A Method for Forecasting Cloud Condensation Nuclei Using Predictions … · 2017-09-29 · A Method...

A Method for Forecasting Cloud Condensation Nuclei Using Predictions ofAerosol Physical and Chemical Properties from WRF/Chem

DANIEL WARD

Department of Earth and Atmospheric Science, Cornell University, Ithaca, New York

WILLIAM COTTON

Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado

(Manuscript received 1 September 2010, in final form 8 February 2011)

ABSTRACT

Model investigations of aerosol–cloud interactions across spatial scales are necessary to advance basic

understanding of aerosol impacts on climate and the hydrological cycle. Yet these interactions are complex,

involving numerous physical and chemical processes. Models capable of combining aerosol dynamics and

chemistry with detailed cloud microphysics are recent developments. In this study, predictions of aerosol

characteristics from the Weather Research and Forecasting Model with Chemistry (WRF/Chem) are in-

tegrated into the Regional Atmospheric Modeling System microphysics package to form the basis of a cou-

pled model that is capable of predicting the evolution of atmospheric aerosols from gas-phase emissions to

droplet activation. The new integrated system is evaluated against measurements of cloud condensation

nuclei (CCN) from a land-based field campaign and an aircraft-based field campaign in Colorado. The model

results show the ability to capture vertical variations in CCN number concentration within an anthropogenic

pollution plume. In a remote continental location the model-forecast CCN number concentration exhibits

a positive bias that is attributable in part to an overprediction of the aerosol hygroscopicity that results from

an underprediction in the organic aerosol mass fraction. In general, the new system for predicting CCN from

forecast aerosol fields improves on the existing scheme in which aerosol quantities were user prescribed.

1. Introduction

Aerosols in the atmosphere have been shown to

modify the droplet distributions in clouds by acting as

cloud condensation nuclei (CCN). In this way, aerosols

contribute to the radiative properties of clouds while

also altering the distribution and amount of precipita-

tion that clouds produce. These impacts have led to the

consideration of aerosols as a significant factor in our

understanding of global climate (Andreae and Rosenfeld

2008), but also one of the most uncertain factors (Charlson

et al. 2001; Ghan and Schwartz 2007). The uncertainty

results in large part because particles in the atmosphere

interact with each other and undergo complex chemical

and physical transformations. The transformations lead

to large variability in the droplet-activating potential of

atmospheric particles across space and time.

For these reasons it is difficult to characterize the

sources and sinks of potentially cloud-active particles

with in situ observations. Instead, atmospheric models

that account for aerosol processes including emission,

phase transitions, transport, and deposition are needed

(Andreae and Rosenfeld 2008). These must then be

combined with cloud microphysics and cloud dynamics

models to investigate aerosol–cloud interactions (Feingold

and Kreidenweis 2002). In such modeling systems, the

process of droplet activation makes up the main inter-

face between the aerosol and cloud components. There-

fore, simulating basic aerosol–cloud interactions requires

an estimate of the four-dimensional distribution of CCN

number concentration Nccn as a function of the environ-

mental temperature T and supersaturation SS (Levin and

Cotton 2009).

The field of aerosol modeling has evolved somewhat

apart from the field of cloud microphysics modeling,

with coupled models, systems with combined cloud

microphysics and aerosol processes, being relatively

new. Zhang (2008) reviews the few examples of these

Corresponding author address: Daniel Ward, Dept. of Earth and

Atmospheric Science, Cornell University, Ithaca, NY 14850.

E-mail: [email protected]

JULY 2011 W A R D A N D C O T T O N 1601

DOI: 10.1175/2011JAMC2644.1

� 2011 American Meteorological Society

comprehensive models and finds that, given the com-

plexity of the processes involved, a need exists for

improvement of the model representation of aerosol

composition and aerosol–cloud interactions. In this paper,

a method for modeling the droplet activation of internally

mixed aerosols with complex composition will be de-

scribed. We use output from the Weather Research and

Forecasting Model with Chemistry (WRF/Chem) aerosol

modules and the hygroscopicity parameter k (Petters and

Kreidenweis 2007) within the droplet-activation scheme

of the latest version of the Colorado State University

Regional Atmospheric Modeling System (RAMS; Cotton

et al. 2003). This scheme in RAMS consists of activated-

fraction lookup tables that predict droplet number in a

computationally efficient manner (Ward et al. 2010)

while retaining the accuracy of an iterative parcel model

(Saleeby and Cotton 2004). The new modeling system

allows for prediction of Nccn based on more representa-

tive forecasts of aerosol characteristics, including num-

ber, size, and composition, and subsequent feedbacks on

the model microphysics and precipitation processes.

Predictions of Nccn from the new system are vali-

dated against aircraft measurements of Nccn along the

Colorado Front Range collected during the Ice in

Clouds Experiment—Layer Clouds (ICE-L) field cam-

paign (Eidhammer et al. 2010) and also ground-based

measurements of Nccn from Storm Peak Laboratory (SPL)

in northwestern Colorado. The results presented in sec-

tion 4 show improvements over the existing scheme while

also highlighting shortcomings and areas for improve-

ment. The goals of this work are twofold: 1) to present

a simple method for representing complex aerosol com-

position in coupled aerosol–cloud microphysics models

and 2) to show the value of this method as implemented in

a regional atmospheric model.

2. Model frameworks

The nonhydrostatic version of RAMS 4.3 was used for

this work. The model is set up on nested Arakawa-C

horizontal grids (Cotton et al. 2003) with a terrain-

following sigma-z coordinate in the vertical dimension.

RAMS predicts six hydrometeor categories on mixing

ratio and number concentration using bulk distribu-

tions within a bin-emulating system. This microphysics

scheme has been shown to be computationally efficient

and yet effective for simulating various cloud regimes

and conditions (Saleeby and Cotton 2008; Cheng et al.

2009; Saleeby et al. 2009).

a. RAMS droplet-activation scheme

As mentioned in the introduction, the initial interac-

tion between aerosols and clouds in the liquid phase

occurs at the point of activation where a solution droplet

grows by vapor deposition to cloud droplet size. Droplet

activation has often been studied by simulating single

aerosol-laden air parcels within an updraft (McFiggans

et al. 2006). These parcel models solve the Kohler equa-

tions of droplet growth within the rising air parcel, re-

turning a precise model estimate of the cloud droplet

number concentration (CDNC).

Parcel models have been used to address the effects

of aerosol composition on droplet activation (e.g., Ghan

et al. 1998; Antilla and Kerminen 2007; Reutter et al.

2009). To avoid the computational expense of imple-

menting a parcel model within a microphysical scheme,

larger-scale numerical models use droplet-activation

parameterizations. Several such parameterizations

have been developed that explicitly define aerosol

chemistry, although often for a limited number of chem-

ical species (e.g., Abdul-Razzak and Ghan 2000; Nenes

and Seinfeld 2003; Ming et al. 2006; Segal and Khain 2006;

Hsieh et al. 2009).

A slightly different approach to droplet activation is

used in RAMS. Saleeby and Cotton (2004) showed that

parcel model output of CDNC, which is used for vali-

dation of the majority of the parameterizations refer-

enced above, could be accessed directly by the model

microphysics using a lookup-table system. The tables

consist of the parcel-model activated-fraction results for

a range in initial parcel T, updraft speed, aerosol size

distribution, and aerosol number concentration. As these

conditions are encountered in RAMS, the appropriate

activated-fraction value is found and is applied to the

model aerosol number concentration. This scheme re-

turns the parcel model result online in RAMS with min-

imal computational expense. For more details about the

construction of the lookup tables, the reader is referred to

Saleeby and Cotton (2004) and Ward et al. (2010).

The Saleeby and Cotton (2004) parameterization has

been applied in several cloud-scale modeling studies

since its introduction (e.g., van den Heever et al. 2006;

van den Heever and Cotton 2007; Cheng et al. 2009;

Saleeby et al. 2009). In these previous studies, however,

aerosols were prescribed an initial, horizontally homo-

geneous concentration that was meant to be represen-

tative of the environment being studied. Aerosol size

distribution median radius was held fixed throughout

individual simulations, and aerosol composition was lim-

ited to a combination of 10% or 50% ammonium sulfate

and a corresponding 90% or 50% insoluble fraction

(Saleeby and Cotton 2004). For these reasons, efforts

to simulate aerosol effects on clouds were limited, for

the most part, to sensitivity studies.

The RAMS droplet-activation lookup tables were

recently extended to include an additional dependence

1602 J O U R N A L O F A P P L I E D M E T E O R O L O G Y A N D C L I M A T O L O G Y VOLUME 50

based on aerosol hygroscopicity (Ward et al. 2010). Hy-

groscopicity was measured using the k parameter, which

is determined empirically and has been reported for

many aerosol chemical components by Petters and

Kreidenweis (2007). The k parameter can be computed

for internally mixed aerosols of known composition us-

ing a simple volume-weighted average of the individual

component k values (Petters and Kreidenweis 2007).

With the added degree of freedom it is now possible to

predict CDNC on heterogeneous fields of aerosol com-

position. These fields are provided to the RAMS mi-

crophysics by the WRF/Chem aerosol modules.

b. WRF/Chem setup

WRF is a nonhydrostatic, atmospheric model that has

been developed collaboratively at several institutions.

Atmospheric chemistry was introduced into WRF by

fitting existing chemical mechanisms into the WRF

framework, creating WRF/Chem (Grell et al. 2005). In

WRF/Chem, version 3.0, separate modules treat emis-

sion of anthropogenic aerosols and precursor gases,

emission of biogenic precursor gases, gas-phase oxida-

tion, aerosol dynamics, and secondary organic aerosol

(SOA) formation. Chapman et al. (2009) recently linked

aerosol and cloud microphysics modules within the WRF/

Chem framework to form a coupled system. They use

a form of the Abdul-Razzak and Ghan (2000, 2002)

droplet-activation parameterization in their scheme.

Abdul-Razzak and Ghan (2000, 2002) included multiple-

mode capability in their droplet-activation parameteri-

zation (Abdul-Razzak et al. 1998). They use a sectional

representation of particle distributions, each with an as-

sociated composition, to capture complex combinations

of aerosol size and chemical makeup. The drawback of a

binned scheme is the substantial computational expense

required when compared with bulk aerosol parameter-

izations.

In the current study, aerosol dynamics were treated in

WRF/Chem using the Modal Aerosol Dynamics Model

for Europe (MADE; Ackermann et al. 1998). Aerosols

in MADE are represented by two lognormal size distri-

butions, corresponding to an Aitken mode and an accu-

mulation mode. The two modes overlap and interact

through coagulation. The MADE Aitken and accu-

mulation modes describe only submicrometer-diameter

particles. Supermicrometer particles are included in

WRF/Chem by the addition of an interactive coarse mode

(Schell et al. 2001).

Particles are added to the lognormal modes either by

direct emission or by secondary formation. New particle

formation in MADE is treated with the Kulmala et al.

(1998) parameterization of sulfuric acid nucleation. New

particles are assigned to the Aitken mode with a diameter

of 3.5 nm, and the size distribution parameters are ad-

justed to retain the lognormal shape of the distribution.

Simulated particles in MADE grow by condensation

and coagulation. Low-vapor-pressure gas-phase species

condense onto existing particles at a rate determined by

the vapor pressure of the species over the aerosol sur-

face. The Kelvin effect, a result of the curvature of the

aerosol surface, is parameterized by a size-dependent

growth factor in this scheme to simplify the computation

of the condensational growth rate (Ackermann et al.

1998; Binkowski and Shankar 1995). Condensation of

organic mass is included by the introduction of the Sec-

ondary Organic Aerosol Model (SORGAM) into the

MADE framework (Schell et al. 2001). Aerosol mass

can be added to existing particles by condensation, but

the particle number remains the same. During coagu-

lation of particles, the aerosol mass is conserved but the

particle number decreases. Throughout, the lognormal

shapes of the size distributions are maintained (Binkowski

and Shankar 1995).

Interspecies coagulation occurs in MADE regardless

of the particle composition, meaning internal mixtures

of multiple species are created. To limit the number of

variables and the computation time that would be re-

quired to follow the various internal mixtures, MADE

essentially assumes that all particles contain the same

proportions of the different species. In other words, a

perfect internal mixture of all aerosol mass is assumed.

Modal schemes that predict particle number and size

separately for each species have been developed (e.g.,

Lohmann et al. 1999; Lohmann and Diehl 2006), but

these are restricted to intraspecies coagulation and the

assumption of externally mixed aerosols. Neither as-

sumption, of a pure internal or external mixture, applies

universally in the atmosphere (McFiggans et al. 2006),

and in some cases an external mixture has been shown to

be more realistic (Cubison et al. 2008). However, ac-

cording to a review of recent observations by McFiggans

et al. (2006), aerosols in locations that are geographically

removed from large-particle-formation areas (such as ur-

ban centers) are likely to be at least somewhat internally

mixed.

The MADE/SORGAM and chemical mechanism

modules within WRF/Chem predict the major known

components of ambient CCN using a modal represen-

tation of aerosols, which minimizes the required com-

putation time. This setup has shown skill at predicting

anthropogenic aerosol mass, particularly in polluted areas

such as the northeastern United States (e.g., McKeen et al.

2007; Grell et al. 2005), but McKeen et al. (2007) found

that WRF/Chem forecasts of the mass of particles with

diameters of less than 2.5 mm (PM2.5) exhibited a nega-

tive bias in rural areas. The bias was attributed to the


inadequate treatment of secondary organic aerosol for-

mation and the lack of consideration of some biogenic

precursor gases.

Gas-phase chemistry in WRF/Chem was represented

with the Regional Atmospheric Chemistry Model (RACM)

developed by Stockwell et al. (1997) as an upgrade to the

Regional Acid Deposition Model, version 2 (RADM2;

Stockwell et al. 1990). Both mechanisms are available

for use in WRF/Chem, but only RACM treats the gas-

phase oxidation of the important biogenic aerosol pre-

cursor gases a-pinene and limonene. McKeen et al. (2007)

note that the neglect of these gas-phase species probably

leads to underestimates of SOA when using RADM2.

The most recent version of the National Emissions

Inventory was used in this study (Environmental Pro-

tection Agency 2010). This inventory is maintained by

the U.S. Environmental Protection Agency and includes

anthropogenic emissions for the continental United States

as well as southern Canada and northern Mexico (at larger

grid spacing). Emission of biogenic aerosol precursor

gases was treated by the Model of Emissions of Gases

and Aerosols from Nature (MEGAN; Guenther et al.

2006). MEGAN was recently added to the WRF/Chem

framework and accounts for the complex dependencies

of biogenic precursor emissions on variable meteorolog-

ical conditions and vegetation.

3. Model development

The WRF/Chem modules predict aerosol mass for

each species and one aerosol size distribution median

radius and particle number concentration containing all

species at every grid point. This information can be used

directly by the RAMS droplet-activation scheme, but

a single-value representation of the particle composition

using the k parameter is still required. Here, the as-

sumption of internally mixed aerosol in WRF/Chem is

especially helpful.

a. Computation of k

To compute single-parameter composition for the

model aerosol, k is weighted by the volume of each

component in the internal mixture and averaged. This

follows Eq. (7) from Petters and Kreidenweis (2007).

The computation requires values of density r for each

WRF/Chem aerosol species or species group to convert

the predicted aerosol mass to volume. In addition, a value

of k must be assigned to each component to compute the

weighted average. Values for both r and k can be found

for specific species in the literature, but several of the

WRF/Chem species are not so easily defined.

For example, in MADE/SORGAM the reaction prod-

ucts of aromatic compounds are represented by two

product groups. These groups presumably contain sev-

eral different species with various values of r and k.

To simplify the many possible combinations of species,

common products of aromatics for which the k is known

were chosen to represent the hygroscopicity of the whole

group. For aromatics, phthalic acid and homophthalic

acid were used. The proportions of the representative

species were, in general, derived from measurements of

Los Angeles–area (California) aerosol mass reported by

Seinfeld and Pandis (2006). Table 1 contains the repre-

sentative species selected for each model group and the

reference for each selection.

Determining the proportions of ammonium nitrate,

ammonium sulfate, and nonammoniated nitrate and sul-

fate for computation of k requires some calculation.

The interactions of these compounds in the atmosphere,

in both the particle and gas phases, are complex, and to

describe them requires a thermodynamic model (Seinfeld

and Pandis 2006). However, Seinfeld and Pandis (2006)

suggest a simple method for estimating the propor-

tions of ammonium nitrate, ammonium sulfate, and non-

ammoniated nitrate and sulfate. Ammonia (NH3) favors

neutralization of sulfate over nitrate when both are

available. In light of this, Seinfeld and Pandis (2006)

recommend assuming that all NH3 reacts first with sul-

fate. Any remaining NH3 is available to form ammonium

nitrate. This simple approach is used to compute k from

the MADE/SORGAM output. In cases in which there

is an insufficient amount of ammonium to neutralize all

of the sulfate or nitrate, nonammoniated, ‘‘extra,’’ sulfate

and nitrate are considered to be sulfuric acid and calcium

nitrate, respectively. These alterations of the aerosol

speciation are used simply for diagnosing k and do

not have an impact on the MADE/SORGAM aerosol

predictions.

Last, k and r must be defined for the two MADE/

SORGAM aerosol groups with unspecified composition.

Primary anthropogenic aerosol was assigned a value

of k 5 0.3 to represent typical continental, internally

mixed aerosol following the conclusions of Andreae

and Rosenfeld (2008). This category includes emissions

of soot (black carbon), industrial dust, and small amounts

of sulfate and nitrate particles. The primary organic

aerosol group is more difficult to characterize. Seinfeld

and Pandis (2006) reported that primary organic aerosol

collected from the Los Angeles urban area consisted

mainly of alkanoic acids and dicarboxylic acids such as

those shown in Table 1 under variable names ORGARO1,

ORGARO2, and ORGALK. The k values of these spe-

cies groups were averaged to give the k for primary or-

ganic aerosol. Values of r and k for all model species and

species groups are given in Table 2 along with associated

references.


b. Nudging scheme

WRF/Chem forecasts of aerosol number concentra-

tion, mass, size distribution median radius, and k, com-

puted using the values in Table 2, were introduced into

RAMS using an internal nudging scheme. At each RAMS

grid point, and at a user-specified time interval, the four

aerosol quantities are updated, or nudged, using the fol-

lowing formula:

Pnew(x, y, z, t) 5 Pold(x, y, z, t) 1 [Pnud(x, y, z, t)

2 Pold(x, y, z, t)]nfac, (1)

where Pold is the value of a nonspecific variable prior to

being updated by the nudging value Pnud and Pnew is the

updated variable. The nudging factor nfac is used to scale

the magnitude of the nudging and to prevent large jumps

in variable values during the nudging time step. The

aforementioned aerosol quantities are nudged for RAMS

coarse-mode and accumulation-mode aerosols. WRF/

Chem predicts aerosol number and mass for Aitken-

mode aerosols as well, but only one mode may be passed

through the RAMS droplet-activation lookup tables, and,

because of their small sizes, these aerosols are not likely

to be as important for Nccn as are accumulation-mode

aerosols. Coarse-mode aerosols are not passed through

the lookup tables but are assumed to be activated at a

100% rate in a supersaturated environment (Saleeby and

Cotton 2004). Neglecting competition for water vapor

between accumulation-mode and coarse-mode aerosols

may be physically inexact, but it was shown to be a non-

factor for predicting CDNC in parcel-model simulations

by Segal et al. (2007) for typical atmospheric concentra-

tions of coarse-mode aerosols.

WRF/Chem and RAMS use different vertical coordi-

nate systems, and so a linear interpolation scheme was

set up to accommodate the vertical level change. Be-

tween nudging time steps, aerosols in RAMS are ad-

vected by the model wind, but no aerosol dynamics or

chemical processes, including emissions, are treated.

Therefore, it is recommended that aerosol variables within

RAMS be tightly constrained to the WRF/Chem fore-

casts using a nudging factor (nfac 5 0.2) and a nudging

frequency of every 5 min, the values that were used

in this study. Higher values of nfac would better main-

tain the WRF/Chem trends in aerosol parameters, but

this would also act to magnify differences between

RAMS and WRF/Chem meteorological fields and mi-

crophysics that cause inconsistencies in the RAMS aero-

sol output.

TABLE 1. A list of the prognostic aerosol mass variables in MADE/SORGAM, and a description of the species that each variable group

includes. For the purpose of computing a constant value of k for each composition variable, each variable is broken down into repre-

sentative species with known k. References for the representative species chosen are also given.

Model species Description Representative species Reference

SO4 Sulfate aerosol mass Sulfuric acid* Seinfeld and Pandis (2006,

hereinafter SP)

NO3 Nitrate aerosol mass Calcium nitrate* SP

NH4 Ammonium aerosol mass Ammonium sulfate*; ammonium nitrate* SP

ANTHA Unspecified primary anthropogenic

aerosol mass

Internally mixed, continental aerosol Andreae and Rosenfeld (2008)

ORGARO1 Reaction products of aromatic

compounds (1)

Phthalic acid SP

ORGARO2 Reaction products of aromatic

compounds (2)

Homophthalic acid SP

ORGALK Reaction products of alkanes Diesel exhaust; palmitic acid**; stearic acid SP

ORGOLE Reaction products of alkenes Oleic acid**; adipic acid; glutoric acid;

malonic acid; succinic acid

SP

ORGBA1 Reaction products of a-pinene (1) Products of a-pinene SP

ORGBA2 Reaction products of a-pinene (2) Products of a-pinene SP

ORGBA3 Reaction products of limonene (1) Limonic acid Virkkula et al. (1999)

ORGBA4 Reaction products of limonene (2) Limononic acid Virkkula et al. (1999)

ORGPA Unspecified primary organic

aerosol mass

Alkanoic acids; dicarboxylic acids SP

EC Elemental carbon aerosol mass Diesel exhaust Weingartner et al. (1997)

SEAS Sea salt aerosol mass Sodium chloride SP

SOILA Mineral dust aerosol mass Saharan dust Koehler et al. (2009)

* The abundance of these species is determined by the availability of ammonium aerosol.

** These species are given 2 times the proportional weight relative to the other species in the group in keeping with their observed

proportions as reported by Seinfeld and Pandis (2006). This is done solely for the purpose of computing a representative value of k for

the grouped species.


c. Limitations

Predicting the aerosol field offline with respect to

RAMS carries with it several major limitations. First,

differences in meteorological conditions and cloud pro-

cesses between WRF/Chem and RAMS predictions could

lead to inconsistencies in the aerosol fields. For example,

a high concentration of particles could appear as dis-

placed from an associated convergence zone in RAMS

if WRF/Chem locates the convergence zone differently.

Large-scale disparities between the two models are un-

likely in this study because each model is constrained to

reanalysis datasets.

A similar issue results from the different RAMS and

WRF/Chem representations of cloud processes. This is

a potentially serious problem because clouds are not

nudged to reanalysis and are likely to be far more vari-

able within the model predictions. Aerosol and cloud

interactions are not limited to droplet activation but in-

clude cloud processing by aqueous-phase chemistry and

resuspension of aerosols and cloud scavenging leading

to possible wet deposition below precipitating clouds.

Wet deposition is not represented in the version of

WRF/Chem described above, which could lead to un-

derestimates of total deposition rates in some areas and

overestimates of the aerosol burden. Therefore, even if

RAMS correctly removes particles from the atmosphere

during a precipitation event, the nudging constrains the

aerosols to possibly inaccurate WRF/Chem output.

The module setup in WRF/Chem used here also lacks

aqueous-phase sulfate production. Recent evidence sug-

gests that the majority of global sulfate aerosol mass is

produced in the aqueous phase (Kanakidou et al. 2005).

To simulate aqueous oxidation, basic online interac-

tion between aerosols and cloud droplets is required.

Chapman et al. (2009) were able to estimate the quantity

of sulfate produced in cloud droplets using their coupled

scheme in WRF/Chem. On the whole, they found that

sulfate aerosol was overpredicted, although this was as-

cribed to overemission of sulfur dioxide.

To reduce potential errors due to the different cloud

processes in RAMS and WRF/Chem, the cases de-

scribed in the following section were chosen for a lack of

cloudiness and precipitation upwind of the study regions.

Online aerosol–cloud interactions would prevent these

potential errors.

Apart from the limitations of the offline setup, there

are large uncertainties in the WRF/Chem forecasts of

aerosols. Naturally produced aerosols are not included,

with the exception of some organic aerosol precursors

emitted from vegetation. The omitted sources include

biomass burning, and omitted species include precursors

TABLE 2. A list of the prognostic aerosol mass variables in MADE/SORGAM with the representative values of k and r assigned to each.

References for each assigned quantity are also given.

Model species k Reference (k) r (g cm23) Reference (r)

SO4 0.71* Petters and Kreidenweis (2007,

hereinafter PK)

1.80 Model for Simulating Aerosol Interactions

and Chemistry (MOSAIC; Fast et al. 2006)

NO3 0.51 Sullivan et al. (2009) 2.50 MOSAIC

NH4 (sulfate) 0.61 PK 1.77 Svenningsson et al. (2006)

NH4 (nitrate) 0.67 PK 1.72 Svenningsson et al. (2006)

ANTHA 0.3 Andreae and Rosenfeld (2008) 1.65 —

ORGARO1 0.051 PK 1.49 Pang et al. (2006)

ORGARO2 0.094 PK 1.49 Pang et al. (2006)

ORGALK 0.005** Virkkula et al. (1999); 0.86 Pang et al. (2006)

Raymond and Pandis (2002)

ORGOLE 0.19 PK 1.46 Pang et al. (2006)

ORGBA1 0.10 Petters et al. (2009) 1.00 MOSAIC

ORGBA2 0.10 Petters et al. (2009) 1.00 MOSAIC

ORGBA3 0.08** Virkkula et al. (1999); 1.00 MOSAIC

VanReken et al. (2005)

ORGBA4 0.08** Virkkula et al. (1999); 1.00 MOSAIC

VanReken et al. (2005)

ORGPA 0.073 PK 1.49 Pang et al. (2006)

EC 0.02** Weingartner et al. (1997) 1.7 MOSAIC

SEAS 1.28 PK 2.17 Svenningsson et al. (2006)

SOILA 0.04 Koehler et al. (2009) 2.6 MOSAIC

* A value of k 5 0.90 for sulfuric acid was mistakenly given in Petters and Kreidenweis (2007). The correct value has been found to be

between 0.68 and 0.74 by Petters, as cited by Shantz et al. (2008), and was included in the model described here after the simulations in

this study were carried out.

** Value was estimated from graphical data in the given reference.


dimethyl sulfide and isoprene, as well as particle-phase

dust and sea salt.

Additional uncertainty remains regarding the accu-

racy of the emissions inventories (Fast et al. 2006) and

the assumptions about aerosol size distribution and new

particle nucleation. Furthermore, while the modal rep-

resentation of aerosol size distribution reduces com-

putation time it also precludes our ability to simulate

complex size distributions and size-dependent compo-

sition. These complexities affect the cloud-active fraction

of a predicted aerosol population as shown by Abdul-

Razzak and Ghan (2002). Until the representation of

these processes is included or improved, forecasts of

CCN will involve large uncertainty.

4. Model evaluation

RAMS forecasts of Nccn are evaluated against two

different datasets of Nccn observations. First, aircraft

measurements of Nccn along the Front Range collected

during the ICE-L field campaign will be used to test the

ability of the model to predict vertical variations in Nccn.

In November of 2007, CCN were counted aboard the

National Center for Atmospheric Research C-130 air-

craft along the Front Range during ICE-L. The main

focus of this field project was clouds that contain ice,

but on 30 November several ‘‘missed approach’’ ver-

tical profiles were flown in clear skies at the Cheyenne,

Wyoming, Greeley, Colorado, and Fort Collins–Loveland,

Colorado, airports.

CCN data were collected during a total of six missed

approaches from the 30 November ICE-L flights between

1900 and 2100 UTC. A thermal-gradient, continuous-

diffusion CCN counter was installed on the aircraft to

record instantaneous Nccn. The instrument is similar in

design to the Droplet Measurement Technologies, Inc.,

(DMT) CCN-100 (Roberts and Nenes 2005) but was de-

signed for use on aircraft and was introduced by Hudson

(1989). The Nccn was measured at five values of SS (0.3%,

0.4%, 0.6%, 1.0%, and 1.5%). The flight tracks for all

missed approaches are shown in Fig. 1. The upward

portions of the missed approaches, which begin at the

marked airports, were used in this analysis.

a. ICE-L case

With clear conditions and southerly winds throughout

the lower troposphere, particles emitted from the Front

Range urban corridor were transported northward to-

ward Cheyenne where four of the missed approaches

took place. To better accommodate the local transport

of pollution in this November case, a horizontal grid

spacing of 4 km was used. A two-grid setup was used for

this case (Fig. 2; Table 3). Simulations were run for 60 h

from 1200 UTC 28 November 2007 to 0000 UTC 1 De-

cember 2007. Additional WRF/Chem and RAMS setup

parameters can be found in Tables 4 and 5. All times in

this and the next section will be given in UTC, with the

time of interest, 1900–2100 UTC, corresponding to

1200–1400 local time.

RAMS was set up on 35 vertical levels, with a mini-

mum spacing of 100 m at the surface. The spacing is

stretched by a factor of 1.1 for each subsequent level

until reaching a maximum spacing of 2000 m. WRF/Chem

was run with 27 vertical levels, dictated by the preprocesser.

These levels rise linearly with pressure, and in this way

they are stretched, similar to the levels in RAMS, with

higher vertical resolution in the planetary boundary

layer (PBL). Note that RAMS meteorological condi-

tions are initialized and nudged to the North American

Regional Reanalysis (NARR) dataset (Mesinger et al.

2006) instead of the high-resolution (40 km) meteo-

rological data from the Eta Model analysis used with

WRF/Chem.

Instead of using constant aerosol k and size distri-

bution median radius within RAMS, as well as the

user-prescribed particle number concentration Ncn, as

in previous studies, these variables are nudged to the

WRF/Chem predictions. Predictions of Nccn were output

FIG. 1. Flight tracks for all six missed-approach vertical profiles

taken during ICE-L on 30 Nov 2007. The airports at which each

missed approach was conducted are marked. The flight tracks are

plotted on the Nccn field (cm23) as predicted by RAMS for

1900 UTC 30 Nov at a height of 200 m AGL.


from RAMS at SS 5 0.3%, 0.4%, 0.6%, and 1.0% for

T 5 258C. RAMS diagnoses Nccn at prescribed values of

SS and T by referencing an additional lookup table that

contains the initial parcel-model updraft speed that leads

to the prescribed SS for the given T, aerosol size, and

number concentration. Only the table for k 5 0.3 is used

because k is an insignificant factor for determination of

the parcel-model maximum supersaturation. The model-

selected updraft speed and prescribed T are then used in

the droplet-activation lookup tables to select Nccn. The

predicted Nccn is strictly diagnostic and depends only on

the model-predicted aerosol properties and not on the

model-predicted environmental conditions.

COMPARISON WITH OBSERVATIONS

At 1900 UTC a substantial plume of pollution from

the Front Range urban corridor was forecast by WRF/

Chem, extending northward from 408N. The highest

particle number concentrations were being transported

at about 2500 m above mean sea level (MSL). The plume

is highly visible at 104.88W and is shown in a vertical cross

section of Ncn and accumulation-mode aerosol mass con-

centration in Fig. 3. The northward extension of the par-

ticle plume is also evident in Fig. 1, which shows the

RAMS forecast of Nccn at 200 m MSL at 1900 UTC

30 November.

The missed-approach vertical profiles were all per-

formed in or near the forecast pollution plume but

at different distances from the main sources. As a first

check, Fig. 4 shows the vertical cross section of the

RAMS forecast of Nccn and k at 1900 UTC and 104.88W.

The pattern of Nccn shows the northward transport from

sources in the Denver, Colorado, area. Values of k in

Fig. 4 are generally greater than 0.5. Above the plume,

k increases and approaches the value of pure sulfuric

acid. In areas where very little aerosol mass was trans-

ported from the surface emissions or created locally,

the initial background sulfate is all that remains. Else-

where, k exceeds typical continental values as defined

by Andreae and Rosenfeld (2008).

The Nccn measured on all six missed approaches is

compared with RAMS predictions of Nccn at SS 5 0.4%.

The measured and predicted Nccn are plotted as vertical

profiles, but note that some horizontal displacement oc-

curs over the course of the aircraft’s ascent. The RAMS

profiles take this horizontal movement into account by

displaying the nearest horizontal grid point to the air-

craft’s location at each model vertical level. Time is

treated as fixed in the model profiles for simplicity, al-

though each vertical profile took about 5 min to com-

plete in the air. The first profile shows Nccn measured

as the aircraft was ascending from the Greeley airport

(Fig. 5a). The observations suggest that the aircraft was

initially below the elevated pollution plume and en-

countered higher values of Nccn at around 2000 m MSL.

RAMS captures the magnitude of the elevated plume

well at this location, although it placed the layer of high

Nccn a few hundred meters lower than was measured.

Profiles 2–5 were measured during ascents to the west

from the Cheyenne airport. Significant differences be-

tween model output and the ICE-L measurements of

CCN are illustrated in profile 2, shown in Fig. 5b. The

aircraft apparently exited the polluted region during its

ascent. This was the only profile that did not include an

elevated layer of increased Nccn.

The elevated layer is apparent in profiles 3–5, but at

different altitudes (Figs. 5c–e). In all of these profiles,

an Nccn of ;1000–1500 cm23 in the boundary layer de-

creases to 500 cm23 or fewer with height until reaching

the pollution plume at a height of 2350 m MSL (profile

3) and 2200 m MSL (profiles 4 and 5). Observed Nccn

FIG. 2. Grid locations for the ICE-L simulations. The same horizontal

grids were used in RAMS and in WRF/Chem.

TABLE 3. Horizontal grid settings for both case-study simulations. These apply to both WRF/Chem and RAMS.

Setting SPL grid 1 SPL grid 2 ICE-L grid 1 ICE-L grid 2

Grid points (X 3 Y) 100 3 85 100 3 90 120 3 100 90 3 100

Horizontal spacing (km) 24 6 16 4

Center lat (8N) 40.0 40.0 39.0 40.5

Center lon (8W) 112.0 108.5 109.0 103.5


increased back to about 1000 cm23 in this layer before

decreasing again with height to a clean, background

concentration. The differences between profiles could

be due to the slightly different flight paths taken in each

ascent and the different times at which each was con-

ducted. RAMS predicts a similar vertical variation in

Nccn with an increase from the lowest level up to 2300 m

MSL and then a steady decrease with height above this

level. Although the height of the polluted layer is not

captured consistently in the model, RAMS predicts the

magnitudes of Nccn within the plume and especially above

the plume to within 30% of the observed values for these

profiles.

The final missed approach was carried out at the Fort

Collins–Loveland airport. The model output agrees very

closely with the observed Nccn in the lowest 500 m of

the vertical profile but fails to pick up the increased

number of CCN above 2200 m MSL that was observed

by the C-130 (Fig. 5f). The aircraft was most likely mea-

suring emissions that had dispersed from the Denver area

as far west and north as Fort Collins. WRF/Chem and

RAMS predicted the plume location to be too far to the

east at this height or, in other words, did not pick up the

extent of the particle dispersion.

b. SPL case

Observations of Nccn were collected at the Desert Re-

search Institute’s Storm Peak Laboratory near Steamboat

Springs, Colorado, from 1 to 14 April 2008. A DMT

CCN-100 was operated continuously at five values of SS

(0.1%, 0.2%, 0.3%, 0.4%, and 0.6%) and rotated the SS

every 5 min. These data were supplemented with par-

ticle number measurements and size distribution data

from a scanning mobility particle sizer (SMPS) that mea-

sured particle diameters between 0.0087 and 0.34 mm. For

model validation, the most cloud-free portion of the mea-

surement period, from 0000 UTC 1 April to 0000 UTC

5 April, was selected.

To simulate CCN at SPL during this case, both WRF/

Chem and RAMS were run for 120 h for the period from

0000 UTC 31 March 2008 to 0000 UTC 5 April 2008.

Identical horizontal polar stereographic grids were used

in both models, with a grid spacing of 6 km in the region

of interest surrounding SPL. A two-grid system was set

up to achieve this grid spacing over SPL. The outer grid

was set to a horizontal grid spacing of 24 km. The lo-

cation and extent of both grids are shown in Fig. 6, and

grid specifics are given in Table 3. RAMS was set up

with parameters that were similar to those outlined in

section 4a. These are listed in Table 5. Again, Nccn fore-

casts were output from RAMS at SS 5 0.3%, 0.4%, 0.6%,

and 1.0% at a temperature of 258C.

After RAMS ran through the 120-h simulation time,

Nccn predictions were extracted from the point on the

inner grid nearest to the latitude and longitude of SPL.

The surface elevation at this grid point was 404 m below

TABLE 4. Important settings used in WRF/Chem to provide the aerosol forecasts for the RAMS simulations described in this study.

Setting Description

Dynamical core Advanced Research WRF (ARW; Grell et al. 2005)

Microphysics Single-moment, bulk microphysics; Lin et al. (1983) scheme

Radiation Shortwave: Dudhia scheme (Dudhia and Moncrieff 1989); longwave:

Rapid Radiative Transfer Model (RRTM; Mlawer et al. 1997)

Surface boundary Monin–Obukhov (Janjic) scheme

Top boundary Rigid lid with damping layer

Time step 20 s

Meteorological nudging Eta Model analysis dataset

PBL Yonsei University (YSU) scheme (Hong and Pan 1996)

Chemical mechanism RACM (Stockwell et al. 1997)

Aerosol option MADE/SORGAM (Schell et al. 2001)

TABLE 5. Important settings used in RAMS for all simulations.

Setting Description

Microphysics Two-moment bulk as in Saleeby et al. (2009) (Meyers et al. 1997); single-cloud-droplet mode

Turbulence closure Horizontal: TKE based on Smagorinsky (1963); vertical: Mellor and Yamada (1982)

Radiation Two-stream (Harrington 1997); aerosol radiative feedback (Stokowski 2005)

Surface boundary Land Ecosystem–Atmosphere Feedback model, version 2 (LEAF-2; Walko et al. 2000)

Top boundary Rigid lid with damping layer

Time step 10 s on grid 1; 2.5 s on grid 2

Meteorological nudging NARR dataset; nudging at lateral and top boundaries at 15-min intervals


the reported MSL elevation of SPL. RAMS output from

vertical level 4 (396 m above ground level) was consid-

ered to be the best approximation to the laboratory lo-

cation and elevation. Although the laboratory is at the

surface, its location on a mountaintop means that it

often samples free-tropospheric air (Borys and Wetzel

1997), and the above-surface model level should be more

representative of free-tropospheric air.

COMPARISON WITH OBSERVATIONS

A time series of Nccn predicted by RAMS at the SPL

location is plotted with the observed Nccn at SPL in

Fig. 7. Measurements and predictions at SS 5 0.4% are

shown. RAMS captured the timing of the high-Nccn event

that occurred at and after 0000 mountain standard time

(MST) on 3 April, but, in general, RAMS overpredicted

Nccn at SPL during this case, with an average Nccn 5

492 cm23 as compared with the observed average Nccn

of 305 cm23. The most extreme example of the over-

prediction occurred on 2 April. Near 0000 MST 2 April,

RAMS predicted an event with Nccn that was nearly 6

times the observed Nccn of approximately 200 cm23. The

predicted event is short lived, with RAMS Nccn returning

to more reasonable values after about 6 h. During this

event, RAMS (nudged from WRF/Chem) forecast high

particle numbers of above 4000 cm23 that compare well

to the average observations at that time (as counted by

the SPL SMPS; see Fig. 8a), but the event occurred 2–3 h

later in the model than was observed. Moreover, the

predicted activated fraction was greater than 30% during

this event, as compared with the observed activated

fraction of about 10% (Fig. 8b). In fact, Fig. 8b shows

a general positive bias in the model activated fraction.

Because we prescribe the T and SS for model prediction

of Nccn to replicate the CCN-100 instrument conditions,

the positive bias is likely a result of overestimating the

aerosol size distribution median radius rg and/or k. Of

course, the discrepancies between model and observa-

tions result in part from the assumptions made in the

model setup, such as the simplified lognormal size dis-

tribution representation. Because the purpose of this

validation is to identify model biases that can be used to

qualify future use of this modeling system, it would be

instructive to investigate how the coupled models per-

formed with regard to prediction of rg and k.

Figure 9 shows the time series of rg for model pre-

dictions and observations. The observed rg was com-

puted by fitting a lognormal distribution to the SMPS

aerosol size bins using the method of maximum likeli-

hood, as was done in Ward et al. (2010). The particle

FIG. 3. Vertical cross section of WRF/Chem-predicted Ncn

(shaded; cm23) and accumulation-mode aerosol mass concentra-

tion (contours; mg m23) for the Front Range running from south to

north along 104.88W at 1900 UTC 30 Nov. Altitude on the y axis is

in meters MSL.

FIG. 4. Vertical cross section of RAMS-predicted Nccn (shaded;

cm23) and k (contours) for the Front Range running from south to

north along 104.88W at 1900 UTC 30 November. Altitude on the

y axis is in meters MSL. Note that the difference between topo-

graphical masks in this figure and in Fig. 3 is a result of post-

processing. The same horizontal resolution was used for topography

in RAMS and in WRF/Chem.


diameters measured by the SMPS were between 0.0087

and 0.34 mm. The average size distribution resulting

from this analysis was characterized by a single mode

with an estimated average rg 5 0.018 mm. The two time

series in Fig. 9 correspond very little and are not well

correlated, but there does not appear to be any bias in

rg for this case. The average RAMS rg of 0.019 mm was

nearly equivalent to the observed rg.

To examine possible bias in the model-predicted aero-

sol hygroscopicity, aerosol composition data from the

Interagency Monitoring of Protected Visual Environ-

ments (IMPROVE) were compared with WRF/Chem

forecasts. Atmospheric aerosol samples are collected

at IMPROVE sites on four separate filters for 24-h

periods once every 3 days. Three filters collect only

PM2.5. The mass ratios of collected particulate species

FIG. 5. Vertical profiles of Nccn (SS 5 0.4%) observed during missed-approach (a) 1, (b) 2,

(c) 3, (d) 4, (e) 5, and (f) 6 on 30 Nov from the ICE-L field campaign (solid) and as predicted by

RAMS for the corresponding time and positions (dashed). Altitude on the y axis is given in

meters MSL.


are determined by several different analysis techniques

(Hyslop and White 2008). It is important to note that

sulfate and nitrate aerosol are assumed to exist as am-

monium sulfate and ammonium nitrate but that mass

measurement of ammonium is not conducted. Also, these

data do not provide information about particle number

concentration.

Particles were sampled at the Mount Zirkel IMPROVE

site (MOZI) in Colorado for 24 h beginning at mid-

night local time on 2 April 2008. This site was chosen for

its proximity to SPL. Results of the analysis are given in

Table 6 and show a higher ratio of carbonaceous aerosol

to inorganic aerosol (as represented by the ammonium

sulfate mass concentration) mass fraction when com-

pared with the WRF/Chem output from the same 24-h

period. The lower organic mass fraction would act to

erroneously increase the model k in this case and would

contribute to the positive bias in predicted Nccn.

5. Conclusions

Modules within the WRF/Chem framework were

used to form a system for modeling aerosol evolution in

the atmosphere beginning at emission by anthropogenic

or biogenic sources. Within these modules aerosol mass

is predicted for 14 species and species groups. The k

parameter is computed as a volume-weighted average of

all 14 aerosol components and assuming an internal aerosol

mixture. The value of k is then passed to RAMS along with

the aerosol number and median radius. Internal nudging

connects the WRF/Chem output to RAMS aerosol vari-

ables, which are transported on the RAMS-simulated

winds and passed to the droplet-activation code.

The end result is a system for predicting CCN and

CDNC in RAMS on the basis of model estimates of

FIG. 6. Grid setup for the SPL case-study simulations. This setup

was used for both WRF/Chem and RAMS.

FIG. 7. Time series plot of Nccn (cm23) observed at SPL (shaded)

and predicted by RAMS (solid line). Periods during which SPL

measured a relative humidity in excess of 96% are shaded in light

gray to represent possible in-cloud events.

FIG. 8. Time series plot of (a) Ncn (cm23) and (b) activated

fraction observed at SPL (shaded) and predicted by RAMS (solid

line). In (b) the Nccn measured and predicted at SS 5 0.4% was

used in computing the activated fraction.

FIG. 9. Time series plot of rg (mm) estimated from the SMPS

observations at SPL (shaded) and predicted by RAMS (solid line).


aerosol properties, including composition. The changes

were made without adding noticeable computation time

to WRF/Chem or RAMS. Of course, significant time is

required to produce the aerosol forecasts from WRF/

Chem that were not needed using the previous scheme

in RAMS.

Despite the limitations and uncertainty inherent in

CCN forecasts and resulting from the offline treatment

of aerosols (discussed in section 3c), the simulations in

this study have shown that the modeling system can

reproduce many aspects of the observed CCN. In the

past, aerosol number, size, and composition, and there-

fore CDNC, were user prescribed in RAMS, as they are

still in many global-scale models (Zhang 2008). The new

method for predicting CCN in RAMS described here

has captured the variability in Nccn in both the time di-

mension and the vertical dimension for the given cases,

which could not have been achieved in the previous

system. As stated by Gustafson et al. (2007), predictions

of CCN that allow for variation in space and time are far

superior to a prescribed uniform and constant CCN

distribution for investigating cloud radiative properties

and precipitation.

The following specific lessons about the performance

of the new modeling system were learned:

1) This system will work better in cases for which organic

aerosols are less dominant because these tend to

decrease the overall aerosol hygroscopicity and are

not well represented in WRF/Chem (McKeen et al.

2007), or in most numerical models (Heald et al. 2005).

2) Issues associated with the difference in cloud pro-

cesses between WRF/Chem and RAMS, including

cloud effects on aerosols, would be resolved with

online computation of aerosol processes in RAMS.

3) The system as a whole generally captured the vertical

variations in Nccn during the ICE-L case. Although

this is only one case, it suggests that the model can

perform well in an area dominated by anthropogenic

emissions and local transport processes.

Perhaps the most important overall lesson to be learned

from the model development work is that CCN can be

predicted using a method similar to the one outlined in

this paper, although the uncertainties are considerable

and are often unquantifiable. Moreover, accurate rep-

resentation of droplet activation is only one piece in a

complex microphysical puzzle. The system presented

here retains simplifications of some atmospheric pro-

cesses and does not address others, such as collision–

coalescence, that may be more important in the formation

of precipitation and are also not well parameterized

(e.g., Segal et al. 2007; Levin and Cotton 2009). Yet, the

design improves on the existing method of predicting

CDNC in RAMS. By integrating the WRF/Chem output

and RAMS, even offline, CDNC predictions in RAMS are

representative of the variability of the ambient aerosol.

Acknowledgments. The authors thank Dr. William

Cheng for his assistance with the model development

aspect of this project. Additional support for this work

was provided by Dr. Christine Wiedinmyer of the Na-

tional Center for Atmospheric Research and Dr. Jim

Hudson of the Desert Research Institute. We also thank

two anonymous reviewers who gave comments that im-

proved this manuscript. Funding was provided by Na-

tional Science Foundation Grant ATM-0835421.

REFERENCES

Abdul-Razzak, H., and S. J. Ghan, 2000: A parameterization of

aerosol activation 2. Multiple aerosol types. J. Geophys. Res.,

105, 6837–6844.

——, and ——, 2002: A parameterization of aerosol activation 3.

Sectional representation. J. Geophys. Res., 107, 4026, doi:10.1029/

2001JD000483.

——, ——, and C. Rivera-Carpio, 1998: A parameterization of

aerosol activation 1. Single aerosol type. J. Geophys. Res.,

103, 6123–6132.

Ackermann, I. J., H. Hass, M. Memmesheimer, A. Ebel, F. S.

Binkowski, and U. Shankar, 1998: Modal aerosol dynamics

model for Europe: Development and first applications. At-

mos. Environ., 32, 2981–2999.

TABLE 6. Composition of PM2.5 aerosol mass analyzed by the IMPROVE project and output by WRF/Chem. The IMPROVE aerosol

samples were collected for the 24-h period beginning at 0000 MST at the Mount Zirkel IMPROVE site. WRF/Chem output was averaged

over the same 24-h period. The estimated hygroscopicity of the IMPROVE composition, using representative species from Table 1, would

be approximately k 5 0.3. The average hygroscopicity from the WRF/Chem output was k 5 0.51. Here, NF indicates that the chemical

species was not forecast.

Mass concentration (mg m23) Mass fraction

Species IMPROVE WRF/Chem IMPROVE WRF/Chem

Ammonium sulfate 1.1 0.89 0.25 0.45

Elemental carbon 0.08 0.11 0.02 0.06

Organic carbon 0.57 0.04 0.13 0.02

Sea salt 0.09 NF 0.02 NF

Soil derived 1.76 NF 0.41 NF


Andreae, M. O., and D. Rosenfeld, 2008: Aerosol–cloud–precipitation

interactions. Part 1. The nature and sources of cloud-active

aerosols. Earth Sci. Rev., 89, doi:10.1016/j.earscirev.2008.03.001.

Antilla, T., and V.-M. Kerminen, 2007: On the contribution of

Aitken mode particles to cloud droplet populations at conti-

nental background areas—A parametric sensitivity study.

Atmos. Chem. Phys., 7, 4625–4637.

Binkowski, F. S., and U. Shankar, 1995: The Regional Particulate

Matter Model: 1. Model description and preliminary results.

J. Geophys. Res., 100, 26 191–26 209.

Borys, R. D., and M. Wetzel, 1997: Storm Peak Laboratory: A

research, teaching, and service facility for the atmospheric

sciences. Bull. Amer. Meteor. Soc., 78, 2115–2123.

Chapman, E. G., W. I. Gustafson Jr., R. C. Easter, J. C. Barnard,

S. J. Ghan, M. S. Pekour, and J. D. Fast, 2009: Coupling

aerosol–cloud–radiative processes in the WRF-Chem model:

Investigating the radiative impact of elevated point sources.

Atmos. Chem. Phys., 9, 945–964.

Charlson, R. J., J. H. Seinfeld, A. Nenes, M. Kulmala, A. Laaksonen,

and M. C. Facchini, 2001: Reshaping the theory of cloud for-

mation. Science, 292, 2025–2026.

Cheng, W. Y. Y., G. Carrio, W. R. Cotton, and S. M. Saleeby, 2009:

Influence of cloud condensation and giant cloud condensation

nuclei on the development of precipitating trade wind cumuli

in a large eddy simulation. J. Geophys. Res., 114, D08201,

doi:10.1029/2008JD011011.

Cotton, W. R., and Coauthors, 2003: RAMS 2001: Current status

and future directions. Meteor. Atmos. Phys., 82, 5–29.

Cubison, M. J., and Coauthors, 2008: The influence of chemical

composition and mixing state of Los Angeles urban aerosol on

CCN number and cloud properties. Atmos. Chem. Phys., 8,

5649–5667.

Dudhia, J., and M. W. Moncrieff, 1989: A three-dimensional nu-

merical study of an Oklahoma squall line containing right-

flank supercells. J. Atmos. Sci., 46, 3363–3392.

Eidhammer, T., and Coauthors, 2010: Ice initiation by aerosol

particles: Measured and predicted ice nuclei concentrations

versus measured ice crystal concentrations in an orographic

wave cloud. J. Atmos. Sci., 67, 2417–2436.

Environmental Protection Agency, cited 2010: 2005 National Emis-

sions Inventory data and documentation. [Available online at

http://www.epa.gov/ttnchie1/net/2005inventory.html.]

Fast, J. D., W. I. Gustafson Jr., R. C. Easter, R. A. Zaveri, J. C.

Bernard, E. G. Chapman, G. A. Grell, and S. E. Peckham,

2006: Evolution of ozone, particulates, and aerosol direct ra-

diative forcing in the vicinity of Houston using a fully coupled

meteorology–chemistry–aerosol model. J. Geophys. Res., 111,

D21305, doi:10.1029/2005JD006721.

Feingold, G., and S. M. Kreidenweis, 2002: Cloud processing of

aerosols as modeled by a large eddy simulation with coupled

microphysics and aqueous chemistry. J. Geophys. Res., 107,

4687, doi:10.1029/2002JD002054.

Ghan, S. J., and S. E. Schwartz, 2007: Aerosol properties and

processes. Bull. Amer. Meteor. Soc., 88, 1059–1083.

——, G. Guzman, and H. Abdul-Razzak, 1998: Competition be-

tween sea salt and sulfate particles as cloud condensation

nuclei. J. Atmos. Sci., 55, 3340–3348.

Grell, G. A., S. E. Peckham, R. Schmitz, S. A. McKeen, G. Frost,

W. C. Skamarock, and B. Eder, 2005: Fully coupled ‘‘online’’

chemistry within the WRF model. Atmos. Environ., 39, 6957–

6975.

Guenther, A., T. Karl, P. Harley, C. Wiedinmyer, P. I. Palmer, and

C. Geron, 2006: Estimates of global terrestrial isoprene emissions

using MEGAN (Model of Emissions of Gases and Aerosols

from Nature). Atmos. Chem. Phys., 6, 3181–3210.

Gustafson, W. I., E. G. Chapman, S. J. Ghan, R. C. Easter, and J. D.

Fast, 2007: Impact on modeled cloud characteristics due

to simplified treatment of uniform cloud condensation nu-

clei during NEAQS 2004. Geophys. Res. Lett., 34, L19809,

doi:10.1029/2007GL030021.

Harrington, J. Y., 1997: The effects of radiative and microphysical

processes on simulated warm and transition season Arctic

stratus. Ph.D. thesis, Colorado State University, 289 pp.

Heald, C. L., D. J. Jacob, R. J. Park, L. M. Russell, B. J. Huebert,

J. H. Seinfeld, H. Liao, and R. J. Weber, 2005: A large organic

aerosol source in the free troposphere missing from cur-

rent models. Geophys. Res. Lett., 32, L18809, doi:10.1029/

2005GL023831.

Hong, S. Y., and H. L. Pan, 1996: Nonlocal boundary layer vertical

diffusion in a medium-range forecast model. Mon. Wea. Rev.,

124, 2322–2339.

Hsieh, W. C., A. Nenes, R. C. Flagan, J. H. Seinfeld, G. Buzorius,

and H. Jonsson, 2009: Parameterization of cloud droplet size

distributions: Comparison with parcel models and observations.

J. Geophys. Res., 114, D11205, doi:10.1029/2008JD011387.

Hudson, J. G., 1989: An instantaneous CCN spectrometer. J. At-

mos. Oceanic Technol., 6, 1055–1065.

Hyslop, N. P., and W. H. White, 2008: An evaluation of Interagency

Monitoring of Protected Visual Environments (IMPROVE)

collocated precision and uncertainty estimates. Atmos. Envi-

ron., 42, 2691–2705.

Kanakidou, M., and Coauthors, 2005: Organic aerosol and global

climate modeling: A review. Atmos. Chem. Phys., 5, 1053–1123.

Koehler, K. A., S. M. Kreidenweis, P. J. DeMott, M. D. Petters,

A. J. Prenni, and C. M. Carrico, 2009: Hygroscopicity and

cloud droplet activation of mineral dust aerosol. Geophys.

Res. Lett., 36, L08805, doi:10.1029/2009GL037348.

Kulmala, M., A. Laaksonen, and L. Pirjola, 1998: Parameteriza-

tions for sulfuric acid/water nucleation rates. J. Geophys. Res.,

103, 8301–8307.

Levin, Z., and W. R. Cotton, 2009: Aerosol Pollution Impact on

Precipitation. Springer, 386 pp.

Lin, Y.-L., R. D. Farley, and H. D. Orville, 1983: Bulk parame-

terization of the snow field in a cloud model. J. Climate Appl.

Meteor., 22, 1065–1092.

Lohmann, U., and K. Diehl, 2006: Sensitivity studies of the im-

portance of dust ice nuclei for the indirect aerosol effect on

stratiform mixed-phase clouds. J. Atmos. Sci., 63, 968–983.

——, J. Feichter, C. C. Chuang, and J. E. Penner, 1999: Prediction

of the number of cloud droplets in the ECHAM GCM. J. Geo-

phys. Res., 104, 9169–9199.

McFiggans, G., and Coauthors, 2006: The effect of physical and

chemical aerosol properties on warm cloud droplet activation.

Atmos. Chem. Phys., 6, 2593–2649.

McKeen, S., and Coauthors, 2007: Evaluation of several PM2.5

forecast models using data collected during the ICARTT/

NEAQS 2004 field study. J. Geophys. Res., 112, D10S20,

doi:10.1029/2006JD007608.

Mellor, G. L., and T. Yamada, 1982: Development of a turbulence

closure model for geophysical fluid problems. Rev. Geophys.,

20, 851–875.

Mesinger, F., and Coauthors, 2006: North American Regional

Reanalysis. Bull. Amer. Meteor. Soc., 87, 343–360.

Meyers, M. P., R. L. Walko, J. Y. Harrington, and W. R. Cotton,

1997: New RAMS cloud microphysics parameterization. Part

II. The two-moment scheme. Atmos. Res., 45, 3–39.


Ming, Y., V. Ramaswamy, L. J. Donner, and V. T. J. Phillips, 2006:

A new parameterization of cloud droplet activation applicable

to general circulation models. J. Atmos. Sci., 63, 1348–1356.

Mlawer, E. J., S. J. Taubman, P. D. Brown, M. J. Iacono, and S. A.

Clough, 1997: Radiative transfer for inhomogeneous atmo-

spheres: RRTM, a validated correlated-k model for the long-

wave. J. Geophys. Res., 102, 16 663–16 682.

Nenes, A., and J. H. Seinfeld, 2003: Parameterization of cloud

droplet formation in global climate models. J. Geophys. Res.,

108, 4415, doi:10.1029/2002JD00291.

Pang, Y., B. J. Turpin, and L. A. Gundel, 2006: On the importance

of organic oxygen for understanding organic aerosol particles.

Aerosol Sci. Technol., 40, 128–133.

Petters, M. D., and S. M. Kreidenweis, 2007: A single parameter

representation of hygroscopic growth and cloud condensation

nucleus activity. Atmos. Chem. Phys., 7, 1961–1971.

——, and Coauthors, 2009: Towards closing the gap between

hygroscopic growth and activation for secondary organic

aerosol—Part 2: Theoretical approaches. Atmos. Chem. Phys.,

9, 3999–4009.

Raymond, T. M., and S. N. Pandis, 2002: Cloud activation of single-

component organic aerosol particles. J. Geophys. Res., 107,

4787, doi:10.1029/2002JD002159.

Reutter, P., J. Trentmann, H. Su, M. Simmel, D. Rose, H. Wernli,

M. O. Andreae, and U. Poschl, 2009: Aerosol- and updraft-

limited regimes of cloud droplet formation: Influence of particle

number, size and hygroscopicity on the activation of cloud

condensation nuclei (CCN). Atmos. Chem. Phys., 9, 7067–7080.

Roberts, G. C., and A. Nenes, 2005: A continuous-flow streamwise

thermal-gradient CCN chamber for atmospheric measure-

ments. Aerosol Sci. Technol., 39, 206–221.

Saleeby, S. M., and W. R. Cotton, 2004: A large-droplet mode and

prognostic number concentration of cloud droplets in the

Colorado State University Regional Atmospheric Modeling

System (RAMS). Part I: Module descriptions and supercell

test simulations. J. Appl. Meteor., 43, 182–195.

——, and ——, 2008: A binned approach to cloud-droplet riming

implemented in a bulk microphysics model. J. Appl. Meteor.

Climatol., 47, 694–703.

——, ——, D. Lowenthal, R. D. Borys, and M. A. Wetzel, 2009:

Influence of cloud condensation nuclei on orographic snow-

fall. J. Appl. Meteor. Climatol., 48, 903–922.

Schell, B., I. J. Ackermann, H. Hass, F. S. Binkowski, and A. Ebel,

2001: Modeling the formation of secondary organic aerosol

within a comprehensive air quality model system. J. Geophys.

Res., 106, 28 275–28 293.

Segal, Y., and A. Khain, 2006: Dependence of droplet concentration

on aerosol conditions in different cloud types: Application to

droplet concentration parameterization of aerosol conditions.

J. Geophys. Res., 111, D15204, doi:10.1029/2005JD006561.

——, M. Pinsky, and A. Khain, 2007: The role of competition effect

in the raindrop formation. Atmos. Res., 83, 106–118.

Seinfeld, J. H., and S. N. Pandis, 2006: Atmospheric Chemistry and

Physics. John Wiley and Sons, 1203 pp.

Shantz, N. C., W. R. Leaitch, L. Phinney, M. Mozurkewich, and

D. Toom-Sauntry, 2008: The effect of organic compounds on

the growth rate of cloud droplets in marine and forest settings.

Atmos. Chem. Phys., 8, 5869–5887.

Smagorinsky, J., 1963: General circulation experiments with the

primitive equations. Part I: The basic experiment. Mon. Wea.

Rev., 91, 99–164.

Stockwell, W. R., P. Middleton, and J. S. Chang, 1990: The second-

generation Regional Acid Deposition Model chemical mech-

anism for regional air quality modeling. J. Geophys. Res., 95,

16 343–16 367.

——, F. Kirchner, M. Kuhn, and S. Seefeld, 1997: A new mecha-

nism for regional atmospheric chemistry modeling. J. Geo-

phys. Res., 102, 25 847–25 879.

Stokowski, D., 2005: The addition of the direct radiative effect of

atmospheric aerosols into the Regional Atmospheric Model-

ing System (RAMS). M.S. thesis, Dept. of Atmospheric Sci-

ence, Colorado State University, 81 pp.

Sullivan, R. C., M. J. K. Moore, M. Petters, S. M. Kreidenweis,

G. C. Roberts, and K. A. Prather, 2009: Effect of chemical

mixing state on the hygroscopicity and cloud nucleation prop-

erties of calcium mineral dust particles. Atmos. Chem. Phys., 9,

3303–3316.

Svenningsson, B., and Coauthors, 2006: Hygroscopic growth and

critical supersaturation for mixed aerosol particles of inorganic

and organic compounds of atmospheric relevance. Atmos.

Chem. Phys., 6, 1937–1952.

van den Heever, S. C., and W. R. Cotton, 2007: Urban aerosol

impacts on downwind convective storms. J. Appl. Meteor.

Climatol., 46, 828–850.

——, G. G. Carrio, W. R. Cotton, P. J. DeMott, and A. J. Prenni,

2006: Impacts of nucleating aerosol on Florida storms. Part I:

Mesoscale simulations. J. Atmos. Sci., 63, 1752–1775.

VanReken, T. M., N. L. Ng, R. C. Flagan, and J. H. Seinfeld, 2005:

Cloud condensation nucleus activation properties of bio-

genic secondary organic aerosol. J. Geophys. Res., 110,

D07206, doi:10.1029/2004JD005465.

Virkkula, A., R. Van Dingenen, F. Raes, and J. Hjorth, 1999: Hy-

groscopic properties of aerosol formed by oxidation of limo-

nene, alpha-pinene and beta-pinene. J. Geophys. Res., 104,

3569–3579.

Walko, R. L., and Coauthors, 2000: Coupled atmospheric–biophysics–

hydrology models for environmental modeling. J. Appl. Me-

teor., 39, 931–944.

Ward, D. S., T. Eidhammer, W. R. Cotton, and S. M. Kreidenweis,

2010: The role of the particle size distribution in assessing

aerosol composition effects on simulated droplet activation.

Atmos. Chem. Phys., 10, 5435–5447.

Weingartner, E., H. Burtsher, and U. Baltensperger, 1997: Hy-

groscopic properties of carbon and diesel soot particles. At-

mos. Environ., 31, 2311–2327.

Zhang, Y., 2008: Online coupled meteorology and chemistry

models: History, current status, and outlook. Atmos. Chem.

Phys., 8, 2895–2932.


Copyright of Journal of Applied Meteorology & Climatology is the property of American Meteorological

Society and its content may not be copied or emailed to multiple sites or posted to a listserv without the

copyright holder's express written permission. However, users may print, download, or email articles for

individual use.

A Method for Forecasting Cloud Condensation Nuclei Using Predictions … · 2017-09-29 · A Method...

Documents

Transcript of A Method for Forecasting Cloud Condensation Nuclei Using Predictions … · 2017-09-29 · A Method...