van Geffen, J., Lorente, A., and Peters,
E.: QA4ECV HCHO tropospheric column data from OMI (Ver-
sion 1.1), Royal Belgian Institute for Space Aeronomy [data set],
https://doi.org/10.18758/71021031, 2017.
De Smedt, I., Theys, N., Yu, H., Danckaert, T., Lerot, C., Comper-
nolle, S., Van Roozendael, M., Richter, A., Hilboll, A., Peters,
E., Pedergnana, M., Loyola, D., Beirle, S., Wagner, T., Eskes, H.,
van Geffen, J., Boersma, K. F., and Veefkind, P.: Algorithm theo-
retical baseline for formaldehyde retrievals from S5P TROPOMI
and from the QA4ECV project, Atmos. Meas. Tech., 11, 2395–
2426, https://doi.org/10.5194/amt-11-2395-2018, 2018.
De Smedt, I., Pinardi, G., Vigouroux, C., Compernolle, S., Bais,
A., Benavent, N., Boersma, F., Chan, K.-L., Donner, S., Eich-
mann, K.-U., Hedelt, P., Hendrick, F., Irie, H., Kumar, V., Lam-
bert, J.-C., Langerock, B., Lerot, C., Liu, C., Loyola, D., Piters,
A., Richter, A., Rivera Cárdenas, C., Romahn, F., Ryan, R.
G., Sinha, V., Theys, N., Vlietinck, J., Wagner, T., Wang, T.,
Yu, H., and Van Roozendael, M.: Comparative assessment of
TROPOMI and OMI formaldehyde observations and validation
against MAX-DOAS network column measurements, Atmos.
Chem. Phys., 21, 12561–12593, https://doi.org/10.5194/acp-21-
12561-2021, 2021.
Duncan, B. N., Yoshida, Y., Olson, J. R., Sillman, S., Martin,
R. V., Lamsal, L., Hu, Y., Pickering, K. E., Retscher, C.,
Allen, D. J., and Crawford, J. H.: Application of OMI obser-
vations to a space-based indicator of NOxand VOC controls
on surface ozone formation, Atmos. Environ., 44, 2213–2223,
https://doi.org/10.1016/j.atmosenv.2010.03.010, 2010.
Emori, S., Nozawa, T., Numaguti, A., and Uno, I.: Impor-
tance of cumulus parameterization for precipitation simulation
over East Asia in June, J. Meteorol. Soc. Jpn., 79, 939–947,
https://doi.org/10.2151/jmsj.79.939, 2021.
Franco, B., Hendrick, F., Van Roozendael, M., Müller, J.-F.,
Stavrakou, T., Marais, E. A., Bovy, B., Bader, W., Fayt, C., Her-
mans, C., Lejeune, B., Pinardi, G., Servais, C., and Mahieu,
E.: Retrievals of formaldehyde from ground-based FTIR and
MAX-DOAS observations at the Jungfraujoch station and com-
parisons with GEOS-Chem and IMAGES model simulations, At-
mos. Meas. Tech., 8, 1733–1756, https://doi.org/10.5194/amt-8-
1733-2015, 2015.
Fu, T. M., Jacob, D. J., Wittrock, F., Burrows, J. P., Vrekous-
sis, M., and Henze, D. K.: Global budgets of atmospheric
glyoxal and methylglyoxal, and implications for formation of
secondary organic aerosols, J. Geophys. Res., 113, D15303,
https://doi.org/10.1029/2007JD009505, 2008.
González Abad, G., Vasilkov, A., Seftor, C., Liu, X., and Chance,
K.: Smithsonian Astrophysical Observatory Ozone Mapping and
Profiler Suite (SAO OMPS) formaldehyde retrieval, Atmos.
Meas. Tech., 9, 2797–2812, https://doi.org/10.5194/amt-9-2797-
2016, 2016.
Guenther, A.: Seasonal and spatial variations in natural
volatile organic compound emissions, Ecol. Appl., 7, 34–
45, https://doi.org/10.2307/2269405, 1997.
Guenther, A., Karl, T., Harley, P., Wiedinmyer, C., Palmer, P.
I., and Geron, C.: Estimates of global terrestrial isoprene
emissions using MEGAN (Model of Emissions of Gases and
Aerosols from Nature), Atmos. Chem. Phys., 6, 3181–3210,
https://doi.org/10.5194/acp-6-3181-2006, 2006.
Ha, P. T. M., Kanaya, Y., Taketani, F., Andrés Hernández, M. D.,
Schreiner, B., Pfeilsticker, K., and Sudo, K.: Implementation of
HONO into the chemistry–climate model CHASER (V4.0): roles
in tropospheric chemistry, Geosci. Model Dev., 16, 927–960,
https://doi.org/10.5194/gmd-16-927-2023, 2023.
Hak, C., Pundt, I., Trick, S., Kern, C., Platt, U., Dommen, J., Or-
dóñez, C., Prévôt, A. S. H., Junkermann, W., Astorga-Lloréns,
C., Larsen, B. R., Mellqvist, J., Strandberg, A., Yu, Y., Galle, B.,
Kleffmann, J., Lörzer, J. C., Braathen, G. O., and Volkamer, R.:
Intercomparison of four different in-situ techniques for ambient
formaldehyde measurements in urban air, Atmos. Chem. Phys.,
5, 2881–2900, https://doi.org/10.5194/acp-5-2881-2005, 2005.
https://doi.org/10.5194/gmd-17-5545-2024 Geosci. Model Dev., 17, 5545–5571, 2024
5568 H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations
He, Y., Hoque, H. M. S., and Sudo, K.: Introducing new
lightning schemes into the CHASER (MIROC) chemistry–
climate model, Geosci. Model Dev., 15, 5627–5650,
https://doi.org/10.5194/gmd-15-5627-2022, 2022.
Hoque, H. M. S., Irie, H., and Damiani, A.: First MAX-
DOAS Observations of Formaldehyde and Glyoxal in
Phimai, Thailand, J. Geophys. Res., 123, 9957–9975,
https://doi.org/10.1029/2018JD028480, 2018a.
Hoque, H. M. S., Irie, H., Damiani, A., Rawat, P., and Naja, M.:
First simultaneous observations of formaldehyde and glyoxal by
MAX-DOAS in the Indo-Gangetic Plain region, Sola, 14, 159–
164, https://doi.org/10.2151/sola.2018-028, 2018b.
Hoque, H. M. S., Sudo, K., Irie, H., Damiani, A., Naja, M., and
Fatmi, A. M.: Multi-axis differential optical absorption spec-
troscopy (MAX-DOAS) observations of formaldehyde and ni-
trogen dioxide at three sites in Asia and comparison with
the global chemistry transport model CHASER, Atmos. Chem.
Phys., 22, 12559–12589, https://doi.org/10.5194/acp-22-12559-
2022, 2022.
Hoque, H. M. S., Sudo, K., and Irie, H.: Model and observa-
tional datasets used for evaluating CHASER simulated formalde-
hyde (HCHO) abundances in 2019 and 2020, Zenodo [data set],
https://doi.org/10.5281/zenodo.10052384, 2024.
Inness, A., Baier, F., Benedetti, A., Bouarar, I., Chabrillat, S., Clark,
H., Clerbaux, C., Coheur, P., Engelen, R. J., Errera, Q., Flem-
ming, J., George, M., Granier, C., Hadji-Lazaro, J., Huijnen,
V., Hurtmans, D., Jones, L., Kaiser, J. W., Kapsomenakis, J.,
Lefever, K., Leitão, J., Razinger, M., Richter, A., Schultz, M. G.,
Simmons, A. J., Suttie, M., Stein, O., Thépaut, J.-N., Thouret, V.,
Vrekoussis, M., Zerefos, C., and the MACC team: The MACC
reanalysis: an 8 yr data set of atmospheric composition, At-
mos. Chem. Phys., 13, 4073–4109, https://doi.org/10.5194/acp-
13-4073-2013, 2013.
Irie, H.: International air quality and sky research remote sens-
ing network (A-SKY): Its development and satellite atmosphere
product validation, J. Remote Sens. Soc. Jpn., 41, 575–581,
https://doi.org/10.11440/rssj.41.575, 2021.
Irie, H., Kanaya, Y., Akimoto, H., Iwabuchi, H., Shimizu, A.,
and Aoki, K.: First retrieval of tropospheric aerosol profiles
using MAX-DOAS and comparison with lidar and sky ra-
diometer measurements, Atmos. Chem. Phys., 8, 341–350,
https://doi.org/10.5194/acp-8-341-2008, 2008.
Irie, H., Takashima, H., Kanaya, Y., Boersma, K. F., Gast,
L., Wittrock, F., Brunner, D., Zhou, Y., and Van Roozen-
dael, M.: Eight-component retrievals from ground-based MAX-
DOAS observations, Atmos. Meas. Tech., 4, 1027–1044,
https://doi.org/10.5194/amt-4-1027-2011, 2011.
Irie, H., Nakayama, T., Shimizu, A., Yamazaki, A., Nagai,
T., Uchiyama, A., Zaizen, Y., Kagamitani, S., and Matsumi,
Y.: Evaluation of MAX-DOAS aerosol retrievals by coin-
cident observations using CRDS, lidar, and sky radiome-
ter in Tsukuba, Japan, Atmos. Meas. Tech., 8, 2775–2788,
https://doi.org/10.5194/amt-8-2775-2015, 2015.
Ito, A. and Inatomi, M.: Use of a process-based model for as-
sessing the methane budgets of global terrestrial ecosystems
and evaluation of uncertainty, Biogeosciences, 9, 759–773,
https://doi.org/10.5194/bg-9-759-2012, 2012.
Jenkin, M. E., Young, J. C., and Rickard, A. R.: The MCM
v3.3.1 degradation scheme for isoprene, Atmos. Chem. Phys.,
15, 11433–11459, https://doi.org/10.5194/acp-15-11433-2015,
2015.
Kaiser, J., Wolfe, G. M., Bohn, B., Broch, S., Fuchs, H., Ganzeveld,
L. N., Gomm, S., Häseler, R., Hofzumahaus, A., Holland, F.,
Jäger, J., Li, X., Lohse, I., Lu, K., Prévôt, A. S. H., Rohrer,
F., Wegener, R., Wolf, R., Mentel, T. F., Kiendler-Scharr, A.,
Wahner, A., and Keutsch, F. N.: Evidence for an unidentified
non-photochemical ground-level source of formaldehyde in the
Po Valley with potential implications for ozone production, At-
mos. Chem. Phys., 15, 1289–1298, https://doi.org/10.5194/acp-
15-1289-2015, 2015.
Khan, M. F., Latif, M. T., Saw, W. H., Amil, N., Nadzir, M. S. M.,
Sahani, M., Tahir, N. M., and Chung, J. X.: Fine particulate mat-
ter in the tropical environment: monsoonal effects, source appor-
tionment, and health risk assessment, Atmos. Chem. Phys., 16,
597–617, https://doi.org/10.5194/acp-16-597-2016, 2016.
Kleipool, Q. L., Dobber, M. R., de Haan, J. F., and Lev-
elt, P. F.: Earth surface reflectance climatology from
3 years of OMI data, J. Geophys. Res., 113, D18308,
https://doi.org/10.1029/2008JD010290, 2008.
Kreher, K., Van Roozendael, M., Hendrick, F., Apituley, A., Dim-
itropoulou, E., Frieß, U., Richter, A., Wagner, T., Lampel, J.,
Abuhassan, N., Ang, L., Anguas, M., Bais, A., Benavent, N.,
Bösch, T., Bognar, K., Borovski, A., Bruchkouski, I., Cede, A.,
Chan, K. L., Donner, S., Drosoglou, T., Fayt, C., Finkenzeller, H.,
Garcia-Nieto, D., Gielen, C., Gómez-Martín, L., Hao, N., Henz-
ing, B., Herman, J. R., Hermans, C., Hoque, S., Irie, H., Jin, J.,
Johnston, P., Khayyam Butt, J., Khokhar, F., Koenig, T. K., Kuhn,
J., Kumar, V., Liu, C., Ma, J., Merlaud, A., Mishra, A. K., Müller,
M., Navarro-Comas, M., Ostendorf, M., Pazmino, A., Peters, E.,
Pinardi, G., Pinharanda, M., Piters, A., Platt, U., Postylyakov,
O., Prados-Roman, C., Puentedura, O., Querel, R., Saiz-Lopez,
A., Schönhardt, A., Schreier, S. F., Seyler, A., Sinha, V., Spinei,
E., Strong, K., Tack, F., Tian, X., Tiefengraber, M., Tirpitz, J.-
L., van Gent, J., Volkamer, R., Vrekoussis, M., Wang, S., Wang,
Z., Wenig, M., Wittrock, F., Xie, P. H., Xu, J., Yela, M., Zhang,
C., and Zhao, X.: Intercomparison of NO2, O4, O3and HCHO
slant column measurements by MAX-DOAS and zenith-sky UV–
visible spectrometers during CINDI-2, Atmos. Meas. Tech., 13,
2169–2208, https://doi.org/10.5194/amt-13-2169-2020, 2020.
Kumar, A., Hakkim, H., Ghude, S. D., and Sinha, V.:
Probing wintertime air pollution sources in the Indo-
Gangetic Plain through 52 hydrocarbons measured rarely
at Delhi and Mohali, Sci. Total Environ., 801, 149711,
https://doi.org/10.1016/j.scitotenv.2021.149711, 2021.
Kupc, A., Williamson, C., Wagner, N. L., Richardson, M., and
Brock, C. A.: Modification, calibration, and performance of the
Ultra-High Sensitivity Aerosol Spectrometer for particle size dis-
tribution and volatility measurements during the Atmospheric
Tomography Mission (ATom) airborne campaign, Atmos. Meas.
Tech., 11, 369–383, https://doi.org/10.5194/amt-11-369-2018,
2018.
Kurucz, R. L., Furenlid, I., Brault, J., and Testerman, L.: Solar flux
atlas from 296 to 1300 nm, National Solar Observatory Atlas
No. 1, Sunspot, New Mexico, 1984.
Kuttippurath, J., Abbhishek, K., Gopikrishnan, G. S., and Pathak,
M.: Investigation of long–term trends and major sources of
atmospheric HCHO over India, Environ. Chall., 7, 100477,
https://doi.org/10.1016/j.envc.2022.100477, 2022.
Geosci. Model Dev., 17, 5545–5571, 2024 https://doi.org/10.5194/gmd-17-5545-2024
H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations 5569
Lee, M., Heikes, B. G., Jacob, D. J., Sachse, G., and Anderson, B.:
Hydrogen peroxide, organic hydroperoxide, and formaldehyde
as primary pollutants from biomass burning, J. Geophys. Res.,
102, 1301–1309, https://doi.org/10.1029/96JD01709, 1997.
Levelt, P. F., Joiner, J., Tamminen, J., Veefkind, J. P., Bhartia, P. K.,
Stein Zweers, D. C., Duncan, B. N., Streets, D. G., Eskes, H.,
van der A, R., McLinden, C., Fioletov, V., Carn, S., de Laat, J.,
DeLand, M., Marchenko, S., McPeters, R., Ziemke, J., Fu, D.,
Liu, X., Pickering, K., Apituley, A., González Abad, G., Arola,
A., Boersma, F., Chan Miller, C., Chance, K., de Graaf, M.,
Hakkarainen, J., Hassinen, S., Ialongo, I., Kleipool, Q., Krotkov,
N., Li, C., Lamsal, L., Newman, P., Nowlan, C., Suleiman,
R., Tilstra, L. G., Torres, O., Wang, H., and Wargan, K.: The
Ozone Monitoring Instrument: overview of 14 years in space, At-
mos. Chem. Phys., 18, 5699–5745, https://doi.org/10.5194/acp-
18-5699-2018, 2018.
Levelt, P. F., Stein Zweers, D. C., Aben, I., Bauwens, M., Bors-
dorff, T., De Smedt, I., Eskes, H. J., Lerot, C., Loyola, D.
G., Romahn, F., Stavrakou, T., Theys, N., Van Roozendael,
M., Veefkind, J. P., and Verhoelst, T.: Air quality impacts of
COVID-19 lockdown measures detected from space using high
spatial resolution observations of multiple trace gases from
Sentinel-5P/TROPOMI, Atmos. Chem. Phys., 22, 10319–10351,
https://doi.org/10.5194/acp-22-10319-2022, 2022.
Liu, F., Zhang, Q., van der A, R. J., Zheng, B., Tong,
D., Yan, L., Zheng, Y., and He, K.: Recent reduction in
NOxemissions over China: synthesis of satellite observations
and emission inventories, Environ. Res. Lett., 11, 114002,
https://doi.org/10.1088/1748-9326/11/11/114002, 2016.
Liu, Y., Wang, H., Jing, S., Peng, Y., Gao, Y., Yan, R., Wang,
Q., Lou, S., Cheng, T., and Huang, C.: Strong regional trans-
port of volatile organic compounds (VOCs) during wintertime
in Shanghai megacity of China, Atmos. Environ., 244, 117940,
https://doi.org/10.1016/j.atmosenv.2020.117940, 2021.
Luecken, D. J., Napelenok, S. L., Strum, M., Scheffe, R., and
Phillips, S.: Sensitivity of Ambient Atmospheric Formalde-
hyde and Ozone to Precursor Species and Source Types Across
the United States, Environ. Sci. Technol., 52, 4668–4675,
https://doi.org/10.1021/acs.est.7b05509, 2018.
Marvin, M. R., Wolfe, G. M., Salawitch, R. J., Canty, T. P., Roberts,
S. J., Travis, K. R., Aikin, K. C., de Gouw, J. A., Graus, M.,
Hanisco, T. F., Holloway, J. S., Hübler, G., Kaiser, J., Keutsch,
F. N., Peischl, J., Pollack, I. B., Roberts, J. M., Ryerson, T. B.,
Veres, P. R., and Warneke, C.: Impact of evolving isoprene mech-
anisms on simulated formaldehyde: An inter-comparison sup-
ported by in situ observations from SENEX, Atmos. Environ.,
164, 325–336, https://doi.org/10.1016/j.atmosenv.2017.05.049,
2017.
Martin, R. V., Fiore, A. M., and Van Donkelaar, A.: Space-
based diagnosis of surface ozone sensitivity to anthro-
pogenic emissions, Geophys. Res. Lett., 31, L06120,
https://doi.org/10.1029/2004GL019416, 2004.
Meller, R. and Moortgat, G. K.: Temperature dependence of the
absorption cross section of HCHO between 223 and 323 K in
the wavelength range 225–375 nm, J. Geophys. Res., 105, 7089–
7102, https://doi.org/10.1029/1999JD901074, 2000.
Mellor, G. L. and Yamada, T.: A hierarchy of turbulence
closure models for planetary boundary layers, J. At-
mos. Sci., 31, 1791–1806, https://doi.org/10.1175/1520-
0469(1974)031<1791:AHOTCM>2.0.CO;2, 1974.
Miyazaki, K., Eskes, H., Sudo, K., Boersma, K. F., Bowman, K.,
and Kanaya, Y.: Decadal changes in global surface NOxemis-
sions from multi-constituent satellite data assimilation, Atmos.
Chem. Phys., 17, 807–837, https://doi.org/10.5194/acp-17-807-
2017, 2017.
Miyazaki, K., Bowman, K., Sekiya, T., Eskes, H., Boersma, F., Wor-
den, H., Livesey, N., Payne, V. H., Sudo, K., Kanaya, Y., Taki-
gawa, M., and Ogochi, K.: Updated tropospheric chemistry re-
analysis and emission estimates, TCR-2, for 2005–2018, Earth
Syst. Sci. Data, 12, 2223–2259, https://doi.org/10.5194/essd-12-
2223-2020, 2020.
Müller, J.-F., Stavrakou, T., Wallens, S., De Smedt, I., Van
Roozendael, M., Potosnak, M. J., Rinne, J., Munger, B., Gold-
stein, A., and Guenther, A. B.: Global isoprene emissions
estimated using MEGAN, ECMWF analyses and a detailed
canopy environment model, Atmos. Chem. Phys., 8, 1329–1341,
https://doi.org/10.5194/acp-8-1329-2008, 2008.
Munro, R., Lang, R., Klaes, D., Poli, G., Retscher, C., Lind-
strot, R., Huckle, R., Lacan, A., Grzegorski, M., Holdak, A.,
Kokhanovsky, A., Livschitz, J., and Eisinger, M.: The GOME-
2 instrument on the Metop series of satellites: instrument design,
calibration, and level 1 data processing – an overview, Atmos.
Meas. Tech., 9, 1279–1301, https://doi.org/10.5194/amt-9-1279-
2016, 2016.
Oomen, G.-M., Müller, J.-F., Stavrakou, T., De Smedt, I., Blu-
menstock, T., Kivi, R., Makarova, M., Palm, M., Röhling,
A., Té, Y., Vigouroux, C., Friedrich, M. M., Frieß, U., Hen-
drick, F., Merlaud, A., Piters, A., Richter, A., Van Roozen-
dael, M., and Wagner, T.: Weekly derived top-down volatile-
organic-compound fluxes over Europe from TROPOMI HCHO
data from 2018 to 2021, Atmos. Chem. Phys., 24, 449–474,
https://doi.org/10.5194/acp-24-449-2024, 2024.
Opacka, B., Müller, J.-F., Stavrakou, T., Bauwens, M., Sinde-
larova, K., Markova, J., and Guenther, A. B.: Global and re-
gional impacts of land cover changes on isoprene emissions
derived from spaceborne data and the MEGAN model, At-
mos. Chem. Phys., 21, 8413–8436, https://doi.org/10.5194/acp-
21-8413-2021, 2021.
Possanzini, M., Palo, V. D., and Cecinato, A.: Sources
and photodecomposition of formaldehyde and acetaldehyde
in Rome ambient air, Atmos. Environ., 36, 3195–3201,
https://doi.org/10.1016/S1352-2310(02)00192-9, 2002.
Price, C. and Rind, D. : A simple lightning parameterization for
calculating global lightning distributions, J. Geophys. Res., 97,
9919–9933, https://doi.org/10.1029/92JD00719, 1992.
Roberts, G., Wooster, M. J., and Lagoudakis, E.: Annual and diurnal
african biomass burning temporal dynamics, Biogeosciences, 6,
849–866, https://doi.org/10.5194/bg-6-849-2009, 2009.
Roscoe, H. K., Van Roozendael, M., Fayt, C., du Piesanie, A.,
Abuhassan, N., Adams, C., Akrami, M., Cede, A., Chong, J.,
Clémer, K., Friess, U., Gil Ojeda, M., Goutail, F., Graves, R.,
Griesfeller, A., Grossmann, K., Hemerijckx, G., Hendrick, F.,
Herman, J., Hermans, C., Irie, H., Johnston, P. V., Kanaya, Y.,
Kreher, K., Leigh, R., Merlaud, A., Mount, G. H., Navarro, M.,
Oetjen, H., Pazmino, A., Perez-Camacho, M., Peters, E., Pinardi,
G., Puentedura, O., Richter, A., Schönhardt, A., Shaiganfar, R.,
Spinei, E., Strong, K., Takashima, H., Vlemmix, T., Vrekoussis,
https://doi.org/10.5194/gmd-17-5545-2024 Geosci. Model Dev., 17, 5545–5571, 2024
5570 H. M. S. Hoque et al.: Evaluating global formaldehyde simulations using multi-platform observations
M., Wagner, T., Wittrock, F., Yela, M., Yilmaz, S., Boersma, F.,
Hains, J., Kroon, M., Piters, A., and Kim, Y. J.: Intercompari-
son of slant column measurements of NO2and O4by MAX-
DOAS and zenith-sky UV and visible spectrometers, Atmos.
Meas. Tech., 3, 1629–1646, https://doi.org/10.5194/amt-3-1629-
2010, 2010.
Ryan, R. G., Silver, J. D., Querel, R., Smale, D., Rhodes, S., Tully,
M., Jones, N.,