Overview of the development of the Aerosol Loading Interface for Cloud microphysics In Simulation

23 Pages·2015·2.77 MB·English

by  Takamichi Iguchi

Iguchietal.ProgressinEarthandPlanetaryScience (2015) 2:45
DOI10.1186/s40645-015-0075-0
REVIEW Open Access
Overview of the development of the
Aerosol Loading Interface for Cloud
microphysics In Simulation (ALICIS)
Takamichi Iguchi1,2*, In-Jin Choi3, Yousuke Sato4, Kentaroh Suzuki5 and Teruyuki Nakajima6
Abstract
Thisreview summarizes the scientific background of and past/prospective update strategies for the development of
theAerosolLoading Interface for Cloud microphysics In Simulation (ALICIS). ALICIS provides a novel approach for
coupling downscaled mesoscale cloud-resolving simulations to large-scale aerosol-transportsimulations. Realistic
aerosol loading,including spatio-temporalaerosolvariations and particle-sizespectra, is implemented inthe cloud-
resolving simulations. Priorstudiesemploying ALICIS have demonstrated how theinterfaceintroduction
significantly improves the reproducibility of thesimulated microphysical cloud structure through better
representation of aerosol effects oncloud.
Keywords: Cloud microphysics, Cloud and aerosol, Regional modeling, Dynamical downscaling
Review particles.Todate,theaerosolinfluenceoncloudhasbeen
Introduction investigatedinanumberofways,fromtheperspectivesof
Aerosoleffectsoncloudinatmosphericmodels weather and climate research, as well as in the context of
The coexistence of the three phases of hydrogen mon- weathermodificationandgeo-engineering,startingfroma
oxide is, within the solar system, unique to the Earth’s series of studies by Twomey and Squires who examined
atmosphere. The liquid and solid phases in the atmos- microphysical cloud structure under various airmass con-
phere (i.e., cloud and precipitation) play critical roles ditions using in situ aircraft measurements in the 1950s
in determining the Earth’s radiation balance and and 1960s (e.g., Squires 1956). A well-known example
hydrological cycle by driving considerable changes in demonstrating the effects of aerosol is that of the system-
the atmospheric hydrometeor distribution and charac- atic difference in microphysical properties between con-
teristics. The hydrometeor behavior is controlled by tinental and oceanic clouds. Continental clouds generally
various factors, not limited to phase changes affected have a smaller mean particle size and a higher particle
by temperature, dry-air and vapor pressure, advection number concentration than oceanic clouds. These differ-
by background winds, and gravitational falling. ences are attributed to the disparity in aerosol concentra-
Particulate matter that is dispersed colloidally with a tionandcharacteristicsbetweentheregions.
variety of the particle sizes, shapes, and chemical compo- Aerosol was not represented explicitly in most numer-
nents in the Earth’s atmosphere is referred to as aerosol. icalmodelsoftheEarth’satmosphereonglobalorregional
Aerosolparticlesarecloselyrelatedtotheliquidandsolid scales until the 1990s, despite its significance for clouds
particlesmakingupcloudsandprecipitation;theyarenot and climate. This was due in part to the fact the models
only necessary for the nucleation process of hydrometeor were generally aimed at analyses of the atmosphere’s
particles but are also a source of impurities in these dynamical structure and at land-surface rainfall predic-
tions.Researchersthusconsideredaerosolunimportantin
the basic framework of atmospheric models, because it
*Correspondence:[email protected]
hadnodirectinfluenceonmassorenergyconservationin
1EarthSystemScienceInterdisciplinaryCenter,UniversityofMaryland,
CollegePark,MD,USA the atmosphere and its fluid dynamics. In addition, com-
2Code612NASAGoddardSpaceFlightCenter,Greenbelt,MD,USA putational resources were insufficient to explicitly include
Fulllistofauthorinformationisavailableattheendofthearticle
©2015Iguchietal.OpenAccessThisarticleisdistributedunderthetermsoftheCreativeCommonsAttribution4.0
InternationalLicense(http://creativecommons.org/licenses/by/4.0/),whichpermitsunrestricteduse,distribution,and
reproductioninanymedium,providedyougiveappropriatecredittotheoriginalauthor(s)andthesource,providealinkto
theCreativeCommonslicense,andindicateifchangesweremade.
Iguchietal.ProgressinEarthandPlanetaryScience (2015) 2:45 Page2of23
aerosol. Aerosol effects on cloud were thus tacitly in- trace long-term transitions in their characteristics (e.g.,
cludedinthemodelsintheformofparametrizationscon- Aerosol Robotic Network (AERONET) Holben et al.
sisting of empirical equations. For example, saturation 1998; US Department of Energy (DOE), atmospheric ra-
adjustment (Soong and Ogura 1973; Tao et al. 1989) is a diation measurement(ARM) Stokesand Schwartz1994).
reasonable assumption for calculations of the nucleation Many intensive field campaigns have also been con-
of cloud droplets and their condensation growth from ducted to produce comprehensive views of aerosol–
water vapor. Even though the presence of aerosol is a cloud interaction by coupling observation results ob-
requirement for the droplets’ nucleation process, the tained from spaceborne satellites with ground-based,
details are not important because of the wealthof aerosol seaborne, and airborne instruments, as well as model
particles that can be activated as cloud condensation simulations (e.g., INDian Ocean Experiment (INDOEX)
nuclei(CCN)inthetroposphere. Ramanathan et al. 2001; Asian Pacific Regional Aerosol
In the 1990s and 2000s, increasing interest in aerosol Characterization Experiment (ACE–Asia) Huebert et al.
brought the issue under consideration even in the 2003; Atmospheric Brown Clouds (ABC) Nakajima et al.
context of atmospheric modeling approaches. The most 2007; Ramanathan et al. 2007; and Distributed Regional
significant turningpoint wastherecognition thataerosol Aerosol Gridded Observation Networks (DRAGON) Eck
might counter the anthropogenic gases that cause global etal. 2012).
warming through their greenhouse effect. Tropospheric
aerosol particles have a direct impact on the Earth’s Modelingapproachestoaerosolanditsactivationin
radiative balance through the scattering and absorption dropletnucleation
of short- and longwave radiation (i.e., aerosol direct Anessentialphysicalprocessconnectingaerosolwithcloud
radiative forcing). In addition, aerosol particles have an isthenucleationofhydrometeorparticles.Allaerosolparti-
indirect influence on the Earth’s radiative balance cles that can potentially be activated as nuclei of cloud
through altering optical cloud properties by serving as dropletsare referred to as condensation nuclei (CN). CCN
CCN (i.e., aerosol indirect radiative forcing). An increase areasubsetofCN;theycanbeactivatedtoinitiatedroplet
in the concentration of tropospheric aerosol particles formationatthelowsupersaturationobservedinthetropo-
may enhance cloud reflectance for solar irradiance as a sphere. In contrast, ice-forming nuclei (IN) are defined as
result of the increased droplet number concentration aerosol particles that act as the base of the deposition nu-
and reduced droplet size for a fixed liquid water path cleation of cloud–ice particles or the heterogeneous freez-
(Twomey1974, 1977;Twomey et al. 1984). On theother ing of supercooled droplets. At present, our knowledge of
hand, a decrease in droplet size may inhibit droplet INislesscompletethanthatofCCNbecauseofinsufficient
growth through collisional coagulation during warm- availability of measurements and experimental data to
rain formation; this effect has an influence on the Earth’s model the complicated ice-nucleation process (Pruppacher
cloud coverage through altering cloud lifetimes owing to andKlett1997).
thesuppressionofrainformation(Albrecht1989). A fundamental theory about CCN activation is sum-
Including the aerosol influence proved necessary in marized in a chart relating the relationship between the
assessments of the Earth’s radiation balance and, conse- vapor saturation ratio and the equilibrium droplet size
quently, the surface air temperature, not only in the formed in CCN (Köhler 1936, Appendix 1). CN size dis-
future but also in the past. Simultaneously, researchers tribution spectra, as well as ambient supersaturation and
havebecome graduallymoreinterestedinchangesinthe temperature, are essential factors in determining the
precipitation amount through modification of cloud dy- droplet nucleation process. Several additional factors
namical/microphysical structures by anthropogenic fac- have a potential impact on the balance of the vapor sat-
tors related to aerosol (Levin and Cotton 2008; Tao et al. uration ratio and the equilibrium droplet size under
2012). Many numerical models have been developed more complex assumptions, other than the basic Köhler
competitively to represent aerosol–cloud interactions theory, such as changes in the amount of dissolving sol-
and their impacts on climate and weather phenomena utes in a droplet, temperature variations through latent
on global and regional scales. In addition, remote and in heating/cooling by vapor condensation and evaporation,
situ measurements have been conducted worldwide to and the existence of insoluble and multiple soluble
provide chances for modelers to justify their results. Sat- chemicalcomponentsinimpactedCN.
ellite measurements have produced global projections of The question of how to numerically handle the two crit-
the distribution and characteristics of aerosol and clouds icalcomponents(i.e.,thesizedistributionspectraofaerosol
on the basis of measurements of their optical properties particlesandambientsupersaturation)isanimportantissue
through spaceborne remote sensors (e.g., Nakajima et al. in the representation of the droplet nucleation process in
2001; Rosenfeld et al. 2014). Continuous observations of cloud microphysics schemes. Existing approaches are gen-
aerosol by ground-based networks allow researchers to erally categorized into the following four types, according
Iguchietal.ProgressinEarthandPlanetaryScience (2015) 2:45 Page3of23
tothetemporalandspatialresolutions,andtheaerosolrep- 2012; Saleeby and van den Heever 2013; Thompson and
resentationinthemodels:(i)theuseofsupersaturationand Eidhammer 2014), whereas the aerosol concentration is
aerosolsizedistributionspectraexplicitlypredictedforeach not predicted but tacitly parametrized in some models
spatial grid point; (ii) the use of parametrization involving (e.g., Rasmussen et al. 2002; Seifert and Beheng 2006;
predicted grid-scale supersaturation or updraft wind vel- Morrison and Grabowski 2007; Seiki and Nakajima
ocity, as well as the mass and/or number concentration of 2014; Seiki et al. 2015). In most of these models, the ac-
aerosol particles; (iii) the use of parametrization including tivated CCN number concentration is parametrized on
the diagnosed maximum supersaturation or updraft wind the basis of the basic Köhler theory, using the predicted
velocityinsteadofpredictedgrid-scalevalues,aswellasthe supersaturation or vertical wind velocity. An empirical
massand/ornumberconcentrationofaerosolparticles;and parametrization was suggested by Twomey (1959a,
(iv) the use of parametrization that implicitly assumes the 1959b) to calculate the activated CCN number concen-
presenceofaerosol. tration from the predicted supersaturation or vertical
The first approach has been employed in a limited wind velocity and prescribed coefficients representing
number of cloud-resolving models (CRMs) that aim at the ambient aerosol loading. A more generalized param-
accurate calculation of cloud microphysics (e.g., Kogan etrization was proposed by Abdul-Razzak et al. (1998),
1991; Chen and Lamb 1994; Feingold et al. 1999; Khain which combines Twomey’s parametrization with a log-
et al. 2000; Geresdi and Rasmussen 2005; Lynn et al. normal representation of the aerosol size distribution
2005a; Kuba and Fujiyoshi 2006; Suzuki et al. 2006, (Ghan et al. 1993); these authors also extended the
2010; Li et al. 2008; Shima et al. 2009; Misumi et al. parameterization to cases including multiple externally
2010; Xue et al. 2010; Fan et al. 2012). The so-called mixed aerosol types (Abdul-Razzak and Ghan 2000). On
sectional representation (e.g., Abdul-Razzak and Ghan the other hand, Saleeby and Cotton (2004) proposed the
2002) is employed to numerically handle aerosol size use of pre-calculated lookup tables through Lagrangian
distribution spectra in most of these models. Köhler parcelmodelcalculations instead ofempiricalparametri-
theory is used to directly calculate the CCN number zations; aerosol size distribution spectra are explicitly
concentration and, consequently, the increased cloud represented using a sectional approach in the parcel
droplet number concentration (CDNC) from the pre- model. The lookup tables are used to calculate the pro-
dicted aerosol size distribution spectra and supersatur- portion of aerosol particles activated for droplet nucle-
ation. Consumption of aerosol particles through the ation by referring to the vertical wind velocity predicted
nucleationprocessisexplicitlyrepresentedinthesizedis- intheirmainmodelframework.
tribution spectrum predicted. Miscellaneous aerosol types The third approach is used in most global or coarse-
are often simplified to monotype CN to reduce the com- resolution regional models for studies of aerosol–cloud
plexity. These CRMs sometimes lack a grid structure of interactions, even on a climatic scale (e.g., Storelvmo et
full three-dimensional Cartesian coordinates, i.e., they al. 2006, 2008; Lohmann et al. 2007; Morrison and Get-
may have only a two- or one-dimensional grid structure. telman 2008; Suzuki et al. 2008; Takemura et al. 2009;
Largecomputationalresourcesarerequiredtoensuresuf- Wang et al. 2011; Wang et al. 2013; Park et al. 2014; Shi
ficient accuracy to explicitly resolve aerosol size distribu- et al. 2014). Supersaturation and the vertical wind vel-
tion spectra. In addition, very fine grid spacings in both ocity cannot be resolved well on model grid points with
timeandspacearerequiredtorepresentminutevariations coarse intervals. Saturation adjustment approaches are
insupersaturation. Such CRMs are generally employed to necessarily employed in the calculation of phase changes
simulate idealized and simplified clouds to investigate the between vapor and cloud water/ice. The approach used
basic mechanisms of cloud and precipitation formation, to introduce aerosol-to-cloud effects is significantly dif-
and to discuss the sensitivity to various factors, including ferentamongmodelsaccordingtothetypeof bulkcloud
aerosol concentration. Typical size distribution forms of microphysics employed. Models using double-moment
aerosol particles are often prescribed using reference bulk cloud microphysics generally employ empirical pa-
measurementdata. rametrizations (e.g., Abdul-Razzak et al. 1998; Abdul-
The second approach is employed in high-resolution Razzak and Ghan 2000) to obtain the relationship be-
regional models that can be used for aerosol–cloud tween the aerosol number concentration and CDNC.
interaction studies based on idealized experiments or Since the grid-scale vertical velocity is not a suitable in-
case studies. Generalized aerosol size distribution func- put into the parametrizations, diagnostic variables (e.g.,
tions are usually assumed instead of their explicitly re- turbulent kinetic energy) are often included in modified
solved spectra. The total or species-distinguished aerosol parametrizations to determine the activated CCN num-
concentration is explicitly predicted in bulk-moment ber concentration through diagnosis of the maxima of
forms in some models (e.g., Hashino and Tripori 2007; the vertical wind velocity and supersaturation (e.g., Mor-
Muhlbauer and Lohmann 2009; Onishi and Takahashi rison et al. 2005). In contrast, single-moment bulk cloud
Iguchietal.ProgressinEarthandPlanetaryScience (2015) 2:45 Page4of23
microphysics cannot directly incorporate the activated concentration of simplified hygroscopic aerosol particles
CCNnumberconcentrationintothemicrophysicscalcu- based on combining all aerosol species except for dust
lations in the form of a CDNC change. Diagnosed and black carbon. The second is that of simplified ice-
CDNC can be included in the calculation of the auto- nucleating particles based on the accumulation of dust
conversion rate from cloud to rain to represent the am- particles withdiameters larger than 0.5 μm. Both number
bientCNeffectonclouds(e.g., LiuandDaum 2004). concentrations were explicitly predicted in the WRF
The fourth approach is adopted in model simulations model simulations as tracers affected by interactions with
onany scale.Since aerosol particles are necessary for the hydrometeors and emission from the land surface. The
nucleation of cloud particles, the presence of aerosol predicted aerosol number concentrations were used in
particles is always assumed in cloud microphysics in a calculating the activation process of aerosol particles
steady form. Explicit representation of aerosol and its in the bulk cloud microphysics scheme. Aerosol acti-
effect on cloud may cause excess computational costs vation (i.e., cloud droplet nucleation) was calculated
and uncertainties in regular operational runs for short- using a method employing pre-calculated lookup ta-
time weather forecasting. Speedy and stable simulation bles (e.g., Saleeby and Cotton 2004). Cloud-ice nucle-
is important in these runs. Even coarse parametrizations ation and heterogeneous freezing of supercooled
based on empirical formulas work reasonably well for droplets were calculated using the parametrizations of
such short-time simulations where aerosol loading has Bigg (1953), Koop et al. (2000), Phillips et al. (2008),
little direct influence and minor feedback on the prog- and DeMott et al. (2010).
nostic variables. Shi et al. (2014) used aerosol and chemical-transport
model simulations as a tool to provide a realistically di-
Solutionstotheproblemrelatedtoaerosol-to-cloud agnosed aerosol field for their cloud microphysics calcu-
modeling lation in the main model framework with the same
The best approach to accurately simulating aerosol-to- domain configuration. They incorporated the following
cloud effects is obviously the first option discussed in two individual models off-line: the GOCART solo-
“Modeling approaches to aerosol and its activation in module version of the WRF coupled with chemistry
droplet nucleation”, i.e., using the explicitly predicted (WRF–Chem; Grell et al. 2005), and the National Aero-
supersaturation and aerosol size distribution. However, nautics and Space Administration (NASA) Unified WRF
predictions of the two components under realistic condi- (NU–WRF; Peters-Lidard et al. 2015). The activated
tionsare contradictoryproblemsinasingle-modelframe- CCN number concentration in the NU–WRF simulation
work because of scale gaps between aerosol and cloud. A was diagnosed based on the air temperature and super-
small computational domain is preferable to define the saturation on the NU–WRF side, and the mass concen-
gridspacingasfinelyaspossibletorepresentminor varia- tration of multiple aerosol species on the WRF–Chem
tions in supersaturation. In contrast, a large computa- GOCARTside. The diagnosed CCN number concentra-
tional domain is preferred tostrictly predict transitions in tion was used to modulate the autoconversion rate from
the aerosol concentration and size distribution from cloud to rainthroughCDNC diagnosis inthe framework
sourcesofaerosol-particleformation.Thelimitedcapabil- of the single-moment bulk cloud microphysics in the
ityofpresentcomputationalresourcesplacesresultsinin- NU–WRFsimulation.
sufficient domain size and grid spacing in a single-model Another compromise adopted to address this problem
framework. was suggested by Iguchi et al. (2008; hereafter IG08).
Several methods can be employed to solve the scale di- Their approach was based on dynamical downscaling of
lemma in aerosol-to-cloud modeling in the present com- realistic aerosol fields from large-scale aerosol (and
puting environment. Thompson and Eidhammer (2014) chemical) transport model simulations to small-scale
used monthly aerosol climatology data as input into their CRM simulations. The basic numerical framework of
Weather Research and Forecasting (WRF) model simula- aerosol downscaling was similar to that of dynamical
tions (Skamarock et al. 2005). The aerosol climatology downscaling for standard prognostic variables such as
data were derived from multiannual global aerosol model wind velocity, temperature, and vapor-mixing ratio. Ini-
simulations using the NASA Goddard Earth Observing tial and time-variant aerosol boundary conditions were
SystemmodelcoupledwiththeGoddardChemistryAero- prepared using four-dimensional output from the large-
sol Radiation and Transport model (GEOS–GOCART, scale simulation. However,aerosolsize distribution spec-
Ginoux et al. 2001; Chin et al. 2002; Colarco et al. 2010). tra were incompletely resolved in the large-scale aerosol
The climatological data were composed of mass-mixing transport model. This deficiency might be problematic
ratios of the five aerosol species with 0.5°×1.25° horizon- for accurate calculation of the activated CCN number
tal grid spacing. The mass-mixing ratios were converted concentration on the basis of Köhler theory on the CRM
intotwo bulknumberconcentrations.One is the number side. To deal with this problem in a reasonable manner,
Iguchietal.ProgressinEarthandPlanetaryScience (2015) 2:45 Page5of23
theauthorsproposedtheuseofanadditionalintermediate Takemuraetal.2000)withregional-scalesimulationsusing
module. The intermediate aerosol loading module com- the Japan Meteorological Agency Non-Hydrostatic Model
plementedtheinformationoftheaerosol sizedistribution (JMA–NHM; Saito et al. 2006) coupled with SBM (IG08).
spectra in the dynamical downscaling framework. The The SBM scheme originates from the corresponding part
aerosol size distribution spectra and their total number/ oftheHebrewUniversityCloudModel(HUCM;e.g.,Khain
mass concentrations should be as realistic as possible, etal.2000).ALICISwasfirstpublishedbyIG08andsubse-
such as those observed at times and locations pertaining quently updated by Sato et al. (2012; hereafter SA12) and
tothesimulations. Choietal.(2014;hereafterCH14).
The objective of this review is to summarize our devel-
opmentactivitiesandprospectiveupdatestrategiesforthe CNmodelingintheoriginalHUCM
aerosol loading interface since the publication of IG08. The representation of aerosol in the original HUCM is
The interface connects small-scale CRM simulations to significantly simplified for idealized CRM simulations
large-scale aerosol transport simulations in a dynamical (e.g., Khain et al. 1999). The CN size distribution
downscaling framework and complements information spectrum for a single chemical component is calculated
about aerosol size distribution spectra. This interface is as prognostic variables for each of the model’s spatial
currently undergoing development in order to couple it gridpoints.The chemical composition ofCNisassumed
with regional models, including a detailed cloud micro- to consist of pure ammonium sulfate. The CN particle
physical scheme, referred to as spectral-bin microphysics shape is assumed to be completely spherical. The size
(SBM; e.g., Khain et al. 2015). Since this paper aims at distribution spectrum is approximated by 33 discrete size
reviewing the aerosol loading interface in detail, detailed bins,coveringaparticleradiusrangingfrom1.23×10−3to
descriptions of SBM are not included. Such descriptions 2 μm. The representative particle radius of each CN size
canbefoundinKhainetal.(2000,2015),amongothers. bin is defined as follows. The CN particle mass of the
maximum size bin (i.e., the 33rd bin) is assumed to be
DescriptionofALICIS identical to the particle mass of the smallest droplet size
Briefoverview bin; the smallest droplet radius is assumed to be 2 μm.
Our aerosol loading module is referred to as the Aerosol The CN particle mass of each size bin is determined by
Loading Interface for Cloud microphysics In Simulation doubling the mass bins; that is, a series of CN bins are
(ALICIS).Figure1providesaschematicillustrationof how characterized in the form of a geometric progression of
ALICIS works as an intermediate module between large- particlemasseswitharatioof2betweensubsequentbins.
scaleaerosoltransportmodelsandregional-scaleCRMs.So The CN particle radius of each size bin is tied to the par-
far,thisinterfacehasbeenemployedintheframeworkcon- ticlemassofthebinandtheCNparticledensity.
nectingglobal-scalesimulationsusingthespectralradiation TheinitialCNsizedistributionspectrumintheHUCM
transport model for aerosol species (SPRINTARS; e.g., is determined using the Köhler equation (Eq. 11) and
Fig.1Conceptualdiagramofthecouplingframeworkbetweenalarge-scaleaerosoltransportmodelandaregional-scalecloudresolvingmodel
usingtheALICISmodule.Graycomponentshavenotyetbeenimplementedbutwillbeincludedinfutureupdates
Iguchietal.ProgressinEarthandPlanetaryScience (2015) 2:45 Page6of23
Twomey’sempiricalparametrizationdescribedbyN = However, most studies have employed a similarly coarse
CCN
N Sk. Here, N is the number concentration of CCN approach to the CN representation (Lynn et al. 2005a,
0 CCN
activated at supersaturation S, and N and k are 2005b; Khain et al. 2010; Fan et al. 2012; Iguchi et al.
0
environment-dependent coefficients (e.g.,Twomey 1959a, 2012a, 2012b, 2014). Such an approach may be insuffi-
1959b). Thefollowingequationfor the CNnumberdens- cient in realistic simulations, because in reality, the
ity function can be obtained by coupling both equations spatial distribution ofaerosol particles isinhomogeneous
(Khainetal.2000): and variable. The environment-dependent parameters in
Eq. 1, N and k, are determined on the basis of limited
(cid:1) (cid:3) 0
dN 4A3 1:5k reference observations (Pruppacher and Klett 1997).
CN ¼1:5N k ; ð1Þ
dðlnr Þ 0 27Br Substitution of arbitrary values for those parameters
CN CN
causes large uncertainties in determining the CN con-
where A and B are calculated from Eq. 8 for a centration and size distribution spectra that should be
temperature of 288.15 K, and the molecular weight, the consistent with their real counterparts. In particular, the
van’tHofffactor,andtheCNparticledensityareidentical uncertainties may be increased under conditions where
to those of a typical ammonium sulfate aerosol particle. anthropogenic pollution has a large impact on the aero-
The spectral shape of the initial CN size distribution is sol distribution. Aerosol–cloud interaction is of great
commontoallspatialmodelgridpoints.Theinitialspatial interest as a research problem under such conditions. In
distribution is vertically inhomogeneous, since the CN addition, the simplified vertical CN distribution may be
number density function decreases exponentially with a problematic. The vertical CN profile assumed in the
scaleheightof2kmaboveanaltitudeof2km(Khainand HUCM is based on a typical observed profile of the ver-
Sednev 1996). CN particles with radii larger than 0.6 μm tical aerosol distribution. However, severe aerosol con-
are eliminated from the size distribution spectrum in tamination is sometimes observed even in the middle
standardsimulationstopreventartificiallyrapidformation troposphere (Marenco et al. 2014); such highly concen-
oflargedropletsattheinitialsimulationstage. trated aerosol is transported from distant intensive
The governing equation of the number density concen- sources through lifting by convection and it even has a
trationineachCNbin(Eq.3.4ofKhainandSednev1996) significant impact on the local cloud properties as well
is calculated numerically for all model grid points and asontheradiation balance.
each time step. Gravitational sedimentation of CN parti- IG08 developed the aerosol loading module ALICIS to
cles is neglected. CN production in the atmosphere or on overcomethedeficienciesinCNmodelingassociatedwith
theterrainsurfaceisignored.NodirectCNsourcetermis the original HUCM. The module provides inhomogen-
assumed.TheCNsinkislimitedtoconsumptionthrough eous initial CN and time-variant lateral boundary condi-
the droplet nucleation process. Scattering and absorption tions with a specified particle-size distribution through a
ofshort-andlongwaveradiativefluxesbyCNparticlesare dynamical downscaling approach using output from glo-
notincluded,becausenoradiationprocessiscalculatedin bal aerosol transport simulations of SPRINTARS (e.g.,
the HUCM. A flat horizontal gradient of the CN number Takemura et al. 2000). The SPRINTARS model includes
density concentration is assumed at the model’s lateral thefivetypesoftroposphericaerosol,i.e.,organiccarbon-
boundaries;theCNsizedistributionspectraarethushori- aceous,blackcarbonaceous,soildust,sulfate,andseasalt.
zontally constant near the lateral boundaries. Conse- Aerosol mass-mixing ratios are predicted in the frame-
quently, CN consumed within the simulation domain is work of the atmospheric general circulation model
replenishedbyadvectionfromthelateralboundaries. (AGCM). Aerosol particle size spectra are incompletely
The microphysical process of CN is limited to activa- resolvedinSPRINTARS.
tion of droplet nucleation. The process is calculated on Aerosol transport simulations using the SPRINTARS
the basis of the following approach. If the predicted model have been validated in various ways. For example,
supersaturation at a grid point is positive, the critical ra- the simulated aerosol optical thickness and Ångström
dius of CCN is calculated based on Eq. 11 using super- exponents have been compared with those retrieved
saturation and the air temperature. All CN larger than from the space-borne Advanced Very High Resolution
the critical radius are immediately activated and con- Radiometer (AVHRR) measurements on a global scale
vertedintoclouddroplets,withtheircorrespondingradii (Takemura et al. 2000); these authors also evaluated the
calculated using Eq. 10. Hygroscopic growth of CN is surface aerosol concentration at multiple global loca-
neglected. tions by comparison with those estimated from optical
measurements of the AERONET network. The single-
AnearlyversionofALICIS:Iguchietal.(2008) scattering albedo of simulated aerosol was also validated
Recently, regional models using SBM have been widely using AERONET data (Takemura et al. 2002). An aero-
applied, even to realistic simulations resolving clouds. sol data assimilation system has been developed on the
Iguchietal.ProgressinEarthandPlanetaryScience (2015) 2:45 Page7of23
basis of a four-dimensional variational data assimilation portion of the sulfate aerosol in the CN size distribution
method (Yumimoto and Takemura 2013). The vertical spectrum is also given by Eq. 2 in units of the number
aerosol distribution was validated through a compari- density concentration, but for parameters typical of
son of the aerosol extinction with lidar measurements sulfateaerosol; rm=0.0695 μmandσ=2.03 isassumed.
at several locations in Japan (Goto et al. 2015). In The particle size distribution function of sea salt aero-
addition, the SPRINTARS model is included in the sol is assumed to be a power law. The corresponding
Aerosol Comparisons between Observations and fraction of sea salt aerosol in the CN number density
Models (AEROCOM) project (e.g., Kinne et al. 2006). concentration isgivenby
Model performance has been evaluated by comparing (cid:1) (cid:3)
tmhoeduglloesb.al modeling outcomes of numerous aerosol ddNðlnCNrðSAÞÞ ¼1:5NbðSAÞk BBSArmr3ðSAÞ k; ð3Þ
CN CN CN
IG08 determined the CN size distribution spectrum of
the initial and time-variant lateral boundary conditions where r =0.1 μm and k=0.6 is assumed. The
m(SA)
in their nested model simulations through the following subscript SA denotes sea salt aerosol. This equation also
process. Among the five types of tropospheric aerosol, includes adjustment of the difference in the B coefficient
organic carbonaceous, sulfate, and sea salt aerosol parti- betweenCNandsea salt aerosol.
cles are assumed to be CN activatable as CCN. Bulk In the version of ALICIS presented here, the CN size
number concentrations of the three types of aerosol in spectrum is discretely approximated by 13 size bins with
SPRINTARS are calculated based on the predicted a radius ranging from 10−3 to 1 μm. The use of a small
mass-mixing ratios for the grid points of the global number of CN bins compared with the 33 bins in the
simulation under the assumption that all aerosol parti- original HUCM yields improved computational effi-
cles are dry. The calculated bulk number concentrations ciency. The CN concentration in each bin is numerically
are then temporally and spatially interpolated at the grid computed every time step, using almost the same gov-
points of the simulation domain of the nested model. erning equations asthestandard scalartracers.
The interpolated bulk number concentrations are The chemical component of CN in the nested model
converted into aerosol size distribution spectra at each can be arbitrarily determined by substituting a proper,
gridpoint.The conversionemploysdifferentbuilt-insize corresponding value into the B coefficient of CN in the
distribution functions, according to aerosol type. Finally, conversion equations (Eqs. 2 and 3). No microphysical
accumulating the three-type aerosol size spectra yields processes causing a change in the CN size spectra are
the CN size distribution spectrum at a given grid point assumed, except for activation to droplet nucleation.
forthe initialandboundaryconditions. Droplet nucleation is calculated using the same ap-
The particle size distribution of organic carbonaceous proach as that employed in the original HUCM; the
aerosol is assumed to be a lognormal function with a value of the B coefficient of CN in Eqs. 2 and 3 is
single mode. The corresponding fraction of organic substituted into Eqs. 10 and 11 to determine the critical
carbonaceousaerosolintheCNsizedistributionspectrum CCN radius and the corresponding radius of the gener-
isgiveninunitsofthenumberdensityconcentration, ated cloud droplet. Although atmospheric radiation is
2 8 (cid:5) (cid:6)9 3 calculatedinthenestedmodel,absorptionandscattering
ddNðlCnNrCðONCÞÞ¼pNffi2ffibffiπffiðffiffiOσCOÞCexp64−12<:ln BC1=N3=B1O=σC3O⋅rCCN=rmðOCÞ =;275; oTaefhrteohseroelraassdopineactfiioveesr ftalhrueixseeislsimbthyinaCtaNttehdearooenpnctoiectatilhnepcrlmuodpueeldrtitpiaeltespoarfeerseoeasncothl.
ð2Þ types have been bundled into standardized CN. In
addition, the other two types of aerosol (i.e., black car-
where N is the bulk aerosol number concentration of bonaceous and dust aerosol) are not included in the
b
the SPRINTARS simulation, and r and σ are the mode nested model, so that this version of ALICIS has a lim-
m
radiusandthestandarddeviationofthelognormalaerosol ited capability to simulate direct aerosol effects on at-
sizedistribution,respectively.Thesubscripts,CNandOC, mosphericradiation.
distinguish the variables pertaining to CN and organic IG08 demonstrated the validity and effectiveness of the
carbonaceous aerosol, respectively; r =0.1 μm and aerosol loading module in model simulations of two pre-
m(OC)
σ =1.8 is assumed. Note that Eq. 2 applies the adjust- cipitation events observed during the 3rd Experiment of
OC
ment of the difference in the B coefficients (Eq. 8) be- the Asian Atmospheric Particulate Environment Change
tween organic carbonaceous aerosol and CN with a Studies (APEX–E3). First, vertical profiles of the CCN
certainsolublecomponent. number concentration in the nested model simulations
Sulfate aerosol is also assumed to have a single-mode were evaluated in direct comparison with those acquired
lognormal particle size distribution. The corresponding by aircraft in situ measurements. The measurement
Iguchietal.ProgressinEarthandPlanetaryScience (2015) 2:45 Page8of23
methods of the CCNnumber concentration using a CCN simulation assuming horizontally homogeneous initial
counterweresummarizedbyIshizaka(2004)andAdhikari and boundary CN conditions failed to reproduce the
et al. (2005); some additional detailed information about structure of the north–south gradients (Fig. 3b). The
themeasurementson the eventdayscan alsobe found in simulated profiles of the average droplet effective radius
IG08.Figure2comparestheresultsforthetwoindividual are almost identical for the northern and southern
cases. The simulations reproduced the observed CCN halves. The simulated north–south gradients in the
concentration profiles reasonablywell. In the April 2 case control run were thus caused by a horizontally inhomo-
(Fig. 2a), the observed CCN concentration ranged from geneous CN distribution introduced by aerosol down-
approximately several tens to close to 1000/cm3. The ob- scaling. This result proved that the implementation of
served concentration decreased a little with increasing realistic, inhomogeneous CN fields is essential for more
altitude. The three-type simulated CCN concentrations accurate simulations of cloud microphysical properties.
rangedfromapproximately100to1000cm−3atheightsof On the other hand, the simulation assuming a doubled
up to 5500 m; the small decrease with increasing altitude CN number concentration compared with the control
wasgenerallyreproduced.IntheApril8case(Fig.2b),the run had smaller effective radii (Fig. 3c). The simulated
observed CCNconcentration wasapproximately1000cm effective radius profiles were in better agreement with
−3atheightsbelow1000m,whichdecreasedwithincreas- the observed profiles than those in the control run. This
ing altitude to roughly 100 cm−3 at 3000 m. The vertical result indicates that the control run underestimated the
profiles of the simulated CCN concentrations showed a CCNnumber concentration.
log-linear decrease with increasing altitude, similar to the
observedprofile.
Second, the horizontal distribution of the liquid cloud UpdatebyChoietal.(2014):Aerosolparticlesize
properties in the nested model simulations was com- distributionrevisionandtypemultiplication
pared with that estimated from data of the Modulate The IG08 version of ALICIS still contained many limita-
Resolution Imaging Spectroradiometer on the Terra tions. Among them was the absence of two aerosol spe-
satellite (Terra/MODIS) based on the satellite’s retrieval cies, i.e., black carbonaceous aerosol and dust particles.
algorithm (Nakajima and Nakajima 1995; Nakajima et al. In reality, they often work as CCN if soluble compo-
2005); plots of the horizontal distribution are shown in nents are combined with the particles, either externally
Fig. 6 of IG08. Figure 3 shows the relationship between or internally (e.g., Sullivan et al. 2009). In addition, the
the liquid cloud-top temperature and the droplets’ ef- modeling of the other three types of aerosol still offered
fective radius near the cloud tops in the satellite mea- room for significant improvements. For example, the ap-
surementsand thesimulation for theApril 2case. These plication of a power-law function to the size distribution
parameters were averaged separately in the northern and spectra of sea salt aerosol might cause an overestimation
southern halves of the simulation domain to highlight of the activated CCN number concentration because of
spatial differences in the droplet radius distribution. The the continuous increase in the CCN number concentra-
north–south difference is likely to have resulted from a tion withincreasing supersaturation.
horizontal bias in the CN concentration. The model The aerosol loading module was comprehensively up-
simulation performed satisfactorily in reproducing the dated by CH14. The conversion from the various types
overall tendency of the north–south gradients of the of aerosol to standardized CN, as implemented by IG08,
observed droplet radius in warm-cloud regimes with has been removed. The same five aerosol species used in
temperatures in excess of 273 K, despite some overesti- the SPRINTARS model are employed in nested models.
mation (Fig. 3a). The observed radius monotonously in- Thediscretizationofthe sizedistribution spectra ofeach
creased with decreasing cloud-top temperature. The aerosol type has been extended to 17 size bins with radii
observed radii averaged in the northern half had larger ranging from 10−3 to 10 μm. The use of more size bins
values than those in the southern half; the maximum and a wider range of particle sizes compared with those
north–south difference was roughly 5 μm. The simu- in the old version allows the representation of aerosol
lated radius generally increased with decreasing cloud- particles with radii larger than 1 μm. Such large aerosol
top temperature down to 275 K and roughly decreased particles are possibly included in the sea salt and dust
or remained constant below 275 K. The simulated radii aerosol categories(Fig. 4).
in the northern half attained larger values than those in The size distribution function of each aerosol type was
the southern half for temperatures greater than 275 K; assumedtobeofamulti-modallognormalformbyCH14.
the north–south difference was roughly 5 μm on aver- Thesizedistributionspectradescribed in thismannerare
age. In addition, two sensitivity simulations were con- consistent with those observed in reality (e.g., Brechtel
ducted to highlight the effects of the aerosol loading et al. 1998). The number density concentration of the
module on the simulation results (Fig. 3b, c). The size distribution of each aerosol type is given by
Iguchietal.ProgressinEarthandPlanetaryScience (2015) 2:45 Page9of23
(a) April 2, 2003
Flight Measurement Sim (cloud-free; 30.5N)
Sim (control; 30.5N) Sim (cloud-free; 31.5N)
Sim (cloud-free; 32.5N)
6000
5000
4000
)
m
(
e
d 3000
u
Altit
2000
1000
0
1 10 100 1000 10000
CCN number concentration (cm-3)
(b) April 8, 2003
Flight Measurement
Simulation (cloud-free; 30.5N)
Simulation (control; 30.5N)
3500
3000
2500
)
m
2000
(
e
d
u
Altit 1500
1000
500
0
1 10 100 1000 10000
CCN number concentration (cm-3)
Fig.2VerticaldistributionoftheobservedandsimulatedCCNnumberconcentrationsovertheEastChinaSeaonaApril2,2003andbApril8,2003.
TheobservedCCNnumberconcentrationintheflightmeasurementswasobtainedforliquid-phasesupersaturationrangesof0.07–0.22%onApril2
and0.09–0.32%onApril8.ThesimulatedCCNnumberconcentrationintheregionalmodelwascalculatedforaliquid-phasesupersaturationof0.1%
intheApril2caseand0.2%intheApril8case.Linesofsimulatedconcentrationshowverticalprofilesofhorizontalaveragesinthreedomainsof
radius±0.5°centeredat129.5°Eand30.5°N,31.5°N,or32.5°N.Thetwotypesofsimulatedconcentrationareplottedseparatelyfromthecontrol
andcloud-freeruns(whereallcloudmicrophysicalprocesseswereswitchedofftopreventCNconsumptionthroughdropletnucleation).Adapted
fromIguchietal.(2008)
Iguchietal.ProgressinEarthandPlanetaryScience (2015) 2:45 Page10of23
(a) Fig.3Dependenceofdropleteffectiveradiinearcloudtopson
Droplet effective radius near cloud top (um)
cloud-toptemperaturecalculatedbasedonsimulationswiththree
0 5 10 15 20 25 30 differentsettingsandfromtheretrievalresultsderivedfromTerra/
250 MODISmeasurementsovertheEastChinaSeaonApril2,2003.
Simulationprofilesarecalculatedinathecontrolrun,btherun
255
characterizedbyahomogenousCNfield,andctherunwith
doubledCN.Plottedradiiareaveragedindividuallyinthenorthern
260
K) andsouthernhalvesoftheanalysisdomain.AdaptedfromIguchi
e ( 265 etal.(2008)
ur
at
er 270
p
m
oud top te 227850 dðdlnNraÞ¼Xin¼m1 NpSffi2ffiPffiπffiRffiffiσfiiexp(cid:1)−lnra2−σlin2rmi(cid:3); ð4Þ
Cl 285 MODIS (north)
MODIS (south) where r is the aerosol particle radius, i the index of a
290 Sim (north) a
given mode, n the total number of modes in the size
Sim (south) m
295 distribution, N the bulk number concentration of
SPR
each aerosol type obtained from the SPRINTARS simu-
(b) Droplet effective radius near cloud top (um) lation, and f the weight factor required to normalize the
0 5 10 15 20 25 30 multi-modal distribution. The variables n , f, r , and σ
m m
250
are the functions of aerosol type as listed in Table 1 of
255 CH14; the values have been determined on the basis of
observational results (d’Almeida et al. 1991; Chuang
260
K) et al. 1997; Penner et al. 1998; Herzog et al. 2004).
e ( 265 Figure 4 compares the aerosol size distribution spec-
ur
at tra. Two size distribution spectra were calculated for
er 270
mp each aerosol type using the assumptions employed by
p te 275 CH14 and IG08. No size distribution was assumed for
d to 280 dust or black carbonaceous aerosol in the study by
ou MODIS (north) IG08. A set of observed spectra was obtained from in
Cl 285 MODIS (south) situ measurements during the ACE–Asia field cam-
290 Sim (uni-CN, north) paign in 2001 (Huebert et al. 2003). The set of spectra
Sim (uni-CN, south) was calculated from the size-segmented aerosol mass
295 concentration measured using the Micro-Orifice
(c) Uniform Deposit Impactor (MOUDI) onboard a re-
Droplet effective radius near cloud top (um)
search vessel; the marine aerosol sampling performed
0 5 10 15 20 25 30
during ACE–Asia 2001 is summarized by Mochida
250
et al. (2007). Another set of observed spectra was
255 obtained from ground-based measurements using
MOUDI at the Gosan site on Jeju Island (Republic of
260
K) Korea) during the Atmospheric Brown Cloud–East
ure ( 265 Asian Regional Experiment (ABC–EAREX) in 2005
erat 270 (Nakajima et al. 2007).
p
m In Fig. 4, the size distribution spectra employed by
p te 275 CH14arein betteragreementwiththosefromthemea-
o
d t 280 surements, although the observed size spectra lack
u
Clo 285 MODIS (north) multi-modal distributions. The observed aerosol num-
MODIS (south) ber densities monotonously decrease with increasing
290 Sim (doubled CN, north) aerosol particle size. The observed size distributions of
Sim (doubled CN, south) sulfate and organic/black carbonaceous aerosol extend
295 to the radii of 2 μm, whereas those of dust and sea salt
aerosol extend to over 5 μm. The spectra of CH14 re-
produce the distinct structure of the observed spectra
according to aerosol type reasonably well, especially