APPLICATION OF A NEW LAND-SURFACE, DRY DEPOSITION, AND PEL
MODEL IN THE MODELS-3 COMMUNITY MULTI-SCALE AIR QUALITY
(CMAQ) MODEL SYSTEM
Jonathan E. Pleim and Daewon W. Byun
Atmospheric Sciences Modeling Division, NOAA, Research Triangle Park,
NC 27711 (on assignment to the National Exposure Research Laboratory,
U.S. Environmental Protection Agency)
INTRODUCTION
The U.S. EPA has developed a new comprehensive air quality modeling system,
known as the Models-3 Community Multi-scale Air Quality (CMAQ) model (Byun and
Ching, 1999). The CMAQ system includes a comprehensive emissions processor, a
Chemical Transport Model (CTM), and a meteorology model. The community modeling
approach aims to provide a focal point for diverse model development research which will
lead to many alternative algorithms and model components that can be inter-compared and
evaluated. An early example of this process is the development of a more advanced
surface exchange, planetary boundary layer (PEL), and dry deposition model and its
incorporation into the CMAQ system.
Accurate simulation of air quality depends on realistic modeling of land surface and
PEL processes. Quantities that exert first-order control on trace chemical concentrations
and photochemistry include PBL height, temperature, wind speed, and dry deposition
velocity. These quantities are all directly related to air-surface exchange processes of heat,
moisture, momentum, and trace chemical species. In addition, cloud cover, which is
important for photolysis rates, is greatly influenced by surface flux and PBL processes. For
these reasons it is especially critical to apply realistic techniques for land-surface and PBL
modeling within an air quality system.
Like most air quality modeling systems, CMAQ divides the treatment of
meteorological and chemical/transport processes into separate models run sequentially. A
potential drawback to this approach is that it creates the illusion that these processes are
minimally interdependent and that any meteorology model with a good reputation is
adequate for air quality work. However, most mesoscale meteorology models are
developed for operational weather forecasting and meteorological research. These foci do
-------
not emphasize all the same critical capabilities as air quality applications. Conversely,
CTMs are often developed to accept basic meteorological inputs from a variety of sources
with little regard to its quality and even less regard to consistency between physical
parameterizations in the meteorology model and the CTM, The work reported here
attempts to address some of these weak links in the system, particularly where
improvements in land-surface modeling in the meteorology model and consistency with
similar components in the CTM can have significant effects on the air quality simulation.
Therefore, this development cuts across several system components. A new land-surface
mode,! (LSM), which features explicit simulation of soil moisture and vegetative
evapotranspiration, has been coupled with the Fifth Generation Penn State/NCAR
Mesoscale Model (MM5). An attendant dry deposition model has been developed to take
advantage of the more sophisticated treatment of surface fluxes, stomatal conductance, and
surface layer diffusion in the new LSM. The Meteorology Chemistry Interface Processor
(MCIP) has been modified to include the new dry deposition model as well as to make the
additional information resulting from the new LSM available to the CTM. Also, a new non-
local closure PEL scheme that is compatible with the modifications made to the MM5 has
been added to the list of vertical diffusion module options of the CMAQ CTM.
This paper is not meant to be an evaluation of the models described here, since many
of the components have been evaluated against field experiment data and reported
elsewhere, but rather a demonstration that such modeling developments are important to air
quality modeling objectives. Comprehensive evaluation of the CMAQ system, including
these components, is ongoing.
MODEL DESCRIPTION
There has recently been heightened appreciation for the importance of more
sophisticated land surface models (LSMs) in mesoscale meteorology models. Currently,
there are at least 3 new LSMs being added to the MM5 system. Pleim and Xiu (1995)
describe the prototype development and testing of an LSM (hereafter referred to as PX) for
use in the MM5 and the CMAQ system. Pleim et al. (1996) describe testing and evaluation
of the LSM as implemented in the MM4 with the new attendant dry deposition model
through comparison to surface flux measurements of heat, moisture, and ozone dry
deposition at two field sites. Xiu and Pleim (2000) describe implementation into the MM5
and evaluation compared to field measurements of surface fluxes, temperature, and PBL
height. The PX LSM includes explicit simulation of soil moisture and temperature in two
soil layers and three pathways for surface evaporation including soil evaporation, canopy
evaporation, and evapotranspiration. The soil moisture model is based on the Interaction
Soil Biosphere Atmosphere (ISBA) model (Noilhan and Planton, 1989; Jacquemin and
Noilhan, 1990), which was specifically designed for mesoscale modeling and has an
extensive record of evaluation and comparison. Stomatal conductance is parameterized
according to root zone soil moisture, air temperature and air humidity, photosynthetically
active radiation (PAR), and several vegetation parameters such as leaf area index (LAI) and
minimum stomatal resistance. Although originally based on the ISBA model the stomatal
and canopy parameterizations are almost entirely new. New features include a canopy
shelter factor to account for shading within denser canopies, new stomatal functions with
respect to environmental parameters, and inclusion of a data assimilation scheme similar to
the technique described by Bouttier et al. (1993). A simple parameterization for describing
seasonal growth of vegetation, including leaf-out of deciduous trees, has also been
developed and tested.
-------
Dry deposition is a good example of the close interaction of chemistry and
meteorology. However, dry deposition models are usually part of the CTM and therefore
inconsistent with the LSM with regard to land-use and vegetation characterization as well
as aerodynamic, canopy, and stomatal resistance parameterizations. Clearly, the treatment
of common parameterizations and datasets should be as consistent as possible between the
LSM used in the meteorology model and the dry deposition scheme used in the CTM.
Therefore, the new dry deposition module for CMAQ (known as MSDry) directly uses the
same bulk stomatal resistance, aerodynamic resistance, and vegetation parameters such as
LAI, roughness length and vegetation coverage as in the PX LSM. Hence, the stomatal
uptake of gaseous trace chemicals is simulated in exactly the same way (adjusting for
molecular diffusivity) as evapotranspiration in the LSM. Other dry deposition pathways,
such as deposition to leaf cuticles and soil, are parameterized according relative reactivity
and solubility of the chemical species (see Byun and Ching, 1999 for more details).
The MSDry dry deposition model also utilizes PBL and surface layer parameters
directly from the MM5-PX. These codependencies not only ensure greater consistency
between meteorology and chemistry models but also benefit the dry deposition calculation
by using more responsive stomatal conductance simulations. In particular, the stomatal
conductance, and therefore the dry deposition, can respond to soil moisture conditions,
which are rarely considered in air quality models. Also, the indirect soil moisture data
assimilation scheme provides realistic constraints on the stomatal conductance.
PBL processes are another critical modeling component for air quality systems. The
PX LSM includes a simple non-local closure scheme known as the Asymmetric Convective
Model (ACM) (Pleim and Chang, 1992). The ACM is quite similar to the Blackadar non-
local scheme (Blackadar, 1978) which has long been the most widely used PBL option in
the MM5 system. Both schemes use non-local transport in the convective boundary layer
to simulate transport by rapidly rising buoyant plumes. The ACM differs from the
Blackadar scheme in its treatment of downward transport, which is prescribed to be local,
one layer at a time, to simulate gradual compensatory subsidence, hence asymmetrical.
MODEL COMPARISON
The modified CMAQ model system using the MM5 with the PX LSM (MM5-PX) and
the new dry deposition model was ran for a study period of July 6-16, 1995 at 36 km grid
resolution and July 11-15, 1995 on a 12 km nested grid. The 36 km domain covered
eastern United States and southeastern Canada while the 12 km nest covered most of the
Northeast US and southern Ontario. Two sets of CMAQ-CTM (CCTM) runs were made
using the MM5-PX and new dry deposition model; one with the standard eddy diffusion
model for vertical mixing and the other using the ACM in the CCTM. These two sets of
runs were compared to a set of base case runs which used the base case MM5 simulations,
with the Blackadar PBL scheme and static moisture availability factors for surface
moisture, and the RADM dry deposition model (Wesely, 1989). Other than the model
modifications outlined above the three sets of runs used the same model options and input
data (emissions, meteorology, and initial/boundary conditions) for both MM5 and the
CCTM as described by Byun and Pleim (2000).
Table 1, Model experiment configurations.
Experiment Meteorology Dry Deposition CCTM Vertical Mixing
Base Base MM5 RADM Eddy Diffusion
-------
PX
PX-ACM
MM5-PX
MM5-PX
MSDry
M3Dry
Eddy Diffusion
ACM
The three sets of model runs, referred to by experiment name as Base, PX, and PX-
ACM as shown in Table 1, were compared to all of the ozone measurements from the
EPA's AIRS monitoring network within the 12 km nested domain (208 sites). Figure 1
shows time series of ground level hourly ozone concentration averaged over the 208
measurements and the collocated 12 km grid cells from the three experiments. Note that
this is not a spatial average since the measurement sites are not evenly distributed. Also,
the AIRS sites tend to be located near urban or suburban areas. Thus, inclusion of the
AIRS data is meant as a rough reference for the models rather than an evaluation goal.
120
100 —
JD
a.
CL
C-
O
O
120
Figure 1. Modeled and measured ozone concentrations averaged over 208 AIRS monitoring sites.
With these disclaimers in mind, Figure 1 shows that the 3 models usually bracket the
observations during the peak ozone periods but over-estimate low concentrations. The
order of model ozone concentration magnitude is very consistent throughout the simulation
with PX, Base, PX-ACM from low to high during the afternoon peaks but the order
changes to PX, PX-ACM, Base at night and early morning. Compared to the observations,
the peaks are best simulated by either the Base or PX while the PX is closest-to the
observation in the troughs. Note that over-prediction of nocturnal ozone minima is an
endemic artifact of Eulerian grid models with coarse vertical resolution (the first layer
thickness is about 35 m). The PX-ACM is generally the highest at the daytime peaks. This
is because of much more rapid upward transport from the surface layer throughout the
convective boundary layer (CBL). Therefore, in the lowest model layer the ACM
simulates lower concentrations of surface emitted precursors, such as NOX and Isoprene,
but higher concentrations in the mixed layer resulting in faster ozone production aloft. The
higher concentration ozone quickly mixes throughout the CBL. As noted by Byun and
Pleim (2000) this apparent over prediction and over-vigorous upward transport may be an
artifact of the simplistic treatment of surface emissions. Since ACM is based on similarity
with sensible heat fluxes (as are most other PBL schemes) the lack of gradient diffusion
-------
surface layer fluxes for chemical emission may result in significant over-estimation of both
surface flux rate and layer 1 concentrations.
To help understand the differences between the Base and PX runs Figure 2 shows
PEL height, surface solar radiation, temperature at 1.5 m, and ozone dry deposition
velocity averaged over the same observation site locations as for Figure 1. On the average,
surface solar radiation is less for the PX case than the Base case. Other studies (Xiu and
Pleim, 2000) have shown that solar radiation is essentially identical between these two
versions of MM5 under clear sky conditions. Therefore, differences in the surface
radiatjon are caused by differences in cloud cover. Averaged over the entire 12-km domain
the cloud cover from the MM5-PX ranged from 0-20% (usually 5-10%) greater than from
the Base MM5 during this 5-day period. It is difficult to say which cloud cover simulation
is better because they both differed significantly from satellite photos. Hence, the
simulation of cloud cover is a critical weak link in current modeling systems. Techniques
for assimilation of satellite data are under development (McNider et al., 1998) that may
lead to substantial improvements in meteorology and air quality modeling.
PEL heights are generally comparable but with significant differences on some days in
either direction. The 1.5 m temperature simulations are similar but with higher
temperatures from the PX runs on the second and third days even though the average
surface radiation is less. It is interesting that the lesser average solar radiation in the PX
experiment does not result in generally lower temperatures or lower PBL height, which
demonstrates that the relationships between surface energy and PBL processes are
somewhat different for the PX LSM.
20 40 60 80 100 120
1000-,
Esoo
g
C600-
o
GJ
'•0400-
£C
ra200 -
6 0
Hour
310
60 80
Hour
Figure 2. Modeled meteorology parameters averaged over AIRS site locations: A) PBL height, B) surface
solar radiation, C) temperature at 1.5 m, D) ozone dry deposition velocity.
The ozone dry deposition velocity is considerably higher from the PX runs during both
day and night, with differences from the RADM model often up to about 0.1 cm/s. The PX
-------
results also show a more consistent morning peak in ozone deposition velocity (usually
about 9 LST) which is more realistic in vegetated areas (Finkelstein et al.s 2000), The
much higher nighttime deposition velocities simulated by the PX runs are partially
attributable to the minimum friction velocity of 0.1 m/s in the MM5.
120
120
Figure 3. Modeled and measured ozone concentration (top) and modeled NO, concentration (bottom) at the
AIRS site near Plymouth, NH.
Cause and effect relationships are hard to understand from analysis of averaged
quantities. Therefore, we have also selected a single site near Plymouth, NH for further
study. This particular site was chosen because of relatively large differences among the
models. Figure 3 shows the ozone and NOX concentrations from the 3 modeling
experiments along with the observations of ozone and Figure 4 shows the modeled
-------
meteorological parameters at this site. The solar radiation plot shows that both models
simulated the second and third days as essentially clear while the other three days had
various amounts of partial cloudiness with the PX simulation generally cloudier. The
concentration plots show considerable difference among the experiments particularly on
the third day. It is interesting that the base run shows very little diurnal change in ozone
compared to the other runs and the observations. The lesser peak NOX concentration on the
morning of the third day in the Base run explains why the Base ran shows no ozone trough.
The lower NOX peak is probably due to difference in the nocturnal PEL between the PX
and B(ase versions and the higher nocturnal ozone dry deposition velocities from the MSdry
model (Figure 4D). On the afternoon of the third day the Base ozone declines while the
observations and other two runs increase to a late afternoon peak. A similar relative dip in
temperature and PBL height is also evident in the Base MM5 output that is not reflected in
the radiation and therefore not caused by clouds. This is an interesting example of how
differences in meteorological simulations propagate through the chemical simulations.
120
20 40 90 80 tOO 120
CED08
toooe- -
Coo 4
CE002--
Q o
60
Hour
Figure 4. Modeled meteorology parameters at the AIRS site near Plymouth, NH: A) PBL height, B) surface
solar radiation, C) temperature at 1.5 m, D) ozone dry deposition velocity.
CONCLUSIONS
Preliminary analysis of the sensitivity of the CMAQ system to inclusion of a more
sophisticated land surface model with consistent treatment of dry deposition shows
substantial effects on one of the most important model outputs, namely ozone
concentration. The PX runs generally produce lower ozone concentrations than the base
runs. This difference is probably mainly due to the tendency of the MM5-PX to produce
greater amounts of cloudiness, which reduces photolysis rates and the greater ozone dry
deposition velocities from the MSDry deposition model. It is interesting that the greater
cloudiness simulated by the MM5-PX does not similarly reduce .temperature and PBL
-------
height compared to the Base MM5, Therefore, the PX LSM not only changes many of the
surface, PEL, and cloud parameters but also alters the relationships between them.
The Plymouth, NH case study does not show such systematic differences in ozone
concentration between the PX and Base runs as seen in the averaged results, particularly
during the clear period on days 2 and 3. This case does show considerable differences
caused by the different meteorological simulations such as the huge difference in
concentration during the early morning of the third day and the difference in the late
afternoon peak. The different dry deposition models may play a significant role in the
nighttime differences but probably not during the daytime when the dry deposition
velocities from the two models were very similar (see Figure 4D). It will take further
analysis to differentiate the effects of the model changes in meteorology and dry
deposition. ,
This preliminary study demonstrates the value of a flexible, comprehensive, air quality
modeling system for developing and evaluating new modeling techniques. Evaluations of
modeling advancements need to be performed at both the component level and the
integrated level to assess relevance to the ultimate products. In this way the greatest
attention can be paid to the weakest links in the system.
This paper has been reviewed in accordance with the US Environmental Protection
Agency's peer and administrative review policies and approved for presentation and
publication. Mention of trade names or commercial products does not constitute
endorsements or recommendation for use.
REFERENCES
Blackadar, A.K., 1978, Modeling pollutant transfer during daytime convection, in: Preprints, Fourth Symp, on
Atmospheric Turbulence, Diffusion, and Air Quality, Reno, NV, Amer. Meteor. Soc, 443-447.
Bouttier, F,, Mahfouf, J.F., and Noilhan, J,, 1993, Sequential assimilation of soil moisture from atmospheric low-
level parameters. Part I: Sensitivity and calibration studies, /. Appl. Meteor. 32:1335-1351.
Byun D.W., and Ching, J.K.S., ed., 1999, Science Algorithms of the EPA Models-3 Community Multiscale Air
Quality (CMAQ) Modeling System, NERL, Research Triangle Park, NC, EPA/600/R-99/03Q.
Byun D.W., and Pleim, J.E., 2000, Sensitivity of ozone and aerosol predictions to the transport algorthms in the
Models-3 Community Multi-scale Air Quality (CMAQ) Model, this volume.
Finkelstein P.L., Ellestad, T.G., Clarke, J.F., Meyers, T.P., Schwede, D.B., Hebert, E.G., and Neal, J.F., 2000,
Ozone and sulfur dioxide dry deposition to forests: Observations and model evaluation, Accepted by J.
Geophys. Res.
Jacquemin, B., and Noilhan, J., 1990, Sensitivity study and validation of a land surface parameterization using the
HAPEX-MOBILHY data set, Bound.-Layer Meteor. 52:93-134.
McNider, R. T., W. B. Norris, D. Casey, J, E. Pleim, S. J. Roselle, W. M. Lapenta, 1998, Assimilation of
satellite data in regional air quality models. In: Air Pollution Modeling and Its Application XII, Gryning
and Chaumerliac, Eds, Plenum Press, New York.
Noilhan, J., and Planton, S., 1989, A simple parameterization of land surface processes for meteorological models.
Mon. Wea. Rev. 117:536-549.
Pleim, J.E., and Chang, J.S., 1992, A non-local closure model for vertical mixing in the convective boundary layer,
Atmos. Environ., 26A:965-981.
Pleim, I.E., Clarke, J.F., Finkelstein, P.L., Cooler, E.J., Ellestad, T.G., Xiu, A., and Angevine, W.M., 1996:
Comparison of measured and modeled surface fluxes of heat, moisture and chemical dry deposition, In:
Air Pollution Modeling and Its Application XI, Gryning and Schiermeier, Eds, Plenum Press, New
York.
Pleim, I.E., and Xiu, A., 1995, Development and testing of a surface flux and planetary boundary layer model for
application in mesoscale models. /. Appl. Meteor., 34:16-32.
Wesely, M.L., 1989, Parameterization of surface resistances to gaseous dry deposition in regional-scale
numerical models, Atmospheric Environment, 23:1293-1304.
Xiu, A., and Pleim, J.E., 2000, Development of a land surface model part I: Application in a mesoscale
meteorology model, Accepted by J. Appl. Meteor. .
-------
NEKL-RTP-AMD-00-055
TECHNICAL REPORT DATA
1. REPORT NO.
EPA/600/A-00/009
3.RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Application of a new land-surface, dry deposition, and PEL
Model in the Models-3 Community Multiscale Air Quality
(CMAQ) Model System
5.REPORT DATE
6.PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Daewon W. Byun and Jonathan E. Pleim
8.PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Same as Block 12
10.PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Office of Research and Development
National Exposure Research Laboratory
Research Triangle Park, NC 27711
13.TYPE OF REPORT AND PERIOD COVERED
Proceedings, FY-00
14. SPONSORING AGENCY CODE
EPA/600/9
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Like most air quality modeling systems, CMAQ divides the treatment of meteorological and chemical/transport processes into separate models run sequentially. A potential drawback to this
approach is that it creates the illusion that these processes are minimally interdependent and that any meteorology model with a good reputation is adequate for air quality work. However, most
mesoscale meteorology models are developed for operational weather forecasting and meteorological research. These foci do not emphasize all the same critical capabilities as air quality
applications. Conversely, CTMs are often developed to accept basic meteorological inputs from a variety of sources with little regard to its quality and even less regard to consistency between
physical parameterizations in the meteorology model and the CTM. The work reported here attempts to address some of these weak links in the system, particularly where improvements in
land-surface modeling in the meteorology model and consistency with similar components in the CTM can have significant effects on the air quality simulation. Therefore, this development cuts
across several system components. A new land-surface model (LSM), which features explicit simulation of soil moisture and vegetative evapotranspiration, has been coupled with the Fifth
Generation Penn State/NCAR Mesoscale Model (MM5). An attendant dry deposition model has been developed to take advantage of the more sophisticated treatment of surface fluxes,
stomatal conductance, and surface layer diffusion in the new LSM. The Meteorology Chemistry Interface Processor (MCIP) has been modified to include the new dry deposition model as well
as to make the additional information resulting from the new LSM available to the CTM. Also, a new non-local closure PBL scheme that is compatible with the modifications made to the MM5
has been added to the list of vertical diffusion module options of the CMAQ CTM.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/ OPEN
ENDED TERMS
c.COSATI
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This
Report)
UNCLASSIFIED
21.NO. OF
PAGES
20. SECURITY CLASS (This
22. PRICE
Page)
UNCLASSIFIED
------- |