xe/EPA
                                United States
                                Environmental Protection
                                Agency
                                Environmental Sciences Researc
                                Laboratory
                                Research Triangle Park NC 277
                                Research and Development
                                EPA-600/S4-81-068 Oct. 1981
Project  Summary
                                Impact  of  Topographic
                                Circulations  on  the
                                Transport  and  Dispersion of
                                Air  Pollutants
                                Richard T. McNider and Roger A. Pielke
                                  A numerical mesoscale model was
                                used to  examine slope flows and
                                classic mountain-plain air circulation
                                for idealized topography. Special
                                emphasis was given to turbulent
                                parameterization in the stable bound-
                                ary layer and to the unique character-
                                istics of turbulent mixing in the slope
                                flows. The numerical simulations for
                                idealized valley-plain configurations
                                produced results consistent with
                                previously described observations,
                                such as shallow sidewall flows, the
                                pooling of cool air in the valley, and a
                                deep mountain flow out of the valley.
                                A Lagrangian particle model, operated
                                in  the terrain-following coordinate
                                system of the mesocale  model, was
                                developed to examine pollutant trans-
                                port in the modeled circulations, while
                                a Markov statistical process was used
                                to evaluate turbulent dispersion.
                                Higher order turbulence parameters
                                needed for the statistical model were
                                directly computed from the numerical
                                model. Results of dispersion tests in
                                the modeled slope flows showed
                                enhanced vertical dispersion  in the
                                slope flows compared to flow over a
                                flat boundary and, importantly, that
                                normal surface scaling parameters for
                                pollutant dispersion, such as friction
                                velocity, were inappropriate for the
                                slope flows.
                                 This research was supported by U.S.
                                Environmenta.1 Protection Agency
                                Grant No. R806207010. Computer
                                support was provided by the National
                                Center for Atmospheric Research,
                                funded by the National Sciences
                                Foundation.
                                  This Project Summary was devel-
                                oped by EPA's Environmental Sci-
                                ences Research Laboratory, Research
                                Triangle Park, NC, to announce key
                                findings of the research project that is
                                fully documented in a separate report
                                of the same title (see Project Report
                                ordering information at back).

                                Introduction
                                  The overall objective of this research
                                was to use a  mesoscale, primitive
                                equation  model [The University of
                                Virginia Mesoscale Model (UVMM)] to
                                produce dynamically consistent flow
                                and turbulence fields in complex
                                terrain. Model-predicted fields were
                                then to be used to examine the transport
                                and dispersion of pollutants. To accom-
                                plish the overall objective, some im-
                                provements to the  mesoscale model
                                were required  to properly simulate
                                drainage flows in complex terrain, and
                                special techniques were developed to
                                fully utilize the model flow and turbu-
                                lence fields to examine point and area
                                sources of pollution.

                                Modeling the Nocturnal
                                Boundary Layer
                                  In the past, use of the UVMM and
                                most other mesoscale models has been

-------
primarily directed towards  examining
the convective boundary layer of classic
mountain-plain air circulation with an
emphasis on sea breeze flows or heated
up-slope flows. For polluted air,  the
stable  boundary  layer and drainage
flows are generally the most restrictive
to pollutant dispersal.  Also,  since
drainage flows are themselves  highly
dependent upon vertical mixing processes,
a good deal  of effort was expended to
incorporate  an  accurate  nocturnal
boundary  layer formulation into  the
model.  The  formation chosen  was  a
local exchange  coefficient scheme
proposed  by Blackadar  (1979). With
some modification, this  scheme  has
been included in the model.
  This technique, in conjunction with a
surface energy budget and a long-wave
radiation parameterization, appeared to
do as well in simulating  the nocturnal
boundary  layer as the more complex
numerical second-order closure schemes.
A  complete description  of the local
exchange coefficient model  is given in
McNider and Pielke (1981). McNider
and Pielke also examine boundary-layer
development over sloping terrain  and
show the importance  of the terrain-
induced pressure  gradient and mixing
processes to the generation of strong
nocturnal jets.
  The local mixing formulation allowed
a  dynamically consistent surface
boundary layer and turbulent field to be
generated over a terrain  obstacle. The
use of local  conditions (based on local
gradient  Richardson number) was
especially important to slope flow
simulation, since the shear zones above
the surface  produce the mixing  and
entrainment into the drainage flow.

Slope Flows
  In  complex terrain,  under synoptic
conditions associated with large scale
air stagnation, local mesoscale circula-
tions provide the primary dilution  and
transport mechanisms for pollutants. To
assess  transport and dispersion of
pollutants in complex terrain, the  first
step is  to understand and  model  the
mountain-valley wind systems. Defant
(1951) provides an excellent summary
of observations and descriptive models,
including the characteristics of slope
flows and the importance of these slope
flows in producing an along-the-valley
axis component (the  true  mountain
wind).
  The modeled valley  has the classic
mountain-plain configuration described
by Defant, and was purposefully accen-
tuated to have a high height-to-width
ratio (19 km wide, 2 km deep) with a
sidewall slope of .25 (14°) (see Figure 1).
Shallow drainage flows developed
quickly over both sidewalls in response
to surface cooling produced by a surface
energy budget. The depth of the slope
flow  and height  of the  downslope
velocity maximum  (Figure 2) appeared
to agree well with generic observations
for similar slopes.  However, the mag-
nitude of the modeled velocity generally
appeared larger  than most observed
maximums. In the model,  an upslope
return flow was also noted above the
shallow downslope flow.
  As mentioned, the slope flows devel-
oped  rapidly, reaching their maximum
within the first hour. These slope flows
gradually produced a pooling of cool air
in the valley. Later, downslope velocities
decreased  throughout the night in
response to increased temperature
stratification in the valley.
  The turbulent  exchange processes
were  directly linked to the developing
flow  fields. Downward  moving air
tended to deform the potential tempera-
ture field, producing a less stable profile
aloft. The return upslope flow accentu-   I
ated this deformation, actually producing
an  unstable layer.  Because the local
turbulent exchange coefficients in the
current model were dependent upon the
local gradient Richardson number, this
unstable layer, in  conjunction with the
wind shear near the top of the drainage
flow, produced a highly turbulent zone
aloft.  Figure 3 shows the exchange
coefficient  profile and Richardson
number over the slope.

Mountain Flows
  According to Defant (1951), the
mountain wind is driven by the slope
flows, filling the valley with cooled air
from the slopes. The  pool of cool air
creates a hydrostatic pressure gradient
between the interior of the valley and
plains outside the valley, causing a deep
flow out of the valley. Because the slope
flows are generally much shallower
than the deep mountain wind, their
importance to  pollutant transport  is
probably  secondary to the  mountain
wind that develops within a valley.
  In the model (after starting with an
adiabatic atmosphere) definite  pooling
 Figure  1.    Three-dimensional perspectives of the topographic configuration used
             in Case A. Ridge height is 2 km; valley width is 19 km. (Note that the
             vertical scale is exaggerated.)
                                                                                                                  4

-------
                  273
      300
                                                 	Downs/ope Component
                                                 	Potential Temperature
      200  -
I
      100  -
 Figure 2.    Downs/ope component and potential temperature profiles over the
             western sidewall of the two-dimensional valley at 0230 LST.
 of the cooled air occurred, as illustrated
 in Figure 4a by the vertical gradient of
 potential temperature. The cooling in
 the valley was unusual because it did
 not appear to  originate at the surface.
 Rather, cooling occurred at all heights
 within  the  valley, with  the  greatest
.cooling occurring near the surface. This
 appears to be consistent with the
 cooling pattern observed in deep valleys
 in Colorado (C. D. Whiteman, 1980). The
 pooling of cool  air  in  the valley,
 consistent with Defant's descriptive
 model, produced a deep flow out of the
 valley. Figure 4b shows this deep
 mountain flow which extends to near
 ridge height.
Pollutant Transport
  Most mesoscale flows, such as sea
breeze and mountain-valley flows, are
characterized by strong  vertical shear
(in  both direction and speed)  and
horizontal variability. In order to examine
pollutant transport in the mean meso-
scale  model wind  field,  a Lagrangian
particle model was developed. The
Lagrangian particle  model  used  a
volume-weighting scheme (Tuescher
and Mauser, 1974) to interpolate model
velocities in the variably  spaced model
grid  to the particle position. Initial
particle positions were transformed into
the terrain-following coordinate system,
and then advected in that system. For
                                        display purposes, particles were back-
                                        transformed into a normal z system. The
                                        particle model could then  produce
                                        streak! ines as well as trajectories from
                                        various positions  in  the  model flow.
                                        Figure 5 shows examples of streaklines
                                        produced  by the particle model  in a
                                        modeled  sea  breeze flow.  A  more
                                        complete description of  the particle
                                        model and related experiments is given
                                        in McNider and Pielke (1979).

                                        Grid Cell Pollutant Transport
                                        and Dispersion
                                          Pollutant plumes with scales larger
                                        than both the model  grid  size and the
                                        dominant turbulent scales can be
                                        transported  and diffused satisfactorily
                                        by using  the advection equation and
                                        gradient transfer theory as follows:
                                           3C _ -uj 3C +  3
                                           3t      3X|   3Xj
                                                                                                                0)
  \n the UVMM,  a  highly  accurate
advection  scheme employing an  up-
stream spline interpolation technique
(Mahrer and Pielke,  1978) was used
together with a forward time-weighted
diffusion scheme (Paegle, et al., 1976)
to solve Equation 1. Mean velocities and
diffusion  coefficients taken at meso-
scale model grid points were used to
transport and diffuse a passive pollutant.
Figure 6 shows particle concentrations
(emitted from an urban strip and area
sources over  the  land) diffused  and
transported within a sea breeze flow
using this scheme.

Point Source Transport and
Diffusion
  As previously mentioned, area sources
can generally be included by using the
conservation equation employing gra-
dient  transfer theory. However, point
sources cannot be directly used in  a
coarse grid numerical  model  using
Equation  1.  This is due to numerical
damping and the fact that plumes (small
when compared to the turbulent scale)
either do  not diffuse  in  a  gradient
fashion, or  require the effective  ex-
change coefficient (K) to be a function of
travel distance.
  Due to the highly sheared environ-
ment  and unsteady character  of  the
local  circulations, the subgrid scale
plume dispersion  could  not be  easily
parameterized using analytical methods.
An  alternative used in the present
investigation consisted of a conditioned
particle dispersion scheme that Hanna

-------
                                                        Turbulent Exchange
                                                        Coefficient

                                                        Local Gradient
                                                        Richardson Number
     10
                                   100
                                    150
               200
                                     ) x 10*
Figure 3.
Turbulent exchange coefficient over the western valley wall showing
pronounced maximum near the top of the slope flow and local gradient
Richardson number showing the  unstable layer aloft created by the
deformation of the temperature field.
(1979)  based  upon Taylor's  1921
theorem for particle dispersion,  in
which the particle velocity (using the u-
component as an example) is expressed
by:
                           an estimate of au and R consistent with
                           the model  boundary  layer.  This was
                           accomplished for the vertical  compo-
                           nent of the convective boundary layer
                           using;
             Up = U + U'
                    (2)
                                                   ixm —
                                                                       (4)
where u =the mesoscale wind, and
      u' = a turbulent component given
          by:
                                       where
       u'(t + r) = u'(t)
                    (3)
  The variable u" is a random compo-
nent dependent  upon the turbulent
energy, 

-------
released from a height of 8 m above the
western sidewall. Figure 9 gives the
root-mean-square vertical spread in the
three experiments.  Results indicated
that dispersion rates were dramatically
accelerated in the  slope flows. The
failure  of  the dispersion  curves to
coincide, which is to say dispersion is
not a function of the friction velocity
alone, points out that surface stress was
not a good indication of turbulent
energy in the slope flows.

Conclusions
  The major result in this study was the
success of a  numerical  model  to
reproduce the essential features of the
classic  mountain-plain circulation,
including the deep mountain flow out of
the valley. The success of the numerical
experiment was readily expected since,
unlike direct sea breeze circulations or
slope  flows, the mountain flow is a
secondary circulation and a true three-
dimensional feature. Part of the success
in the simulations must be attributed to
the turbulent parameterization scheme
incorporated into the  model, which
avoided physically and  numerically
troublesome superadiabatic layers aloft
as the cool air pools in the valley.
  The second major result of this study
was the  finding that Lagrangian particle
methods, coupled with the numerical
mesoscale model could directly examine
pollutant transport in the complex flow
field, and could provide a robust method
for  explicitly incorporating parameters
characterizing the turbulent dispersion.
Techniques described in this investi-
gation allowed these turbulence  pa-
rameters to be computed directly from
the mesoscale model  so they were
consistent with the dynamic solution.
  The models  and methodologies de-
veloped in this  research should be
useful for  guiding the design or inter-
pretation of dispersion experiments in
complex terrain. Also, the models could
be  used to  shorten or extend  the
parametric range of limited observations
in complex terrain.


References
Blackadar, A.K., 1979. High Resolution
Models of the Planetary  Boundary
Layer. In:  Advances  in  Environmental
and Scientific Engineering, Vol. 1,
Gordon  and Breach,  pp. 50-81.

Deardorff, J., 1974. Three-Dimensional
Numerical  Study of the Height and
Mean Structure of the Planetary
                            13.  15.   17.   19.  21.  23.  25.  27.   29.   31.
Figure 4a.    Contours of potential temperature at 2330 LST (y = 23). Contour
              interval is 1 K.
Boundary Layer. Bound.-Layer Meteor.,
15, 1241-1251.

Defant,  F.,  1951.  Local  Winds.  In:
Compendium  of Meteorology. Amer.
Meteor. Soc.,  Boston, Mass. pp.  655-
675.

Hanna, S.R., 1979.  Some Statistics of
Lagrangian and Eulerian Wind  Fluctu-
ations. J. Appl. Meteor., 18. pp.  518-
531.

Mahrer, Y. and R.A. Pielke, 1978. Test of
an Upstream Spline Interpolation Tech-
nique for the Advective Terms  in  a
Numerical Mesoscale Model. Mon.
Wea. Rev., 106. pp. 818-830.

McNider, R.T.  and R.A.  Pielke, 1979.
Application of the University of Virginia
Mesoscale Model to Pollutant Transport.
Proceedings of Fourth Symposium on
Turbulence, Diffusion and Air Pollution.
Amer. Meteor. Soc.,  Reno, Nev.

McNider,  R.T., S.R.  Hanna, and  R.A.
Pielke, 1980.  Subgrid Scale  Plume
 Dispersion in  Coarse Grid Mesoscale
 Models. Second Conference on Appli-
 cations of Air Pollution  Meteorology.
 New Orleans,  La.

 McNider,  R.T., and R.A. Pielke, 1981.
 Diurnal Boundary Layer  Development
 Over  Sloping Terrain. Submitted to J.
 Atmos. Sciences.

 Paegle, J., W.G. Zdunkowski, and R.M.
 Welch, 1976.  Implicit Differencing of
 Predictive Equations  for the Boundary
 Layer. Mon. Wea. Rev., 104. pp. 1321-
 1324.

 Smith, F.B., 1980. Personal Communi-
 cation on June 15, 1980.

 Tuescher, LH. and L.E. Hauser, 1974.
 Development  of Modeling Techniques
 for Photochemical Air Pollution.  EPA-
 650/4-74-003. pp. 91.

Whitenidn, C.D., 1980. Personal Com-
 munication to University of Virginia.
Nov. 19, 1980.

-------
        5.    7.
11.   13.   15.  17.  19.  21.   23.   25.  27.  29.   31.
Figure 4b.     Contours of the mountain wind (v-component) for the same time and
              location as 4a. Contour interval is 1 m/s.

-------
        I  1  I I  I  I  I  I  I I  1  I  I  I 1 I  I  I  I  I  I I  I  1  I  1  I
       3.  5.   7.  9.  11. 13. 15. 17. 19. 21. 23. 25. 27.  29. 31.
                                 X
(A) Plan view showing x-y distributions of particles.
Figure  5a.     Particle streaklines at 19OO EST from 50-m release heights.
       3.  5.   7.  9. 11. 13.  15. 17. 19.  21.23. 25.  27. 29. 31.
i
N
   0.

                                                    (B)
                           r
(Bj Vertical view, looking northward, showing x-y distributions of particles. Particles
were released at 10-min intervals, beginning at sunrise.

Figure 5b.    Particle streaklines at 19OO EST from 50-m release heights.

-------
I
N
       3   -
2   -
N
       2   -
                                                                    Sea
                                        Figure 6b.
Figure 6.    Contours of panicle concentrations in a two-dimensional version of the
             mesoscale model. Synoptic flow is from the east (R). Particles are emitted
             over land areas and the urban strip. Figure 6a shows particles well-
             mixed within the convective boundary layer,  while Figure 6b shows
             capture of particles within the nocturnal boundary layer carried out to
             sea in a shallow land breeze.

-------
       70"
N
                                                        Irwin (1979)

                                             X X X X  Deardorff (1974)

                                             • • •  • Model - Wangara 33

                                                        1400 LST



                                             I    I   I   I  I   I
                                            .5
                                               1.0
                                  o-w/W*
Figure 7.
Plot of model-extracted, scaled 
-------
      28.

      26.

      24.

      22.

      20.

      18.

      16.
^-
      14.

      12.

      10.

       8.

       6.

       4.

       2.



Figure 8.
       0.17
                                     0.15
2.   4.   6.   8.  10.  12.  14. 16. 18. 20.  22. 24. 26. 28. 30.  32. 34. 36.
                               X
    Trajectories of particles released an hour apart from a height of 2 m.
    Final heights (in km) are given on the side of the end of each trajectory.
          .1
                                               o  Flat boundary, no wind
                                               •  Slope case, no wind
                                               X  Slope case with wind
                                           10
                                       U*f/H
100
Figure 9.    Vertical dispersion of the three cases described.  Root-mean-square
             spread is scaled by the boundary layer height.
                                  10

-------
Richard T. McNider and Roger A. Pielke are with the Department of Environ-
  mental Sciences. University of Virginia, Charlottesville, VA 22903.
George C. Holzworth is the EPA Project Officer (see below).
The complete report, entitled "Impact of Topographic Circulations on the Trans-
  port and Dispersion  of Air Pollutants," (Order No. PB 82-102 435; Cost:
  $17.00, subject to change) will be available only from:
        National Technical Information Service
        5285 Port Royal Road
        Springfield, VA 22161
        Telephone: 703-487-4650
The EPA Project Officer can  be contacted at:
        Environmental Sciences Research Laboratory
        U.S.  Environmental  Protection Agency
        Research  Triangle Park, NC 27711
                                                                              11
                                                                           U.S. GOVERNMENT PRINTING OFFICE: 1981 --559-092/3352

-------
f!  >-r

  WOf
    ^
  c   C
 m  c
     
                                                                   o>
                                                                   oo
                                                            >Sf s.

-------