JOURNAL OF GEOPHYSICAL RESEARCH: ATMOSPHERES, VOL. 118, 3794–3806, doi:10.1002/jgrd.50262, 2013
Development and evaluation of an ammonia bidirectional flux
parameterization for air quality models
Jonathan E. Pleim,1 Jesse O. Bash,1 John T. Walker,1 and Ellen J. Cooter1
Received 7 September 2012; revised 16 December 2012; accepted 8 February 2013; published 8 May 2013.
[1] Ammonia is an important contributor to particulate matter in the atmosphere and
can significantly impact terrestrial and aquatic ecosystems. Surface exchange between the
atmosphere and biosphere is a key part of the ammonia cycle. New modeling techniques
are being developed for use in air quality models that replace current ammonia emissions
from fertilized crops and ammonia dry deposition with a bidirectional surface flux model
including linkage to a detailed biogeochemical and farm management model. Recent field
studies involving surface flux measurements over crops that predominate in North America
have been crucial for extending earlier bidirectional flux models toward more realistic
treatment of NH3 fluxes for croplands. Comparisons of the ammonia bidirection flux
algorithm to both lightly fertilized soybeans and heavily fertilized corn demonstrate that
the model can capture the magnitude and dynamics of observed ammonia fluxes, both net
deposition and evasion, over a range of conditions with overall biases on the order of the
uncertainty of the measurements. However, successful application to the field experiment
in heavily fertilized corn required substantial modification of the model to include new
parameterizations for in-soil diffusion resistance, ground quasi-laminar boundary layer
resistance, and revised cuticular resistance that is dependent on in-canopy NH3
concentration and RH at the leaf surface. This new bidirectional flux algorithm has been
incorporated in an air quality modeling system, which also includes an implementation of
a soil nitrification model.
Citation: Pleim, J. E., J. O. Bash, J. T. Walker, and E. J. Cooter (2013), Development and evaluation of an ammonia
bidirectional flux parameterization for air quality models, J. Geophys. Res. Atmos., 118, 3794–3806, doi:10.1002/jgrd.50262.
1.
Introduction
[2] Ammonia is an important precursor to fine-scale
particulate matter (PM2.5) which is known to be a serious
human health hazard [Pope, 2000], and PM2.5 is subject
to regulation through the National Ambient Air Quality
Standards (NAAQS). Gas-phase ammonia reacts with sulfuric
acid (H2SO4) to form ammonium sulfate (NH4)2SO4 or
ammonium bisulfate aerosol (NH4)HSO4, depending on the
ratio of available sulfate to ammonia. Any remaining ammonia
can react with gas-phase nitric acid (HNO3) to form ammonium nitrate aerosol (NH4NO3). Both of these reactions create
aerosol mass by conversion of gaseous NH3 to aerosol NHþ
4,
but the nitric acid reaction creates additional aerosol mass by
converting gaseous HNO3 to NO
3 aerosol while sulfuric acid
will be mostly in the aerosol phase regardless of how much
NH3 is available for reaction. In addition to the health effects
of increased PM due to ammonia emissions, there are also
significant climate effects by direct scattering of shortwave solar radiation and indirect alteration of cloud albedos and
1
U.S. Environmental Protection Agency, Research Triangle Park, North
Carolina, USA.
Corresponding author: J. E. Pleim, U.S. Environmental Protection
Agency, Research Triangle Park, NC 27711, USA. ([email protected])
©2013. American Geophysical Union. All Rights Reserved.
2169-897X/13/10.1002/jgrd.50262
lifetimes through increased cloud condensation nuclei concentrations [Haywood and Boucher, 2000; Ramanathan et al.,
2001]. Both of these effects cause negative radiative forcing
(cooling) and increase diffuse radiation which is more efficient
than direct for photosynthesis. Atmospheric ammonia and
ammonium aerosol contribute a large fraction of reactive nitrogen deposition that is a source of nutrient enrichment and one
of the sources of acidification that cause deleterious impacts
on terrestrial and aquatic ecosystems such as eutrophication,
forest health decline, and biodiversity loss [Lovett and Tear,
2008; Dennis et al., 2007; Driscoll et al., 2001]. The importance of reduced forms of nitrogen deposition is expected to
increase as NOx emissions are further controlled and agricultural emissions of ammonia continue to increase [Pinder
et al., 2008].
[3] About 86% of the ammonia emissions in the U.S. are
from agricultural sources [Gilliland et al., 2003] with about
34% of that coming from fertilizer application (2002 EPA
National Emissions Inventory, http://www.epa.gov/ttn/chief/
eiinformation.html). Measurements over managed agricultural
fields have shown that there are NH3 emission pulses following
fertilizer applications lasting only a few days that account for
the bulk of the annual evasive flux [Flechard et al., 2010].
NH3 emission inputs in air quality models usually have been
developed to capture the general seasonal variations [Gilliland
et al., 2006]. Currently, most air quality models do not model
NH3 emissions and air-surface exchange in sufficient physical,
3794
PLEIM ET AL.: AMMONIA BIDIRECTIONAL FLUX MODEL
temporal, and spatial detail to accurately represent fertilizer
emission pulses or agricultural management practices to credibly model NH3 mitigation strategies. Improved techniques for
modeling NH3 emissions from agricultural sources have been
recently developed in both North America [Zhang et al., 2010]
and Europe [Hamaoui-Laguel et al., 2012]. Also, Massad et
al. [2010] compiled an extensive review of field measurement
studies and proposed generalized parameterizations suitable
for air quality modeling. The main objectives of this paper
are to describe and evaluate the parameterizations of a bidirectional NH3 exchange model for use in an air quality model
coupled to an agricultural management model for improved
physical and temporal detail. The bidirectional NH3 flux
model described below has recently been incorporated and
tested in the coupled meteorology and air quality model composed of the Weather Research and Forecast model and the
Community Multiscale Air Quality model (WRF-CMAQ) as
described by Bash et al. [2013].
2.
NH3 Bidirectional Exchange Model
2.1. Calculation of the NH3 Compensation Point
[4] When exposed to liquid water, ammonia gas (NH3(g))
will dissolve and dissociate in solution and establish an equilibrium between ammonia gas and ammonium ion ( NHþ
4 )
and hydroxide ion (OH). When combined with the equilibrium dissociation of water, the net equilibrium is between
ammonia gas plus hydrogen ion and aqueous ammonium ion:
NH3 ðgÞ þ H 2 O↔NH3 H 2 O
(1)
NH3 H 2 O↔NHþ
4 þ OH
(2)
þ
OH þ H ↔H 2 O
NH3 ðgÞ þ H
þ
↔NHþ
4
A
exp B=Ts;g Γs;g
Ts;g
Ft ¼
ðwc wa Þ
¼ Fg þ Fst þ Fcut ;
Ra þ 0:5Rinc
(5)
(6)
[H+] is the concentration of hydrogen ion, and NHþ
4 is the
(7)
and the component fluxes are
Fg ¼
wg wc
0:5Rinc þ Rbg þ Rsoil
ðws wc Þ
Rb þ Rst
wc
¼
Rb þ Rw
Fst ¼
Fcut
(4)
where ws,g is the compensation point concentration of NH3
in the air space inside the leaf stomata or soil pores (mg
NH3 m3), A (2.7457 1015) and B (10,378) are constants
derived from the equilibria constants, Ts,g is the leaf or soil
temperature (K), and Γs,g is the dimensionless NH3 emission
potential in the leaf stomata or soil where
þ
NH4
Γ¼
½H þ 2.2. Bidirectional Exchange Resistance Model
[6] Like most dry deposition models, the bidirectional flux
model is based on an electrical resistance analog where flux
is analogous to current and concentration difference is analogous to voltage [Wesely, 1989] (Figure 1). The total flux
between the plant canopy and the overlying atmosphere is
the sum of two bidirectional pathways, to the leaf stomata
(Fst) and the soil (Fg), and one uni-directional deposition pathway, to the leaf cuticle (Fcut). Because each bidirectional
flux pathway depends on concentration difference across
resistances, an in-canopy concentration wc is computed at
the junction of the ground, stomatal, and cuticle pathways
(equation 20). Thus, the total flux is defined as
(3)
[5] Such equilibria exist in leaves where NH3(g) in the
stomatal cavity is in equilibrium with the NHþ
4 and OH in
the water contained in the apoplast within the leaf and in the
soil where NH3(g) in the soil pore air space is in equilibrium
with the NHþ
4 and OH dissolved in soil water. By combining
the equilibrium expressions for reactions (1–3), including
the temperature dependent expressions for their equilibrium
constants (i.e., Henry’s Law KH and dissociation equilibria
for ammonia Ka and water Kw), the concentration of NH3(g)
in the stomatal cavities (ws) and the soil air space (wg) can be
+
related to the aqueous concentrations of NHþ
4 and H in the
leaves and soil water as follows [Nemitz et al., 2000]:
ws;g ¼
concentration of NHþ
4 ion in the apoplast water of the canopy leaves or soil water. The direction of ammonia flux to or
from the leaf apoplast or soil water depends on the gradient between the compensation points and the air concentration in the
canopy, wc, such that ammonia will volatilize (emit) when ws,
g > wc and deposit when ws,g < wc.
(8)
(9)
(10)
[7] Some of the resistances are the same as are used for
evapotranspiration in meteorology models and dry deposition in the WRF-CMAQ model, such as Ra, Rb, and Rst
which have been defined in previous papers describing those
models [e.g., Pleim, 2006; Pleim and Xiu, 2003; Xiu and
Pleim, 2001; Pleim and Ran, 2011]. The resistances that
needed to be developed or refined for this work primarily
involve the soil pathway because in fertilized agricultural
fields, the evasive flux from the ground usually dominates
all other flux pathways. For heavily fertilized crops, such
as for the field experiment in a North Carolina corn field that
is described in section 3.2 and a fertilized grassland in
Germany described by Personne et al. [2009], measured
values of Γg were often on the order of 100,000 which,
depending on the temperature and pH of the soil, results in
soil compensation concentrations of ammonia on the order
of 1000 mg m3. Thus, the resistance to transport of gaseous
NH3 in the soil pore spaces must be very large to describe
the observed NH3 fluxes.
[8] The ground pathway involves three serial resistances
between the ammonia at the soil water interface (wg) and
the in-canopy concentration (wc) as shown in equation (8):
in-canopy aerodynamic resistance (Rinc), quasi-laminar
boundary layer resistance at the ground (Rbg), and the insoil resistance (Rsoil). The formulation for Rinc is the same
as has been used in the dry deposition model in the Community
3795
PLEIM ET AL.: AMMONIA BIDIRECTIONAL FLUX MODEL
Ra
Rb
Rinc
Rbg
Rw
Rst
χa
c
χc
χg
χst
Rsoil
Resistance to
deposition
Aerodynamic
resistance
Laminar boundary
layer resistance
In-canopy
aerodynamic
resistance
Ground laminar
boundary layer
resistance
Cuticular resistance
Stomatal resistance
Atmospheric
concentration
Canopy
compensation point
Soil compensation
point concentration
Stomatal
compensation point
concentration
Soil resistance
Figure 1. Resistance model schematic for bidirectional NH3 flux with leaf and soil compensation point
concentrations.
Multiscale Air Quality (CMAQ) regional chemical transport
model for many years, as suggested by Erisman et al. [1994]:
Rinc ¼ bLAI
hcan
;
u
(11)
where u* is the friction velocity, LAI is the one-sided leaf
area index, hcan the canopy height, and b an empirical
constant taken as14 m1. Note that Rinc is split such that half
of the resistance is applied between the soil and the air
concentration in the canopy (wc) and the other half from
the canopy to the atmosphere above, where wc acts like a
canopy compensation point. The splitting of the Rinc is
important for bidirectional modeling since NH3 can flux into
the canopy from either above or below; thus, some symmetry
in the resistance to the stomatal and cuticle sinks from either
direction seems appropriate.
[9] Similar to the Rb in the canopy pathway equations
(equations 9 and 10), Rbg represents the resistance to transfer
across the quasi-laminar sublayer adjacent the surface. The
difference is that the wind speeds and therefore the intensity
of turbulence at the ground surface underneath the canopy
are small fractions of their above canopy counterparts. Thus,
the ground level friction velocity u*g is defined as
ug ¼ u expðLAI Þ
Rbg ¼
kug
Rsoil ¼
(14)
Ldry
Dp
(15)
where Ldry is a function of soil water content relative to saturation [Sakaguchi and Zeng, 2009]:
Ldry ¼ ds
h
i
exp ð1 θs =θsat Þ5 1
e1
(16)
where ds is the depth of the soil layer, θs is the volumetric
soil water content, θsat is the volumetric soil water content
at saturation, and e is Euler’s number. The diffusivity
through the soil pores is
θr 2þ3=b
Dp ¼ DNH3 θ2sat 1 θsat
(13)
where Sc is the Schmidt number for NH3, k is the von
Karman constant, and do is the thickness of the laminar layer
at the ground surface given by
v
kug
where n is the kinematic viscosity of air. The below canopy
reference height zr represents the top of the logarithmic wind
profile layer where eddy size is proportional to height above
the surface. The value of zr is estimated to be 0.1 m because
above this approximate height, eddy size is likely to be
influenced by stems and leaves. Note that changes in zr of
a factor of two cause only 10% change in Rbg.
[10] The third serial resistance in the ground pathway is the
resistance to diffusion through the air within the soil matrix
from the soil water to the ground surface (Rsoil). Since the
source of gas-phase NH3 in the soil is NHþ
4 dissolved in the
soil water, the soil resistance for NH3 is analogous to the soil
resistance to evaporative flux from moist soil. Thus, the Rsoil
for NH3 can be represented as the characteristic length scale
from the soil surface to the soil water Ldry divided by the
diffusivity of NH3 through air-filled pores of the soil, Dp, as
(12)
based on in-canopy measurements in the same corn field that
is described in section 3.2 [Bash et al., 2010]. The ground
quasi-laminar boundary layer resistance is then computed
as [Schuepp, 1977]
Sc ln dZor
do ¼
(17)
where DNH3 is the diffusivity of NH3 gas in air, b is the slope
of the retention curve parameterized as a function of soil
3796
PLEIM ET AL.: AMMONIA BIDIRECTIONAL FLUX MODEL
type following Clapp and Hornberger [1978], and θr is
the residual soil water content [Rawls et al., 1982]. The
expression for Dp, equation (17), was derived by Moldrup
et al. (1999) and recommended by Moldrup et al. (2004) on
the basis of comparisons to diffusivity measurements through
soil cores of various soil texture types. For their experiments
with varying matrix water potentials, θr was replaced with the
experimental value of the soil moisture content θ in equation
(17). However, since we are using equation (17) to describe
the diffusivity through the dry portion of the soil from the soil
water to the surface, θ is set to its residual value. Note that Rsoil,
and parameter values for b, θr, and θsat, are dependent on soil
texture [e.g., Cosby et al., 1984; Noilhan and Lacarrère,
1995]. Thus, this formulation is well suited for air quality modeling systems such as WRF-CMAQ where soil texture parameters
are already provided for the entire modeling domain for the land
surface modeling components of the system. Note that Personne
et al. [2009] used a very similar approach for Rsoil based on a
model described by Choudhury and Monteith [1988].
[11] The soil resistance described by equations (15)–(17) is
essentially the same as that implemented in the Community
Land Model (CLM) component of the Community Earth
System Model (CESM) for evaporation from bare soil
[Sakaguchi and Zeng, 2009] where the diffusivity for water
vapor is replaced by the diffusivity of NH3 in equation (17).
Since the NH3 flux from soil is extremely sensitive to Rsoil
for heavily fertilized crops, a realistic and robust formulation
for this resistance formulated with dependence on soil texture
is essential for air quality modeling in domains that include
intensive agriculture. Sakaguchi and Zeng [2009] demonstrated that various other formulations for Rsoil that have been
used in land surface models produce vastly different values
as functions of soil moisture. While such differences may have
limited affects on evaporation, especially in vegetated areas
where transpiration dominates, the effects on NH3 fluxes for
fertilized crops can be extreme. Indeed, sensitivity tests using
some of the other Rsoil formulations in the bidirection flux
model resulted in larger errors in NH3 flux for the field experiment over fertilized corn as shown below in section 3.2.
[12] The cuticle resistance was also substantially modified
from the simple reactivity and solubility scaling approach used
in the CMAQ dry deposition model [Pleim and Ran, 2011].
Jones et al. [2007a and 2007b], using moorland vegetation
in a gas flux chamber, showed that cuticle resistance to NH3
increases substantially with increasing in-canopy concentrations. They postulate that this relationship between wc and
Rw, which their experiments suggest, is roughly linear and is
due to increasing saturation of the cuticle and cuticular surface
water by depositing NH3. Many experiments have also shown
a strong dependence of NH3 cuticle resistance on relative
humidity where resistance declines as humidity increases
because of the relatively high solubility of NH3. While this
relationship is often expressed as an exponential function of
RH [e.g., Wyers and Erisman, 1998, Massad et al., 2010,
Flechard et al., 2011], a linear function is used for this study
based on the experimental data reported by Jones [2006].
Thus, the cuticle resistance for NH3 is represented as
Rw ¼
1 Rwo wc
þ 1 þ ah ð1 fRHs Þ
LAI Heff wref
(18)
where Heff = KH (1.0 + Ka / [OH-]) is the dimensionless
effective Henry’s law coefficient for ammonia where KH is
the Henry’s law equilibrium constant for ammonia gas and
Ka is the dissociation equilibrium constant for the aqueous
ammonia-ammonium equilibrium reaction (equation 2).
fRHs is the fractional relative humidity at the leaf surface,
which is computed as a compensation point for air humidity
at the leaf surface as shown by Xiu and Pleim [2001]. There
are three empirical constants that are set: Rwo = 125,000 s m1,
ah = 100.0 s m1, and wref = 1.0 mg m3. For the portions of
leaf surfaces covered with water from rain or dew, fRHs is assumed to be 1.0 and the saturation effect is neglected such that
equation (18) collapses to
Rw ¼
1 Rwo
LAI Heff
(19)
which is the same as the equation used for wet cuticle resistance
for all other chemical species in the CMAQ dry deposition
model. The value of ah in equation (18) is based on the
experiments performed by Jones [2006] but with empirical
adjustments to give realistic results for the experiments
described below.
[13] The computation of the total flux is a two step process
where first equations (7–10) are combined and solved for wc.
Since Rw is a function of wc the solution for wc is a quadratic
equation which is solved as
0:5
wc ¼
b þ ðb2 4acÞ
2a
(20)
where
a ¼ Rwet Gt
b ¼ Rwb Gt þ LAI ð1 fwet Þ Rwet Ga wa þ Gsb ws þ Ga wa þ Gg wg
c ¼ Rwb Ga wa þ Gsb ws þ Ga wa þ Gg wg
and where,
Ga
Gsb
Gg
Gt
Gcw
Rwet
Rwb
¼ ðRa þ 0:5Rinc Þ1
¼ ðRst þ Rb Þ1
1
¼ Rbg þ 0:5Rinc þ Rsoil
¼ Gsb þ Gg þ Ga þ fwet Gcw
LAI
¼
Rb þ Rwet
Rwo
¼
Heff
¼ Rwet þ ðah ð1 fRHs Þ þ Rb ÞLAI
and fwet is the canopy wetness fraction. Once wc is known,
the total flux is computed according to the first term on the
right-hand side of equation (7).
2.3. Field Scale Application
[14] A stand alone box model of dry deposition based on
the CMAQ dry deposition resistance parameterization
[Pleim et al., 2001] and the bidirectional exchange resistance
model described in section 2.2 was developed to run using
meteorological and ambient NH3 data collected in the field.
For this study, the model was run using observed data from
flux experiments located in Warsaw and Lillington, North
Carolina. The Warsaw site was in a lightly fertilized soybean
field located in the vicinity of animal husbandry facilities
and had high background NH3 concentrations [Walker
et al., 2006], while the Lillington site was heavily fertilized
3797
PLEIM ET AL.: AMMONIA BIDIRECTIONAL FLUX MODEL
corn and had low background NH3 concentrations [Walker
et al., 2012]. These experiments effectively bound the ambient
NH3 concentrations and nutrient management options that
would be expected in managed agricultural cropping settings.
For the Warsaw modeling study, stomatal and soil Γ values
were constants based on observations from the field. For
the Lillington simulations, additional measurements of soil
properties allowed for the parameterization of a dynamic Γg
and linkage to a detailed biogeochemical and farm management model following Cooter et al. [2010].
3.
Evaluation Against Field Experiments
[15] Air-surface exchange of NH3 was measured over a
lightly fertilized soybean canopy in Duplin County, North
Carolina, in 2002 and a fertilized corn (Zea mays, pioneer
varieties 31G66 and 31P41, 70,000 plants ha1) canopy
in Harnett County, North Carolina, in 2007. In 2002, flux
measurements were taken at a 90 ha soybean field fertilized
with ~65 kg N ha1 of poultry litter amended to the soil
2 weeks prior to planting bordered by mature pine trees to
the north, east, and south and bordered by additional agricultural fields to the west and southwest, the predominant wind
direction [Walker et al., 2006]. Ancillary measurements of
þ
soil NO
3 and NH4 , soil moisture, and LAI along with
estimates of the canopy compensation point were taken
[Walker et al., 2006]. In 2007, flux measurements were
taken over a 200 ha corn field fertilized with 135 kg
N ha1 of surface applied urea ammonium nitrate (UAN)
solution bordered by mature deciduous forest [Bash et al.,
2010; Walker et al., 2012]. A urease inhibitor, AgrotainW,
was added to the UAN solution. Ammonia fluxes were
measured using the modified Bowen-ratio (MBR) technique
at both sites [Hicks and Wesely, 1978]. NH3 concentration gradients were measured using a chemiluminescence NOx/NH3
analyzer (Model 17C, Thermo Electron Corporation, Franklin,
MA) at the soybean field and with a continuous flow ammonia
measurement by ANular Denuder sampling with online
Analysis (AMANDA) [Wyers et al., 1993] at the corn field
[Walker et al., 2012]. Ancillary measurements in 2007
included soil water solution, dew, and leaf apoplast chemistry
measurements. The soil chemistry measurements were linked
with a mechanistic model of soil NHþ
4 transformation processes taken from the USDA Environmental Policy Integrated
Climate (EPIC) model [Williams et al., 1985; Izaurralde et al.,
2006] to develop estimates of daily soil compensation points
[Cooter et al., 2010]. In addition, in-canopy profiles of ambient NH3, temperature, and wind speed were used to assess
in-canopy sources and sinks [Bash et al., 2010]. Table 1 summarizes the evaluation statistics while the following two sections provide details of model evaluation for both field studies.
3.1. Soybeans
[16] Flux measurements were taken at the Warsaw, NC,
soybean field from 18 June to 24 August 2002. A constant
value for Γs of 1054 was used in the model as suggested
by Walker et al. [2006] from estimates of the compensation
concentration when observed flux was close to zero. A value
of Γg = 800 was used as a reasonable estimate for lightly
fertilized soybeans. The mean observed and modeled fluxes
during this period were 9 64 ng m2 s1 and 16 47
ng m2 s1, respectively, and the mean ambient concentration was 9.45 mg m3. The model captured the observed
diurnal trends well (Figure 2) with deposition dominating
during the night generally peaking at about 9–10 A.M.,
and evasion usually occurring during the rest of the day
peaking around the time of the maximum daily air temperature (2–3 P.M.). However, the model seemed to lag behind
the measurements during the late morning when the
observed evasive fluxes were increasing rapidly. The
measurements often had very high emission spikes during
these hours that the model could not match. Note however
that both model and measurements had very wide distributions with at least 25% overlap between model and measurements for each hour and that these flux measurements were
most uncertain during transition periods when the heat flux
and temperature gradients were small [Walker et al., 2012].
[17] The model was significantly correlated with the observations, p < 0.001, during this measurement campaign and
the mean normalized bias in the model estimates was
78.6% (7.23 ng m2 s1). Note that the normalized mean
bias seems large compared to the mean bias because the mean
flux was near zero during the measurement period. A 50%
uncertainty was estimated for the observations due to sequential sampling of the gradients by the chemiluminescence
NOx/NH3 analyzer [Walker et al., 2006]. The bidirectional
resistance model estimated that stomatal evasion from NHþ
4
in the apoplastic solution was the dominant emission pathway
followed by a small net evasive flux from bidirectional
exchange at the soil surface ranging from 29 to 113 ng
m2 s1 and a net deposition to the plant cuticles in agreement
with the findings of Walker et al. [2006]. Soybeans rarely
receive fertilizer applications after planting, and by the time
the measurement campaign began, very little fertilizer would
be left to transform and volatilize so the addition of a dynamic
Γg following Cooter et al. [2010] would not likely change the
results. This was confirmed by sensitivity simulations that
perturbed the Γg by 50% resulting in less than 1% change
in the air-canopy flux.
3.2. Corn
[18] Two simulation periods were chosen to represent the
full range of flux conditions. The first 1 week simulation
Table 1. Summary of Statistical Evaluation of Model Results
Field Experiment
Soybeans
Corn (21–29 June)
Corn (11–19 July)
Mean SD (Median)
Obs Flux (ng m2 s1)
Mean SD (Median)
Model Flux (ng m2 s1)
Mean SD (Median)
Ambient Conc (mg m3)
Mean (Median)
Normalized Bias
Mean (Median)
Bias (ng m2 s1)
9 64 (3)
431 714 (190)
31 65 (4)
16 47 (14)
219 271 (107)
31 47 (9)
9.4 5.3 (9.0)
7.5 3.6 (6.9)
1.9 1.0 (1.5)
78.6% (182%)
49% (29%)
1% (42%)
7.2 (5.9)
211 (56)
0.5 (2)
3798
PLEIM ET AL.: AMMONIA BIDIRECTIONAL FLUX MODEL
Figure 2. Diel hourly box plots of observations from the 18 June to 24 August 2002 flux measurements
over a soybean canopy at Warsaw, NC (green), paired with CMAQ box model results, red. The 5th and
95th quantiles are represented by the whiskers, the 25th and 75th quantiles are enclosed in the box, the median
is represented by the horizontal line through the box, and the mean is represented by the black triangle.
was run from 21 to 29 June 2007, 3 weeks after the application of urea ammonium nitrate fertilizer. Daily Γg values for
this period computed as described by Cooter et al. [2010]
ranged from 84,500 to 222,000 with only 1 day below
100,000. The median observed and modeled fluxes during
this period were 190 ng m2 s1 and 107 ng m2 s1,
respectively, and the mean ambient NH3 concentration was
7.5 mg m3. The model generally reproduced the observed
time series but underestimated the observed evasion events
larger than 1000 ng m2 s1 (Figure 3). These large morning
evasion events accounted for 8.7% of the observations during this week. During this period, the model significantly
correlated, p < 0.001, with the observations and the median
normalized bias was 29% (56 ng m2 s1), similar to the
43% median uncertainty estimated for the NH3 flux measurements [Walker et al., 2012]. Two to three weeks later
(11–19 July), during the second simulation period, the daily
гg values fell to 35,900–85,800. The median observed and
modeled flux had dropped by more than an order of magnitude to 4 ng m2 s1 and 9 ng m2 s1, respectively, and the
median ambient concentration was 1.87 mg m3 (Figure 4).
The model was again significantly correlated, p < 0.001,
with the observations. The normalized median bias during
this period was 42%, but the median bias was only 2 ng
m2 s1. The relative bias in this case was larger due to a
small median flux during this measurement period, but the
absolute value was similar to the estimated uncertainty in
the flux measurements [Walker et al., 2012]. Further development of the resistance model to reduce these biases may
not be productive since the observations probably include
measurement artifacts similar to the standard error reported
by Milford et al. [2009] which may represent the precision
at which NH3 fluxes over a managed agricultural field can
be measured using flux gradient systems.
[19] During both modeling periods, the diurnal dynamics
of the flux were captured well with the exception of the
extreme morning evasion events observed from 8:00 to
11:00 local time that were measured during the earlier period
following fertilization (Figure 3). While this time period
corresponded to the evaporation of dew from the canopy,
measurements of the dew NHþ
4 concentrations were not high
enough to account for these measured emission fluxes [Bash
et al., 2010]. Evaluation of the soil flux portion of the model
for this experiment indicates that the EPIC biogeochemical
model parameterized for this location consistently estimated
soil moisture, ammonium, and hydrogen ion concentrations
as well as nitrogen transformation rates within the range of
field observations [Cooter et al., 2010]. These large morning
emission events may be due to a combination of upward
mixing of NH3 in the top soil layers that accumulated near
the surface on calm nights, which would not be captured
by the model that uses daily constant values of гg, and
drying of soil surface moisture. For example, on the night
of DOY 172–173, wind speed and friction velocity averaged
0.45 and 0.04, respectively, between midnight and 6:00.
Such an extended duration of calm conditions would allow
NH3 emitted from the soil to accumulate below the lowest
NH3 measurement sensor and allow moisture containing
NHþ
4 to accumulate at the soil surface. At this stage in the
growing season, before peak LAI, a significant fraction of
the subsequent emission pulse generated by the rapid atmospheric mixing and surface drying post-sunrise may escape
the canopy without re-adsorption by overlying vegetation.
Indeed, measurements on DOY 173 show that friction velocity jumps from 0.047 to 0.24 m s1 from 06:30 to 07:30,
while the observed NH3 flux jumps from 170 to 4300 ng
m2 s1 demonstrating that the large morning emission
spikes are coincident with the rapid onset of morning turbulent mixing. Another possibility for the under prediction
of very large emission fluxes on the morning of DOY 173
that the Rsoil is too large in this particular instance is investigated by setting Rsoil to zero. While the Rsoil = 0 run
did roughly double the net flux on DOY 173 at 9:30
(760 versus 320 ng m2 s1), the increase was not
3799
PLEIM ET AL.: AMMONIA BIDIRECTIONAL FLUX MODEL
a
b
Figure 3. (a) Diel hourly box plots of flux measurements over a fertilized corn canopy at Lillington, NC
(green), paired with CMAQ box model results (red) from 21–29 June 2007. The 5th and 95th quantiles are
represented by the whiskers, the 25th and 75th quantiles are enclosed in the box, the median is represented
by the horizontal line through the box, and the mean is represented by the black triangle. (b) Time series of
observed (black diamonds) and modeled NH3 fluxes above the corn canopy (red) with the modeled fluxes
from the soil (green), the stomata (blue), and the cuticle (orange).
sufficient to explain the measured flux of 6220 ng m2
s1 at this time.
[20] As shown in Figure 3b, the large upward flux during
the measurement period following fertilization was largely
from NHþ
4 in the soil water solution (ground flux) because
гg ranged from approximately 100,000 to 200,000. The
net fluxes for the stomatal and cuticle pathways were almost
entirely deposition/negative indicating that a portion of the
ground flux was taken up by the canopy. Model simulations
2–3 weeks later during the in-canopy measurement period,
shown in Figure 4b, suggest that 78% of the ground flux
was intercepted by the canopy, which is in approximate
agreement with the 73% canopy uptake estimated using
in-canopy ammonia measurements made by Bash et al.
[2010] during this same period. Note that the canopy uptake is split between the stomatal and cuticular pathways
during the daytime with cuticle uptake slightly greater
than stomatal uptake during the earlier modeling period
but almost twice the stomatal flux during the later modeling period. This difference is due primarily to the greater
NH3 air concentrations during the earlier period which
cause higher cuticular resistance in accordance with
equation (18).
[21] In order to check whether the stomatal pathway is
realistic for this case, the latent heat flux computed by the
model is compared to the measurements for the 11–19 July
period and shown in Figure 5. The model tends to show a
slight high bias which may indicate a slight low bias in Rst.
However, note that the Rst and evapotranspiration calculations come directly from the Pleim-Xiu land surface model
(PX LSM) as implemented in WRF [Pleim and Xiu, 1995]
without any tuning for this experiment. Furthermore, the
3800
PLEIM ET AL.: AMMONIA BIDIRECTIONAL FLUX MODEL
a
b
Figure 4. (a) Diel hourly box plots of flux measurements over a fertilized corn canopy at Lillington, NC
(green), paired with CMAQ box model results (red) from 11–19 July 2007. The 5th and 95th quantiles are
represented by the whiskers, the 25th and 75th quantiles are enclosed in the box, the median is represented
by the horizontal line through the box, and the mean is represented by the black triangle. (b) Time series of
observed (black diamonds) and modeled NH3 fluxes above the corn canopy (red) with the modeled fluxes
from the soil (green), the stomata (blue), and the cuticle (orange).
measured soil moisture at 23 cm depth is used in the stomatal
resistance parameterization that is designed for modeled soil
moisture representing an average over a 1 m deep layer. Thus,
it is not surprising the there is some bias in evapotranspiration
when applied to field study measurements.
[22] At night, the stomata are closed, but there are substantial cuticle fluxes that mostly offset evasion fluxes from the
soil. During the later period, the modeled net flux is close to
zero during most nighttime hours, which agrees well with
the measurements. However, small amounts of nighttime
evasion were more prevalent during the flux measurements
following fertilization earlier in the field campaign, while the
modeled net fluxes were close to zero. This under prediction
of nighttime evasion during this early period could be due to
either underestimated soil flux (underestimated soil г or
overestimated soil resistance) or overestimated cuticular
uptake (underestimated cuticular resistance). Seeing that the
measured total fluxes were often greater than the modeled
soil fluxes (Figure 3b) underestimated soil flux is more
likely the cause of these nighttime biases than errors in
cuticular resistance.
[23] To better understand the behavior of the cuticular
resistance, computed from equation (18), during these two
flux measurement periods, the relationships between Rw
and wc and between Rw and RHs for both modeling periods
are shown in Figure 6. For both periods, Rw increases
roughly linearly with increasing in-canopy concentration in
qualitative agreement with Jones et al. [2007b] although
for the earlier period (Figure 6a), the slope is steeper for
the highest concentrations with a hint of upward curvature.
3801
PLEIM ET AL.: AMMONIA BIDIRECTIONAL FLUX MODEL
Figure 5. Modeled and measured latent heat flux for 11–19 July 2007.
Figure 6. (a and c) Relationships between in-canopy concentration (wc) and cuticular resistance (Rw) and
(b and d) relationships between relative humidity at the leaf surface (RHs) and cuticular resistance (Rw). (a and b)
Model results for the early period (21–29 June) and (c and d) model results for the later period (11–19 July).
3802
PLEIM ET AL.: AMMONIA BIDIRECTIONAL FLUX MODEL
The times when wc is highest occur during the hottest part of
the early afternoon which is also when temperature is highest
and RHs is lowest (Figure 6b). Both high temperature (Heff decreases as T increases) and low humidity contribute to greater
cuticular resistance. However, according to equation (18), the
most that the humidity term (second term) can contribute to Rw
is 100 s m1 when RHs = 0.0. Thus, when Rw is greater than
about 300 s m1, the first term in equation (18) dominates.
Since the highest values of wc occur at the highest temperatures, the steeper slope and upward curvature at the high end
of the graph is caused by the nonlinear dependence of Heff
on T combined with the linear dependence on wc in the first
term. The dominance of the first term can also be seen in Figure 6b where at the lowest RHs values (~30%), Rw shows no
relationship to RHs. Conversely, at high RHs when temperature and wc tend to be low, the second term dominates as indicated by the lack of scatter at the high humidity ends of
Figures 6b and 6d. Furthermore, if the x axis were expressed
as a fraction rather than %, the slope at high humidities
(> ~70%) is ~100 s m1, which is the value of ah in equation (18). The Rw relationship to in-canopy concentration
does not show the same steepening and upward curvature
for the later period (figure 6c) since wc and Rw never get
to the levels where term one dominates.
[24] Note that both Rw and wc were much greater during the
earlier period shortly after fertilization when soil flux was also
much greater. Although there were no measurements of incanopy concentrations during the early period the modeled wc
values during the later period (Figure 6c) are mostly in the measured range reported by Bash et al. [2010] (0–20 mg m3).
Thus, validation of modeled in-canopy concentrations combined with a reasonable degree of agreement of modeled net
fluxes with measurements for both periods, with an order of
magnitude difference in fluxes between the periods, supports
the formulation of Rw represented by equation (18).
[25] The other critical parameter that had to be added to the
base dry deposition scheme for NH3 bidirectional modeling,
especially for highly fertilized crops, is Rsoil. Therefore,
several sensitivity runs were made where different formulations for Rsoil were tested. Figure 7 shows the observed and
base modeled median values of NH3 flux for both measurement periods at Lillington, which are identical to the center
line on each diel plot shown in Figures 3a and 4a, along with
three sensitivity runs: Zero, where Rsoil = 0; Kando, which
uses the empirical formula for loam derived by Kondo et al.
[1990]; and Sellers, which represents another empirical
formula by Sellers et al. [1992]. For the earlier period
(Figure 7a), the Zero and Kando cases greatly over predict
the median of the measurements during the afternoon hours
(after 1300), while the Sellers case was substantially too low
except for the late afternoon. The Base case agrees with the
measurements quite well in the afternoon but is too low in
the morning when the Zero and Kando cases are closer to
observations. Time series of NH3 flux comparing the Zero
run to the Base run and observed fluxes is shown in Figure 8.
As noted above, the measurements often showed very large
spikes during morning hours that were not well represented
by the model even by the Zero case. On some days, for example, DOY 173, the Zero run predicted very high fluxes that
were similar in magnitude to the observations, but they
occurred at the wrong time of day such that the Zero run
showed a large low bias in the morning and a large high bias
in the afternoon. Note that on days when soil moisture is low
(also shown in Figure 8), the Zero run often grossly over
predicts, while on days with higher soil moisture, the Zero
and Base runs are similar. Thus, Rsoil is a particularly important parameter when the soil is dry. Comparison of the sensitivity cases for the later period (Figure 7b) shows that the
Base case generally agrees best with the measurements with
Zero and Kando mostly too high and Sellers mostly too low.
4.
Integration Into the CMAQ Model
[26] The bidirectional resistance model for NH3 exchange
developed for these field scale applications (section 2) was
incorporated into the CMAQ dry deposition routines. The soil
+
ratio of NHþ
4 to H (Γg) was computed in CMAQ based on
fertilization application rates and soil pH from a continental
U.S. simulation of EPIC [Cooter et al., 2012]. EPIC nitrification parameterizations were incorporated into CMAQ to fully
Figure 7. Median NH3 fluxes for (a) 21–29 June and (b) 11–19 July for the corn field in Lillington, NC,
comparing model sensitivity runs for Rsoil to observations. The definitions for Base, Zero, Kando, and
Sellers are provided in the text.
3803
PLEIM ET AL.: AMMONIA BIDIRECTIONAL FLUX MODEL
Figure 8. Sensitivity of NH3 flux to Rsoil for 21–29 June for the corn field in Lillington, NC. The Base
case is the same as shown in Figure 3b, and the Zero case is for Rsoil= 0.0. The soil moisture represents the
top 1 cm thick layer which is derived from EPIC model calculations.
couple the ammonium soil budget with NH3 evasion and deposition. CMAQ was provided with crop management scenarios,
crop area, crop type, and fertilizer application timing, method,
and rates. The vegetative apoplastic emission potential (Γs) and
the soil emission potential (Γg) in unmanaged non-agricultural
areas are modeled as a function of the land cover types similar
to Zhang et al. [2010]. The parameterization of the NH3 bidirectional exchange reduced the bias and error in wet deposition
results for an annual simulation (2002) with 12 km grid resolution over the continental U.S. [Appel et al., 2011]. A description of the implementation, annual simulation results, and
evaluation against network observations of NHx (NH3 + NH+)
wet deposition and inorganic aerosol observations of a regional
scale application to a photochemical 3-D air quality model
coupled to the EPIC agro-ecosystem model are reported by
Bash et al. [2013].
5.
Conclusions and Future Work
[27] Bidirectional NH3 surface flux capability has been
added the land surface model (LSM) and dry deposition model
components of the WRF-CMAQ coupled meteorology and
air quality modeling system. Existing parameterizations for
aerodynamic resistance, quasi-laminar boundary layer resistance, in-canopy aerodynamic resistance, and bulk stomatal
resistance used for dry deposition and evapotranspiration were
also used for NH3 bidirectional flux model. Surface concentrations in the soil and stomatal cavities are computed from
aqueous equilibrium relationships according to equation (5).
For initial testing in box model form applied to the soybean
field study, nothing else was changed from the LSM/dry
deposition model other than the flux solution technique which
requires an intermediate calculation of canopy compensation
concentration. This simple model was able to well reproduce
the diurnally bidirectional fluxes measured in the soybean
field. However, for the fertilized corn field study, several
modifications to the resistance model were needed including
the cuticular resistance which was revised to be dependent
on in-canopy NH3 concentration and RH at the leaf surface.
Also, new parameterizations were added for the quasilaminar boundary layer resistance at the soil surface (below
canopy) and the in-soil diffusion resistance. For heavily
fertilized crops, such as corn, these two serial resistances are
critical for restricting the soil flux to realistic levels. The
revised cuticular resistance also plays a crucial role for these
highly fertilized crops since both the model and in-canopy
measurements reported by Bash et al. [2010] show that a large
fraction of the soil flux is absorbed by the canopy. The earlier
experiment for soybeans was revisited with the revised
model developed for the corn experiment to make sure that
the model’s ability to simulate the soybeans had not been
degraded. Recognizing that the model includes many parameters with large uncertainties, this iterative process of model
refinement and evaluation will continue as the model is
applied to more field measurement data. In this way, the model
will continue to improve and uncertain parameters get further
constrained by increasingly diverse measurements.
[28] The modified LSM/dry deposition/bidirectional box
model has been shown to compare well to NH3 flux measurements for both a high flux period about 3 weeks after fertilizer
application (Figure 3) and a period about 3 weeks later when
fluxes had dropped by about an order of magnitude (Figure 4).
The modified model was also re-applied to the soybean study
with similarly good results compared to measured fluxes
(Figure 2). These experiments, while limited to two sites,
give confidence in the realism of the model since they span a
wide range of fertilizer amounts including heavily fertilized
corn which covers very large acreage in North America
and contributes a substantial portion of the NH3 input to
the atmosphere.
[29] The bidirectional exchange NH3 resistance model is
part of a comprehensive modeling system which includes
an air quality model coupled to a farm and nutrient management model. This modeling system estimates the temporal
3804
PLEIM ET AL.: AMMONIA BIDIRECTIONAL FLUX MODEL
and diurnal dynamics of NH3 exchange from managed
agricultural sites with a range of fertilizer inputs. In addition
to the measurement campaigns over two commercial crops
in North Carolina, the evaluation of these models over
a wider range of natural (e.g., grassland and forest) and
managed land cover types are underway, and expansion to
more diverse climates and model intercomparison studies
will likely lead to the development of more robust NH3
bidirectional exchange parameterizations. The new NH3
bidirectional flux capability in the WRF-CMAQ regional air
quality modeling system enhances the scientific credibility of
nitrogen surface exchange processes. In addition, the new
WRF-CMAQ-EPIC coupled system will facilitate the identification of soil and nutrient management options for managed
agricultural systems.
Disclaimer
[30] Although this work was reviewed by EPA and approved for publication, it may not necessarily reflect official
Agency policy. Mention of commercial products does not
constitute endorsement by the agency.
References
Appel, K. W., K. M. Foley, J. O. Bash, R. W. Pinder, R. L. Dennis, D. J. Allen,
and K. Pickering (2011), Multi-resolution assessment of the community
Multiscale Air Quality (CMAQ) model v4.7 wet deposition estimates for
2002–2006, Geosci. Model Dev., 4, 357–371.
Bash, J. O., J. T. Walker, G. G. Katul, M. R. Jones, E. Nemitz, and W. Robarge
(2010), Estimation of in-canopy ammonia sources and sinks in a fertilized
Zea Mays field. Environ. Sci. Technol., 44, 1683–1689.
Bash, J. O., E. J. Cooter, R. L. Dennis, and J. E. Pleim (2013), Evaluation of
a regional air-quality model with bidirectional NH3 exchange coupled to
an agro-ecosystem model, Biogeosciences, 10, 1635–1645.
Clapp, R. B., and G. M. Hornberger (1978), Empirical equations for some
soil hydraulic properties, Water Resour. Res., 14, 601–604.
Choudhury, B. J., and J. L. Monteith (1988), A four-layer model for the
heat budget of homogeneous land surfaces, Q. J. Roy. Meteorol. Soc.,
114, 373–398.
Cooter, E. J., J. O. Bash, J. T. Walker, M. R. Jones, and W. Robarge (2010),
Estimation of NH3 bi-directional flux from managed agricultural soils,
Atmos. Environ., 44, 2107–2115.
Cooter, E. J., J. O. Bash, V. Benson, and L. Ran (2012), Linking agricultural
crop management and air quality models for regional to national-scale
nitrogen assessments, Biogeosciences, 9, 4023–4035.
Cosby, B. J., G. M. Hornberger, R. B. Clapp, and T. R. Ginn (1984), A
statistical exploration of the relationships of soil moisture characteristics
to the physical properties of soils, Water Resour. Res., 20, 682–690.
Dennis, R., R. Haeuber, T. Blett, J. Cosby, C. Driscoll, J. Sickles, and
J. Johnston (2007), Sulfur and nitrogen deposition on ecosystems in the
United States. EM: Air and Waste Management Association’s Magazine
for Environmental Managers, 12–17.
Driscoll, C. T., G. B. Lawrence, A. J. Bulger, T. J. Butler, C. S. Cronan,
C. Eagar, K. F. Lambert, G. E. Likens, J. L. Stoddard, and K. C. Weathers
(2001), Acidic deposition in the northeastern United States: sources and
inputs, ecosystem effects, and management strategies, Bioscience,
51, 180–198.
Erisman, J.W., A. van Pul, and P. Wyers (1994), Parameterization of dry
deposition mechanisms for the quantification of atmospheric input to
ecosystems, Atmos. Environ., 28, 2595–2607.
Flechard, C. R., et al. (2011), Dry deposition of reactive nitrogen to European
ecosystems: A comparison of inferential models across the NitroEurope
network, Atmos. Chem. Phys., 11, 2703–2728.
Flechard, C. R., C. Spirig, A. Neftel, and C. Ammann (2010), The
annual ammonia budget of fertilised cut grassland—Part 2: Seasonal
variations and compensation point modeling, Biogeosciences, 7, 537–556,
doi:10.5194/bg-7-537-2010.
Gilliland, A.B., R. L. Dennis, S. J. Roselle, and T. E. Pierce (2003),
Seasonal NH3 emission estimates for the eastern United States based on
ammonium wet concentrations and an inverse modeling method,
J. Geophys. Res., 108, 4477, doi:10.1029/2002JD003063.
Gilliland, A. B., K. W. Appel, R. W. Pinder, and R. L. Dennis (2006),
Seasonal NH3 emissions for the continental United States: Inverse model
estimation and evaluation, Atmos. Environ., 40, 4986–4998.
Hamaoui-Laguel, L., F. Meleux, M. Beekmann, B. Bessagnet, S. Génermont,
P. Cellier, L. Létinois (2012), Improving ammonia emissions in air quality
modelling for France, Atmos. Environ., doi: 10.1016/j.atmosenv.2012.08.002.
Haywood, J. M., and O. Boucher (2000), Estimates of the direct and indirect
radiative forcing due to tropospheric aerosols: A review, Rev. Geophys.,
38, 513–543.
Hicks, B. B., and M. L. Wesely (1978), An examination of some micrometeorological methods for measuring dry deposition. U.S. EPA Report, EPA-600/
7-78-116. Research Triangle Park, North Carolina.
Izaurralde, R.C., J. R. Williams, W. B. McGill, N. J. Rosenberg, and
M. C. Quirogas-Jakas (2006), Simulating soil C dynamics with EPIC:
Model description and testing against long-term data, Ecol. Model. 192,
362–384.
Jones, M. R. (2006), Ammonia deposition to semi-natural vegetation, Ph.D.
Dissertation, University of Dundee, Nethergate, Dundee, UK.
Jones, M. R., I. D. Leith, D. Fowler, J. A. Raven, M. A. Sutton, E. Nemitz,
J. N. Cape, L. J. Sheppard, R. I. Smith, and M. R. Theobald (2007a),
Concentration-dependent NH3 deposition processes for mixed moorland
semi-natural vegetation, Atmos. Environ., 41(10), 8980–8994.
Jones, M. R., I. D. Leith, J. A. Raven, D. Fowler, M. A. Sutton, E. Nemitz,
J. N. Cape, L. J. Sheppard, and R. I. Smith (2007b), Concentrationdependent NH3 deposition processes for moorland plant species with
and without stomata, Atmos. Environ., 41(10), 8980–8994.
Kondo, J., N. Saigusa, and T. Sato (1990), A parameterization of evaporation
from bare soil surfaces, J. Appl. Meteorol., 29, 385–389.
Lovett, G. M., and T. H. Tear (2008), Threats from above: Air pollution
impacts on ecosystems and biological diversity in the Eastern United States,
The Nature Conservancy and the Cary Institute of Ecosystem Studies.
Massad, R.-S., E. Nemitz, and M. A. Sutton (2010), Review and
parameterisation of bi-directional ammonia exchange between vegetation
and the atmosphere, Atmos. Chem. Phys., 10, 10,359–10,386, doi:10.5194/
acp-10-10359-2010.
Milford, C., et al. (2009), Ammonia fluxes in relation to cutting and fertilization
of an intensively managed grassland derived from an inter-comparison of
gradient measurements, Biogeosciences, 6, 819–834.
Moldrup, P., T. Olesen, T. Yamaguchi, P. Schjønning, and D. E. Rolston
(1999), Modeling diffusion and reaction in soils: IV. The BuckinghamBurdine-Campbell equation for gas diffusivity in undisturbed soil, Soil
Sci., 164, 542–551.
Moldrup, P., T. Olesen, S. Yoshikawa, T. Komatsu, D. E. Rolston (2004),
Three-porosity model for predicting the gas diffusion coefficient in
undisturbed soil, Soil Sci. Soc. Am. J., 68, 750–759.
Nemitz, E., M. A. Sutton, J. K. Schjoerring, S. Husted, and G. P. Wyers
(2000), Resistance modelling of ammonia exchange above oilseed rape,
Agric. Forest Meteorol., 105(4), 405–425.
Noilhan, J., and P. Lacarrère (1995), GCM grid-scale evaporation from
mesoscale modeling, J. Clim., 8, 206–223.
Pinder, R.W., A. B. Gilliland, and R. L. Dennis (2008), The environmental
impact of NH3 emissions under present and future, Geophys. Res. Lett.,
35, L12808.
Personne, E., B. Loubet, B. Herrmann, M. Mattsson, J. K. Schjoerring,
E. Nemitz, M. A. Sutton, and P. Cellier (2009), SURFATMNH3: A model
combining the surface energy balance and bidirectional exchanges of
ammonia applied at the field scale, Biogeosciences, 6, 1371–1388.
Pleim, J. E., and A. Xiu (1995), Development and testing of a surface flux
and planetary boundary layer model for application in mesoscale models.
J. Appl. Meteor., 34, 16–32.
Pleim, J. E., A. Xiu, P. L. Finkelstein, and T. L. Otte (2001), A coupled land
surface and dry deposition model and comparison to field measurements
of surface heat, moisture, and ozone fluxes, Water Air Soil Pollut. Focus,
1, 243–252.
Pleim, J. E., and A. Xiu (2003), Development of a land surface model. Part
II: Data assimilation, J. Appl. Meteorol. Climatol., 42, 1811–1822.
Pleim, J. E. (2006), A simple, efficient solution of flux-profile relationships
in the atmospheric surface layer, J. Appl. Meteor. and Clim., 45, 341–347.
Pleim, J. E., and L. Ran (2011), Surface flux modeling for air quality
applications, Atmosphere, 2, 271–302.
Pope, C. A. (2000), 3rd epidemiology of fine particulate air pollution and
human health: Biologic mechanisms and who’s at risk? Environ. Health
Perspect., 108Suppl 4, 713–723.
Ramanathan, V., P. J. Crutzen, J. T. Kiehl, and D. Rosenfeld (2001), Aerosols,
climate, and the hydrological cycle, Science, 294, 2119–2124, doi:10.1126/
science. 1064034.
Rawls, W. J., D. L. Brakensiek, and K. E. Saxton (1982), Estimation of soil
water properties, Trans. of the ASAE, 25, 1316–1320.
Sakaguchi, K., and X. Zeng (2009), Effects of soil wetness, plant litter, and
under-canopy atmospheric stability on ground evaporation in the
3805
PLEIM ET AL.: AMMONIA BIDIRECTIONAL FLUX MODEL
Community Land Model (CLM3.5), J. Geophys. Res., 114, D01107,
doi:10.1029/2008JD010834.
Schuepp, P. H. (1977), Turbulent transfer at the ground: On verification of a
simple predictive model, Boundary-Layer Meteorology 12, 171–186.
Sellers, P. J., M. D. Heiser, and F. G. Hall (1992), Relations between
surface conductance and spectral vegetation indices at intermediate (100
m2 to 15 km2) length scales, J. Geophys. Res., 97(D17), 19,033–19,059.
Walker, J. T., M. R. Jones, J. O. Bash, L. Myles, W. Luke, T. P. Meyers,
D. Schwede, J. Herrick, E. Nemitz, and W. Robarge (2012), Processes
of ammonia air-surface exchange in a fertilized Zea Mays canopy, Biogeosciences Discuss., 9, 7893–7941.
Walker, J. T., W. P. Robarge, Y. Wu, and T. P. Meyers (2006), Measurement of bi-directional ammonia fluxes over soybean using the modified
Bowen-ratio technique, Agr. Forest. Meteorol. 138, 54–68.
Wesely, M. L. (1989), Parameterization of surface resistances to gaseous
dry deposition in regional-scale numerical models, Atmos. Environ., 23,
1293–1304.
Williams, J. R., A. D. Nick, and J. G. Arnold (1985), SWRRB, a simulator
for water resources in rural basins, ASCE Hydr. J., 111, 970–986.
Wyers, G. P., R. P. Otyes, and J. A. Slanina (1993), A continuous-flow
denuder for the measurement of ambient concentrations and surface
exchange of ammonia. Atmos. Environ., 27A, 2085–2090.
Wyers, G. P., and J. W. Erisman (1998), Ammonia exchange over coniferous
forest, Atmos. Environ., 32, 441–451.
Xiu, A. J., and J. E. Pleim (2001), Development of a land surface model.
Part I: Application in a mesoscale meteorological model, J. Appl.
Meteorol., 40, 192–209.
Zhang, L., L. P. Wright, and W. A. H. Asman (2010), Bi-directional airsurface exchange of atmospheric ammonia: A review of measurements
and a development of a big-leaf model for applications in regional-scale
air-quality models, J. Geophys. Res., 115, D20310, doi:10.1029/
2009JD013589.
3806