Content uploaded by George Kallos
Author content
All content in this area was uploaded by George Kallos on Sep 26, 2017
Content may be subject to copyright.
This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/authorsrights
Author's personal copy
Natural and anthropogenic aerosols in the Eastern Mediterranean and
Middle East: Possible impacts
G. Kallos ⁎, S. Solomos, J. Kushta, C. Mitsakou, C. Spyrou, N. Bartsotas, C. Kalogeri
University of Athens, Division of Physics of Environment —Meteorology, Atmospheric Modeling and Weather Forecasting Group (AM&WFG), University Campus, Building PHYSICS V,
Athens 15784, Greece
HIGHLIGHTS
•A modeling study on aerosol–cloud–precipitation interactions is presented.
•Formation of CCN and IN is examined with respect to aerosol properties.
•Semi and indirect effects of aerosols are more important than direct ones.
•Air quality model performance improved by incorporating new mechanisms.
•Integrated modeling study showed the benefits of using fully coupled atmospheric/air quality models.
abstractarticle info
Article history:
Received 24 July 2013
Received in revised form 10 October 2013
Accepted 7 February 2014
Available online 12 March 2014
Keywords:
Aerosol
Radiation
Cloud
Precipitation interactions
Atmospheric modeling
Microphysics
The physical and chemical properties of airborne particles have significant implications on the microphysical
cloud processes. Maritime clouds have different properties than polluted ones and the final amounts and types
of precipitation are different. Mixed phase aerosols that contain soluble matter are efficient cloud condensation
nuclei (CCN) and enhance the liquid condensate spectrum in warm and mixed phase clouds. Insoluble particles
such as mineral dust and black carbon are also important because of their ability to act as efficient ice nuclei (IN)
through heterogeneous ice nucleation mechanisms. The relative contribution of aerosol concentrations, size dis-
tributions and chemical compositions on cloud structure and precipitation is discussed in the framework of
RAMS/ICLAMS model. Analysis of model results and comparison with measurements reveals the complexity of
the above links. Taking into account anthropogenic emissions and all available aerosol–cloud interactions the
model precipitation bias was reduced by 50% for a storm simulation over eastern Mediterranean.
© 2014 Elsevier B.V. All rights reserved.
1. Introduction
The Mediterranean region and Middle East are well known places of
high aerosol and ozone concentrations. The characteristic paths and
scales of transport and thechanges in the physical and chemical proper-
ties of the aerosols along the followed paths have been the subject of
several past studies (e.g. Kallos et al., 1998, 2007). Aerosol levels have
several impacts on various gaseous pollutants but the most important
effects are associated with radiation, clouds and precipitation (direct
and indirect effects). Airborne particles of natural or anthropogenic or-
igin, may act as efficient cloud condensation nuclei (CCN), depending
on their concentrations, their size distributions and their chemical com-
position. Some particles —suchas mineral dust and soot —may also act
as ice nuclei (IN) and contribute in the formation of ice particles in high
clouds. For example, sea salt, sulphates, and nitrates (soluble aerosols),
are responsible for the formation of cloud droplets as the air rises and
the relative humidity increases to slightly above saturation near the
cloud base (Levin and Cotton, 2009). Moreover, therole of natural or an-
thropogenic particles, such as mineral dust and black carbon, as IN has
been demonstrated in several studies (e.g. Pruppacher and Klett,
1997; Liu et al., 2009). There is increased evidence that the suspension
of sulphate and soot particles produced during fossil fuel use and bio-
mass burning increases the ice number concentrations through ice
nucleation mechanisms (Penner et al., 2009). On the other hand, forma-
tion of secondary particles andatmospheric aging of aerosol lead to par-
ticles with substantially different properties than those at source
regions. For example, desert dust particles being initially not very solu-
ble and ineffective CCN can become coated with soluble material turn-
ing them into effective gigantic CCN (GCCN) (Levin et al., 2005). The
physical processes and interactions that define the above links can be
addressed within the framework of an integrated atmospheric and air
pollution model. In this study, several experimental runs were
Science of the Total Environment 488–489 (2014) 3 89–397
⁎Corresponding author. Tel.: +30 2107276835; fax: +30 2107276765.
E-mail address: kallos@mg.uoa.gr (G. Kallos).
URL: http://forecast.uoa.gr (G. Kallos).
http://dx.doi.org/10.1016/j.scitotenv.2014.02.035
0048-9697/© 2014 Elsevier B.V. All rights reserved.
Contents lists available at ScienceDirect
Science of the Total Environment
journal homepage: www.elsevier.com/locate/scitotenv
Author's personal copy
performed with an enhanced version of the RAMS/ICLAMS model ini-
tially described in Solomos et al. (2011) focusing mostly on the amounts
and chemical compositions of the available airborne particles that could
be activated as CCN or IN for each particular case. A quick description of
this modeling capabilities and configuration is provided in Sections 2.1
and 3.1. Theeffects on precipitation and the microphysical structure in-
side the clouds are discussed with regard to the various types of air
masses and aerosol mixtures.
2. Effects of air quality on cloud development
2.1. Model configuration
The model includes an advanced microphysical scheme (Meyers
et al., 1997) with eight categories of water (vapor, cloud droplets, rain
droplets, pristine ice, snow, aggregates, graupel and hail) and also an in-
teractive mineral dust and sea salt cycle, biogenic and anthropogenic pol-
lutant emission/transport/depletion processes, gas and aerosol chemical
reactions. The radiative transfer scheme in the model (RRTM —Mlawer
et al., 1997; Iacono et al., 2000) includes the aerosol feedbacks on radia-
tion fluxes. Activation of CCN into cloud droplets is explicitly computed
with the scheme of Fountoukis and Nenes (2005) based on the proper-
ties of airborne particles. The formation of IN is also calculated online
with the scheme of Barahona and Nenes (2009) based on the modeled
air quality properties.
Initially the model is set up in a 2-D configuration assuming flat ter-
rain so thatthe effects of microphysics in cloudprocesses can be isolated
from the possible topographic forcing. A single sounding that is repre-
sentative of unstable atmospheric conditions is used to initialize the
runs and a warm and moist bulb is applied in the center of the domain
in order to trigger convection. Full-scale 3-D runs have been performed
and discussed in Section 3.
Conceptual 2-D model runs for twelve different mixtures of aerosol
particles are implemented, as seen in Table 1. Each run lasts for 6 h.
The distribution of the particles in the model is represented by a
three-modal lognormal distribution (fine–accumulated–coarse) with
constantgeometric dispersion (σ= 2) and modal diameters varyingac-
cording to the needs of each experiment. The chemical composition and
the aerosol concentrations also vary for each model run (see Table 1).
The airborne particles are assumed to be chemically-inert during these
simulations.
2.2. Formation of CCN from airborne particles
Activation of the different aerosol types as CCN results in significant
variation in the total accumulated precipitation as indicated in Fig. 1.
The aerosols that consist of dust particles and externally coated with a
soluble material produce similar precipitation amounts —within the
range of 300–350 mm. These results seem not to be affected by the
chemical composition of the soluble fraction as seen for Cases 1–3,
where the soluble fraction is sodium chloride (NaCl) and Cases 9–11,
wherethesolublematerialisammoniumsulphate(NH
4
)
2
SO
4
.Howev-
er, by increasing the hygroscopicity of the particles (the percentage of
soluble material), the accumulated precipitation is reduced (i.e. Case
3, Case 11). This is due to the increased number of cloud droplets
resulting in slower auto-conversion rates of cloud to rain sizes. In
Cases 6–8, the particles are assumed to be completely soluble (NaCl)
with different size distribution characteristics (see Table 1). The results
indicate aslight reduction in total precipitation for the aerosols with the
greater particle diameters per size mode that is possibly related to the
formation of bigger cloud and rain droplets and the suppression of the
precipitation from the ice phase of the cloud.
An interesting result is the almost 100% increase in total precipita-
tion that is found in Cases 4–5 (completely soluble particles) compared
to Cases 1–3 (partially soluble particles). In both Case 4 and Case 5 runs,
the precipitation rates remain relatively high even during the latest
stages of cloud development. This is an indication of significant contri-
bution of ice processes to the overall precipitable water. In order to illus-
trate the differences that are attributed solely to chemical composition
we selected to compare the cloud properties between Case 4 and Case
12 as depicted in Fig. 2. These cases are representative of totally soluble
aerosol particles with the same size and concentration properties but
with different chemical characteristics, namely NaCl (Case 4) and
(NH
4
)
2
SO
4
(Case 12). As seen in Fig. 2a, after the first modeling hour
the precipitation rate for Case 12 falls below 4 mm/h while after 3 h
the rainfall has been ended. In contrary, the rate remains very high for
Case 4 (blue line in Fig. 2a). These results illustrate the crucial role of
the chemical properties (Table 2) in the temporal evolution of the mi-
crophysical processes. For example, the effectiveness of ammonium sul-
phate particles to form CCN results in increased cloud droplet
concentrations throughout the modeling period (Fig. 2b). This
Table 1
Aerosol characteristics for the twelve modeling scenarios.
Case Chemical composition Soluble fraction Concentration (cm
−3
)Meandiameter(fine–accumulated–coarse) (μm)
1 Dust + NaCl 0.2 1000 0.02–0.2–2
2 Dust + NaCl 0.5 1000 0.02–0.2–2
3 Dust + NaCl 0.7 1000 0.02–0.2–2
4 NaCl 1.0 1000 0.02–0.2–2
5 NaCl 1.0 2000 0.02–0.2–2
6 NaCl 1.0 1000 0.05–0.2–2
7 NaCl 1.0 1000 0.02–0.5–2
8 NaCl 1.0 1000 0.02–0.2–5
9 Dust + (NH
4
)
2
SO
4
0.2 1000 0.02–0.2–2
10 Dust + (NH
4
)
2
SO
4
0.5 1000 0.02–0.2–2
11 Dust + (NH
4
)
2
SO
4
0.7 1000 0.02–0.2–2
12 (NH
4
)
2
SO
4
1.0 1000 0.02–0.2–2
Fig. 1. Domain total ac cumulated (6 h) precipitation (mm) for the tw elve aerosol
scenarios.
390 G. Kallos et al. / Science of the Total Environment 488–489 (2014) 389–397
Author's personal copy
suppresses the formation of precipitable-size rain droplets due tothe in-
creased competition between smaller droplets for the available
moisture.
Moreover, the most significant difference between these two runs
comes from the ice stage of the cloud. As seen in Fig. 2c the cloud up-
drafts in Case 4 are significantly higher than in Case 12. This cloud con-
tains more ice near the cloud top (Fig. 2d). Melting and rimming of the
frozen elements invigorated the production of graupel and hail at the
middle and higher cloud layers and the major part of precipitation dur-
ing the 4–6 h of simulation is generated from this stage of the cloud. Ad-
ditionally, one more run is performed for the same air quality properties
as in Case 4. In this runthe initial conditions are slightly changed and the
dew point temperature is reduced by 1 °C in the initial sounding. The
total accumulated precipitation in this case is more than two times
lower thanin Case 4. Such results imply the importance of the synerget-
ic effects between air quality and meteorology since a minor change in
any of the two can lead in significant precipitation variability.
2.3. Formation of IN from airborne particles
The interplay between air quality and high clouds has also been test-
ed within the framework of the model, assuming an initial sounding
that is representative of a cold cloud structure. The amount of ice parti-
cles that will activate during cloud formation depends on the IN concen-
tration, atmospheric conditions and also on the competition between
homogeneous and heterogeneous ice processes. For example, by
considering twelve different concentrations of dust or soot particles
that can be activated as IN, the respective accumulated precipitation
performs great variance as seen in Fig. 3a. In general, the precipitation
remains similar for both species until an aerosol concentration
of 50 μgm
−3
. After this threshold, the results vary considerably. Maxi-
mum precipitation values are found for the 500 μgm
−3
and for the
1000 μgm
−3
of soot and dust particles respectively. A significant
amount of precipitation for the 1000 μgm
−3
of soot scenario is hail.
Further increase of the aerosol concentrations results in less precip-
itation as these clouds contain great amounts of small ice elements and
finally burn off before these condensates manage to grow up to precip-
itable sizes. The sensitivity of precipitation towards IN properties is also
tested for four different IN spectra namely: MY92 (Meyers et al., 1992),
PDG07 (Phillips et al., 2007), PDA08 (Phillips et al., 2008), and CNT
(Pruppacher and Klett, 1997; Barahona and Nenes, 2008). These distri-
butions have been widely used in similar studies in the past. As seen in
Fig. 3b, explicitly resolving of the competition between homogeneous
and heterogeneous freezing results in about 25% more precipitation
a b
dc
Fig. 2. a) Maximum precipitation rate (mm/h). b) Maximum cloud droplet concentration (cm
−3
). c) Maximum updrafts (m/s). d) Maximum ice concentration (cm
−3
) for Case 4 (blue
lines) and Case 12 (red lines).
Table 2
Chemical properties of the aerosol soluble fraction.
Density
(kg m
−3
)
Molar mass
(kg mol
−1
)
Van't Hoff factor
(ions molec
−1
)
NaCl 2165 0.058 2
(NH
4
)
2
SO
4
1760 0.132 3
391G. Kallos et al. / Science of the Total Environment 488–489 (2014) 389–397
Author's personal copy
according to these conceptual runs. The precipitation rate is similar for
all IN spectrums during the first modeling hour (Fig. 3c). During this
stage all of the precipitation comes from the warm phase of the cloud.
Significant variability is only evident after 2 h of run when the
production of rain from iceprocesses becomes dominant. Similar results
regarding the variability of the condensate mixing ratios and precipita-
tion due to INvariations have also been found in earlier relevant studies
(e.g. Phillips et al., 2008; Barahona and Nenes, 2009; Teller et al., 2012).
In this context, it is expected that the proper knowledge and parameter-
ization of the air quality properties in an integrated modeling study can
lead to a better understanding of the aerosol indirect effect. Adaptation
of the above air quality/meteorology approach for the interpretation of
the physical interactions during a past event over Mediterranean is
discussed in the following section.
3. Cloud–aerosol interactions over Mediterranean
3.1. Case study description and model setup
On 29 January 2003, an event of heavy precipitation took place over
Eastern Mediterranean. Previous studies on this event (Levin et al.,
2005; Solomos et al., 2011; Teller et al., 2012) indicated that the clouds
in this area were affected by mineral dust and sea salt particles penetrat-
ing the base of the clouds (Fig. 4). A considerable percentage of these
particles was activated as CCN. The selected case is representative of a
weather pattern that is responsible for significant precipitation heights
along the Middle East coast. Such storms are essential for the water sup-
plies of the population at these areas. Moreover during this event the
MEIDEX experiment was active providing a valuable set of aircraft and
ground observations thus making this event a suitable benchmark for
testing new model developments. In the present study, this event is ex-
amined taking also into consideration the anthropogenic aerosols over
the region (sulfate particles) and their role on cloud and precipitation.
The gas, aqueous and aerosol phase chemistry mechanisms namely
SAPRC99 (Carter at al., 2003) and ISORROPIA (Nenes et al., 1998) are di-
rectly coupled with the RAMS/ICLAMS model described inSolomos et al.
(2011) and are activated for these runs. The model configuration for
these simulations is described in Table 3. The sulfate aerosol properties
are calculated in the aqueous phase chemistry sub-module of the model
and these particles are also included in the CCN activation calculations.
The aerosol size-spectrum is assumed to fit in a 5-modal lognormal dis-
tribution where the first three modes represent mineral dust, accumu-
lated and coarse sea salt particles respectively. The last two modes
represent the accumulated and coarse sulphates.
The number concentrations of all five modes are prognostic model
variables. The first mode (dust) is assumed to fit a distribution with con-
stant deviation (σ= 2) and the median diameter is recalculated at
every model step. The secondand third modes (accumulated and coarse
sea salt) have median diameters of 0.36 μm and 2.85 μm and geometric
deviations of 1.80 and 1.90 respectively. Geometric standard deviation
and geometric mean diameter for the accumulated sulfate areexplicitly
calculated. For the coarse sulfate, the geometric mean diameter is also
explicitly calculated and the standard deviation is set to 2.2.
The Israeli Mediterranean coast is one of the areas with significant
concentrations of sulfate aerosols due to major sulfur emission sources
(Luria et al., 1996). Other sources of anthropogenic pollutants in the
Eastern Mediterranean are shipping and major urban conglomerate up-
wind from the area under consideration. These sources are included in
the emission inventory used in this study. The emission inventory
used has been prepared at the framework of the EU-funded Framework
Program 6 project CIRCE (http://www.circeproject.eu). Measurements
of fine aerosol particles (diameter b1μm) during this event (Levin
et al., 2005) showed that, below 1000 m, anthropogenic sulfate repre-
sented 11.6% of the total sample. For measurements above 1000 m the
respective ratio reached 35%. The average ratio of the modeled sulfate
aerosols to the total particles is 11.9% below 1000 m and 39.6% above
1000 m indicating a good agreement with the observations. As seen
also in Fig. 5 comparison of modeled PM10 with ground measurements
at the stations of Afule (R = 0.866), Modiin (R = 0.907) and Beer Sheva
a
b
c
Fig. 3. a) Total accumulated precipitation (mm) for various concentrations of soot (black
line in μgm
−3
) and dust (red line in μgm
−3
). b) Total domain precipitation (mm). c) Ten
minutes of accumulated precipitation (mm).
392 G. Kallos et al. / Science of the Total Environment 488–489 (2014) 389–397
Author's personal copy
(R = 0.895) denotes the abrupt increase of particulate matter concen-
trations in the area during 28–29 January 2003. This increase is mostly
related to the dust event.
3.2. CCN activation of natural and anthropogenic aerosols
Three sensitivity studies were performed for this event: SIM0 —
reference simulation with no impact of aerosols on clouds, SIM1 —
simulation with impact of natural aerosols (mineral dust and sea salt)
and SIM2 —simulation with impact of natural and anthropogenic (sul-
fate) aerosols. The averaged values of hourly precipitation during 29
January 2003 show similar variation for SIM1 and SIM2. As shown in
Fig. 6a, slightly higher precipitation rates are found in the SIM2 case,
however the average 24 h accumulated precipitations do not differ sig-
nificantly (SIM1: 60.69 mm, SIM2: 63.81 mm).
The anthropogenic aerosol forcing is more noticeable in the hourly
precipitation maxima and timing (Fig. 6b) and in the dislocation of the
rainfall areas. As shown in Fig. 7a–b including the sulfate aerosols in the
computation of the cloud droplet activation results in the suppression of
cloud development over the sea. The clouds in this case exhibit more
vigorous development inland towards the eastern part of the domain. In
the next hour of the simulation the cloud formation is horizontally
constrained and vertically enhanced over the coastal zone while ice for-
mation is less favored compared to the SIM1 cloud structure (Fig. 7c–d).
Despite the differences in the horizontal and vertical structures the two
cloudsystemsreachedthesametopheight(10km)at10:00UTC.
3.3. Comparison with measurements
In order to assess the model performance for the different aerosol
scenarios, the 24-hour accumulated precipitation is compared with ob-
servations from 86 stations located over Northern Israel (Z. Levin, pers.
comm., 2010). The bias analysis of precipitation is based on the behavior
of the modeled exceedances versus observational exceedances. The
model bias is calculated as:
bias ¼aþb
aþc
where, a is the number of exceedances that was correctly modeled, b is
the number of modeled exceedances that was not observed and c is the
number of the observed exceedances missed by the model. Bias greater
than one indicates overprediction while bias less than one indicates
underprediction. Unbiased model results have bias equal to 1. The bias
scores are calculated for all three cases for 29 January 2003 (Fig. 8). In
general, the passive tracer experiment (SIM0) underestimates precipi-
tation while the two feedback runs (SIM1 and SIM2) overestimate the
precipitation at high thresholds. In order to extract a single metric for
the model performance at each aerosol scenario the bias performance
Fig. 4. Satellite image over Eastern Mediterranean on 28 January 2003, at 11:00 UTC (MODIS-Aqua). Colocation of dust and clouds is evident over SE Mediterranean (http://modis.gsfc.nasa.gov/).
Table 3
Model configuration.
Grid number Grid 1/grid 2/grid 3
Horizontal spacing 12 km/4 km/1 km
Horizontal X grid points Nx: 188/182/154
Horizontal Y grid points Ny: 116/146/106
Domain center (33° N, 28° E)/(33° N, 33° 30′E)/(33° N, 35° 10′E)
Statistics time period 29 January 2003
Cloud microphysics SIM0: no impact of aerosols on clouds
SIM1: impact of natural aerosols (mineral dust and sea
salt) on clouds
SIM2: impact of natural and anthropogenic (sulfate)
aerosols on clouds
393G. Kallos et al. / Science of the Total Environment 488–489 (2014) 389–397
Author's personal copy
index (BPI) is calculated as the sum of the absolute differences of each
thresholds biases from unity:
BPI ¼X
N
i¼1
biasi−1
jj
;
where i is the number of the precipitation thresholds. The closer the BPI
value is to zero the greater the agreement between the model and the
observed values of precipitation (threshold biases closer to 1). The re-
spectiveBPI value for each case is shown in parentheses next to the leg-
end labels in Fig. 8. Values closer to zero indicate better agreement
between model results and observations. The improvement of about
a
b
c
Fig. 5. Comparison of PM10 (μgm
−3
) measurements (blue dots) and modeled (red lines) at the stations of: a) Afule, b) Modiin and c) Beer Sheva during 27–31 January 2003.
394 G. Kallos et al. / Science of the Total Environment 488–489 (2014) 389–397
Author's personal copy
ab
Fig. 6. Time series of the hourly precipitation rate in mm/h. a) Averaged over North Israel and b) maximum values on January 29th, 2003 with natural aerosols (redline) and with natural
and anthropogenic aerosols (blue dashed line).
ab
cd
Natural Natural + Sulphates
Natural Natural + Sulphates
Fig. 7. West–East crosssection of liquidwater mixing ratio(color scale in g kg
−1
) and ice mixingratio (black line in g kg
−1
) over Haifa(latitude = 32.81, longitude= 34.99)on 29 January
2003 at a) 09:00 UTC with natural aerosols, b) 09:00 UTC with natural and anthropogenic aerosols, c) 10:00 UTC with natural aerosols and d) 10:00 UTC with natural and anthropogenic
aerosols.
395G. Kallos et al. / Science of the Total Environment 488–489 (2014) 389–397
Author's personal copy
18% that is found for SIM2 (BPI = 0.95) compared to SIM1 (BPI = 1.16),
is possibly attributed to the more detailed representation of the aerosol
field due to the inclusion of the anthropogenic aerosol. Compared to the
reference run the model performance is improved by 39% in SIM1 (only
natural sources) and by 50% in SIM2 (natural and anthropogenic
sources).
4. Conclusions
Based on the results from the conceptual model runs and the real
event analysis we can derive the following concluding remarks: Cloud
and precipitation processes are found to be very sensitive to variations
in both concentrations and chemical compositions of the airborne parti-
cles. These feedbacks are attributed to the complex role of the aerosols
in cloud microphysics for both the warm and cold stages of the clouds.
Model results indicate that even minor changes in the aerosol field
can result in considerable changes in cloud structure and precipitation.
In addition, the efficiency of several aerosols to form IN depends on
their composition. Similar concentrations and size distributions of min-
eral dust and anthropogenic soot modify the cloud structure in different
ways, thus adding to the complexity of the system.
The aerosol forcing is more evident in the spatial and temporal dis-
tribution of clouds and precipitation and less in the total amount of rain-
fall. In general, the amount of precipitation is not significantly affected
by the air quality properties since convective and stratiform precipita-
tions are mostly governed by atmospheric dynamics. However, the tem-
poral evolution of a cloud system and the corresponding time and type
of precipitation (rain, snow, hail) are found to be highly dependent on
the available CCN and IN. Certain combinations of atmospheric compo-
sition (mainly aerosols) and atmospheric conditions are likely to trigger
flood events or, in contrary, to suppress rainfall. For example, when the
ice processes are favored in the presence of dust or soot IN, the accumu-
lated precipitation was in some cases found to increase upto five times.
The sensitivity tests reveal a clear link between aerosol physical and
chemical properties and clouds but this link is not always towards the
same direction while the intensity of the associated phenomena varies
significantly.
The detailed analysis of a characteristic case in the Eastern Mediter-
ranean where both anthropogenic and natural aerosol sources were
considered resulted in a model precipitation bias that is half the bias
of the non-interactive model. Moreover, the bias is improved by 18% be-
tween the only natural and the natural and anthropogenic runs. The
CCN and IN activation schemes used in this study are based on funda-
mental physical principles. Thus the continuous improvement of the re-
sults with the inclusion of the additional information is an indication
that a number of physical processes related to the indirect effect are
now covered on a satisfactory way. Following this approach, more
work needs to be done in order to a) further improve our interpretation
of similar phenomena and b) investigate other interactions between
aerosols and clouds (for example the role of bacteria in IN that seems
to be the next important step in cloud microphysics). Additional
model runs utilizing also in situ and laboratory experimental measure-
ments are currently underway in order to improve our understanding
on the indirect effect.
Acknowledgments
This work has been carried out at the framework of the European
Union 6th Framework Program CIRCE IP, contract# 036961 and the
EUROCONTROL research studentship agreement no. CO6/22048ST.
References
Barahona D, Nenes A. Parameterization of cirrus formation in large scale models: homo-
geneous nucleation. J Geophys Res 2008;113:D1 1211. http://dx.doi.org/10.1029/
2007JD009355.
Barahona D, Nenes A. Parameterizing the competition between homogeneous and het-
erogeneous freezing in ice cloud formation —polydisperse ice nuclei. Atmos Chem
Phys 2009;9:5933–48.
Carter WPL, Sauer CG, Fitz DR, BumillerK, Bufalino C, Malkina I, et al. A new environmen-
tal chamber facility for atmospheric chemical mechanism evaluation. Presented at
the GordonConference on Atmospheric Chemistry, Big Sky Resort, Montana, Septem-
ber 7–12; 2003.
Fountoukis C, Nenes A. Continued development of a cloud droplet formation parameter-
ization for global climate models. J Geophys Res 2005;110:D11212. http://dx.doi.org/
10.1029/2004JD005591.
Iacono MJ, Mlawer EJ, CloughSA, Morcrette JJ. Impact of an improved longwave radiation
model, RRTM, on the energy budget and thermodynamic properties of the NCAR
Community Climate Model, CCM3. J Geophys Res 2000;105:14873–90.
Kallos G, Kotroni V, Lagouvardos K, Papadopoulos A. On the long-range transport of air
pollutants from Europe to Africa. Geophys Res Lett 1998;25(5):619–22.
Kallos G, Astitha M, Katsafados P, Spyrou C. Long-range transport of anthropogenically
and naturally produced particulate matter in the Mediterranea n and North Atlantic: cur-
rent state of knowledge. J Appl Meteorol Climatol 2007;46(8):1230–51. http://dx.doi.org/
10.1175/JAM2530.1.
Levin Z, Cotton WR . Aerosol pollution impact on precipi tation —a scientificreview.
Springer; 2009.
Levin Z, Teller A, Ganor E, Yin Y. On the interactions of mineral dust, sea-salt particles and
clouds: a measurement and modeling study from the Mediterranean Israeli Dust Ex-
periment campaign. J Geophys Res 2005;110:D20202. http://dx.doi.org/10.1029/
2005JD005810.
Liu X, Penner JE, Wang M. Influence of anthropogenic sulfate and black carbon on upper
tropospheric clouds in the NCAR CAM3 model coupled to the IMPACT global aerosol
model. J. Geophys. Res. 2009;114:03204. http://dx.doi.org/10.1029/2 008JD010492.
Luria M, Peleg M, Sharf G, Tov-Alper DS, Spitz N, Ben Ami Y, et al. Atmospheric sulfur
over the east Mediterranean re gion. J Geophys Res 1996;101(D20):25,917 –30.
http://dx.doi.org/10.1029/96JD01579.
Meyers MP, DeMott PJ, Cotton R. New primary ice nucleation parameterization in an ex-
plicit cloud model. J Appl Meteorol 1992;31:708–21.
Meyers MP, Walko RL, Harrington JY, Cotton WR. New RAMS cloud microphysics param-
eterization. Part II: the two-moment scheme. Atmos Res 1997;45:3–39.
Mlawer EJ, Taubman SJ, Brown PD, Iacono MJ, Clough SA. Radiative transfer for inhomo-
geneous atmosp heres: RRTM, a validated correlated-k model for the lo ngwave.
J Geophys Res 1997;102(D14):16663–82.
Nenes A, Pilinis C, Pandis SN. ISORROPIA: a new thermodynamic model for multiphase
multicomponent inorganic aerosols. Aquat Geochem 1998;4:123–52.
0.5
0
1
Bias
24610
Precipitation thresholds (mm)
16 24 36 54
Fig. 8. Bias of the 24-hour accumulated precipitation for 86stations for each threshold and for three cases of aerosol composition. The number of stations exceeding each threshold is de-
noted in parenthesis: 0.5 (86), 2 (86), 4 (86), 6 (86), 10 (84), 16 (81), 24 (57), 36 (34) and 54 (14) mm h
−1
. The BPI for each case is specified within parenthesis after the legend label.
396 G. Kallos et al. / Science of the Total Environment 488–489 (2014) 389–397
Author's personal copy
Penner JE, Chen Y, Wang M, Liu X. Possible influence of anthropogenic aerosols on cirrus
clouds and anthropogenic forcing. Atmos Chem Phys 2009;9:879–96.
PhillipsVTJ, Donner LJ, Garner ST.Nucleation processesin deep convection simulated by a
cloud-system-resolving model with double-moment bulk microphysics. J Atmos Sci
2007;64:738–61. http://dx.doi.org/10.1175/JAS3869.1171.
Phillips VTJ, DeMott PJ, Andronache C. An empirical parameterization of heterogeneous
ice nucleation f or multiple chemical species of aerosol. J Atmos Sc i 2008;65:
2757–83. http://dx.doi.org/10.1175/2007JAS2546.2751.
Pruppacher HR, Klett JD. Microphysics of clouds and precipitation. Dordrecht: Kluwer
Academic Publishers; 1997.
Solomos S, Kallos G, Kushta J, AstithaM, Tremback C, Nenes A, et al. An integrated model-
ing study on the effects of mineral dust and sea salt particles on clouds and precipi-
tation. Atmos Chem Phys 2011;11:873–92.
Teller A, Xue L, Levin Z. The effects of mineral dust particles, aerosol regeneration and ice
nucleation parameterizatio ns on clouds and precipitation. Atmos Chem Phys
2012;12:9303–20. http://dx.doi.org/10.5194/acp-12-9303-2012.
397G. Kallos et al. / Science of the Total Environment 488–489 (2014) 389–397