GMDGeoscientific Model DevelopmentGMDGeosci. Model Dev.1991-9603Copernicus GmbHGöttingen, Germany10.5194/gmd-8-1885-2015Integration of prognostic aerosol–cloud interactions in a chemistry transport model coupled offline to a regional climate modelThomasM. A.manu.thomas@smhi.seKahnertM.AnderssonC.KokkolaH.https://orcid.org/0000-0002-1404-6670HanssonU.JonesC.LangnerJ.DevasthaleA.Research Department, Swedish Meteorological and Hydrological Institute, Folkborgsvägen 17, 60176 Norrköping, SwedenDepartment of Earth and Space Sciences, Chalmers University of Technology, 41296 Gothenburg, SwedenFinnish Meteorological Institute, Kuopio, FinlandNational Centre for Atmospheric Science, School of Earth and Environment, University of Leeds, LS2 9JT Leeds, UKM. A. Thomas (manu.thomas@smhi.se)30June2015861885189825November201403February201508June201512June2015This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/This article is available from https://gmd.copernicus.org/articles/8/1885/2015/gmd-8-1885-2015.htmlThe full text article is available as a PDF file from https://gmd.copernicus.org/articles/8/1885/2015/gmd-8-1885-2015.pdf
To reduce uncertainties and hence to obtain a better estimate of aerosol
(direct and indirect) radiative forcing, next generation climate models aim
for a tighter coupling between chemistry transport models and regional
climate models and a better representation of aerosol–cloud interactions. In
this study, this coupling is done by first forcing the Rossby Center regional
climate model (RCA4) with ERA-Interim lateral boundaries and sea surface
temperature (SST) using the standard cloud droplet number concentration
(CDNC) formulation (hereafter, referred to as the “stand-alone RCA4
version” or “CTRL” simulation). In the stand-alone RCA4 version, CDNCs are
constants distinguishing only between land and ocean surface. The meteorology
from this simulation is then used to drive the chemistry transport model,
Multiple-scale Atmospheric Transport and Chemistry (MATCH), which is coupled
online with the aerosol dynamics model, Sectional Aerosol module for Large
Scale Applications (SALSA). CDNC fields obtained from MATCH–SALSA are then
fed back into a new RCA4 simulation. In this new simulation (referred to as
“MOD” simulation), all parameters remain the same as in the first run
except for the CDNCs provided by MATCH–SALSA. Simulations are carried out
with this model setup for the period 2005–2012 over Europe, and the
differences in cloud microphysical properties and radiative fluxes as a
result of local CDNC changes and possible model responses are analysed.
Our study shows substantial improvements in cloud microphysical properties
with the input of the MATCH–SALSA derived 3-D CDNCs compared to the
stand-alone RCA4 version. This model setup improves the spatial, seasonal and
vertical distribution of CDNCs with a higher concentration observed over
central Europe during boreal summer (JJA) and over eastern Europe and Russia
during winter (DJF). Realistic cloud droplet radii (CD radii) values have
been simulated with the maxima reaching 13 µm, whereas in the
stand-alone version the values reached only 5 µm. A substantial
improvement in the distribution of the cloud liquid-water paths (CLWP) was
observed when compared to the satellite retrievals from the Moderate
Resolution Imaging Spectroradiometer (MODIS) for the boreal summer months.
The median and standard deviation values from the “MOD” simulation are
closer to observations than those obtained using the stand-alone RCA4
version. These changes resulted in a significant decrease in the total annual
mean net fluxes at the top of the atmosphere (TOA) by -5 W m-2 over
the domain selected in the study. The TOA net fluxes from the “MOD”
simulation show a better agreement with the retrievals from the Clouds and
the Earth's Radiant Energy System (CERES) instrument. The aerosol indirect
effects are estimated in the “MOD” simulation in comparison to the
pre-industrial aerosol emissions (1900). Our simulations estimated the domain
averaged annual mean total radiative forcing of -0.64 W m-2 with a
larger contribution from the first indirect aerosol effect
(-0.57 W m-2) than from the second indirect aerosol effect
(-0.14 W m-2).
Introduction
The scientific understanding of the climate effects of the different aerosol
species as well as their representation in models and their physical and
chemical transformation under different meteorological conditions is still
low . Aerosols have a direct radiative effect by scattering and
absorbing shortwave and long-wave radiation, thereby changing the reflectivity,
transmissivity and absorptivity of the atmosphere. They can further act as
cloud condensation nuclei (CCN), thereby influencing the microphysical
properties of clouds. This, in turn, can impact the optical properties and
lifetimes of clouds, thus indirectly affecting the radiative properties of
the atmosphere . Apart from ambient conditions, the ability of
the aerosols to act as CCN depends on the size distribution
and, for particles in the size range between 40 and 200 nm, on the chemical
composition and mixing state .
The direct effect of aerosols and, even more so, their indirect impact on
radiative forcing have been identified as the largest sources of uncertainty
in quantifying the radiative energy budget and its impact on climate system
. An accurate estimate of these effects requires the coupling of
atmospheric chemistry/aerosols to global circulation models (GCMs); however,
due to their coarse resolution, their accuracy reduces when one starts to zoom
into regional scales. Hence, the recent generation of models uses the regional
climate models at a higher horizontal and vertical resolution instead of
GCMs, for example, WRF-Chem , ENVIRO-HIRLAM ,
RegCM3-CAMx . Recently,
summarized the status of the online/offline European coupled meteorology and
chemistry transport models with varying degrees of complexity in the
representation of dynamical and physical processes, aerosol–cloud–climate
interactions, radiation schemes, etc. The main conclusion was that an online
integrated modelling approach is the future and can be adapted to several
modelling communities such as climate modelling and air-quality-related studies
depending on the objective of the study (, and the references
therein). The study also showed that for climate modelling, the inclusion of
feedback processes is the most important and significant improvements were
noticeable in climate–chemistry/aerosols interactions. Whether the coupling
need to be online or offline depends on the specific study. For example,
showed that in long-lived greenhouse gas forcing experiments,
the online approach did not give significant improvements, whereas for
short-lived climate forcers, aerosols in particular, the online approach is very
beneficial. The aerosol–cloud interactions, in particular, are either
implicity or explicitly included in all online models. Schemes (e.g.
) that explicitly resolve the activation of CCN to cloud
droplets are currently included only in a handful of online coupled models
(ENVIRO-HIRLAM, WRF-Chem, etc). Instead, the droplet number concentrations are
derived empirically and are used in the parameterization of droplet radii and
autoconversion processes.
Here, we attempt a similar approach by adapting the Rossby Center regional
climate model (RCA4) for the offline ingestion of cloud droplet number
concentrations (CDNCs) from the cloud activation module embedded in the
chemistry transport model, Multiple-scale Atmospheric Transport and Chemistry
(MATCH), that is coupled online to the aerosol dynamics model, Sectional
Aerosol module for Large Scale Applications (SALSA). Such a setup is useful
in many ways:
A more detailed description of the emissions, transport, particle
growth, deposition and aerosol processes can be included so as to obtain an
accurate evaluation of aerosol radiative effects on a higher spatial
resolution compared to global models .
It is possible to assess the level of detail that is required to
describe the effects on a regional scale.
It can be used to assess the effects of future climate change on air
quality.
In this paper, we present the results from a full fledged working version of
the coupling between a chemistry transport model (CTM) with a detailed
aerosol dynamics model and a regional climate model. The coupling between
these two model systems is offline and is done through CDNCs calculated by
the CTM. The drawback of offline coupling is that there is no feedback on the
simulation of chemistry and aerosols from changes in meteorology due to
altered CDNC/radiation and no coupling to sea surface temperature (SST). In
the following subsections, we introduce the models used in this study, their
coupling and the improvements made in the cloud microphysical properties and
radiative forcing.
Description of the models and experimental setupsDescription of the models
Schematic showing the different model components and their
couplings.
The schematic of the model coupling is shown in Fig. . In this
study, we use the Multiple-scale Atmospheric Transport and Chemistry (MATCH)
model which is an Eulerian CTM that accounts for
transport, chemical transformation and deposition of chemical tracers in the
atmosphere based on EMEP (European Monitoring and Evaluation Programme) emissions
(http://www.ceip.at). The MATCH model is online coupled to the aerosol
dynamics model, SALSA that takes into account physical
processes such as nucleation of particles, growth of particles by
condensation and coagulation and computes the size distribution, number
concentration and chemical composition of the aerosol species. A sectional
representation of the aerosol size distribution is considered and has three
main size regimes (a: 3–50 nm, b: 50–700 nm and c: > 700 nm) and each
regime is again subdivided into smaller bins and into soluble and insoluble
bins adding up to a total of 20 bins. A schematic of the sectional size
distribution and the aerosol species considered in each bin is shown in
Fig. . Anthropogenic emissions such as primary particulate matter
(PM), NOx, NMVOC (Non-methane volatile organic compounds), SOx, NH3
and carbon monoxide, volcanic and DMS (Dimethyl sulfide) emissions are taken
from the EMEP expert emissions inventory for the year 2003. The aerosol and
gaseous concentrations at the lateral and top model boundaries are set as
described in . The boundary concentrations are based on both
observations at background locations and large-scale model calculations and
are prescribed as monthly or seasonally varying fields. However, the boundary
concentrations of organic matter (OM) are set to the seasonally varying mass
size distributions and totals of marine OM as described in . The
aerosol number concentrations are also introduced at the southern, western
and northern lateral boundaries. These values are prescribed at the first
model level and interpolated linearly to the top and eastern boundaries where
the concentrations are set to zero. Primary PM is divided into EC (elemental
carbon), OC (organic carbon) and other emissions. This division of the
primary PM is based on the TNO-MACC (TNO-Monitoring Atmospheric Composition and
Climate)
emissions of EC and OC (; see also the MACC project web
page http://www.gmes-atmosphere.eu/). The emissions were given as
annual totals. Seasonal, weekday and diurnal variations of the emissions are
sector specific and based on results from the GENEMIS (Generation and Evaluation of Emission Data) project (http://genemis. ier.uni-stuttgart.de/;
). The vertical distribution is also sector specific and based
on the vertical distribution used by the EMEP model. The particle emissions
of EC and OC are distributed over different particle sizes according to
sector resolved mass size distributions described by ; see
for more details on how the emissions are distributed.
Particulate nitrogen is described outside SALSA, i.e. ammonium salts are not
taken into account in the modelling of the aerosol microphysical processes.
The lack of ammonium nitrate condensation in the aerosol microphysics could
cause underestimation of CDNC. Currently there are no parameterizations
available that take into account co-condensation of ammonia and nitric acid.
Isoprene emissions are modelled online depending on the meteorology based on
the methodology by . The terpene emissions (α-pinene) are
taken from the modelled fields by the EMEP model. Sea salt is parameterized
following the scheme of Foltescu et al. (2005) but modified for varying
particle sizes. This means that is used if the particle
diameter is ≤1 µm otherwise is used.
Schematic of sectional distribution of aerosol size bins and the
chemical components in the bins (taken from Kokkola et al., 2008).
The coupling of MATCH with SALSA and the evaluation of this model setup is
described in detail in . A cloud activation model that computes
3-D CDNCs based on the prognostic parameterization scheme of
specifically designed for aerosol representation with sectional bins is
embedded in the MATCH–SALSA model. In addition to the updraft velocity and
supersaturation of the air parcel, this scheme simulates the efficiency of an
aerosol particle to be converted to a cloud droplet depending on the number
concentration and chemical composition of the particles. The updraft velocity
is computed as the sum of the grid mean vertical velocity and turbulent
kinetic energy (TKE) for stratiform clouds derived from the
RCA4 simulation. These CDNCs are then offline coupled to a regional climate
model, RCA4 , that provides us information on cloud
microphysical properties such as cloud droplet radii, cloud liquid-water path
as well as radiative fluxes. In the stand-alone version of RCA4, the total
number of cloud particles were set to constant values over the whole domain
based on whether the surface is oceanic (150 cm-3) or land
(400 cm-3) and scaled vertically. These constant values were further
used in calculation of effective radius of cloud droplets and in the
autoconversion process (conversion of cloud droplet to rain). In this work,
the 3-D CDNC fields obtained from the cloud activation model in MATCH–SALSA
are now used in the RCA4 simulation.
Experimental setup – 1
For the simulations, RCA4 is run with 6-hourly ERA-Interim meteorology on
lateral boundaries and sea surface temperature and the 3-hourly RCA4
meteorological fields along with fields necessary to compute the updraft
velocity are used to drive the MATCH–SALSA-cloud activation model. The
aerosol properties are used in the cloud activation model to derive the
6-hourly CDNCs which are then employed to re-run RCA4 with dynamic rather
than prescribed CDNCs to obtain cloud microphysical properties and radiative
effects. The CDNC data are interpolated at every time step in the RCA4 model.
Simulations were carried out with this model setup at
44 km × 44 km spatial resolution for the European domain and 24
levels in the vertical (up to 200 hPa) for an 8-year period (2005–2012).
Here, we look into the improvements in the cloud microphysical properties and
radiative fluxes with the incorporation of dynamic CDNCs where only local
land surface fluxes can respond to these changes, hereby referred to as the
MOD simulation. These
results are compared with the control simulation, hereby referred to as the
CTRL simulation in which the stand-alone version of RCA4 is used.
Experimental setup – 2
To evaluate the indirect aerosol effects due to the present-day (PD)
anthropogenic aerosols, the pre-industrial (PI) emissions required for this
simulation were taken from those developed for the ECLAIRE project (effects
of climate change on air pollution impacts and response strategies) for the
year 1900 (http://www.eclaire-fp7.eu). The PI anthropogenic precursor
emissions were provided for CO, NH3, NOx, SOx and volatile organic
compound (VOC). Other emissions such as biogenic emissions, DMS and volcanic
are kept the same as in the original model setup. The particulate organic
matter emissions are reduced to 14 % of the current emission levels in the
pre-industrial setup based on the study by . Simulations were
carried out at the same spatial resolution as in the previous setup and for
the same European domain and for the same meteorology from 2005 to 2012.
Here, we address the changes in cloud properties with respect to the
emissions from the PI period without the climate feedbacks and we analyse the
total radiative forcing. To evaluate the individual contribution from the
first and second indirect aerosol effects to the total radiative forcings,
two additional simulations each (for PI and PD climate) were carried out. We
turn off the individual indirect aerosol effects (IAEs) by prescribing the
constant CDNC values for the calculation of one IAE at a time; for example,
to evaluate the sole contribution from the first IAE, 3-D CDNC fields are
used in the computation of cloud droplet (CD) radius to assess the changes in
cloud albedo (first IAE) and constant CDNC values are used in the scheme for
the autoconversion process (second IAEs) and vice versa.
Model evaluation and resultsAerosol number concentrations
Figure shows the spatial distribution of the seasonal mean
accumulation mode aerosol number concentrations from MATCH–SALSA model
simulations driven by RCA4 meteorology. The main contribution to accumulation
mode particles are from SO42-, EC, OC, sea salt and mineral dust. In
the figure, emphasis is given to accumulation mode particles as they can act
as CCN and, depending on the water availability and updraft velocity, be
efficiently converted into cloud droplets. A clear seasonality is noticeable
with the highest concentrations during the summer months and lowest during the
winter months. Concentrations reach as high as 600 cm-3 over the southern
European subcontinent during summer. This may be partly due to relatively
large emissions of primary fine particles and gaseous SOx
(), and partly due to less precipitation in southern
Europe compared to the rest of Europe resulting in longer residence times of
these particles in the atmosphere .
Averaged aerosol number concentrations in the accumulation mode
(cm-3).
Cloud droplet number concentrations
The seasonal mean in-cloud averaged CDNCs for the European domain is
presented in Fig. . During boreal summer, CDNCs are extremely high,
reaching as high as 225 cm-3 over central Europe mainly because of
summer time vertical mixing, high probability of liquid clouds and is also
consistent with high aerosol number concentrations simulated during this
time. The land–sea contrast is more prominent during the summer months
compared to the other seasons. A closer comparison with Fig.
reveals that regions of high CDNCs correspond mostly with regions of high
accumulation mode aerosol particles. However, in some regions, this
correlation is not very noticeable, especially over the Mediterranean in
summer. This may be due to subsidence and lack of clouds in these regions.
Residential biomass burning is more prominent in late autumn–winter–early
spring months over eastern Europe and Russia. Whereas, biogenic VOC emissions
are higher in these regions during the summer season .
This is reflected in the higher CDNCs over these regions during these
seasons.
showed that differences in the cloud droplet activation would
account for about 65 % of the total spread in shortwave forcing. So, it is
important to see if this model setup reproduces the spatial distribution of
CDNCs realistically. The high droplet concentrations simulated over land
compared to oceans agrees well with the previous studies of and
. analysed the CDNCs from the Moderate Resolution Imaging Spectroradiometer (MODIS) and Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) sensors
and it is evident that the droplet concentrations over the European subcontinent
would be on an average between 150 and 200 cm-3, and over oceans the
concentrations are as low as 100 cm-3. This agrees well with our
simulations.
Seasonal mean in-cloud averaged cloud droplet number concentrations
(cm-3).
To understand the vertical distribution of CDNCs, we select the following four regions as in Fig. – R1: Mediterranean R2: eastern Europe R3:
central Europe and R4: northern Atlantic. Figure shows the joint
histograms of CDNC and height (in km) over the four selected regions for
winter (DJF) and summer (JJA) months. The colour scale shows the
normalized frequency and the deeper shade means there is a high probability
for the occurrence of those values. It can be seen that the majority of the
CDNC values are less than 500 cm-3, but occasionally values also reach
as high as 800 cm-3 irrespective of season and region. The figures also
show an interesting seasonality of the PDFs (probability distribution
functions) of CDNC. For example over the Mediterranean (R1) two peaks can be
observed in summer, one around 3 km and another in the boundary layer;
however, in winter, only one peak is present and is located below 2 km. This
is mainly due to the stronger vertical mixing of aerosols together with
increased buoyancy and convection in summer. It is well known that the
frequency of occurrence of very low level water clouds is high over this
region during summer, while, in winter, the baroclinic disturbances lead to
northward transport of winter storms over this region. Over eastern Europe
(R2) and central Europe (R3), the height at which maxima of CDNCs occur is
around 1.5 km in summer, whereas during winter, the maxima is in the
boundary layer. The droplet concentrations can vary widely over eastern
Europe compared to over central Europe where the concentrations are mostly
towards the higher side irrespective of the seasons. The opposite is observed
over the northern Atlantic (R4) with a maxima of CDNCs simulated at around 1.5 km in
winter and in the boundary layer in summer. This can be attributed to the
long-range transport of pollutants from across the Atlantic observed
during the winter months .
Schematic of the four regions selected for this study. R1 –
Mediterranean; R2 – eastern Europe; R3 – central Europe; R4 – northern
Atlantic.
Seasonality in the vertical distributions of CDNC (cm-3) shown
as joint histograms averaged over winter averaged (DJF) months (first column)
and summer averaged (JJA) months (second column) for the selected four regions.
The colour scale indicates the normalized frequency.
Cloud droplet radii
As mentioned in Sect. 2, both the CTRL and MOD simulations follow the same
parameterization scheme of in the radiation and are formulated
in Eq. (). The effective radius for spherical droplets is
estimated as
re,water=3CC4πρlkN,
where CC is the cloud condensate content in kg m-3, ρl is
the density of liquid water, N is the number concentration of cloud
droplets in m-3 and k is a constant depending on marine or continental
clouds. In the CTRL simulation, N was assigned 150 and 400 cm-3
depending on the underlying surface; however, in the MOD simulation dynamic
CDNCs are used.
The winter and summer mean cloud droplet radii in liquid water clouds for the
MOD simulation (top row) and the CTRL simulation (second row) is discussed in
Fig. . At first glance, apart from the spatial differences, one
notices the strong disparity in the range of the radii values. In the MOD
simulation, CD radii reach as high as 13 µm, whereas in the CTRL
simulation, the maxima observed is 5 µm. It can be seen that the
radii of the droplets are much lower in the summer months compared to the
winter months basically due to the increased pollution during summer
resulting in smaller droplets.
Seasonal mean CD radii (µm) averaged over the entire water
cloud for DJF mean and JJA mean in the MOD simulations (top row) and CTRL
simulation (second row). Note the difference in scale.
In Fig. , the joint histograms of CD radii and height during the
summer over the Mediterranean and eastern Europe in MOD simulation are shown. The
corresponding pattern in the CTRL simulation is shown as the solid line. It
can be seen that the CTRL simulation does not exhibit any variability.
However, the MOD simulation shows a distinct variability in height and range.
It can be seen that over the Mediterranean and over Eastern Europe, a wide
range of droplet radii can be observed from as low as 5 up to 16 µm
and the larger droplets are present at around 2.0–4.0 km. However, there is
a higher probability of observing larger droplets over eastern Europe
compared to the Mediterranean.
The critical droplet radius at which large-scale precipitation occurs is set
to 10 µm in the microphysics scheme. This would mean that with
these low CD radii values obtained in the stand-alone RCA4 model,
precipitation would be absent. It is important to note that the plotted CD
radii values are derived from the model radiation scheme (i.e. this is the
radiatively active CD radii). For more details, refer Appendix A.
Cloud liquid-water path
Figure refers to the percentage change in column integrated cloud
water (cloud liquid-water path – CLWP) in the MOD simulations compared to CTRL simulations averaged
over the winter (DJF) and summer (JJA) months. Positive values mean that the
CLWP increased in the MOD simulations compared to the CTRL and vice versa. It
can be seen that during the winter months, there is a significant decrease
(up to 25 %) in the vertically integrated cloud liquid water over land and
a slight increase over the water bodies when the 3-D CDNCs are used. However,
during the summer months, the pattern is reversed with a noticeable increase
in the CLWP over most of the European subcontinent. The decrease in the CLWP with
increase in CD radii is consistent and may be partly attributed to
precipitation removal. In the model, the critical droplet radius at which
autoconversion becomes efficient is set to 10 µm. When the CD
effective radius exceeds this critical droplet radius, precipitation occurs.
However, over land during summer, an increase in the CLWP is observed and can be
partly due to the fact that the increase in CD radius is not sufficient to
trigger precipitation.
Joint histograms of CD radii (µm) vs height averaged over
JJA months in MOD simulation and the solid line shows the same in CTRL
simulation. The colour scale shows the normalized frequency.
The difference in the vertically integrated cloud water in the MOD
simulation compared to the CTRL simulation expressed as a percentage over the
winter (DJF) and summer (JJA) months.
We evaluated model simulated cloud liquid-water path estimates using
satellite sensor retrievals. We selected the liquid water path (LWP) for evaluation as it is tightly
connected not only to other microphysical properties of clouds but also to
the first and second indirect aerosol effects, which are the main application
focus of the coupling attempted here. We used a decade-long data record
(2003–2012) from the MODIS
sensor flying onboard NASA's Aqua satellite since 2002 .
The monthly level 3 data from the collection 5 are analysed over the study
area for the boreal summer months of JJA (June, July and August), when liquid
clouds are most prevalent and the quality of satellite retrievals is also
best. The comparison is shown as spatial distributions in Fig.
and as probability density functions of the LWP in Fig. as they cover
the whole range of LWP values. We observe substantial improvement in the
distribution of LWPs in the MOD simulation compared to the CTRL-simulation.
The inclusion of the chemistry–aerosol–cloud microphysical link leads to a more
realistic distribution of LWPs that is closer to the observations. At the
lower end of the distribution, the model simulates more optically thin liquid
water clouds compared to the observations, more predominantly over southern
Europe as can be seen in Fig. . But the LWP of optically thick
clouds, which are most abundant and play a key role in the radiation budget
over the study area, shows substantial improvements in the MOD simulation.
Spatial comparison of the simulated cloud liquid-water path
(g m-2) (CTRL and MOD) with observations from the MODIS sensor onboard
the Aqua satellite for the JJA months.
Total radiative fluxes at the top of the atmosphere
The difference in the annual mean total net fluxes at the top of the atmosphere (TOA) due to these
changes in the cloud microphysical properties is shown in Fig. . A
significant change is seen over most of the domain with decreases up to
-5 W m-2 when the CDNC values are assigned fixed numbers depending
on the underlying surface.
To evaluate the TOA radiative fluxes, we used a decade-long data for
comparison from the Clouds and the Earth's Radiant Energy System (CERES)
sensor that is flying onboard Aqua satellite as well (More
information is available at
http://ceres.larc.nasa.gov/documents/DQ_summaries/CERES_EBAF_Ed2.8_DQS.pdf).
The level 3 Energy Balanced and Filled estimates of all-sky net top of the
atmosphere (EBAF-TOA, Edition 2.8) fluxes are analysed for comparison with
MOD and CTRL simulations and shown in Fig. . The distribution of
fluxes in the MOD simulation is closer to the CERES observations compared to
CTRL simulation.
Aerosol radiative forcing at the TOA
In this section, we evaluate the effect of changing cloud albedo and cloud
lifetime due to the PD anthropogenic aerosols which is widely known
as the first indirect aerosol effect and the second indirect
aerosol effect , respectively. In a review paper,
summarized the aerosol radiative effects into positive and negative
perturbations to the radiation budget. Both the indirect aerosol effects tend
to cool the Earth system by increasing the cloud optical depth and cloud
cover, thereby resulting in the reduction of net radiation reaching the
TOA and surface.
The local radiative forcing associated with these IAE are in most cases
estimated as the difference between the perturbed and unperturbed radiative
fluxes. The perturbed case is the current climate scenario and in the
unperturbed case, the fluxes are calculated based on a pre-industrial or
pristine scenario with meteorology and SST remaining the same in the both
cases. In this study, for the unperturbed case (PI), we use the
pre-industrial emissions from 1900s as explained in Sect. 2.3. The PD perturbed case climate scenario is using the MOD simulation setup.
In the following paragraphs, we analyse the changes in the CDNC and CLWP in the
PD–PI differences and the TOA aerosol radiative forcing over Europe which
arises mainly from the response of the land surface without other climate
feedbacks. Figures , and show the annual
mean difference in aerosol number concentrations, CDNC and CLWP,
respectively,
with respect to the PI simulation expressed as a percentage.
Comparison of cloud liquid-water path (g m-2) with
observations from the MODIS sensor onboard the Aqua satellite. The histograms
are computed over the entire study area and for the JJA months. The median
and standard deviation (in brackets) values are shown for all cases.
Difference in the annual mean total net fluxes (W m-2) at the
TOA in the (MOD–CTRL) case.
It can be seen that there is an approximately 50–80 % increase in the aerosol
number concentrations with respect to PI era over the southern and eastern
European subcontinent and around 10–30 % over the rest of Europe and over
the oceans. The steep increase in the aerosol concentrations may be
attributed to the increase in anthropogenic pollutant precursor emissions in
these countries in the PD. These differences seen in the
spatial distribution are reflected as an increase of almost up to 70 %
in CDNCs and, correspondingly, an increase of up to 10 % in the CLWP.
Figure shows the spatial distribution of the annual mean indirect
aerosol forcing when using the 1900 reference value over the study region.
The European domain averaged annual mean radiative forcing is
-0.64 W m-2, with values reaching as high as -1.3 W m-2.
This is comparable to the estimate of -0.96 W m-2 obtained by
for the global mean forcing using the same reference period.
Comparison of total TOA fluxes (W m-2) with observations from
the CERES sensor onboard the Aqua satellite. The histograms are computed over
the entire study area and for the JJA months. The median and standard
deviation (in brackets) values are shown for all cases.
Difference in the annual mean aerosol number concentrations for the
(PD–PI) simulation expressed as a percentage.
Difference in the annual mean CDNC for the (PD–PI) simulation
expressed as a percentage.
Difference in the annual mean CLWP for the (PD–PI) simulation
expressed as a percentage.
Annual mean aerosol radiative forcing at the TOA (W m-2).
We also estimated the individual contribution of the first and second IAE to
the total aerosol radiative forcing. This is done by turning off the
individual IAEs by prescribing the fixed values for CDNCs as in the
stand-alone RCA4 version in the radiation and cloud microphysics calculation
.
In IPCC-AR5 (Intergovernmental Panel for Climate Change – 5th Assessment
Report) it was pointed out that the estimated values of the
first IAE constitute the largest uncertainty and vary significantly between
the different models. The impact of changes of aerosols on cloud albedo
through the changes in droplet radius (first IAE) is estimated from our model
setup to be -0.57 W m-2 when averaged over the European domain. IPCC
models estimated the global annual mean first IAE to be -0.7 W m-2
and can vary widely from -1.8 to -0.3 W m-2. However, the impact
of changes of aerosols on cloud lifetime via the modification of
precipitation efficiency is estimated to be -0.14 W m-2 when
averaged over the European domain used in this study. Studies have shown that
the uncertainties in this are even larger compared to the first IAE.
Depending on the autoconversion schemes used in the global models,
showed that the global mean second IAE can vary from -0.71 to
-0.28 W m-2, of which the scheme used to obtain the value of
-0.28 W m-2 is more realistic.
Conclusions
In this study, we coupled the Rossby Center regional climate model (RCA4) for
the offline ingestion of CDNCs from the cloud activation module that is
embedded in the aerosol–chemistry transport model, MATCH–SALSA. Such a setup
is beneficial mainly because a more sophisticated representation of aerosol
distribution (emissions, transport and microphysical processes) can be
included at a higher resolution compared to global models. Simulations were
carried out with this setup for the period 2005–2012 over Europe using
present-day emissions (PD) from EMEP for the year 2000 as well as using the
stand-alone version of RCA4. We carried out two sets of analysis:
Investigate the improvements in a regional climate model simulation of
the cloud microphysical properties, using spatially and temporally resolved
3-D CDNC fields from a detailed aerosol and cloud activation model.
Evaluate the indirect aerosol effects using this integrated modelling
approach using the PI emissions taken from the ECLAIRE project. The
particulate matter in the PI period are reduced to 14 % of the current
levels based on .
This model setup improves the spatial, seasonal and vertical distribution of
CDNCs with higher concentrations observed over central Europe during summer
(JJA averaged) and over eastern Europe and Russia during winter (DJF
averaged). Realistic cloud droplet radii numbers have been simulated with the
maxima reaching 13 µm, whereas in the stand-alone version, the
values reached only 5 µm. The stand-alone version of RCA4
overestimated the vertically integrated cloud water by up to 25 % in
winter and underestimated by a similar magnitude in summer over the European
subcontinent. Comparisons with satellite retrievals from MODIS reveals a
significant improvement in the LWP distribution; the median and standard
deviation values obtained from the MOD simulation is much closer to
observations than the CTRL simulation. A significant decrease by up to
-5 W m-2 in the total TOA net fluxes is simulated owing to these
changes. The TOA net fluxes obtained with the new model setup show a better
agreement with net flux retrievals from the CERES instrument than those
computed with the old model setup. This confirms the importance of employing
a realistic, dynamic description of aerosol number distribution fields in
regional climate models.
Using the pre-industrial emissions from 1900s, we estimated an increase of
around 50–70 % CDNCs over southern and eastern Europe and below 30 %
over the rest of Europe in the PD climate consistent with the increases in
aerosol number concentrations, and correspondingly an increase in the CLWP is
simulated over our study area. These changes resulted in an annual mean TOA
radiative forcing over Europe of -0.64 W m-2 which is comparable to
previous estimates for the same reference period. It was also estimated that
the contribution from the first IAE (-0.57 W m-2) is larger than the
contribution from the second IAE
(-0.14 W m-2). This study shows a
substantial improvement in the cloud microphysical properties and radiative
fluxes with the offline coupled model setup. Hence, we recognize the need for
an online coupled model system and we plan to implement SALSA directly into
RCA4 in the future.
The calculations were performed on a HP Cluster Platform 3000 with SL230s
Gen8 compute nodes, each with two 8-core Intel Xeon E5-2660 “Sandy Bridge”
processors at 2.2 GHz. Using three nodes and 48 MPI-ranks, a 1-year
simulation with the online coupled MATCH–SALSA including the cloud activation
module takes 20 h (wall-clock time). On the other hand, RCA4 takes
approximately 1.5 h for 1-year simulation using two nodes and 32
MPI-ranks.
In a given time step (t=n), the thermodynamic tendencies resulting from
resolved dynamics and parameterized vertical turbulent fluxes are first
calculated. These terms are stored in the relevant tendency terms
(∂χ∂t at t=n). Next, the radiation scheme is
called using the thermodynamic values (temperature, T; humidity, q; cloud
water, cw) valid at t=n. After the radiation fluxes have been
calculated, the model calls the surface scheme, followed by convection and, finally,
condensation and cloud microphysics are called. On entering the microphysics
scheme, the thermodynamic variables valid at t=n are updated by the
tendencies from dynamics and turbulence. For example,
T=Tn+Δt×∂T∂t|dyn+∂T∂t|turb,
and these variables are used to calculate a new value of cw, q, etc.,
consistent with the updated saturated state of the
atmosphere.
With this, calculated cloud water can increase
locally due to the dynamical and turbulent terms, and this re-calculated CD
radii for the autoconversion process in the microphysics scheme may be
substantially larger than in the radiation scheme (exceeding the threshold
for precipitation onset of 10 µm). With the microphysics scheme,
precipitation removal of generated cloud water is also parameterized; hence,
the fraction of the newly generated cloud water is removed as rain. The
resulting cw tendency due to condensation and subsequent precipitation
removal is then incremented to t=n values, along with all other tendencies
from t=n to give new values for the next time step. In this manner, CD
radii in the microphysics scheme can increase above the threshold for onset
of precipitation and then, as a result of this precipitation removal of cloud
water, decrease again below the threshold. The low (precipitation affected)
value of cloud water is what the radiation scheme “sees” in each time step
leading to a low estimate in CD radii.
Code availability
The different model source codes and the coupled system used in this study
are not entirely available for open-source distribution at present, but can
be made available to interested users/investigators upon request. The aerosol
microphysics code SALSA is distributed under the Apache 2.0 license, while the
regional climate model RCA4 and the chemistry transport model MATCH are
available upon request from the SMHI.
Acknowledgements
This work was supported and funded by the Swedish Environmental Protection
Agency through the CLEO (CLimate change and Environmental Objectives) and
partly by the SCAC (Swedish Clean Air and Climate research program) projects.
We also acknowledge the funding from FORMAS through MACCII (Modeling
Aerosol–Cloud–Climate Interactions and Impacts) project. M. Kahnert
acknowledges funding from the Swedish Research Council (Vetenskapsrådet)
under project 621-2011-3346. Edited by:
A. Colette
ReferencesAbdul-Razzak, H. and Ghan, S. J.: A parameterization of aerosol activation 3.
Sectional representation, J. Geophys. Res., 107, AAC1.1–AAC1.6, 10.1029/2001JD000483,
2002.
Albrecht, B.: Aerosols, cloud microphysics, and fractional cloudiness,
Science, 245, 1227–1230, 1989.
Andersson, C., Langner, J., and Bergström, R.: Interannual variation and
trends in air pollution over Europe due to climate variability during
1958-2001 simulated with a regional CTM coupled to the ERA-40 reanalysis,
Tellus, 59B, 77–98, 2007.Andersson, C., Bergström, R., Bennet, C., Thomas, M., Robertson, L.,
Kokkola, H., Korhonen, H., and Lehtinen, K.: MATCH-SALSA: Multi-scale
Atmospheric Transport and Chemistry model coupled to the SALSA aerosol
microphysics model, SMHI RMK Report no 115, available at:
http://www.smhi.se/publikationer/match-salsa-multi-scale-atmospheric-transport-and-chemistry-model-coupled-to-the-salsa-aerosol-microphysics-model-1.34623
(last access: 26 June 2015),
2013.Andersson, C., Bergström, R., Bennet, C., Robertson, L., Thomas, M.,
Korhonen, H., Lehtinen, K. E. J., and Kokkola, H.: MATCH-SALSA – Multi-scale
Atmospheric Transport and CHemistry model coupled to the SALSA aerosol
microphysics model – Part 1: Model description and evaluation, Geosci. Model
Dev., 8, 171–189, 10.5194/gmd-8-171-2015, 2015.
Baklanov, A., Korsholm, U., Mahura, A., Petersen, C., and Gross, A.:
ENVIRO-HIRLAM: online coupled modelling of urban meteorology and air
pollution, Adv. Sci. Res., 2, 41–46, 2008.Baklanov, A., Schlünzen, K., Suppan, P., Baldasano, J., Brunner, D.,
Aksoyoglu, S., Carmichael, G., Douros, J., Flemming, J., Forkel, R.,
Galmarini, S., Gauss, M., Grell, G., Hirtl, M., Joffre, S., Jorba, O., Kaas,
E., Kaasik, M., Kallos, G., Kong, X., Korsholm, U., Kurganskiy, A., Kushta,
J., Lohmann, U., Mahura, A., Manders-Groot, A., Maurizi, A., Moussiopoulos,
N., Rao, S. T., Savage, N., Seigneur, C., Sokhi, R. S., Solazzo, E., Solomos,
S., Sørensen, B., Tsegas, G., Vignati, E., Vogel, B., and Zhang, Y.:
Online coupled regional meteorology chemistry models in Europe: current
status and prospects, Atmos. Chem. Phys., 14, 317–398,
10.5194/acp-14-317-2014, 2014.Barahona, D., Sotiropoulou, R., and Nenes, A.: Global distributionof cloud
droplet number concentration, autoconversion rate, andaerosol indirect effect
under diabatic droplet activation., J. Geophys. Res., 116, D09203,
10.1029/2010JD015274, 2011.
Boucher, O., Randall, D., Artaxo, P., Bretherton, C., Feingold, G., Forster,
P., Kerminen, V.-M., Kondo, Y., Liao, H., Lohmann, U., Rasch, P., Satheesh,
S. K., Sherwood, S., Stevens, B., and Zhang, X. Y.: Clouds and Aerosols, in:
Climate Change 2013: The Physical Science Basis. Contribution of Working
Group I to the Fifth Assessment Report of the Intergovernmental Panel on
Climate Change, edited by: Stocker, T. F., Qin, D., Plattner, G.-K., Tignor, M., Allen,
S. K., Boschug, J., Nauels, A., Xia, Y., Bex, V., and Midgley, P. M., Cambridge
University Press, Cambridge, United Kingdom and New York, NY, USA, 2013.
Carslaw, K. S., Lee, L. A., Reddington, C. L., Pringle, K. J., Rap, A.,
Forster, P. M., Mann, G. W., Spracklen, D. V., Woodhouse, M. T., Regayre,
L. A., and Pierce, J. R.: Large contribution of natural aerosols to
uncertainty in indirect forcing, Nature, 503, 67–71, 2013.Colarco, P., da Silva, A., Chin, M., and Diehl, T.: Online simulations of
global aerosol distributions in the NASA GEOS-4 model and comparisons to
satellite and ground-based aerosol optical depth, J. Geophys. Res.-Atmos., 115, D14207, 10.1029/2009JD012820, 2010.Dusek, U., Frank, G. P., Hildebrandt, L., Curtius, J., Schneider, J., Walter,
S., Chand, D., Drewnick, F., Hings, S., Jung, D., Borrmann, S., and Andreae,
M. O.: Size matters more than chemistry for cloud nucleating ability of
aerosol particles, Science, 213, 1375–1378, 10.1126/science.1125261, 2006.
Folberth, G. A., Rumbold, S., Collins, W. J., and Butler, T.: Regional and
Global Climate Changes due to Megacities using Coupled and Uncoupled Models,
D6.6, MEGAPOLI Scientific Report 11-07, MEGAPOLI-33-REP-2011-06, 18 pp.,
2011.
Forster, P., Ramaswamy, V., Artaxo, P., Berntsen, T., Betts, R., Fahey, D.,
Haywood, J., Lean, J., Lowe, D., Myhre, G., Nganga, J., Prinn, R., Raga, G.,
Schulz, M., and Dorland, R. V.: Changes in Atmospheric Constituents and in
Radiative Forcing, in: Climate Change: The Physical Science Basis.
Contribution of Working Group I to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change, edited by: Solomon, S., Qin, D., Manning,
M., Chen,
Z., Marquis, M., Averyt, K. B., Tignor, M., and Miller, H. L., Cambridge
University Press, Cambridge, United Kingdom and New York, NY, USA, 2007.
Friedrich, R. and Reis, S. E.: Emissions of air pollutants – measurements,
calculations and uncertainties, Springer, ISBN 978-3-540-00840-8, 2004.
Grell, G. A., Peckham, S. E., Schmitz, R., McKenn, S. A., Frost, G.,
Skamarock, W. C., and Eder, B.: Fully coupled “online” chemistry within the
WRF model, Atmos. Environ., 39, 6957–6975, 2005.
HTAP: Hemispheric Transport of Air Pollution 2010-Part A: Ozone and
Particulate Matter, edited by: Dentener, F., Keating, T., and Akimoto, H.,
United Nations, New York and Geneva, Air Pollution Studies, 17, 135–198,
2010.
Hubanks, P. A., King, M. D., Platnick, S., and Pincus, R.: MODIS Atmosphere
L3 Gridded Product Algorithm Theoretical Basis Document – Collection 5,
Version 1.1., 1–96, 2008.
Huszar, P., Miksovsky, J., Pisoft, P., Belda, M., and Halenka, T.:
Interactive coupling of a regional climate model and a chemical transport
model: evaluation and preliminary results on ozone and aerosol feedback,
Clim. Res., 51, 59–88, 2012.Kokkola, H., Korhonen, H., Lehtinen, K. E. J., Makkonen, R., Asmi, A.,
Järvenoja, S., Anttila, T., Partanen, A.-I., Kulmala, M., Järvinen,
H., Laaksonen, A., and Kerminen, V.-M.: SALSA – a Sectional Aerosol module
for Large Scale Applications, Atmos. Chem. Phys., 8, 2469–2483,
10.5194/acp-8-2469-2008, 2008.
Kuenen, J., Denier van der Gon, H., Visschedijk, A., van der Brugh, H., and
van Gijlswijk, R.: MACC European emission inventory for the years 2003–2007,
TNO report, TNO-060-UT-2011-00588, 2011.
Lohmann, U. and Feichter, J.: Can the direct and semi-direct aerosol effect
compete with the indirect effect on a global scale?, Geophys. Res. Lett., 28,
159–161, 2001.Lohmann, U. and Feichter, J.: Global indirect aerosol effects: a review,
Atmos. Chem. Phys., 5, 715–737, 10.5194/acp-5-715-2005, 2005.
Lohmann, U., Feichter, J., Chuang, C. C., and Penner, J. E.: Prediction of
the number of cloud droplets in the ECHAM GCM, J. Geophys. Res., 104,
9169–9198, 1999.Mårtensson, E. M., Nilsson, E. D., de Leeuw, G., Cohen, L. H., and
Hansson, H.-C.: Laboratory simulations and parametrization of the primary
marine aerosol production, J. Geophys. Res., 108, 4297,
10.1029/2002JD002263, 2003.McFiggans, G., Artaxo, P., Baltensperger, U., Coe, H., Facchini, M. C.,
Feingold, G., Fuzzi, S., Gysel, M., Laaksonen, A., Lohmann, U., Mentel, T.
F., Murphy, D. M., O'Dowd, C. D., Snider, J. R., and Weingartner, E.: The
effect of physical and chemical aerosol properties on warm cloud droplet
activation, Atmos. Chem. Phys., 6, 2593–2649, 10.5194/acp-6-2593-2006,
2006.
Monahan, E. C., Spiel, D. E., and Davidson, K. L.: A model of marine aerosol
generation via whitecaps and wave disruption, in: Oceanic Whitecaps and Their
Role in Air-Sea Exchange 688, edited by: Monahan, E. C. and Mac Niocaill, G., D
Reidel, Norwell, MA, 167–174, 1986.Moore, R. H., Karydis, V. A., Capps, S. L., Lathem, T. L., and Nenes, A.:
Droplet number uncertainties associated with CCN: an assessment using
observations and a global model adjoint, Atmos. Chem. Phys., 13, 4235–4251,
10.5194/acp-13-4235-2013, 2013.
O'Dowd, C. D., Facchini, M. C., Cavalli, F., Ceburnis, D., Mircea, M.,
Decesari, S., Fuzzi, S., Yoon, Y. J., and Putaud, J.-P.: Biogenically driven
organic contribution to marine aerosol, Nature, 431, 676–680, 2004.
Penner, J. E., Dong, X., and Chen, Y.: Observational evidence of a change in
radiative forcing due to the indirect aerosol effect, Nature, 427, 231–234,
2004.Platnick, S., King, M. D., Ackerman, S. A., Menzel, W. P., Baum, B. A.,
Riedi, J. C., and Frey, R. A.: The MODIS cloud products: Algorithms and
examples from Terra., IEEE T. Geosci. Remote Sens., 41, 459–473,
10.1109/TGRS.2002.808301, 2003.
Pouliot, G., Pierce, T., Denier van der Gon, H., Schaap, M., Moran, M., and
Nopmongcol, U.: Comparing emission inventories and model-ready emission
datasets between Europe and North America for the AQMEII project, Atmos.
Environ., 53, 4–14, 2012.
Qian, Y. and Giorgi, F.: Interactive coupling of regional climate and sulfate
aerosol models over east Asia, J. Geophys. Res., 104, 6477–6499, 1999.
Qian, Y., Giorgi, F., and Huang, Y.: Regional simulation of anthropogenic
sulfur over East Asia and its sensitivity to model parameters, Tellus, 53B,
171–191, 2001.
Robertson, L., Langner, J., and Engardt, M.: An Eulerian Limited-Area
Atmospheric Transport model, J. Appl. Meteorol., 38, 190–210, 1999.Rotstayn, L. D. and Liu, Y.: A smaller global estimate of the second indirect
aerosol effect, Geophys. Res. Lett., 32, L05708,
10.1029/2004GL021922, 2005.
Samuelsson, P., Jones, C., Willen, U., Ullerstig, A., Gollvik, S., Hansson,
U., Jansson, C., Kjellström, E., Nikulin, G., and Wyser, K.: The Rossby
Center Regional Climate model RCA3: model description and performance, Tellus
A, 63, 4–23, 2011.
Simpson, D., Guenther, A., Hewit, C., and Steinbrecher, R.: Biogenic
emissions in Europe. 1. Estimates and uncertainties, J. Geophys. Res., 100,
22875–22800, 1995.Spracklen, D. V., Carslaw, K. S., Merikanto, J., Mann, G. W., Reddington, C.
L., Pickering, S., Ogren, J. A., Andrews, E., Baltensperger, U., Weingartner,
E., Boy, M., Kulmala, M., Laakso, L., Lihavainen, H., Kivekäs, N.,
Komppula, M., Mihalopoulos, N., Kouvarakis, G., Jennings, S. G., O'Dowd, C.,
Birmili, W., Wiedensohler, A., Weller, R., Gras, J., Laj, P., Sellegri, K.,
Bonn, B., Krejci, R., Laaksonen, A., Hamed, A., Minikin, A., Harrison, R. M.,
Talbot, R., and Sun, J.: Explaining global surface aerosol number
concentrations in terms of primary emissions and particle formation, Atmos.
Chem. Phys., 10, 4775–4793, 10.5194/acp-10-4775-2010, 2010.Storelvmo, T., Lohmann, U., and Bennartz, R.: What governs the spread in
shortwave forcings in the transient IPCC AR4 models?, Geophys. Res. Lett.,
36, L01806, 10.1029/2008GL036069, 2009.
Twomey, S.: The nuclei of natural cloud formation part II: the
supersaturation in natural clouds and the variation of cloud droplet
concentration, Pure Appl. Geophys., 43, 243–249, 1959.
Visschedijk, A. J. H., Denier van der Gon, H., Dröge, A. C., Van der
Brugh, H., and Kooter, I. M.: A European high resolution and
size-differentiated emission inventory for elemental and organic carbon for
the year 2005, TNO report, TNO-034-UT-2009-00688-RPT-ML, 2009.
Wielicki, B. A., Cess, R. D., King, M. D., Randall, D. A., and Harrison,
E. F.: Mission to Planet Earth: Role of Clouds and Radiation in Climate, B.
Am. Meteorol. Soc., 77, 853–868, 1996.
Wyser, K., Rontu, L., and Savijärvi, H.: Introducing the Effective Radius
into a Fast Radiation Scheme of a Mesoscale Model, Contr. Atmos. Phys., 72,
205–218, 1999.Yttri, K. E., Dye, C., Braathen, O.-A., Simpson, D., and Steinnes, E.:
Carbonaceous aerosols in Norwegian urban areas, Atmos. Chem. Phys., 9,
2007–2020, 10.5194/acp-9-2007-2009, 2009.Yttri, K. E., Lund Myhre, C., Eckhardt, S., Fiebig, M., Dye, C., Hirdman, D.,
Ström, J., Klimont, Z., and Stohl, A.: Quantifying black carbon from
biomass burning by means of levoglucosan – a one-year time series at the
Arctic observatory Zeppelin, Atmos. Chem. Phys., 14, 6427–6442,
10.5194/acp-14-6427-2014, 2014.Yu, F. and Luo, G.: Simulation of particle size distribution with a global
aerosol model: contribution of nucleation to aerosol and CCN number
concentrations, Atmos. Chem. Phys., 9, 7691–7710,
10.5194/acp-9-7691-2009, 2009.
Zeng, S., Riedi, J., Trepte, C. R., Winker, D. M., and Hu, Y.-X.: Study of
global cloud droplet number concentration with A-Train satellites, Atmos.
Chem. Phys., 14, 7125–7134, 10.5194/acp-14-7125-2014, 2014.