Implementation of an Urban Canopy Parameterization for Fine-Scale Simulations


IMPLEMENTATION OF AN URBAN CANOPY PARAMETERIZATION FOR FINE-SCALE SIMULATIONS
Tanya L. Otte1'" arid Avraham Lacser2
1 Atmospheric Sciences Modeling Division, NOAA ARL, Research Triangle Park, North Carolina
2lsrael institute for Biological Research, Ness Ziona, Israel
"On assignment to the National Exposure Research Laboratory, U.S. EPA
1.	INTRODUCTION
The Pennsylvania State University/National Center
for Atmospheric Research Mesoscale Model (MM5)
(Grell et al. 1994) has been modified to include an urban
canopy parameterization (UCP) for fine-scale urban
simulations (~1-km horizontal grid spacing). The UCP
accounts tor drag exerted by urban structures, the
enhancement of turbulent kinetic energy (TKE)
especially near the tops of the buildings, and the
modification of the energy budget within the urban
canopy (i.e., from the surface to the tops of buildings)
This UCP is applied to grid cells in MM5 that have a
non-zero fraction of urban land use. This refinement o1
MM5 is targeted to enable the Community Multiscale Air
Quality (CMAG) Modeling System (Byun and Ching
1999) to capture the details o( pollutant spatial
distributions in urban areas, Preliminary results will be
presented below,
2.	URBAN CANOPY PARAMETERIZATION
2.1 Momentum
The horizontal components ol the momentum
equations are modified to account for the area average
effect of the sub-grid urban elements following Brown
(2000). The modifications are implemented in MM5 via
the TKE-based Gavno-Seaman planetary boundary
layer (PBL) parameterization scheme (e.g., Shatran
et al. 2000). The momentum equations accounting loi
the urban elements are:
dU	I 2 pxO-5
— = f u - 0,5/urbCd A z) U (U'2 + V2)
Y = Fv - Q.5UC^z)v(i/ + VZf5
= F» + 0.5furbC^(z)(c/J + V* + W3f'
where the Fs are the general forcing terms in each
equation; U, V, and V\f are the wind components; and
TKE is the turbulent kinetic energy. In this formulation,
it is assumed thai the buildings aftect the flow but oo noi
take up any volume within the grid cell. is a drag
coefficient. The urban traction of the grid cell i;
described by fmt.. fi(z) is the canopy area density, or
the surface area of the obstacle (e.g., building}
perpendicular to the wind, per unit volume of the urban
canopy, expressed in rrf1. There are several
approaches to describe A,i) (e.g., Uno et al, 1985:
Brown 2000) where the function reaches its maximum at
the ground level, but vanishes at the top of the
obstacles (so the drag term vanishes also at that level).
The integral of At,z) from the ground level to the tops of
the tallest buildings (H) is Kf, which corresponds to the
ratio of the frontal area to the total surface area of the
buildings. In general A,z) is a function of the location
within the domain as it depends on building morphology.
A.z) can be estimated from \f and H assuming some
functional form for A(z); here we use a linear function.
To solve the modified momentum equations (the
added new term), we follow the analytical solution
suggested by Byun and Arya (1986), The TKE equation
is solved explicitly. To take proper account of the
influence of A(z), the vertical resolution in MM5 is
increased in the domain such that several prognostic
layers are below H.
2.2	Energy Budget
To account for the impact of urban settings on the
energy budget, the anthropogenic heat flux is included
in the heat equation and not in the surface energy
budget (e.g., Chin et al. 2000). The anthropogenic heat
flux is set as a function of urban land use subcategory,
and it has a temporal weighting function (e.g., Taha
1999). The heat equation also includes the heat
contribution of the city canyons following Yamada
(1982); the contribution due to rooftops has not yet been
implemented. The surface energy balance includes the
shadowing/trapping effect of the net radiation reaching
the ground in the city canyons modified by extinction of
the radiation through the urban canopy using a simple
exponential function (e.g.. Brown 2000).
2.3	Urban Morphology
Parameters that are required for the UCP (e.g., H
and Kf) can be extracted Irom digital imagery (e.g.,
Ratti et al. 2001) which is commercially available from
several vendors tor various cities and with different
degrees of accuracy and precision. We did not have
access to a true urban morphology database for our
area of interest, so we modified the MM5 land use
database for our domain to have seven subcategories of
urban areas adapted Irom R. Eilefsen (personal
communication 2001). These categories crudely
represent urban zones such as high-rise, industrial, and
urban residential. These urban zones are used to
create pseudo-morphology lor the region of interest, and
the physical characteristics of these areas are used In
the UCP. Each of the urban subcategories has a
different value for H, Xf, canyon traction, and maximum
anthropogenic heal flux.
'Corresponding author sddress: Tanya Ctte, U.S. EPA.
MD E243-03. RTP, NC 27711; otte.tanyaigepa.gov

-------
3. PRELIMINARY RESULTS
The unmodified MM5 was run in a one-way nested
mode for several days in July 1995 during which there
was a high-ozone episode in the Northeastern U.S. that
coincided with a photochemical field study. The MM£
domains included a five-domain configuration (108, 36,
12, 4, and 1.33 km grid spacing). The first four domains,
were run with 30 vertical layers (about 12 layers in tht
PBL, and lowest level at 19 m) and physics options
appropriate tor each resolution. To include the influence
of smaller obstacles, the UCP was used on the 1,33-km
domain (Figure 1} with 40 layers that included ten new
layers in the lowest 100 m (lowest level at 2 m).
Several simulations have been made with the
1.33-km domain to determine the impact of the UCF
and the morphology on the simulation. Four 1.33-krr.
simulations are discussed here. The first experiment,
nocari30, is a 30-iayer experiment without the UCF.
The second experiment, nocan40, is a 40-layei
experiment without the UCP. Both nocdn30 and
nocanAO are shown so that the impact of the UCP on
the simulation can be evaluated independent of the
increase in vertical resolution. Expt. morphAO includes
the UCP with urban morphology. Expt. homog40
includes the UCP and a homogeneous representation ot
the city defined by the ' average" of the morphology.
Figure 2 is a comparison of the surface (~2-m)
temperature for Philadelphia International Airport (PHL)
from the four 1.33-km experiments compared tc
observations lor 14 July 1995. Figure 2 shows that the
two experiments with the UCP improved the simulated
nighttime temperatures by ~2"C and improved the
maximum temperatures by -0 5°C. In the mid-
afternoon, the addition of the ten new layers improver
the verification, as the 40-layer experiment werf
superior to nocari30 Overall, the diumat temperature
patterns of the simulations with the UCP compare morf
favorably with observations than the two experiment
without the UCF
As expected tor a non-urban location, Millville, New
Jersey (MIV), the use of the UCP had little effect on tht
diurnal pattern of the temperature (not shown). The tou;
simulations are very similar tor nighttime lemperaluret
tn addition, the same mid-afternoon impact of the ten
new layers occurs at MIV as well as PHL. Thi:
suggests that there are no detrimental effects of tht
UCP on areas beyond the urban core.
Verification statistics were calculated against thf
National Weather Service standard observation site:
shown in Figure 1. Root-mean-square error (RMSE)
statistics reflect the nighttime improvements in Iht
temperature in 1he urban ateas in the UCP simulations
Over the 24-h simulation period, the two experiment:
with the UCF outperformed the two experiment!
without. Expt. homogAO has an RMSE of 1.9X, and tht
RMSE in morphAO is 2.0'C. Expt. nocanSO has ar
RMSE of 2.2'C while nocanAO has an RMSE of 2.4'C
The index of agreement (e.g.. Willrnott, 1982) for thf
same period reflects the same ordinal ranking of tht
experiments and a larpe separation between the UCF
and no-UCP simulations. These statistics tend to show
that the use ot the UCP generally improves tht
temperature simulations throughout the domain
compared to the "standard* 30-iayer configuration, and
the improvements are not solely due to the increase in
vertical resolution from 30 to 40 layers.
The hourly RMSE comparison for moisture (not
shown) indicates that the 40-layer experiments are
significantly better than nocan30. However, morpMO
verified best of the four experiments (2.11 g/kg), while
homogAO verified third (2.21 g/kg).
The wind speed timeseries for various sites in the
domain do not show a clear "winner", potentially due to
the more erratic hourly changes in wind speed.
However, there is clearly an effect on the wind speed
timeseries both in the urban areas (e.g., PNE, not
shown) and downstream (e.g., WRI, not shown). The
RMSE over the 24-h simulation (not shown) indicates
that morphAO is the best of the experiments (1.32 m s"1),
Expt. homogAO was the worst of the four experiments
(1.58 ms"1) for wind speed. The bias (model minus
observation) for the same time period (Figure 3) clearly
shows that morphAO is the best of the experiments
(-0.31 ms"1), followed by nocartSQ (+0.75 m s"1),
nocanAO (-0.91 m s"'}, and homogAO (-0.96 m s"1). This
is consistent with the RMSE scores discussed above.
The UCP had very little effect on wind direction in the
timeseries and the statistical scores over the same
simulation period. Thus the UCP with morphology has a
positive impact on the simulations of winds at 1.33-km.
4. CLOSING THOUGHTS
Initial tests with the UCP for simulations on 14 July
1995 show that the UCP at 1,33-km tends to produce
the desired effects on the wind, temperature, and TKE
fields. The temperature simulations with the UCP tend
to be superior to 30-layer and 40-layer simulations
without the UCP. in addition, the UCP simulations with
pseudo-morphology lend to be superior to the UCP
simulations with an "average' homogeneous
morphology (i.e., no urban subcategories), particularly
tor wind speed. Further evaluation of the UCP in MM5
on additional days is underway.
Future plans include expanding the UCP to add the
roof energy budget and to use more detailed land use
databases and morphology databases so that the urban
at«..{is can be more accurately characterized. Also, the
UCP will be coupled with a land-suriace model and
urban soil model.
Disclaimer, The information in this manuscript has been
piepared under funding by the United States
Environmental Protection Agency. It has been
subjected to Agency review and approved for
publication. Mention of trade names or commercial
products does not constitute endorsement or
recommendation for use.
6. REFERENCES
biown, M, J,, 2000: Urban parameterizatioris for
mesoscale meteorological models Mcsoscale
Atmospheric Dispersion. Ed., Z, Boybeyi,
Byun, D. W., and S. P. S. Arya, 1986: A study of mixed
layer momentum evolution. Atmos. Environ., 20,
715-728.

-------
Byun, D, W,, and J. K. S. Ching, 1999: Science
algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System,
EPA-600/R-99/030, U.S. EPA.
Chin, H. S , M. L, Leach, and M J. Brown, 2000: A
sensitivity study of the urban effect on a regional
scale model: an idealized case. Preprints, Third
Sym. on the Urban Environment, Davis, CA, Amer.
Meteor. Soc.
Greli, G„ J. Dudhia, and D. R. Stauffer, 1994: A
description of the fifth-generation Penn State/NCAR
Mesoscale Model (MM5). NCAR Tech. Note,
NCAR/TN398+STR, 138 pp
Ratti, C„ S. Di Sabatino, R. E. Britter, M. J. Brown, F.
Caton and S. Burian, 2001: Analysis of 3-D urban
databases with respect to pollution dispersion for &
number of European and American cities.
Proceedings, Third Int. Conf. on Urban Air Quality,
Loutraki, Greece,
Shafran, P. C.. N. L Seaman, and G, A. Gayno, 2000:
Evaluation of numerical predictions of boundary
layer structure during the Lake Michigan Ozone
Study. J. Appl. Meteor., 39, 412-426
Taha, H., 1999: Modifying a mesoscale meteorological
model to better incorporate urban heat storage: A
bulk parameterization approach. J. Appl. Meteor.,
38, 466-473.
Uno, I., H. Ueda, and S, Wakamaisu, 1989: Numerical
modeling of the nocturnal urban boundary layer.
Bound-Layer Meteor., 49, 77-98.
Willmott, C. J., 1982: Some comments on the evaluation
of model performance. Bull Amer. Meteor. Soc.,
63, 1309-1313.
Yamada, T., 1982: A numerical model study of turbulent
airflow in and above a forest canopy. J. Met. Soc.
Japan, 60 (1}, 439-454.
TTN
NEL
PNl
WRI
ILG
ACY
MIV
Figure 1. The s'aooard National Weather Servicf
observation sites used for verification on the 1 33-krr,
domain.
4
2
0
•2
¦4
0
24
time (h)
Figure 2. Temperature timeseries (model minus
observation) of tour 1.33-km experiments compared to
observations lor Philadelphia International Airport (PHL)
for 14 July 1995. Time is in U1C.
1.5
0.5
If
-0.5
ft
-1.5
time 
-------
TECHNICAL REPORT DATA
1. REPORT NO.
2,
3 .
4, TITLE AND SUBTITLE
Implementation of an urban canopy parameterization
for fine-scale simulations
5.REPORT DATE
6.PERFORMING ORGANIZATION CODE
7, authorCS)
T.L. Gtte, and A. Lacser
8.PERFORMING ORGANIZATION REPORT NO.
9, PERFORMING ORGANIZATION NAKE 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
14. SPONSORING AGENCY CODE
EPA/600/9
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The Pennsylvania State University/National Center for Atmospheric Research Mesoscale Model
(MM5) (Grell et al. 1994) has been modified to include an urban canopy parameterization (UCP)
for fine-scale urban simulations (~1-km horizontal grid spacing}. The UCP accounts for drag
exerted by urban structures, the enhancement of turbulent kinetic energy (TKE) especially
near the tops of the buildings, and the modification of the energy budget within the urban
canopy (i.e., from the surface to the tops of buildings). This UCP is applied to grid cells
in MM5 that have an non-zero fraction of urban land use. This refinement of MM5 is targeted
to enable the Community Multiseale Air Quality (CMAQ) Modeling System (Eyuis and Ching 1999)
to capture the details of pollutant spatial distributions in urban areas. Preliminary results
will be presented below.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/ OPEN ENDED
TERMS
c.COSATI



18, DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This
Report)
UNCLASSIFIED
21. NO. Ci' PAGES
20. SECURITY CLASS (This
Page)
UNCLASSIFIED
22. PRICE
-------