I   600385064
                                                (7
 GREEN RIVER AIR QUALITY MODEL DEVELOPMENT
   VALMET  - A Valley Air Pollution Model
  ATMOSPHERIC SCIENCES RESEARCH LABORATORY
     OFFICE OF RESEARCH AND  DEVELOPMENT
    U.S.  ENVIRONMENTAL PROTECTION AGENCY
RESEARCH  TRIANGLE PARK, NORTH  CAROLINA 27711

-------
 GREEN RIVER AIR QUALITY MODEL DEVELOPMENT
   VALMET - A Valley Air Pollution Model

                     by
     C. D. Whiteman,  and K.  J. Allwine
        Pacific Northwest Laboratory
         Rich!and, Washington 99352
      Interagency Agreement AD89F20970
     with the U.S. Department of Energy
              Project Officer

               Alan H. Huber
    Meteorology and Assessment Division
  Atmospheric Sciences Research Laboratory
    U.S.  Environmental Protection  Agency
   Research Triangle Park,  NC 27711
  ATMOSPHERIC SCIENCES RESEARCH LABORATORY
     OFFICE OF RESEARCH AND DEVELOPMENT
    U.S.  ENVIRONMENTAL PROTECTION AGENCY
RESEARCH  TRIANGLE PARK, NORTH CAROLINA 27711

-------
                                 DISCLAIMER
     The research described in this report has been funded wholly or in part
by the United States Environmental Protection Agency through Interagency
Agreement AD-89-F-2-097-0 to the Pacific Northwest Laboratory.  This docu-
ment has been reviewed in accordance with U.S. Environmental Protection
Agency policy and approved for publication.  Mention of trade names or
commercial products does not constitute endorsement or recommendation for
use.

-------
                                   PREFACE
     This final report is submitted as part of the Green River Ambient Model
Assessment (GRAMA) project conducted at the U.S. Department of Energy's
Pacific Northwest Laboratory for the U.S. Environmental Protection Agency.
The GRAMA Program has, as its ultimate goal, the development of validated
air quality models that can be applied to the complex terrain of the Green
River Formation of western Colorado, eastern Utah and southern Wyoming.  The
Green River Formation is a geologic formation containing large reserves of
oil shale, coal, and other natural resources.  Development of these
resources may lead to a degradation of the air quality of the region.  Air
quality models are needed immediately for planning and regulatory purposes
to assess the magnitude of these regional impacts.  This report documents
one of the models being developed for this purpose within GRAMA—
specifically a model  developed to predict worst case air pollutant
concentrations caused by elevated continuous point sources of pollution
located in deep valley terrain.  The model shows promise for use as a
planning tool and eventually as a regulatory tool.  Further testing and
evaluations of the model are needed to gain a measure of confidence in the
model's performance.

-------
                                  ABSTRACT
     An air quality model  is described for predicting air pollution con-
centrations in deep mountain valleys arising from nocturnal down-valley
transport and diffusion of an elevated pollutant plume, and the fumigation
of the plume on the valley floor and sidewalls after sunrise.  Included is a
technical description of the model, a discussion of the model's applica-
tions, the required model  inputs, sample calculations and model outputs, and
a full listing of the FORTRAN computer program.
                                      IV

-------
                              EXECUTIVE SUMMARY
     Pacific Northwest Laboratory has developed an air quality model for
application in valleys for the U.S. Environmental Protection Agency  (EPA)
under their Green River Ambient Model Assessment (GRAMA) program.  This
program was initiated in response to the need for air quality assessment
tools applicable in the Green River Oil Shale Formation region of western
Colorado, eastern Utah, and southern Wyoming.  This region has the potential
for large-scale growth because vast energy resources, especially oil shale,
are located in the region.

     Following a thorough analysis of meteorological data obtained from deep
valleys of western Colorado, a modular air pollution model has been
developed to simulate the transport and diffusion of pollutants released
from an elevated.point source in a well-defined mountain valley during the
nighttime and morning transition periods.  This initial version of the
model, named VALMET, operates on a valley cross section at an arbitrary
distance down-valley from a continuous point source.  The model has  been
constructed to include parameterizations of the major physical processes
that act to disperse pollution during these time periods.

     The nighttime simulation uses a modified Gaussian plume to calculate
concentrations on the valley floor and sidewalls.  The Gaussian plume model
uses Briggs1 plume rise equations and Pasquill-Gifford diffusion coeffi-
cients, as modified for complex terrain diffusion enhancement.  Dilution of
the plume due to tributary flows can be handled in the model, and a  new
method has been incorporated to ensure conservation of plume mass as the
plume is transported down the valley axis.  The purpose of the nighttime
simulation is to provide air pollution concentrations on a cross-valley
section at sunrise, as an initial  condition for the daytime simulation.

     The daytime simulation uses numerical techniques and a valley energy
budget to predict how concentrations will vary in time as pollutants from
the elevated plume are fumigated onto the valley floor and sidewalls during
the post-sunrise temperature inversion breakup period.  The effects  of CBL
growth, temperature inversion subsidence, transport and diffusion in uoslope
flows, warm air advection above the valley, albedo, sensible heat flux, and
other physical processes are incoroorated in the model.  The daytime
simulation is driven by solar radiation and a simplified surface energy
budget, taking account of valley topography.

     Valley meteorological observations are required in order to obtain the
necessary input parameters to the model.  The required observations  can be
obtained using standard meteorological instrumentation, and the costs for

-------
the observations necessary to perform a simple screening analysis  are
expected to be within the means of a small industrial applicant wanting  a
permit for operation in a mountain valley.  The VALMET model could  form  the
basis for a more comprehensive site-specific model if further meteorological
and tracer data were available in a given valley.  The model outputs are in
a form suitable for regulatory decision making.  The model predicts the
maximum 1- and 3-hr-average concentrations for locations on the valley floor
and sidewalls.

     The model shows promise for use as a planning tool and, eventually  as a
regulatory tool.  Further development, testing, and tracer evaluations of
the model will be necessary before sufficient confidence can be gained to
justify the model's use in a regulatory setting.  The priorities for further
development and testing have been provided in the body of the report.  Test-
ing of the model's sensitivity to input parameters, and an initial  evalua-
tion of the model  with tracer experiment data are high priority tasks.
These tests will,  no doubt, result in future modifications to the  initial
version of the model.

     The authors stress that the model's ultimate utility in addressing  and
providing solutions to potential  air pollution problems in mountain valleys
will depend on the further evaluation of the model.  In order to have
confidence in model predictions it is necessary to test the model  against
actual air pollution data.  Several parameters in the model (AQ, k, a  ,  and
a ) are, at present, poorly understood for mountain valleys due to  a aearth
of experimental data and theoretical research.  We hope, by pointing out
these deficiencies, that attention will  be focused on the need for
information on both turbulent diffusion and valley energy budget studies.
The use of full physics models may help in providing some of the answers
necessary to improve the present model.

-------
                                  CONTENTS
Preface	  i i i
Abstract	   1v
Executi ve Summary	    v
Figures	   ix
Tab! es	  xi i
Acknowledgments	 xiii

     1.  Introduction	    1
     2.  Background	    4
               Nocturnal  Val 1 ey Meteorol ogy	    4
               Valley Meteorology During the Temperature Inversion
               Breakup Period	    5
     3.  Technical  Discussion	   14
               Coordinate System	   14
               Topographic Cross Section	   14
               Nocturnal  Model	   17
                    Pollutant Source	   17
                    PIume Rise	   18
                    Gaussi an PI ume Model	   21
                    Dispersion Coefficients	   22
                    Channel ing	   22
                    PI ume Di 1 uti on	   24
                    Reflections	   26
                    Calculation of Steady-State Nighttime
                        Concentrations	   31
               Daytime Model	   32
                    Extraterrestrial Solar Radiation	   33
                    Parameterization of Sensible Heat Flux	   36
                    Model Grid	   36
                    Thermodynamic Equations for CBL Ascent and
                        Inversion Descent	   38
                    Advection in the Slope Flows	   42
                    Pollution Concentration Calculation Method	   44
                    Exponential  Decay of Concentration	   47
                    Maximum 1- and 3-Hour Average Concentrations	   48
     4.  Overvi ew of Modul ar VALMET Model	   49
               Features of the Computer Code	   50
               Model  Inputs	   50
               Specification of the Model  Inputs by the User	   53
                    Valley Characteristics	   53
                    Date	   55
                    Site Location	   56

-------
                    Model  Characteristics..	   56
                    Val 1 ey Atmosphere	   57
                    Stack  Characteristics.	   57
                    Inversion Characteristics...	   58
                    Gaussian Plume Parameters	   58
                    Sensible Heat Flux	   60
               Error Messages	   62
     5.   Technical  Description of Individual  Modules..	   64
               VALMET-Mai n Program	   64
               INPUT-Input Module	   69
               JULIAN-Julian Day Module	   71
               PRISE-PIume Rise Module....	   72
               DILUTE-PIume Dilution Module	   74
               INGRAT-Valley Plume Reflection Module	   75
               NORMAL-Normal or Gaussian Curve Integration Module	   77
               GAUSS-Gaussi an PI ume Modul e	   78
               SOLAR-Extraterrestrial Solar  Radiation Module	   79
               EBDGT-Surf ace Energy Budget Modul e	   81
               DESCNT-Inversion Descent and  CBL Growth  Module	   82
               PROFIL-Concentrati on Prof i 1 e  Modul e	   84
               VELOCY-Upsl ope Fl ow Vel oci ty  Modul e	   86
               BRKUP-Pol 1 utant Mass Budget Modul e	   88
               PSTPRC-Post Processor Modul e	   90
     6.   Samp!e Model Runs	   92
               Simulation  Number 1	   92
               Simulation  Number 2	   97
     7.   Further Work	  100
               Guidance from the 1982 Experiments	  100
               Suggested Modifications to the VALMET Model	  103
                    Deposi ti on	  103
                    Emission Above or Below  Stable Core	  103
                    Energy Budget	  104
                    Cross  Vail ey Fl ows	  104
                    Turbulent Erosion of the Valley Inversion by
                        Overlyi ng Fl ows	  104
                    Effect of Tributary Flows on  the Enhancement of
                        Di f f usi on	  105
                    Diffusion Coefficients	  105
                    Time-Varying Wind Speeds in the Stable Core	  105
                    Temperature Inversion Buildup	  105
                    Differential Heating	  105
                    Stacks Located on SidewalIs	  106

References	  109
Appendices
     A.    FORTRAN Listing  of VALMET	  115
     B.    FORTRAN Listing  of VALMET Output Plotting Program	  147
     C.    Research Paper Entitled "Breakup  of Temperature Inversions
          in Deep Mountain Valleys:  Part II.  Thermodynamic Model"..  155
     D.    Summary of Modifications to VALMET	  173
                                    vn

-------
                                   FIGURES


Number                                                                  Page
   1    Three patterns of temperature structure evolution  during  the
          i nversi on breakup peri od	    7

   2    Typical mid-morning wind structure over and within  a  deep
          valley on the western slope of the Rockies,  illustrating
          the five interrelated wind systems identified  in  field
          studi es	    8

   3    Relationship between temperature structure layers  and wind
          systems	    9

   4    Dual  soundings from a valley floor and a valley  sidewall
          illustrating the up-slope flow found within  the  CBL over
          the si dewal 1	    9

   5    Illustration of the hypothesis of inversion destruction.
          In the center of the diagram cross sections  of a valley
          are shown at times t, , t^, t,, t^, and tg	   10

   6    Air pollution implications of CBL growth and inversion
          top descent	   12

   7    Illustration of the nocturnal down-valley transport and
          dispersion of a pollutant plume	   15

   8    Illustration of the local coordinate system on a cross
          val 1 ey secti on	   16

   9    Parameters used in the VALMET model to approximate  a
          valley topographic cross section	   16

  10    Illustration of the definitions of secondary topographic
          variables used in the model	   16

  11    Cross section of pollutant plume and valley topography
          illustrating the integral  method of calculation  for
          plume reflection and diffusion out the top of  the
          val 1 ey temperature i nversi on	   28
                                      ix

-------
Number                                                                 Page

  12    VALMET grid configuration on a valley cross section
          illustrating the nomenclature used in the model	  32

  13    Extraterrestrial  solar heat flux as a function of time,
          showing solar model nomenclature	  37

  14    Cross section of the valley floor and sidewalls illustrating
          the grid elements whose height corresponds to the CBL
          heights	  38

  15    Illustration of the effect of topography in controlling the
          heating rates of the air within a valley temperature
          inversion versus the air within an inversion over the
          pi ai ns	  39

  16    Diagram showing the changes in CBL and inversion depth at a
          given time step above a valley floor grid element	  43

  17    Representation of the volumetric element of mass incorporated
          into the growing CBL above a valley floor grid element at
          each model time step	  43

  18    Schematic diagram of an individual model  grid element
          illustrating the pollutant mass balance	  45

  19    Flow diagram of VALMET model, showing the modular structure
          of the model and the physical processes parameterized
          in the modules	  51

  20    Illustration of the VALMET model input table as it appears
          on the user's interactive screen	  54

  21    Listing of summary output file generated by Sample
          Simulation 1	  93

  22    Plots of Sample Simulation 1:  (a) nocturnal vertical
          concentration profile through plume centerline, (b) CBL
          height (long dashes) and inversion height (short dashes)
          as a function of time, and (c) nocturnal  cross-valley
          concentration profile through plume centerline	  95

  23    Pollutant concentration versus time for selected grid
          elements for Sample Simulation 1	  96

  24    Same as Figure 22, for Sample Simulation 2	  98

  25    Same as Figure 23, for Sample Simulation 2	  99

-------
Number                                                                 Page

  26    Illustration of differential solar flux on opposing sidewalls
          for Brush Creek Colorado on August 4, 1982	 107

  27    Brehm's (1981) conceptual model  of temperature inversion
          destruction in Austria's Inn Valley, showing differential
          CBL growth over the opposite sidewalls and continued
          down-valley flow in the elevated stable core	 108
                                      XI

-------
                                   TABLES

Number
  1     Values of the Constants I,  J,  and K,  for ay as a Function
        of Downwind Distance,  for Six  Stability  Conditions	   23
  2     Values of the Constants I,  J,  and K,  for az as a Function
        of Downwind Distance,  for Six  Stability  Conditions	   23
  3     Relationship Between Weather Conditions  and Stability
        Categori es	   24
  4     Equation  of Time Correction	   35
  5     Default values of VALMET Input Parameters	   55
  6     Limits on the Values of Input Parameters.	   56

-------
                               ACKNOWLEDGMENTS
     The work reported here had its origins in research conducted at the
Department of Atmospheric Science at Colorado State University while one  of
the authors (COW) was a student there.  Dr. Thomas B. McKee's contributions
to the work at that time, as a thesis advisor, are greatly appreciated.

     The authors' colleagues at PNL contributed suggestions and criticisms
as the work progressed that led to significant improvements in the model's
design and in the clarity of the text.  Drs. Thomas W. Horst and
J. Christopher Doran are especially thanked for their help in that regard.

     Mr. Alan Huber, the EPA Project Officer, and Mr. Rich Fisher, EPA
Region VIII meteorologist, are thanked for their guidance, support, and
encouragement of the work.  Mr. Fisher tested early versions of the model
and provided a number of very useful suggestions from the viewpoint of a
model  user that resulted in improvements of the model code.  The final set
of technical revisions to this report was made following suggestions by
Mr. Bill Bernardo of EPA's Region VIII office, arising from sensitivity
tests he conducted on the model in late 1983.  We wish to thank Mr. Bernardo
for his help.
                                     xi n

-------
                                  SECTION 1

                                 INTRODUCTION


     This report documents an air quality model that was developed to  pre-
dict concentrations of nonreactive pollutants  arising from elevated con-
tinuous point sources that emit pollutants within well-defined deep mountain
valleys.  The model, termed VALMET, is intended to simulate the effects on
pollutant transport and diffusion of various meteorological processes  that
are thought to result in worst-case pollutant  concentrations.  The model  is
run for situations when pollutants are carried in locally developed circula-
tions within a valley when these circulations  are "decoupled" from prevail-
ing circulations above the valley.  The primary physical processes included
in the model are

Nocturnal Simulation:

     Transport by down-valley drainage flows
     Plume channeling within the valley
     Enhanced horizontal and vertical diffusion due to topography
     Plume reflections off valley floor and sidewalls
     Pollutant diffusion out the top of the valley
  o  Dilution of the plume due to clean air inflow from tributaries.

Post-Sunrise Simulation During Temperature Inversion Breakup Period:

  •  Convective boundary layer growth
  o  Plume subsidence in the valley inversion
  •  Fumigation into growing convective boundary layers
  •  Transport and diffusion in upslope flows  over the sidewalls.

     The model, while including a variety of meteorological processes, is
highly parameterized so that it is simple in concept and easy to run.  The
model is composed of 13 modules, or subroutines, arranged in such a way that
an improved understanding of individual valley meteorological phenomena can
be easily incorporated in future versions of the model.  The modules within
the model can be replaced by data if they are  available.  Thus, the model
can be used in one of two modes.  It can be used in a "screening" mode to
calculate pollutant concentrations within a valley when little site specific
data is available, or the model can be "calibrated" with site-specific data
so that it can be used as a site-specific model.

-------
     The two-dimensional model was developed primarily to predict pollutant
concentrations on the valley floor and sidewalls on a valley cross section
an arbitrary distance down-valley from a pollutant source during the post-
sunrise temperature inversion breakup period.  It is necessary, however, to
know the air pollution concentration within the valley cross section at sun-
rise, as an initial condition for the post-sunrise simulation.  The model is
therefore comprised of two parts—a nighttime part whose purpose is to pre-
dict concentrations on the valley cross section at sunrise, and the daytime
part which predicts concentrations on the valley floor and sidewalls during
the post-sunrise temperature inversion breakup period.  The temperature
inversion breakup period has been identified by previous investigators [1]
as a period when diurnal fumigations [2] can produce high pollutant con-
centrations in valleys.

     The nighttime simulation, which applies during the steady-state period
after valley temperature inversions and drainage wind systems have become
established, uses a modified valley-following Gaussian plume algorithm to
calculate air pollution concentrations for points on the valley floor and
sidewalls.  A plume rise formulation is used to simulate the initial rise of
a pollutant plume at the stack due to momentum and buoyancy of the efflu-
ent.  Pasqui11-Gifford diffusion coefficients are modified to account for
enhanced nocturnal diffusion caused by rough terrain.  The Gaussian plume is
also modified to allow for dilution of the plume during its down-valley
transport caused by clean air flowing into the plume from valley tributaries
or by converging downslope drainage flows.  An integral constraint on pol-
lutant mass is applied to ensure that pollutant mass is conserved during the
plume's transport down the valley and within any valley cross section down-
valley from the emission source, except for pollution diffusion out the top
of the valley.

     The daytime simulation uses numerical techniques that simulate the
fumigation of the nocturnal plume onto the valley floor and sidewalls as a
convective boundary layer grows upwards from the heated valley surfaces and
as subsiding motions occur over the valley center after sunrise.  The rate
of growth of convective boundary layers and subsidence within the valley
temperature inversion are simulated using the bulk thermodynamic model of
Whiteman and McKee [3].  This model is driven by sensible heat flux, esti-
mated as a fraction of the solar radiation using a highly parameterized
surface energy budget.  The effects of such factors as snow cover, soil
moisture, cloud cover, or surface albedo are not explicitly included in the
model but can be incorporated into the model in the future through an
expanded energy budget module.  The shape of the topographic cross section
of the valley is explicitly included in the model through the specification
of valley floor width and sidewall inclination angles at the valley cross
section of interest.  The retarding effect on temperature inversion breakup
and pollution dispersion due to warm air advection above the inversion is
also included in the model.  Fumigated pollutants are transported from the
valley cross section in upslope flows that develop within the convective
boundary layers over the slopes.  Pollutants are diffused through model grid
elements during this transport up the slopes in the growing convective

-------
boundary layer.  Pollutant concentrations decay exponentially within  indi-
vidual grid elements high on the sidewall as they are dropped from  the  simu-
lation as the inversion top subsides below them.

     The output from the nighttime simulation includes the steady-state
pollutant concentration at valley floor and sidewall grid elements  on the
valley cross section of interest.  The fraction of plume mass that  has  dif-
fused out the top of the valley during the plume's travel is also an  output
of the model.  Since an analytical formula describes the concentrations
within a valley cross section, cross valley and vertical profiles of  pollu-
tant concentration can be calculated and plotted.  The plume centerline con-
centration is an output of the model.

     The primary output of the daytime simulation is the maximum 1- and 3-hr
average pollutant concentrations in each of the model grid elements on  the
valley floor and sidewalls.  The time varying 5-min-average concentrations
for each of the grid elements between sunrise and the time of inversion
destruction is also an output of the model.  In addition to these primary
outputs, intermediate model outputs come from individual modules in the
program.  The local standard time of sunrise, the duration of the daylight
period, and the solar flux on a horizontal surface at solar noon come from
the solar module.  The convective boundary layer height and inversion top
height as a function of time come from the temperature inversion breakup
modu1e.

     Twenty-seven input parameters are necessary to drive the model.  These
input parameters include the date, site location, topographic characteris-
tics of the valley cross section, temperature inversion characteristics at
sunrise, emission and stack characteristics, down-valley wind speed(s),
atmospheric stability, grid element length, and sensible heat flux  para-
meters.  If known, the rate of warm air advection above the valley  can  be
input.  The necessary model inputs can be obtained from topographic maps,
engineering information on the pollutant source, and one or more seasonal
meteorological data collection campaigns in the valley of interest  using
tethered balloon data collection systems and/or doppler acoustic sounders.

     This report is written following U.S. Environmental Protection Agency
(EPA) reporting guidelines [4,5] and is organized in the following way.
First, meteorological observations taken in a number of deep valleys  in
western Colorado are summarized in order to provide a brief description of
the salient meteorological features that must be included in a  realistic air
quality model.  Second, a technical discussion is given of the  mathematical
representation of these processes used in VALMET.  Third, a brief overview
is given of the air quality model's structure.  Fourth, the individual
modules of the model are described in detail, including the FORTRAN code,
inputs, outputs, etc.  Fifth, sample runs of the model are presented.
Finally, the need for further data and verification studies is  indicated and
an overall summary of the modeling approach is presented.  Appendices
include a complete listing of the VALMET source code, a listing of  a  sample
plotting program for depicting model outputs, and a reprint of  a scientific
paper which describes the theoretical basis of one of the major components
of the modeling approach.

-------
                                  SECTION 2

                                  BACKGROUND
     The VALMET model is a further development of ideas that were published
in research papers and reports over the last 5 years, based on a comprehen-
sive set of valley meteorological field experiments conducted in western
Colorado.  These earlier papers summarize a case study of temperature inver-
sion buildup [6]; describe the data analysis, hypothesis, and mathematical
foundation of the temperature inversion destruction part of the model
[3,7-10]; describe the initial idea of an air pollution model framework able
to incorporate upslope winds and fumigations arising after sunrise [11]; and
present a general summary of Colorado valley meteorology based on data
analyses [12].  Progress in the development of VALMET has been described in
two progress reports submitted to EPA in 1981 and 1982 [13,14].

     The present version of VALMET, as described in this report, must be
considered as an initial or preliminary version of the model.  This  report
is written to facilitate the limited distribution of the model to users who
will provide comments and suggestions regarding the further development of
the model.  The present version has not yet undergone a full set of  sensi-
tivity tests, has not been fully evaluated with respect to its performance
in simulating air pollution concentrations in actual valleys, and contains
modules that have not yet been fully developed.

     The user is cautioned against applying the air quality model for
meteorological or topographical conditions that are contrary to assumptions
made in the model.  In order for the user to understand fully the assump-
tions and hypotheses used in the model, it is necessary to summarize briefly
the results and hypotheses that arose from the observational program and the
work of previous investigators.  In this section we present a condensed sum-
mary of this material for both the nocturnal and temperature inversion
breakup periods, emphasizing the physical mechanisms in the valley that are
responsible for the transport and diffusion of pollutants.  It is these
individual mechanisms that are parameterized in the air pollution model.

NOCTURNAL VALLEY METEOROLOGY

     The most universal features of nighttime valley meteorology, reported
in valleys all over the globe [15-19, and others], is the drainage of cooled
air down the valley sidewalls and valley axes.  The initiation of these
flows, which is closely related to changes in the temperature structure
above valley surfaces, has received inadequate study to date and cannot be
adequately parameterized for inclusion in the model.  After the short (2 to

-------
4 hr) initiation period, the down-valley flows become well-established
through the valley depth, and attain a speed that is characteristic of the
valley and is reasonably steady-state.  Observations [9,12] show that the
strength of the down-valley flows is quite variable from valley to valley,
and no model  is yet available which can accurately predict their strengths
in individual valleys.  In Colorado's valleys, however, observations show
that windspeeds on a clear undisturbed night in one season are typical of
windspeeds on similar nights in other seasons.  Observations thus seem to be
the best way of specifying down-valley wind system strength.

     Down-slope flows on the valley sidewalls become weaker and less
organized after the valley inversion develops, since air flowing down the
slopes must work against a very stable ambient stratification.  Slope flows
can more easily become detached from the sidewalls during this period,
providing an important stirring mechanism which causes enhanced horizontal
diffusion of pollutant plumes within the valley inversion.  The effects of
terrain roughness and mechanically generated turbulence also play an impor-
tant role in enhancing dispersion in the down-valley flows.

     Pollutant concentrations can be strongly affected by dilution of the
plume due to clean air flowing in from valley tributaries.  This feature of
meteorology,  unique to valleys, can make substantial changes in pollutant
concentrations, especially where high volume fluxes of clean air are
involved, as  when a pollutant plume originates in a small tributary and
emerges into a major valley.

VALLEY METEOROLOGY DURING THE TEMPERATURE INVERSION BREAKUP PERIOD

     Observation in Colorado's deep valleys [9,10] have provided detail on
the changes in atmospheric structure during the morning transition period
when the nocturnal down-valley flows are reversed to daytime up-valley
flows.  The transition period is much longer than expected in these deep
valleys and the physical processes leading to the transition are expected to
be of great importance for air pollution transport and dispersion.  In this
section, we will summarize the observations of inversion destruction, point-
ing out typical characteristics of the meteorology of these valleys, and the
physical processes that must be included in realistic air pollution models
simulating the inversion destruction period.

     At sunrise nocturnal inversions in most of the valleys were built up to
about the level of the surrounding ridgetops.  The average depth, based on
21 cases studies, was 604 m.  Vertical potential  temperature gradients
within the inversions averaged 0.0295°K m" , but ranged from 0.0187 to
0.0566°K m" .  The strength and direction of prevailing winds aloft, as
determined from the Grand Junction, Colorado, morning rawinsonde sounding,
had no demonstrable effect on the characteristics of valley inversions.  The
valley inversions showed much less variability from day to day and from
season to season than inversions over Grand Junction.  This suggests that at
least for a narrow range of synoptic conditions, valley topography produces
more consistent inversions, perhaps by protecting them from winds aloft.

-------
     When the along-valley winds were weak, the valley vertical potential
temperature profiles frequently were hyperbolic, especially near the
ground.  In most valleys where along-valley winds were of at least moderate
strength, the inversion profiles were nearly linear with height.

     Temperature inversions in all of the valleys investigated were
destroyed after sunrise following one of three patterns of temperature
structure evolution (Figure 1).  The first pattern, observed in the widest
valley studied, approximates inversion destruction over flat terrain, in
which the nocturnal inversion is destroyed after sunrise by the upward
growth from the ground of a warming convective boundary layer  (CBL).  The
second pattern, observed in snow covered valleys, differs significantly  from
the first.  Here the growth of the CBL, which begins after sunrise, is
arrested once the CBL has attained a depth of 25 to 50 m.  The inversion is
then destroyed as the top of the nocturnal  inversion descends  into the
valley.  Successive profiles of the valley atmosphere show a warming
consistent with a simple subsidence of the previous profiles.  The third
pattern of temperature structure evolution was observed in all the valleys
when snow cover was not present and describes the majority of  case studies
observed in field experiments.  Following this pattern, inversions are
destroyed by two processes:  the continuous upward growth from the valley
floor of a warming CBL and the continuous descent of the top of the
nocturnal temperature inversion.  Warming of the elevated inversion layer
above the CBL is consistent with a simple subsidence of the previous
profiles.  In the Colorado valleys studied, the time required  to break an
inversion and establish a neutral atmosphere within the valley was typically
3-1/2 to 5 hr after sunrise.  Temperature structure evolution  during clear,
undisturbed weather was surprisingly uniform from day to day and season  to
season.  Thus, in future work, one may be fairly confident of  observing
typical inversion breakup during short field studies in undisturbed weather.

     The common element of all three patterns of temperature structure
evolution is the development of a CBL over the valley floor after it is
illuminated by direct sunlight.  Observations taken from the sidewall of one
valley also show the development of a CBL after direct sunlight illuminates
the sidewall.  Due to the shading effects of surrounding topography, the
different valley surfaces can be illuminated at significantly  different
times, thus affecting the initiation of CBL growth.  The temperature struc-
ture of the sidewall CBL is similar to that of the CBL over the valley
floor, but winds blow up the sidewall CBL at speeds of up to 3 m sec  .

     Five different temperature structure layers have been observed during
the inversion destruction period.  Above the valley floor CBL  and the
sidewall CBL just mentioned is the stable core of the potential temperature
inversion.  A neutral  stability layer above the stable core appears to be
part of a large-scale convective boundary layer that forms over the western
slope of the Rocky Mountains.  Above this layer is the free atmosphere.

-------
              zt
                 (a)
                            (b)
                                                 H(t)
e
                                               t
                               Pattern 1.  Growth of CBL.
              zt
                 (0
                            (d)
                                                 \
                                                   h(t)
                                                     \

                                                          \
                                                          \
                                            	H(t)	
                                  8      tj     t2     t3     t0    t

                  Pattern 2.  Descent of inversion top and arrested growth of CBL.
                                      Zt
                                  to
                                                               (f)
                                                  h(t)
                                                      \
                                                        \
                                               .H(t)
                                                     j	i
                                  9      t|     tz     t3     t0    t

                 Pattern 3.  Descent of inversion top and continuous growth of CBL.

  Figure  1.   Three patterns  of  temperature structure  evolution during the
              inversion breakup  period.  Potential  temperature profiles
              are on the  left,  and  time-height analyses of CBL height (H)
              and inversion top  height (h) are on  the  right.
     Each  of  the five temperature structure  layers, identified  primarily by
their potential  temperature  structure, can also  be identified by  the winds
that prevail  within them  (see  Figures 2, 3 and 4).  During inversion
destruction,  the CBLs over the valley floor  and  sidewalls contain winds
which blow up the floor of the valley and up  the slopes.  The CBL, or

-------
                           UP-FLOOR
              A. DOWN-VALUEY
 Figure 2.  Typical mid-morning wind structure over and within a deep  valley
            on the western slope of the Rockies, illustrating the five
            interrelated wind systems identified in field studies.
neutral layer, above the valley inversion has winds which blow up the
inclined Western Slope of the Rocky Mountains during the day.  Winds in  the
stable core typically continue to blow down-valley after sunrise until the
stable core is nearly destroyed.  Winds in the stable free atmosphere may
blow from any direction with speeds determined by snyoptic-scale pressure
gradients.  Despite variability in the strength and timing of reversal of
the winds, the temperature structure evolved uniformly from day to  day in
individual valleys.

     On the basis of the wind and temperature observations summarized above,
an hypothesis (Figure 5) has been developed to explain the temperature
structure evolution.  Since energy is required to change the temperature
structure, and the changes begin at sunrise, it is reasonable to hypothesize
that solar radiation is the driving force.  A fraction of the solar radia-
tion, received on the valley floor and sidewalls, is converted to the sen-
sible heat flux that provides energy to the valley atmosphere.  Sensible
heat flux from a surface, as over flat terrain, causes a convective boundary

-------
              t
                 FREE
                 ATMOSPHERE
                NEUTRAL LAYER
                 STABLE
                 CORE
                CBL
        GRADIENT
      DP-INCLINE
   DOWN-VALLEY
UP-FLOOR
                                             e
Figure  3.   Relationship between temperature  structure layers
           and wind systems.  The  temperature profile is a
           typical mid-morning sounding  from the floor of a
           deep valley.
              Z f  NEUTRAL
                   LAYER
                  STABLE
                  CORE
                  CBL
                           UP-SLOPE
                                           ff
     Figure 4.  Dual soundings  from  a  valley floor and a
               valley sidewall  illustrating the up-
               slope flow found within the CBL over
               the sidewal1.

-------
                  9
Figure 5.  Illustration of the hypothesis of inversion destruction.  In the
           center of the diagram cross sections of a valley are shown at
                 h.
times t-i ,  t~, t.,, t., and tn.  On the left are corresponding
           potential  temperature profiles as taken from the valley center.
           On the right are corresponding up-valley wind components as a
           function of height.  At sunrise, tn-, an inversion is present in
           the valley.  At t?, a time after sunlight has illuminated the
           valley floor and slopes, a growing CBL is present over the val-
           ley surfaces.  Mass and heat are entrained into the CBLs from
           the stable core above and carried up the sidewalls in the
           upslope flows.  This results in a sinking of the stable core and
           growth of the CBLs (t3 and t^) until the inversion is broken
           (t,,) and a turbulent well-mixed neutral atmosphere prevails
           through the valley depth.  Down-valley winds continue to blow in
           the stable core during the inversion breakup period.  Winds in
           the CBL below, and in the region above the stable core often
           blow up-valley during this same period.
                                     10

-------
layer to develop over the surface.  Mass and heat are entrained into the CBL
from the stable core above.  Mass entrained into the valley floor and side-
wall CBLs, however, is carried from the valley in the upslope flows that
develop in the convective boundary layers over the sidewalls.  This removal
of mass from the base and sides of the stable core causes the elevated
inversion to sink deeper into the valley and to warm adiabatically due to
subsidence, and decreases the rate of growth of the CBLs.  Following this
hypothesis, the rate of warming depends directly on the rate of energy input
into the valley atmosphere.  This energy may be used to deepen the CBLs or
to move mass up the sidewalls, allowing the stable core to sink.  From this
hypothesis a thermodynamic model  of temperature inversion destruction has
been developed [3].  This model forms the basis for parameterizations of
inversion breakup in the VALMET model.

     The valley inversion destruction mechanism has important implications
for the dispersion of valley air pollution.  These implications can be
investigated by assuming that the nocturnal inversion at sunrise contains
pollutants emitted from a continuous elevated source within the valley.
During the night such pollutants would be carried down the valley in the
along-valley wind system, undergoing both vertical and horizontal disper-
sion.  The horizontal dispersion in these flows is known to be much greater
than over flat terrain [20].  After sunrise a convective boundary layer
forms over the valley floor and sidewalls (Figure 6a).  The subsequent dis-
position of pollutants in the stable core will  be affected by two competing
processes—the sinking of the stable core and the growth of the CBL.  The
air pollution implications of the two processes can be considered by study-
ing a valley cross section down-valley from the source at a later time.  The
three inversion breakup patterns discussed above have the following air
pollution implications [8]:

 1.  Pattern 1 - Growth of convective boundary layer (Figure 6b):  Pure
     growth of a CBL will result in fumigation [2] of pollutants at the
     valley floor as the CBL grows upward into the polluted stable core.
     This process is favored when the slope flows are ineffective in
     removing mass from a valley, and thus will occur in very wide or
     shallow valleys.

 2.  Pattern 2 - Sinking of stable core (Figure 6c):  Failure of the CBL to
     grow once it has formed over the valley floor and sidewalls results in
     inversion destruction by sinking of the stable core.  Thus, the pol-
     lutants sink into the top of a shallow mixed layer, producing high con-
     centrations at the ground.  The pollutant plume, once entrained into
     the CBL, is advected up the sidewalls and dispersed into the neutral
     layer aloft.  This process is favored for narrow-to-wide valleys when
     sensible heat flux is weak.

 3.  Pattern 3 - Combination (Figure 6d):  A combination of CBL growth and
     stable core descent results in the sinking of pollutants into the top
     of a growing mixed layer.  Pollution concentrations should be interme-
     diate between the previous two cases.  This pattern is the most common
     one in Colorado mountain valleys.
                                      11

-------

     Figure 6.  Air pollution implications of CBL growth and inversion
                top descent.
     Another process, that of cross-valley advection of the pollutant plume
due to differential  heating of the valley sidewalls, may be important in
determining the cross-valley position of maximum pollutant concentration.
Unfortunately, relatively little observational evidence is presently avail-
able on cross-valley advection or the effects of cross valley advection on
pollutant dispersion.

     Observations in snow-free valleys of Colorado show that inversions are
typically destroyed within 3-1/2 to 5 hr after sunrise, regardless of
season.  Thus, high pollutant concentrations at the ground may be expected
following sunrise in polluted Colorado valleys.  But inversions are typ-
ically broken every day and, once the inversion is destroyed, good mixing
will prevail in an up-valley flow regime that is coupled to flows above the
valley.

     Moist surface conditions or high albedo due to snow cover may change
the surface energy budget components so that sensible heat flux is
reduced.  In these conditions inversion destruction will be delayed or an
inversion may persist all day.  In Colorado's snow-covered Yampa Valley on
February 23, 1978, the CBL failed to grow deeper than 35 m and the observa-
tions were characterized by a slow descent of the stable core.  The inver-
sion failed to break during this day, although the top of the stable core
descended to within 150 m of the ground.

     To summarize, the primary factors affecting a valley pollutant plume
after sunrise are

  »  continued transport of the pollutant plume down the valley in the
     stable core

  •  sinking motions in the stable core as inversion destruction progresses
     after sunrise, bringing the plume closer to the valley floor
                                      12

-------
  •  fumigation of the plume into convective boundary layers growing over
     the valley floor and sidewalls

  •  transport and diffusion of pollution in upslope flows that develop over
     the sidewalls.

These processes continue from sunrise until  the valley temperature inversion
is destroyed and the valley atmosphere becomes coupled with the overlying
circulations.  The amount of time required for temperature inversion
destruction in Colorado Mountain Valleys is  typically 3-1/2 to 5 hr, but it
depends on many factors including the depth  and strength of the initial
temperature inversion, the incoming solar energy, the sensible heat flux,
the pattern of inversion destruction followed, etc.
                                      13

-------
                                  SECTION 3

                             TECHNICAL  DISCUSSION
     In this section we present a technical discussion of the VALMET  valley
air pollution model.  We begin with a short discussion of the coordinate
system and the means used in the model to  incorporate topography.   Discus-
sions of the mathematical equations used in both the nighttime and  daytime
portions of the model then follow along with a statement of any assumptions
used in the model  development.  Unless otherwise noted, numerical values
of the parameters in the equations are assumed to be in the MKS (meter-
kilogram-second) system of units, except for pollutant concentrations
expressed in micrograms per cubic meter.

COORDINATE SYSTEM

     VALMET is a two-dimensional, air pollution transport and diffusion
model in which pollutant concentrations are determined on a valley  cross
section located an arbitrary distance down-valley from a pollutant  source.
The distance x is measured following the axis of the valley down the  center-
line of the valley floor (Figure 7).  The  local coordinate system utilized
on the valley cross section is shown in Figure 8.  The x-coordinate system
specifies the distance of the cross section down-valley from the pollutant
source, as measured on the valley floor centerline following any turns  in
the valley.  The y-coordinate system has its origin at the valley floor
centerline, is orthogonal to the x-axis, and increases to the right when
viewed from the mouth of the valley.  The  z-coordinate system is a  vertical
coordinate system with origin at the valley floor centerline.

TOPOGRAPHIC CROSS SECTION

     For a particular valley cross section, the model assumes that  the  val-
ley topography can be simply represented by a horizontal valley floor of
width, w, and two inclined sidewalls of inclination angle a^ and o^,  as
shown in Figure 9, below.

     To simulate a real valley, the modeler must approximate the topography
of any cross section of interest using a topographic map, from which  w, a  t
and <*2 are estimated.  Given the simplified valley cross section and  the
local coordinate system, one may calculate several secondary topographic
parameters that will be used in the model.  For example, using definitions
shown in Figure 10, the area of the cross  section of a valley containing a
temperature inversion of depth h is
                                      14

-------
Figure 7.  Illustration of the nocturnal down-valley transport and
           dispersion of a pollutant plume.  The terrain following
           coordinate system, with cross valley section an
           arbitrary down-valley distance x, is illustrated.
                                  15

-------
      Figure 8.   Illustration of the local  coordinate system on a
                 cross  valley section.
Figure 9.  Parameters used in the VALMET model  to approximate a valley
           topographic cross section.   The three parameters include
           the valley width and the two sidewall inclination angles.
                            t
 Figure 10.  Illustration of the definitions of secondary topographic
             variables used in the model.
                                    16

-------
                                 h2
                       A = h w +• j- (cot c^ + cot <*2)                     (1)
The y-distance to the left and right sidewalls from the vertical coordinate
axis is
                           yL(z)  = -(£+ z cot a2J                         (2)
and
                                   w
                              z) = j + Z cot a, .                         (3)
The topographic cross section may be considered to have a thickness of
unity.  In this case the area AT of the top of the inversion (z = h) is
                   = yR(h)  - y  (h)  =  w + h(cot  a  + cot a )  .              (4)
NOCTURNAL MODEL

Pollutant Source
     The model begins with an elevated pollutant source on the valley floor
at the topographic cross section where x=0.  The pollutant source may be
located anywhere on the valley floor.  The y-position of the source is
specified relative to the coordinate origin.  For example, the y-coordinate
of a pollutant source on the valley floor 200 m off the valley floor center-
line would be specified as yQ=+200 m if the source were in the positive y
direction from the centerline, or as y =-200 m if in the negative direc-
tion.  The present version of the mode? cannot handle sources on the valley
sidewalls.  The source is assumed to emit a continuous elevated pollutant
plume into a nocturnal valley temperature inversion of depth h=hi.  This
pollutant plume is approximated in the model by a modified Gaussian plume
algorithm.  The potential temperature gradient within the inversion is
T=86/3z, and the down-valley wind speed at plume carrying height is u$.
Wind speed is assumed not to vary with height.  The model, being run for
worst-case dispersion under nocturnal temperature inversion conditions,
would be run for Pasquill-Gifford stability categories D, E or F.  In later
sections, further details of the Gaussian plume formulation will be given
along with details concerning dispersion coefficents.
                                      17

-------
Plume Rise

     Plume rise is treated in the model according to formulas described  by
Briggs [21-24] and used in the MPTER model [25].

     The "2/3 law" of plume rise is used  for all stability categories  up  to
a certain downwind distance Xf to calculate the plume rise using the
equation


                                  1  6  F1/3 x  2/3
                            Ah  = U6  F    x	                          (5)
where F is the plume's buoyancy flux, x is the downwind distance  (m),  and  u,
is the wind speed at stack height (m/s).  The parameter F  is  calculated  as
                                           -)                            (6)
                                                  _p
where g is the acceleration due to gravity (9.8 ms  ), Ve is the exit
velocity of gases emitted from the stack  (m/s), R is the stack  inside  radius
(m), and T  and T are the exit and ambient temperatures (°K), respectively.
The ultimate rise Ah-r and the downwind distance x^ beyond which  it  applies
are calculated as shown below.

Unstable or Neutral  Atmospheric Conditions—Calculations of Xf  and  Ahf
depend on whether the plume rise is "buoyancy-dominated" or "momentum-
dominated."  This is determined by a crossover temperature difference,
calculated as
                                 1/3     2/3
             (AT). = 0.0297 T  V_ ' /(2R) '         for F < 55
                 \*           j  C


             or (AT)  = 0.00575 T  V 2/3/(2R)1/3    for F >_ 55.
                    v«            j  C
(7)
The actual temperature difference AT = T -T is compared to the  crossover
temperature difference (AT)C and if AT _<_ (AT)  then the plume  rise  is
momentum-dominated.  If AT > (AT)  then the plume is buoyancy-dominated.
                                      18

-------
     Momentum-Dominated Plume Rise—For a momentum-dominated plume  rise the
distance to final  rise is zero, such that


                                   xf  =  0.                              (8)


The final  plume rise is given by


                               Ahf = 6RVe/us.                          (9)


     Buoyancy-Dominated Plume Rise—For a buoyancy-dominated plume  rise the
downwind distance of final plume rise is x.f - 3.5 x*, where x* is the
distance at which atmospheric turbulence begins to dominate entrainment,
given by
                    x* =
                     14 F    when F < 55 m  sec
                         9/c;              A     T                   (10)
                     34 F '   when F > 55 nT sec    .
The final plume rise under these conditions is



                         1  C. C1/3 /O C  *>
                   Ah    1.6 F    (3.5 x*)
                         -
                                                                         Mlx
                                                                         (11)
Stable Atmospheric Conditions—Calculations of x^ and Ahf, as for unstable
conditions, requires a determination of whether the plume rise is buoyancy-
dominated or momentum-dominated.  This is determined by a crossover
temperature difference, calculated as


                          (AT)C  =  0.01958  T  Vg  s1/2                      (12)


where s is a stability parameter given by


                                       ae_

                                s  = ~^- ,                             (13)
                                      19

-------
and ae/az is the vertical gradient of potential temperature within  the
valley temperature inversion at sunrise.  The actual temperature difference
AT = T$-T is compared to the crossover temperature difference and if AT  <_
(AT)C the plume rise is momentum-dominated.  If AT > (AT)C the plume rise" is
buoyancy-dominated.

     Momentum-Dominated Plume Rise—For a momentum dominated plume  rise  the
distance to final  rise is zero, such that
                                   xf =  0.
                                      (14)
The final plume rise is given by
                            =1.5
                                     2  2
                                   ve R T
                                    VsJ
                                           1/3
            .-1/6
                                      (15)
     Equation (9) for unstable-neutral plume rise is also evaluated and  the
smaller of the two plume rise calculations is used as the final plume rise.

     Buoyancy-Dominated Plume Rise—When the plume rise is buoyancy-
dominated, the distance to final rise depends on wind speed at stack
height.  The actual  wind speed is compared to a certain critical wind speed,
uc, and the distance to final rise is calculated using one of two formulas
depending on whether the actual wind  speed is greater than or less than  the
critical speed.

     The critical wind speed is calculated as follows:
                            uc  =  0.2746  f1/4  s1/8
                                      (16)
The distance to final rise is given by

                           ,3/2
                                F-i/8,-9/1S
                  or
xf =
                           1.6.
Us s
for u  >
                                      20

-------
The final plume rise is given by
                                 1.6 F1/3 xf2/3
                           Ahf =	1	  .                        (18)
Gaussian Plume Model

     Steady-state nocturnal pollutant concentrations within  a  valley  cross
section are determined by means of a Gaussian plume model which  is modified
to account for channeling and reflections off the  sidewalls  within the  val-
ley, dilution due to clean air volume flux convergence within  the valley,
and diffusion of pollutants out the top of the  valley  inversion.

     We begin with the basic Gaussian plume equations  having no  reflection
terms:
                  XG(x,y,z;5;y0) = 2u    a  u
                                       y  z  s
where
Gt(y) = exp [-^
                                                                         (20)
                          G2(z) - exp {-(-}}                       (21)
and where XQ = (x,y,z;C;y0) is the pollutant concentration at a  receptor
located at point P(x,y,z) due to a plume with center!ine  height  C  and  offset
distance from the valley floor centerline yQ.  The  factor 10  is the conver-
sion factor from kilograms to micrograms.  a  and a  are  the dispersion
coefficients, Q is the source strength and u  is the  transporting  down-
valley wind speed at the stack.  The pollutant stack  is constrained to be  on
the valley floor, so that

                                      21

-------
Dispersion Coefficients

     Nocturnal dispersion coefficients a  and az for the Gaussian  plume
equations are determined from the empirically derived Pasquill-Gifford
curves [26,27], as developed from atmospheric tracer experiments conducted
over homogeneous terrain.  The empirical Pasquill-Gifford curves relate the
values of a  and az to downwind distance and stability category.   A  numer-
ical approximation to the curves has been provided by McMullen [28]  and is
given as:


                     a = exp [I + J (In x) + K (In x)2]                 (23)


where
         a = standard deviation of the concentration in the horizontal  (a  )
             or vertical (az) in meters,                                 y

         x = downwind distance (km), and

     I,J,K = empirical constants for a given stability condition for a  and
             V

Tables 1 and 2 provide McMullen's [28] values of the constants I,  J,  and K
for determination of o  and az> respectively.  Table 3 is provided to deter-
mine the proper stability, category from meteorological observations.

     The above method of determining a  and az is appropriate for  homoge-
neous terrain.  Previous work [20,29-31] shows that dispersion is  enhanced
in valleys over that experienced over homogeneous terrain where the  original
empirical Pasquill-Gifford dispersion coefficients were determined.   This
enhancement is particularly marked for stable conditions [20].  Enhanced
dispersion in valleys is due to a number of physical mechanisms that have,
to date, been inadequately studied or characterized.  In the VALMET  model  we
have taken the approach, which we do not consider entirely  satisfactory, of
calculating valley dispersion coefficients from the original empirical
Pasquill-Gifford curves [26] by simply adjusting the stability categories.
From existing data it appears that the enhancement of the horizontal  disper-
sion can be approximated by dropping the stability two categories  (e.g., F
becomes D).  The vertical enhancement of dispersion is approximated  by  drop-
ping the stability one category (e.g., F becomes E).  The use of these  sim-
ple adjustments to the stability categories should be considered a stopgap
measure.  Further work will be required to evaluate its appropriateness and
seek more appropriate means of handling nocturnal dispersion in deep valley
terrain.

Channeling

     Calculations with the Gaussian plume model are made assuming  the
Gaussian plume follows the valley axis, despite meanders in the valley's
course.  The model user must therefore use a topographic map to determine
the down-valley distance of a given cross section.
                                      22

-------
TABLE 1.  VALUES OF THE CONSTANTS I, J, AND K, FOR a  AS A
          FUNCTION OF DOWNWIND DISTANCE, FOR SIX STABILITY
          CONDITIONS [28]

Stability
Condition*
A
B
C
D
E
F
*As defined
I
5.357
5.058
4.651
4.230
3.922
3.533
in Table 3.
J
0.8828
0.9024
0.9181
0.9222
0.9222
0.9181

TABLE 2. VALUES OF THE CONSTANTS I, J, AND
FUNCTION OF DOWNWIND DISTANCE, FOR
CONDITIONS [28]
K
-0.0076
-0.0096
-0.0076
-0.0087
-0.0064
-0.0070

K, FOR CF_ AS A
SIX STABILITY

Stability
Condition*
A
B
C
D
E
F
I
6.035
4.694
4.110
3.414
3.057
2.621
J
2.1097
1.0629
0.9201
0.7371
0.6794
0.6564
K
0.2007
0.0136
-0.0020
-0.0316
-0.0450
-0.0540
          *As defined in Table 3.
                              23

-------
      TABLE 3.  RELATIONSHIP BETWEEN WEATHER CONDITIONS AND STABILITY
                CATEGORIES [26]




Day

Surface Wind
Speed
m





(at 10 m),
sec'1
<2
2-3
3-5
5-6
6
Incomi
Strong
A
A-B
B
C
C
ng Solar Radi
Moderate
A-B
B
B-C
C-D
D
ation
Slight
B
C
C
D
D
Night
Thinly Overcast
or >4/8
Low Cloud

E
D
D
D


<3/8
Cloud

F
£
D
D
  The neutral class, D, should be assumed for overcast conditions during
  day or night.
Plume Pilution

     An important factor affecting plume concentrations is the dilution of
the plume during its travel  down a valley due to tributary flows that bring
clean air into the plume.  This process could be extremely important in
decreasing plume concentrations, especially when the volume of diluent air
is large relative to the plume volume, as when a pollutant's source is
within a small tributary to a larger valley.  Other processes occur in val-
leys that can affect plume dilution, including converging downslope drainage
flows and entrainment of clean air into a valley from above.  Any process
capable of producing volume flux divergence between two vertical valley
cross sections is capable of diluting a plume.  At present, the physical
understanding of these processes is insufficient to incorporate them
explicitly into an air pollution model.  The extent to which volume flux
divergence is a feature of a particular valley's meteorology can be deter-
mined, however, by wind observations.

     The approach used in the present version of the model is to modify the
Gaussian plume equation in a simple way that makes an initial correction for
the dilution of a pollutant plume during its down-valley travel.  The
correction is applied only when data are available to show that a plume
would actually be diluted in the valley being modeled.  The correction is
applied to the plume transporting wind speed at the pollution source cross
section to obtain a modified or virtual wind speed uv-  The virtual wind
speed is a function of the ratio of the volume fluxes across the source and
receptor cross sections, such that
                                      24

-------
                                      ucAc
                                      usAs
                                                                         (24)
where uc is the measured plume-transporting wind speed at the  receptor  cross
section and
source cross
Ac and AS are
 sections,
                          the cross sectional areas at the receptor and
	._	 	, respectively.  The value of uv calculated using this
equation is simply substituted for u? in the denominator of the Gaussian
plume equation to calculate pollutant concentrations at the receptor cross
section.  To avoid physically unrealistic results (i.e., plume concentra-
tions increasing with transport distance) uv must always be equal to or
greater than u .  This is accomplished by specifying
                                  ucAc
                                  usAs
                           > 1
                                                                         (25)
Using this plume dilution approach, pollutant concentrations at  the  receptor
cross section are calculated using the equation
                                y  =
                                xv
                                                             126)
The plume dilution method described above is an initial approach  that will
need improvement as the physics of the dilution process are more  fully
understood.  The advantages of the method are that it  should produce more
realistic concentrations than when plume dilution is not considered, and the
method can be supported by simple measurements.  The above method  suffers
from several drawbacks, however.  First, the separate  effects of  individual
physical processes that can lead to dilution are not clearly separated using
the current method.  Second, the dilution method does  not consider  the
effects on plume dilution of the enhanced lateral and  vertical mixing where
tributary flows merge into a main valley or of the drift of a olume center-
line towards one sidewall caused by incoming tributary flows.  Third, the
method does not consider stratification effects that may arise in merging
flows of different temperature.  Finally, dilution air is assumed in the
model to be clean with respect to the pollutant species.
method seems appropriate, however, as an initial means of
dilution until  a better understanding is obtained for  the
tion (and dispersion) processes.  The wind observations necessary to use the
dilution method could be obtained from a doppler acoustic sounder or special
pibal and/or tethersonde campaigns.  A user should use data to show that the
plume would be diluted due to volume flux divergence.
                                              The dilution
                                              handling plume
                                              individual dilu-
                                      25

-------
Reflections
     Dispersion of pollutants during the plume's nocturnal transport  down
the valley will result in a spreading of the plume in both the vertical  and
horizontal, resulting in the plume's eventual contact with the valley  sur-
faces (valley floor and sidewalls).  Additional spreading of the plume,
using the basic Gaussian equations, would give the physically unrealistic
result of appreciable concentrations occurring "beyond" the valley  sidewalls
or "below" the valley floor.  Concentrations calculated for receptors  within
the valley cross section would be too low and the total pollutant mass of
the plume would not be conserved within the valley.  Several methods  have
been proposed to remedy this situation [26,32].  The valley may, for
example, be treated as a channel having vertical sidewalls.  Then,  single  or
multiple reflections of the plume from the sidewalls can be simulated  by the
method of "image plumes", in exact analogy to the way that reflections off
an elevated inversion surface are now handled in many Gaussian models
[26].  Other similar means of handling such plume reflections have  been
tried by the authors giving consideration to the inclined sidewalls of real
valleys, but all such methods thus far investigated have suffered from draw-
backs of one type or another.  The most satisfactory solution came  from  a
consideration of pollutant mass conservation within the plume, to be
discussed in the following paragraphs.

     The Gaussian plume equations are a solution to the classical diffusion
equation obtained by means of a number of restrictive assumptions (steady-
state motion, au/9z=0, etc.) and application of certain boundary condi-
tions.  Among the boundary conditions is an expression of plume mass  conser-
vation, given by
                    Q  =    /     /  u   x  dy dz for z > o .                 (27)
This integral constraint, which applies to the basic mathematical  form  of
the Gaussian equation [Equation (9)], has a simple physical meaning.
Suppose that a source emits a pollutant at the rate Q of 1 kg/s  into  a
horizontal wind of 1 m/s.  This will result in a kilogram of pollutant  mass
being carried in a meter of transported air.  Following the Gaussian  plume
equation this kilogram of pollutant is simply distributed in the  1-m  slice
of air following a bi-normal distribution.  The total plume mass  within the
1-m slice, however, does not change.  It is 1 kg regardless of whether  the
kilogram is relatively concentrated into a tightly confined plume or  widely
dispersed about the plume center.  Thus, when one integrates the  pollutant
concentration over the volume of the slice from y=-o> to y=+°= and  from z=-<=°
to z=+°°, and uv is constant, we obtain
                                      26

-------
                      £--   /    /  X.. dydz=i^.                   (28)
                       "y   -co   _co
In terms of a worst-case model of pollutant  dispersal  in  a  valley,  one must
consider ways of preserving or conserving the  total  pollutant  mass  so that
this relation still holds.

     The total amount of pollutant mass  in a unit  cross  section  perpendicu-
lar to the plume path is determined by Q and uv, but  the  distribution of
pollutant mass about the plume centerline is determined  by  the bi-normal
probability density function.  The characteristics  of this  function  are such
that
            /    /  ~T^exp  U       --— exp   {- % j  dYdH      (29)
                                               a
                                               z
where
                           V -      and „ .
A valley cross section can be superimposed  on  the  theoretical  plume  cross
section (Figure 11) and, through integrations,  one  can  determine  the frac-
tion of total plume mass within the  cross section  within  the  valley  and
below the inversion top, as well as  the mass that  has diffused  above the
inversion top and the mass that has  diffused beyond the valley  sidewalls and
valley floor.  For example, the fraction of the  total mass  in  the cross sec-
tion that is within the valley cross section below  the  inversion  top hi is



              ^   /R     1        /   A       1        /  H2\
         f,  =  /     /   ^r"— exp    - 7 J  • ^^- exp   (- f  J  dYdH    (30)
          1    o    Y,                   z        -             '
                                      27

-------
                                  GAUSSIAN PLUME
                            TEMPERATURE INVERSION TOP
            ^>;^;lj  POLLUTION DIFFUSING OUT THE INVERSION TOP

            || | |  ||  POLLUTION REFLECTED BACK INTO VALLEY CROSS SECTION
    Figure 11.  Cross section  of  pollutant  plume and valley topography
                illustrating the  integral method of calculation for
                plume reflection  and  diffusion  out  the top of the
                valley temperature  inversion.
where
                     h  '
                      y  " y
                                                    y  " y
The fraction of plume mass that  has  diffused  out  the top of the inversion is
fo =
Hi  -
—— exp
                                           	exp
                                           /2-n: a
                                            -     /  dYdH
(31)
The fraction of plume mass which  has  diffused  "beyond"  the valley sidewalls
(below the inversion top) and  "below"  the  valley  floor  is
                                  =  1-frf2  •
                                                             (32)
                                      28

-------
In our simple model of plume reflection we must  find  a means  of  folding  the
mass fraction f3 back into the valley cross section so that pollutant  mass
will be conserved within the valley.  Although the pollutant  mass  could  be
distributed into the valley cross section any number  of ways, we choose  the
simple expedient of mixing it uniformly within the valley  unit cross section
volume, so that concentrations within the cross  section include  an  offset
concentration and are given by
                     x(x,y,z;r;yo) =  fXy +                               (33)
where the factor PQT/pT  adjusts the concentrations  to  standard  atmospheric
conditions, with pQ = 1013 mb, T  = 293.16°K  and T and  P  are  the ambient
temperature (°K) and pressure (mB).  The offset concentration  is  given  as
fol 1 ows:
                              v    - f
                              yoff   T3
Equation (33) applies within the spatial domain where

                            i)  x >  o,
                          ii) yLfz) £ y _£ yR(z), and
                          iii)  o <_ z <_ h.j.

By mixing the lost mass uniformly we are assured that,  in  the  limit  of  long
travel  distances x, plume concentrations within the valley cross  section
will approach uniformity.

The above method provides a  simplified means of allowing some  plume  mass  to
escape upwards out of a valley  during its down-valley transport,  as  well  as
providing for "reflections"  of  the  plume from  sidewalls and  the valley
floor.   An objection to the  volumetric mixing  of the reflected plume mass
through the entire cross-sectional  volume seems necessary, however.  This
mixing of reflected plume mass  through the whole valley cross  section will
result in an underestimate of concentrations near the sidewalls and  valley
floor.   The amount of underestimation is not easily calculated due to a lack
of understanding of the physics of  the reflection processes  at the bounda-
ries during nighttime conditions.   The effects of deposition on valley  sur-
faces may reduce ambient air concentrations near the valley  surfaces,
however, and will tend to counteract these errors.  Further  work  is  deemed
necessary to evaluate both the  effects of deposition and the presence of
concentration underestimates on the valley surfaces.
                                      29

-------
     The integrations  in  Equations  (30)  and  (31)  are accomplished numeri-
cally in VALMET using  a polynomial  equation  given  by Abramowitz and Stegun
[33] attributed to Hastings  [34],   The  Gaussian  or Normal  probability
density function is given by
                           Z(x) =--exp (- f-) .                       (35)
where variable x is the standard deviation.

     The area under the Gaussian function  from -°° to x is given by
                             x          2           x
               P(x) = —   /  exp  (- f-]  dt  =    /  Z(t)  dt .            (36)
                      /2lt -°°                    -05
Using these definitions, we  note that  Z(-x)  =  Z(x)  and P(-x)  = l-P(x).

     Following Hastings [34] we may  approximate  P(x),  for 0 <_ x _<_ °°, by the
polynomial equation



          P(x) = 1 - Z(x)  [bxt + b2t2  +  b3t3 + b4t4 +  b5t5] + e(x)       (37)
where


                  1
       t =
           1 + 0.2316419 x  '

                       -8
      |e(x)| < 7.5 x 10
and
     b-,  =  0.319381530
     bi = -0.356563782
     b3 =  1.781477937
     b4 = -1.821255978
     b5 =  1.330274429.
                                      30

-------
The method for the numerical integration  is  illustrated  below  for
Equation (30).
                                                                         (38)
                           exp L- 7
                                                    Az
where
      n = h/Az
and  z.j = iAz.

The numerical integration for f^ is accomplished  in  an  analogous  manner  from
Equation (31).  The approximation is made in VALMET  using +5  standard  devia-
tions as the limits of integration for y as this  makes  the  calculations
tractable and produces no appreciable error in the results.

Calculation of Steady-State Nighttime Concentrations

     Calculation of nighttime pollutant concentrations  at receptor  point
P(x,y,z) can be made in a straightforward way with Equation  (27)  assuming
steady-state conditions.  The receptor must be located  within  a valley cross
section, so that Equation (33) can be applied for
  i)
 11
any
0 <
<_ h.j,  and
iii) yLTz) <_y <_yR(z)
In practice nighttime calculations are performed for  a  limited  number  of
points on the valley surfaces (floor and sidewalls),  choosing points that
correspond to locations where daytime pollutant concentrations  will  be
calculated.  The fixed daytime grid configuration  is  shown  below  in
Figure 12.  Since the model is run on a half-valley cross section, the cross
section sidewall angle a is chosen to be the average  of the observed side-
wall angles a, and ou-  Points for which nighttime pollutant concentrations
are calculated are snown on the figure with x's.

     If the grid element numbers are given by n =  1,2, ...NBOX,  locations at
points P(y,z) on the valley floor are at


                y  =  BOX,LEN  + (n  -  1)  BOXLEN   and    z  = 0   .
                                      31

-------
                                  w/2
Figure 12.  VALMET grid configuration on a valley cross section  illustrating
            the nomenclature used in the model.  The cross section  is
            representative of the model configuration  at  sunrise, when  the
initial  CBL-height (l-h), inversion top-height (h^
center-line height (?) are as shown.  The x's on th
illustrate the locations where nocturnal pollutant
concentrations are determined.
                                                               and  plume
                                                              e  figure
Calculation points on the valley sidewalls (n = NBOX + 1, NBOX + 2,
NTOTI) are at points P(y,z) where
                  y = 2" + z cot a ,
                      BOXLEN tan a
                  z = - + (n - NBOX - 1) BOXLEN tan a  ,
             and  a =
DAYTIME MODEL

     The daytime portion of VALMET begins with solar radiation calculations,
a parameterization of the surface energy budget, and estimation of the time
variation of sensible heat flux which destroys the temperature inversion
after sunrise.  The valley energy budget model of Whiteman and McKee [3] is
invoked to simulate CBL growth and inversion descent.  The nighttime
                                     32

-------
Gaussian plume on a valley cross section is used as the  initial condition
from which the numerical model proceeds to calculate concentrations  in grid
elements fixed to the valley floor and sidewalls.  The depth of these verti-
cally growing grid elements is identified with the CBL height.  Calculations
are made of concentrations within the grid elements taking account of fumi-
gation of the Gaussian plume into the elements and upslope advection from
grid element to grid element.  Pollutants are mixed uniformly within each
grid element, roughly simulating atmospheric diffusion processes in  the con-
vective (actually, advective) boundary layer.

Extraterrestrial  Solar Radiation

     Solar radiation on a plane surface above the earth's atmosphere can be
calculated using basic principles of spherical trigonometry.  The basic
formula for solar flux on a horizontal surface is



                             Q=S(j   cosZ                         (39)
                                       _
where SQ is the solar constant (1367 Wm  ), defined as the solar flux on  a
plane surface perpendicular to the sun's rays at the mean earth-sun dis-
tance, d.  The factor (d/d ) modifies the solar constant to account for the
fact that the earth-sun distance, d, varies during the year as the earth
travels in an elliptical orbit about the sun.  The cosine term accounts for
the projection of solar flux onto the horizontal plane.  This term is
clearly a function of earth-sun geometry, depending on the time of day, day
of year, and latitude of the surface.  Z is the angle between the normal  to
the surface and the direction to the sun, known as the zenith angle.  From
spherical trigonometry [35,36] this  is given as

                   cos  Z  =  sin    sin  &  +  cos  = latitude
      6 = sun's declination
      X. = sun's hour angle (i.e., an angular measure of  time  reckoned from
          solar noon where 15° represents one hour).  For example, 30°
          represents 2 hours after solar noon and -30° represents 2 hours
          before solar noon.

                                                        —  2
In these equations both the declination and the factor (d/d ) are a function
of the day of year (D, from 1 to 365).  McCullough [37]  gives some approxi-
mate analytical formulas that allow  the calculation of these  terms, utiliz-
ing the longitude of the earth (A) in its orbit around the sun as reckoned
from the earth-sun radius vector at  vernal  equinox (D =  DQ).  The formulas
are
                                      33

-------
          <5 = sin"  (sin £ sin \)   ,
         = (D - DQ) + 2e(sin cuD - sinuDQ),
and
 -T 2
(±)
                 - e cos
(41)

(42)




(43)
where
      E = 23°26' = maximum solar declination,
      01 = 2ir/365,
     DQ = 80 = day of vernal equinox, and
      e = 0.0167 = eccentricity of earth's orbit.

Using these equations, the extraterrestrial solar flux on an arbitrarily
oriented plane surface can be calculated as a function of hour angle, n.
Input to the equations must include the latitude and the day of year for
which calculations are desired.

     The equations are used to calculate the solar flux at any time from the
hour angle of sunrise, £_R, to the hour angle of sunset, £<-(-.  For a
horizontal plane, the times of sunrise and sunset are generally specified  in
terms of half-day length, R, calculated from Equation (40) for cos Z = 0,  or
Then
                          R=  cos"1  (-tan    tan  6)
                             = -R   and
                                                                 (44)


                                                                 (45)
Calculations using the above set of equations are made in terms of hour
angle and are thus referenced to the time of solar noon at a particular
site.  To convert the hour angles to local standard time it is necessary to
determine the local standard time of solar noon, and to adjust all the times
accordingly.  This is accomplished by making two corrections to the time of
solar noon (12h 00m OOs) in the local solar time coordinate system [38].
Then the standard time of solar noon is given by
   tNOON = 12h 00m OOs +   Equation of   + Longitude
                         time correction   correction
         = 12h 00m OOs
                          ET
                                                                 (46)
First, the equation of time correction is applied by adding the appropriate
number of minutes and seconds from Table 4.  Second, a longitude correction
accounting for the difference in geographical longitude between the site and
the reference meridian of its time zone is applied.  The correction is
+4 minutes for each degree of longitude west of the meridian.  For example,
                                      34

-------
                    TABLE 4.  EQUATION OF TIME CORRECTION

0* CFT
1 3m
16 9
31 13
46 14
61 12
76 8
**
12s
32
24
16
23
39
D
91
106
121
136
151
166
C
4m
-0
-2
-3
-2
0
ET
08s
01
51
44
33
10
0
181
196
211
226
241
256
C
3m
5
6
4
1
-3
ET
21s
46
21
44
07
50
D
271
286
301
316
331
346
C
-9m
-13
-16
-15
-12
-6
ET
06s
33
06
53
34
34
D CET
361 Om 47s





 * D = day of year
** Cry = equation of time correction (minutes and seconds).


the reference meridian of the Mountain Time Zone is 105°W longitude.   For  a
site at 106°W longitude, the correction is CL = 4(106-105) = +4 minutes.

     The above equations can be used to determine extraterrestrial solar
flux at any site on a given day for any given time  (i.e., hour angle).
Calculations using these equations for a latitude of 40°N and longitude of
109°W show that the solar flux QSb as a function of time on a given  day
closely follows a sine curve.  This simple analytical form proves so accu-
rate at the latitudes and longitudes of interest and is so easy to calculate
that it is used in the model instead of the full formulation.  The sine
approximation is given by


                          Qsh = Ax sin ~ (t - tSR)                      (47)


where
      Ai = solar flux at solar noon
       T = length of daylight period
     tSR = time °^ Sunnse» and
       t = time.

One can calculate A-^ by evaluating Equation (40) with 2 = 0, such that

                          _ 2
                  Al  =  So  tfj  (sin    sin  S  +  cos   cos  <5)  .


The length of the daylight  period  is given by


                      T  =  2R =  2  cos"   (-tan  £ tan  6)  .                  (49)
                                      35

-------
The time of sunrise is evaluated in local standard time using
If the model equations are used far from the latitudes of the central  Rocky
Mountains, the model user should verify that the sine approximation  is  a
valid representation of solar flux, by comparing its calculations with  the
full analytical equations.

Parameterization of Sensible Heat Flux

     The sensible heat flux which constitutes the basic driving term  of the
daytime portion of the model is parameterized from the extraterrestrial
solar flux using the formula
= A0[Al
sin
                                  I (t  - tSR)]  » AQ Qsh                 (51)
where A0 is a fraction between 0 and 1.  This formula should be  recognized
as a crude parameterization for sensible heat flux, simply stating that the
sensible heat flux is a constant fraction of the time-dependent  extrater-
restrial solar flux, as illustrated below in Figure 13.

     AQ depends on a great many factors which are not yet included
explicitly in the model.  In general, A  should be a time-varying function
dependent on the

     transmissi vity of the earth's atmosphere,
     cloud cover,
     surface albedo,
     soil moisture,
     surface cover or vegetation,
     longwave radiation budget, and
     other factors.

A future version of VALMET should include some of these factors  explicitly,  at
least the ones which may affect worst-case pollutant concentration, using as  a
basis the extensive work already conducted on these topics by other
investigators.

Model Grid
     The daytime portion of VALMET uses a numerical technique to  simulate  time
varying pollutant concentrations in grid elements that are fixed  to the  valley
floor and sidewalls.  Calculations are made for individual time steps  during
the temperature inversion breakup period beginning at sunrise and using  the
nocturnal steady-state concentrations within the valley cross section  as an
initial  condition.
                                      36

-------
        HEAT FLUX
                                      Q
                                       Sh
       A0Ai _
 Figure 13.  Extraterrestrial solar heat flux as a function of time, showing
             solar model nomenclature.  The figure illustrates how the
             sensible heat flux F is parameterized as a fraction AQ of the
             extraterrestrial solar heat flux on a horizontal  surface above
             the site.  The times of sunrise and sunset
             daylength T is defined in the model as the
             between sunrise and sunset.
are shown.  The
time difference
     The geometry of the valley cross section, illustrating the numerical
grid, is shown in Figure 14.  Calculations in the model are performed on a
half-valley cross section.

     Grid elements are arranged from the valley center across the valley floor
and up the sidewall, with grid element height representing the height of the
CBL, assumed not to vary from grid element to grid element.  At sunrise the
model is initiated with a 25 m CBL height, and an inversion top height that is
obtained from observations in the modeled valley.  These initial heights
change after sunrise as sensible heat flux drives the destruction of the
valley inversion.  Specifically the height of the CBL, H(t), increases and the
height of the inversion top, h(t), decreases.  The simulation is accomplished
by treating the Gaussian plume within the cross section as being "frozen"
within the inversion.  After sunrise, concentrations within individual grid
elements change as the grid elements grow upwards into the frozen plume, as
the inversion sinks causing pollutants to sink into the tops of the grid ele-
ments, and as upslope flow develops within the CBL.  These upslope flows
develop as air parcels are heated by sensible heat flux over the sidewalls and
rise up the slope.  The speed of the slope flows is calculated under a con-
tinuity of mass constraint within the cross section below the (sinking) inver-
sion top.  That is, sinking of the top of the inversion with time implies that
mass is removed from the cross section below the inversion top level.  In the
                                      37

-------
 Figure 14.  Cross section of the valley floor and sidewalls illustrating the
             grid elements whose height corresponds to the C8L heights.  The
             rate of sinking of the top of the inversion (arrow) is  related
             to the upslope transport of mass (other arrow) by the equation
             of mass continuity.


model, the assumption is made that the upslope flows carry the mass  required
to allow the inversion to sink at the rate predicted from Whiteman and McKee's
[3] bulk thermodynamic model. This model is driven by sensible heat  flux
considering the thermodynamic energy budget of the valley cross section.

     One of the assumptions in the model is that the top of the valley inver-
sion at sunrise extends horizontally across the valley cross section.  This
assumption is supported by data from an observational program conducted in
Colorado mountain valleys [9],  In that study several experiments were con-
ducted by dual  tethersondes located on a cross valley section.  These obser-
vations showed that the inversion top was nearly horizontal.  This is not
unexpected, since hydrostatic forces should produce settling of cold air
masses in topographic depressions or valleys and would result in horizontal
isentropes in the absence of major cross valley inflows.  This effect would be
more pronounced for strong inversions where the hydrostatic forces would be
stronger.  Gravity waves are known to propagate frequently on the upper
boundary of such inversion layers and are frequently noted in acoustic sounder
records.  These waves are generally of small amplitude and should not greatly
affect the validity of the assumption of horizontal homgeneity.

Thermodynamic Equations for CBL Ascent and Inversion Descent

     For details of Whiteman and McKee's thermodynamic model of temperature
inversion breakup in mountain valleys the reader Is referred to the  original
article (Appendix C).  Here a brief explanation is given.
                                      38

-------
     One can view a valley temperature inversion  as  a pool  of  cold  air  trapped
within a valley.  This pool  has an energy  deficit relative  to  the air above
the valley, and the temperature inversion  can be  destroyed  only  by  warming the
pool  to the same potential temperature as  the air above  the valley.  The
amount of energy required to warm the pool  can be calculated from the First
Law of Thermodynamics knowing the depth of  the pool,  the vertical temperature
gradient within the pool  (assumed constant)  and the  cross valley dimensions  of
the pool, by means of a volumetric integration through the  valley cross sec-
tion.  This energy deficit is overcome by  sensible heat  flux,  which  can be
estimated as a fraction of the downward solar radiation  coming across the area
of the top of the valley  temperature inversion.  Conversion from solar  energy
to sensible heat occurs at the ground.  Downwards solar  heat flux across the
inversion top will be intercepted on a valley surface below the  inversion top
and will be available as  sensible heat flux  to warm  the  inversion airmass.

     The topography of the valley strongly  affects the course  of the inversion
destruction by determining which of the patterns  of  inversion  destruction will
be followed and by controlling the overall  rate of heating  of  the valley atmo-
sphere.  This control on  the rate of heating can  be  explained  by Figure 15.
Consider, for example, a  valley that is completely filled by a temperature
inversion.  Solar energy  coming downwards  across  the  top of the  valley
inversion will be converted to sensible heat flux at  valley surfaces and will
be used to heat the inversion volume.  The  ratio  of  the  area at  the  inversion
top through which this radiation comes to  the inversion  volume governs  the
rate of warming of the valley atmosphere and, consequently, the  time required
to destroy the inversion.  Given an equally  strong and deep inversion over the
plains, the valley inversion will  be destroyed more  rapidly since the same
energy input goes to heat a smaller volume  of air in  the mountain valley.  The
thermodynamic model includes this important  factor,  which arises from the
volumetric integration in the derivation of  the model  equations.
           I    J_J   LI
L_L_L_L_LLJ
                                                                     1
                 VALLEY
             PLAIN
   Figure 15.  Illustration of the effect  of topography  in  controlling  the
               heating rates of the air within  a valley  temperature
               inversion versus the air within  an inversion  over  the
               plains.  The same incoming  energy heats  a smaller  volume of
               air in the valley case.


     Whiteman and McKee's model is a "bulk"  thermodynamic model  in that it
does not differentiate between sensible heat flux over  different  valley sur-
faces, dealing only with a bulk heat flux.   The heat  flux which  drives  the
valley inversion destruction is partitioned  or  distributed  in  a  fundamentally
different way from that for an inversion over homogeneous terrain.  There the
                                     39

-------
sensible heat flux destroys the inversion by driving an upward growth  from  the
ground of a convective boundary layer, which warms the inversion air mass from
below until the temperature deficit is overcome and the inversion  is
destroyed.  In contrast, in a valley the upward heat flux over valley  surfaces
develops a convective boundary layer but also, due to sensible heat flux
convergence over the slopes, causes warmed air parcels to flow up  the  slopes
in an upslope flow that develops within the convective boundary  layer.  These
upslope flows remove mass from the base of the temperature inversion in the
shallow slope flows, and through mass continuity, results in a general subsid-
ing motion over the valley center.  The atmospheric energy budget  approach
used by Whiteman and McKee is capable of partitioning energy between these  two
different processes to produce inversion destruction solely by CBL growth,
solely by inversion descent (assuming a non-growing CBL is present initially
in a simulation), or by a combination of the two processes.  The partitioning
is controlled by a single parameter, k, defined as the fraction  of sensible
heat flux going to CBL growth.  The remaining fraction, 1-k, is  assumed to  be
responsible for mass transport up the CBL which results in inversion
descent.  The proper partitioning of energy (i.e., the most appropriate value
of k) in a particular valley can be determined by fitting the thermodynamic
model to observations in a particular valley.  This was done for a Colorado
valley by Whiteman and McKee [9], although they cautioned that the value of k
will depend on other factors besides valley geometry, including  the magnitude
of sensible heat flux.

     Further research on the functional value of k is suggested  as a high
priority in this modeling approach, although the model is functional using  the
curve fitting approach to determining k.

     The thermodynamic model is composed of two coupled equations.  The first
equation is a prediction equation for CBL height where, in accordance  with  the
bulk nature of the model, the CBL depth H is assumed not to differ over the
valley floor and sidewalls.  The first equation is



                                    =  HJ  + AHJ                           (52)
where j is a time step index and


                      n   ,   w  +  H .C   A  A,
                                               f
Note that a non-zero initial CBL height is necessary to make this  prognostic
numerical equation tractable.  In the model, this requirement  is met  by  using
an initial CBL height at sunrise of 25 m.  The second equation, describing  the
inversion top height h is
                                      40

-------
                                                                        (54)
where
                                                             h.+n .
                                  h.      g
                          h,Y(w + f- C) -j (t-t.)(w+h.C)
                           J      £•       ^-      '     J
                                                                    At' (55)
  T"
       1000, .286

  p = atmospheric pressure,

 c  = specific heat of air at constant pressure,

  C = cot ctj + cot a2,

 Ho E Hi * °»

 ho = hi'

  t = (j+1) At + t-j, and

  j = 0,1,2,3,....n.

     The terms in the numerator and denominator of Equation.. (54) involving the
warming coefficient for the air above the inversion, 6(°Ks~ ), allow the model
to incorporate the retarding effect on temperature inversion breakup caused by
warm air advection above the valley temperature inversion.  Extra energy is
required to destroy the valley temperature inversion if this warming occurs
during the temperature inversion breakup period since the inversion cannot be
broken until the entire valley atmosphere is warmed to the temperature of the
air above the valley.

     The numerical simulation using these coupled equations proceeds with
discrete time steps and is completed when the inversion is destroyed at the
first time step n at which the CBL height becomes greater than the inversion
top height, such that
                                                                        (56)
In the limit of an infinitely wide valley (i.e., a plain) and k=l (i.e., all
sensible heat flux going into CBL growth) the equations  reduce to a single
equation describing CBL growth over homogeneous terrain.  In the limit as k
goes to 0, the equations describe inversion destruction where the CBL fails to
                                      41

-------
grow after attaining an initial height, and inversion destruction occurs due
to sinking of the inversion and adiabatic warming of the valley atmosphere.
Whiteman and McKee [3] have shown that this approximates inversion destruction
in snow covered valleys.  Values of k between 0 and 1 allow the coupled
equations to describe intermediate methods of inversion breakup in which the
sensible heat flux is partitioned between CBL growth and inversion sinking.  A
typical inversion destruction in the fall in the Eagle Valley of Colorado was
simulated well  using a value of k=0.14.  The appropriate value of k for other
valleys would be selected based on careful analysis of valley atmospheric
data, as discussed further in Section 4.

     Sinking of the inversion layer will result in a sinking of the frozen
plume within the inversion.  The pollution mass can be calculated by consider-
ing the vertical Gaussian pollution profile above each of the grid elements.
As shown by Equations (52) and (53) for k*0 the CBL depth (and hence grid
element height) increases with each time step.  This CBL growth entrains
pollutant mass  from the inversion's stable core into the top of each grid
element.  These two processes, sinking of the frozen plume into the top of
each grid element and growth of the grid element upwards into the frozen
plume, combine  to determine the amount of pollutant mass introduced into each
grid element during each time step.

     As illustrated in Figure 16 for a grid element on the valley floor, these
two processes result in a volume of air of depth AH. + Ah- and length y, - y~
being entrained into the growing grid element during time step j.  The average
pollutant concentration in this air volume can be calculated from the Gaussian
plume equation  [Equation (33)] by averaging the pollutant concentrations at
the centerpoints of the four sides of this volumetric element (Figure 17).  A
z-coordinate transformation is useful in these calculations so that one may
think of the coordinate origin on the valley floor rising into the fixed
plume, rather than the frozen plume sinking towards the valley floor.

Advection in the Slope Flows

     Conservation of total air mass on the two-dimensional valley cross sec-
tion of interest implies that there must be a relationship between the sinking
of the inversion top and removal of mass from below the inversion top due to
upslope transport of mass in the convective boundary layer over the slope.
Thus, at the level of the inversion top, a slow subsidence of the broad stable
core must be balanced by stronger upward motions through this level due to
flow up the narrower sidewalls CBLs.

     The continuity equation for air of constant density on the two-dimen-
sional cross section may be written as
                                  _ _                                     .  }
                                 3y  '  •  3l •                              (57)
                                      42

-------

                     Ah;
 I
'AH?
                                 CBL
J+1
                    //////////^^^^^
Figure 16.  Diagram showing the changes in CBL and  inversion depth
            at a given time step above a valley floor grid
            element.  The amount of air volume incorporated into
            the growing CBL at each time step is seen to be repre-
            sented by (y2 - y-^'Uh. + AH.) cubic meters.
                            Ah:
                           AH:
                  v                             v
                  M                             '2

   Figure 17.  Representation of the volumetric element of mass
               incorporated into the growing CBL  above a valley
               floor grid element at each model time step.  The
               x's in the figure represent the positions where
               pollutant calculations are performed at each time
               step  to estimate the average pollutant concentra-
               tion  carried within the volumetric element during
               the time step.
                                 43

-------
This equation may be used to calculate wind velocities  at  the  boundaries  or
septa of the grid elements in the slope flow layer.  For the first  grid
element on the valley floor, we assume that the wind velocity  on  the  left
septum at the valley center is zero (i.e., there  is no  mass transferred across
the valley center within the CBL).  At a given time step a certain  amount of
mass sinks into the grid element of length L and  depth  D.  Integrating,
    ,                     ,
 /   §7 dy  dz  =  -   /    /  |f dy dz
\    *            r\   r\
                                                                         (58)
or
                                        L -
                                                                         (59)
Thus, an inversion sinking at the rate of  .01 m/s  into the  top  of  a  100-m-long
grid element of depth 50 m will produce a wind speed  increase from the  left
septum of the grid element to the right septum of  the grid  element of
VL - Vo =
                                  100
               .01  =  .02  m/s  .
                                                                         (60)
The velocity, v, normal to a given septum of any grid element  can  thus  be
determined by summing the velocity changes calculated across each  grid  element
beginning at the valley center and ending at the septum  of  interest.

Pollution Concentration Calculation Method

     Pollution concentration calculations are made  for each time  step  for  each
grid element following the general outlines of a numerical method  described  by
Whiteman and McKee [11].  Following this method, calculations  at  each  time
step are performed sequentially for all of the grid elements,  beginning with
the grid element at the valley center  and progressing up  the valley  sidewall.
Calculations within each grid element  (Figure 18) are made under  a pollutant
mass conservation assumption.  The pollutant mass sinking into  the top  of  each
growing grid element is calculated using the method described  above.   Simi-
larly, the advection of pollutant mass into and out of each grid  element is
accomplished using the velocity calculations at grid element septa as  deter-
mined using a total air mass conservation constraint as  described  above. The
concentration Cn t of pollutant in grid element n at the  end of the  t-th time
step of length At is given by a simple pollutant mass balance:
                                      44

-------
                                                           »»M5
                                      M3

     Figure 18.  Schematic diagram of an individual  model  grid element
                 illustrating the pollutant  mass balance.
                               M.  .„.  .  + M.   - M  ,.
                             _  initial     in    out
                        'n,t
                                                                        (61)
                    M1n,t + M2n,t + M3n,t + M4n,t " M5n,t
where

 Vn  is the volume of the nth grid element

Ml   is the mass of pollutant coming into the growing grid element  n  during
     time step At

M2n  is the mass of pollutant advected into box n from the adjacent upstream
     grid element (n-1)  during time step At

M3n  is the net mass of  pollutant introduced into grid element  n during  time
     step At from pollutant sources and sinks within the grid element
                                     45

-------
M4n  is the initial mass of pollutant  in  grid element  n  at  the  beginning of
     time step At

M5n  is the mass of pollutant advected out  of grid  element  n  into the adjacent
     downstream grid element (n+1) at  the end of time  step  At after complete
     mixing of masses Ml through M4 has taken place within  the  grid element.

The separate terms are calculated as follows:
                          Ml
                            n,t
                       AHJ
(62)
where xn is the average pollutant concentration within  the  volume  element
(Ah, + AH.) that is incorporated into the  grid element  n  at  time  step  t  (or,
following our previous nomenclature, j).   Ai , the  area  of the  grid element t
(Figure 18), should not be confused with the A, used  previously  in
Equation (40).
                                                    top
                              n .. = C  .
                              n,t    n-l,
                                                 (63)
where

A2 = area of box septum

vi = wind velocity component into grid element  normal  to  box  septum.
    M3
      n,t
                                       - S) AlAt
(64)
where
                                 O   1
 Q = pollutant emission rate [ML   T  ]
                               _p   I
 S = pollutant removal rate [ML"  T  ].
                              M4n,t = Cn,t-l Vn
                                                 (65)
               M5
                 n,t
Ml    + M2    + M3    + MA
  n,t	n .t	n ,t	n.t
             V
              n
(66)
                                      46

-------
where

V£ = wind velocity component out of box normal to box septum.

Exponential  Decay of Concentration

     Calculation of concentrations in individual grid elements follows the
procedure outlined.  A modification of this procedure is necessary, however,
for individual grid elements on the sidewalls when the top of the  inversion
sinks below the level  of the grid element, as the depth of the CBL suddenly
increases when this occurs.  Concentrations would then be expected to decrease
rapidly.  This is handled in the model by simply allowing the last concentra-
tion calculated in the grid element to decay exponentially with time after the
inversion top descends below the grid element.  In view of the lack of
observational evidence, the time constant of the exponential decay is chosen
arbitrarily so that the concentration decreases by 2% at each 10-s time step,
such that


                           r    -go At/10 r                             (67]
                           cn,t   >yB       Ln,t-l   '                     lb/;


As with any numerical  model, one must be cognizant of the possibility of
numerical instabilities affecting model results.  The criteria for maintaining
stability [39,40] constrains the model user to maintain a certain  relationship
between the model time step and the grid element length, depending on the
upslope wind speed encountered in the simulation.  This relationship is


                               vmax  At < y2  -  YI  .                         (68)


Thus, the maximum upslope wind speed  simulated in the model, when multiplied
by the time step length must be less  than the grid element length  in order to
maintain computational stability.  In the VALMET model, the user may choose
(yo-y-i) but has no control over At and v   .  Calculations are made auto-
matically within VALMET to ensure computational stability.  Since the maximum
upslope wind velocities should never  exceed 10 m/s,  we may ensure  computa-
tional stability by setting

                                      y9 - y,
                                 At =  *1Q  X .                         (69)


Thus, for example, a time step of 10  s is sufficiently small to ensure com-
putational stability in a model simulation with 100  m grid elements.

     Following the numerical method outlined above,  concentration may be
calculated for every time step for every grid element.  Rather than storing
each of these concentrations, which would take a great deal of computer
                                      47

-------
memory, 5-min-average concentrations are stored.  This greatly reduces
computer storage requirements.  The 5-min-average concentrations constitute
one of the main outputs of the model.

Maximum 1- and 3-Hour-Average Concentrations

     Another ouptut of the model is the maximum 1- and 3-hr-average concentra-
tions and their times of occurrence in each of the grid elements.  These
calculations are made from the basic 5-min-average concentration array for
each grid element by means of a moving average method.  Initial tests have
been conducted to confirm that the maximum 1- and 3-hr-average concentrtions
calculated from the 5-min-average array do not differ significantly from
concentrations calculated from a 10-s-average concentration array.
                                      48

-------
                                   SECTION 4

                       OVERVIEW OF MODULAR  VALMET MODEL
     The initial valley air quality model called VALMET  has been designed  and
constructed based on hypotheses presented in Section 2 that arose  from  an
observational study of Colorado valley meteorology.  The model uses  the  tech-
nical  approach and equations outlined in Section 3.

     The VALMET model can be used to simulate the transport and diffusion  of
pollutants released from an elevated source in a well-defined mountain  valley
during the nighttime and morning transition periods. The model operates  on  a
valley cross section an arbitrary distance down-valley from an air pollution
source and has been constructed to include parameterizations of the  major
physical processes that act to disperse pollution during these time  periods.

     Before a modeler attempts to use VALMET to simulate dispersion  in  a
particular valley, he should critically review Sections  1 through  3  of  this
report.  Since the modeling approach is phenomenological, and individual
physical processes affecting pollutant dispersion are parameterized, the
modeler should be wary of applying the model to a meteorological or  topo-
graphical  situation where model assumptions are invalidated or physical  pro-
cesses parameterized in the model are clearly not occurring.  The  modeler
should carefully review existing meteorological data for the valley  of
interest to see whether the model applies.

     Even if the model's assumptions seem valid, the present state of the  art
in such modeling results in the use of rather arbitrarily specified  parameters
within the model.  Further work is necessary to refine our estimates of  the
values of these parameters (e.g., k, a   and a ).  This work would benefit
from comparisons of model results witlr actual diffusion  trials in  real  val-
leys.   In view of our rather tentative understanding of  valley meteorology,
the model  has been constructed in a modular fashion.  This modularization  of
the code is a major design feature of the model and should allow the code  to
be modified easily as we learn more about the meteorology of valleys and find
better parameterizations for individual  physical processes.  The modules have
been designed, where possible, so that the analytical or numerical calcula-
tions  in the modules can be replaced by observational data, when available,
with minimal modifications in the computer code structure.  For example, if
acoustic sounder data are available to measure CBL height and inversion  top
height, these data can replace the analytical scheme for predicting  these
heights.


                                     49

-------
     The modular structure of the model is illustrated in Figure 19, where  the
modules are named and the functions of the modules are indicated.   In this
section we give the model user an overview of the VALMET model, explaining  its
structure and its input requirements.  A full technical description of the
modules, including variable name definitions and module inputs and  outputs  is
given in Section 5 for the interested computer programmer.  In Section 6 the
model outputs are illustrated for two sample simulations.  The full VALMET
code is given in Appendix A.  A brief historical summary of modifications to
the code is given in Appendix 0.

FEATURES OF THE COMPUTER CODE

     The VALMET model is coded following a FORTRAN 77 standard [41] and should
be useable without modification on any computer system having an up-to-date
FORTRAN compiler.  The model is documented internally through the liberal use
of comment statements.  These comments explain the purpose of sections of code
or individual FORTRAN statements, or serve as variable name definition tables.

     Several  special  features or protocols are included in the code.  The
dimensions of arrays in the code are generally set using PARAMETER  state-
ments.  Thus, if the user finds that the array sizes are too small  for the
length of the simulation or too large to fit in a smaller computer  he may
easily change the dimensions by modifying PARAMETER statements in the main
program and in selected subroutines.  COMMON blocks are fully utilized in the
code to reduce the memory requirements of the operating progam.  Most of the
parameters passed to the subroutines are generally passed through the sub-
routine argument lists.  Arrays, however, are usually passed through COMMON
blocks.

     The model was developed on a VAX/VMS 11/780 computer.  After compilation,
a single model run generally takes 1 minute of central processor time on the
VAX, loads in 95,000 bytes (25K words) of core, and costs approximately $1.00.
Costs of optional plotting will vary from installation to installation.

MODEL INPUTS

     The VALMET program runs in an interactive mode in which the user controls
program execution from a remote interactive terminal.  The user enters model
inputs by following directions given to the user on the terminal screen.

     As with any model, model performance will be a function of the suitabil-
ity of the input data entered by the user.  Where worst-case values of the
input parameters are estimated on the basis of few data, the user should
consider the model results to be a screening analysis only.  On the other
hand, an industrial  installation having a great deal of pollutant concentra-
tion and meteorological data could use the model as a site-specific model by
making specific modifications to the model to include observed plume dilution,
dispersion coefficients, etc.  This would require considerable field experi-
mentation and data analysis.
                                     50

-------
                                          PROCESSES
                                          PLUME RISE
                                          DILUTION
                                          POLLUTANT MASS
                                          CONSERVATION
                                          STEADY STATE TRANSPORT
                                          AND DIFFUSION
           I
         EBDGT
         DESCNT
                                          SOLAR FORCING
                               SURFACE ENERGY BUDGET
                               INVERSION DESCENT AND
                               CBL GROWTH
Figure  19.
Flow diagram of VALMET model, showing the  modular structure  of
the model  and the physical  processes parameterized in the
modules.   The modules indicated between the  two-horizontal
dashed  lines constitute the nocturnal portion of the model.
The daytime portion of the  model follows the second dashed
line.
                                  51

-------
y
N=0
"t ,N=N+1
NO -X"
f iv
PROFIL
\
VELOCY
I
BRKUP
A
l-IMCTPDC

__ GAUSS

FU
> ^s
          FUMIGATION
    WRITE
   SUMMARY/
 *\ OUTPUT,
     FILE

    WRITE
   OPTIONAL
•^.PLOTTING/
     FILE
\
                         POLLUTANT MASS FUMIGATION
                          TOTAL MASS CONSERVATION
                          AND ADVECTION
                          POLLUTANT MASS CONSERVATION
                         MULTI-HOUR-AVERAGE
                         CONCENTRATIONS
        Figure  19.  (contd)
               52

-------
     When the user runs VALMET from an interactive terminal,  a data  table
appears on the user's screen.  This table gives sample  input  values  for  a
model run.  The user may follow directions given on the  screen to  change any
of the input values to simulate air pollution dispersion in any valley of
interest, or he may simply specify that the sample (or  default) simulation  be
run.

     The input table (Figure 20) is arranged according  to  categories  of  input
data.  Table input values for the sample simulation are  generally  specified in
the MKS system of units, although pressures are given in millibars and
parameters involving potential temperatures are given in degrees Kelvin  in
conformance with general meteorological practice.

     Table 5 below gives a full listing of the appropriate units of  all  input
parameters used in VALMET as well as the default values  of these parameters.
VALMET is programmed so that the user can change input  values as many times as
necessary until the input table is properly completed.   Once  the input table
is satisfactory to the user, he instructs VALMET to proceed with the model
run.

SPECIFICATION OF THE MODEL INPUTS BY THE USER

     As a convenience to the user, the VALMET program checks  to see  that any
parameter values input by the user are within a normal  atmospheric range of
values for that parameter in the units required by VALMET.  See Table 6  for a
listing of the ranges of values assigned to each parameter.   If the  input
value is outside normal atmospheric ranges, a message is sent to the user to
respecify the input value.  The intent of this feature  is  to  help  the user
identify gross input errors caused, for example, by using  the wrong  units for
input variables.

     In this section of the report we will instruct the  model user how to
determine the input parameters that are required by the  model.  The  parameters
are entered into the model by the user as he follows instructions  given  on  the
interactive terminal screen.  The input parameters are  discussed in  groups, as
follows:

Valley Characteristics

     The characteristics of the valley topographic cross section must be
obtained from topographic maps.  The user should plot the  terrain  cross  sec-
tion at the down-valley distance of interest.  From this cross section the
user should estimate the valley floor width (estimates  to  the nearest 100 m
should be sufficient) and the two sidewall inclination  angles.  The  user must
use his judgment when making these estimates, depending  on the representative-
ness of the topographic cross section plotted.  Since the  model will  deal with
the valley sidewalls only to the initial depth of the temperature  inversion,
the user should ignore the sidewalls that extend above  this height when  esti-
mating representative sidewall angles.
                                     53

-------
           •     O     CM     O O O               CO
          in     o     co     o o o               Ed
          r-     O             .   •   •               C5
                  .            ,=r .=r on              z
                 o                                   <
                                                     x
                                                     o

                                                     EM        0>
                                                     O        -=

CO        JEdXCJCOCC               CC        c
Ed        «a!CQMO3eO               Ed        o
a                                                   CQ
                                                     S        ^

                                                     2:        a;
                                                               Q.
                                                       •        a.
II
CM
<
X
CU
J
«a!
in

in
n


«s
H
Ed
ca
<*
in
CM
n



CC
>H
M
*
O «-
O CM
II II




CJ CO
a a
cr> CM
<- CM
o •
o o
n


Q
<4
CC
CO
in
CM
O •
O O
TO THE FOLLOWINi
EH
Ed
Ed
CC

CO
CC
Ed
EH
Ed
S

CC

cu

z
o
M

N
M
t •)
•a!
H
EH
M

M

Z
•a!
ce
o
o
cc
a,

Ed
33
EH
in
o
n
CM
CM
e
O


II




o
ai

in
n
33
CM
<
.=r
0
0
VD


II

33
EH
Q
H
3

in
CM
o
o
n




•
CC
Ed
EH
• O
CC *aj
Ed CC
H «J
CO 35
«! O
CC
•a! Z
33 0
CJ H
CO
J CC
Ed Ed
Q >
0 Z
S M
1 URLONG=105.00
I IDAY = 21
«- on
O O>
0
«
0
jy

II II



EH
•a! O
J S

O CM
»- t-









z
0
H
EH
«aj
O
O
J

Ed Ed
H £H
H <
CO Q
O •
O O
O CM
in
c-
II II II
CO O
CO CC CQ
Ed Ed <
CC N EH
CM >M CO
vO CO ^
T- T- f\j
O « O
O 0 i-
• O O
000
t- o •
T- O
II II II


cu
s
Cd 0
EH X O

in c— o
r~ r- CM





Ed
CC
Ed
33 Ed
*** i«
CO 0
O J
S CU
EH
aj z
<$
SH M
Cd CO
J CO
J »
«< <£
> O
O •
O O
0
o
II II
2
EH CO
CO «B
sr i>-
CM CM
O 0
o o
• »
0 ^T
in CM
CM
II II


J
Ed

33 CO

ro vo
CM CM






•
(X
Ed
E-i
CJ
•a
OS
•a
33
O

M
CJ
•a!
EH
CO
cn
a:
Ed
EH
Z
Ed
1
1
1
Ed
•1
>— 1
ca
H
CO
Ed
p-i
J
•a;
{>

Sg
M

Ed
O
^

CQ

O
g-t

Cd
CC
*3j

CO
Ed
O
y%
«3J
X
O

o
z

EM
M


co

CC
Ed
EH
y*
Ed

1
1
1
1


0

X
CJ

Ed
CQ

0
EH

CO
M

S
CQ


c-


**—
O
c
o
4-1
tO
•!->
I/I
3
r—
'""
1—1

•
O
CM
CL)

^
C71
>r~
lj-








.
CJ
0)
{_
0
in
OJ
>
•i—
-^J
O
2
O)
c
•r—
to
c_
d)
to
3







1









                                    54

-------
             TABLE 5.  DEFAULT VALUES OF VALMET INPUT PARAMETERS

Variable
Name
1.

2.

3.
4.
5.
6.

7.
8.

9.

10.
11.
12.
13.
14.
15.
16.
17.
18.

19.

20.
21.
22.

23.
24.
25.
26.
27.
AO

K

WIDTH
ALPHA1
ALPHA2
NBOX

HZERO
GAMMA

BETA

LAT
URLONG
MO
IDA
IYR
TEMP
PRESS
X
YZERO

UC

Q
STAB
US

H
STMP
SRAD
SVEL
AS
Definition
Fraction of solar flux converted
to sensible heat flux
Fraction of sensible heat flux used
to grow CBLs
Val ley floor width
Sidewall #1 elevation angle
Sidewall #2 elevation angle
Number of grid elements on valley
floor half-width
Initial inversion depth
Inversion vertical potential
temperature gradient
Rate of potential temperature
increase above valley
Latitude
Longitude
Month
Day of month
Year
Ai r temperature
Air Pressure
Down-valley distance from stack
Stack offset distance from valley
floor centerline
Plume-carrying wind speed at
receptor cross section
Stack emission rate
Atmospheric stability
Plume-carrying wind speed at
source cross section
Plume centerline height
Stack emission temperature
Stack radius
Stack emission velocity
Cross-sectional area at source
Default
Units Value
0-1

0-1

m
o
o
-

m
°K/m

°K/s

°N
°W
1-12
1-31
00-99
°C
mb
m
m

m/s

kg/s
D-F
m/s

m
°C
m
m/s ,,
10J nr
0.24

0.15

600
15
15
3

500
0.025

0

40
105
9
21
82
10
750
10000
0

4

0.001
F
4

250
100
3
0
0

Date

     The user must specify the date of the simulation by giving the month, day
and year.  The date is necessary in the solar model to calculate day length,
time of sunrise, etc.  The year is necessary to account for the minor effect
of leap years on the simulation.
                                     55

-------
            TABLE 6.  LIMITS ON THE VALUES OF INPUT PARAMETERS

Variable
Name
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
AO
K
WIDTH
ALPHA1
ALPHA2
NBOX
HZERO
GAMMA
BETA
LAT
URLONG
MO
IDA
IYR
TEMP
PRESS
X
YZERO
UC
Q
STAB
US
H
STMP
SRAD
SVEL
AS
Units
_
-
m
o
o
-
m
°K/m
°K/s
°N
°W
-
-
-
°C
mb
m
m
m/s
kg/s
D-F
m/s
m
°C
m
m/s 0
10V
Lower Limit
0
0
0
0
0
1
0
0
0
-90
-180
1
1
0
-50
600
100
-50000
0
0
D
0
0
0
0
0
0
Upper Limit
1
1
50000
90
90
30
2000
0.050
0.0005
90
180
12
31
99
50
1050
100000
50000
15
1
F
15
1000
600
5
50
150000

Site Location

     Both latitude and longitude of the site must be
order to support solar radiation calculations within
value is necessary to account for corrections to the

Model Characteristics
 specified by the user in
 the model.  The longitude
 time of sunrise.
     The user must specify the number of grid elements on the valley floor
half width in order to define the model's numerical
considerations that the user should keep in mind when
number.  The implications of the specification can be
to Figure 12.
grid.  There are several
  specifying this
  considered by referring
                                     56

-------
     The  number  of  grid  elements  on  the  valley  floor half width (NBOX)  is used
in the model to  figure the  length  of the grid elements.   The  length is  given
by

                               BOXLEN=w/(NBOX*2)

where w is the valley floor width.   Once BOXLEN  is  calculated,  the  model  then
calculates how many grid  elements  of this length (NTOTI)  will  be necessary in
order to  reach the  top of the  inversion  when the grid elements  are  arrayed end
to end.   NTOTI must not  exceed 30, as  currently  dimensioned  in  the  model.  On
the other hand,  NTOTI should not  be  too  small,  since this parameter affects
the spatial resolution of concentrations calculated within the  model.   As a
rule of thumb, NBOX may  be  chosen  to produce grid elements of  100 or 200  m
length.

Valley Atmosphere

     The  mean temperature and  pressure within the valley  temperature inversion
at sunrise should be estimated using the altitude of the  site  and knowledge of
seasonal  changes in atmospheric temperature.  Observational  data would  be use-
ful in estimating both of these parameters, but  the accuracy of these
estimates is not critical to the model results.   The values  are used to
correct the model concentration values to standard  pressure  and temperature
and to calculate the atmospheric  density for use in the  inversion breakup or
energy budget calculations.

Stack Characteristics

     The  stack characteristics should  be obtained from engineering  calcula-
tions for the specific air pollution  source.  The required parameters include
physical  stack height, effluent temperature, stack  inside radius, and exit
velocity.  If the user does not wish  to  engage the  plume  rise module, the
stack radius or exit velocity  should  be  specified as zero.  The plume center-
line height over the entire length of  the valley will then be equal  to  the
value specified as the physical stack  height.  The  final  parameter  that must
be specified by the user  is the area  of  the valley  cross  section at the
stack.  This area, bounded by  the  valley floor,  the two  sidewalls and the
temperature inversion top at sunrise,  is  used in the plume dilution module to
calculate clean air volume flux across the cross section  at the pollutant
source.  The area should  be calculated from Equation (1),  but is input  in
thousands of square meters  (i.e.,  the  results of the calculation with
Equation  (1)  should be divided by  1000 before being  input into  the  model).
The model user will need  to know the  inversion depth at  sunrise in  order  to
calculate the area.  If the user does  not wish to engage  the plume  dilution
module or does not have the wind speed observations  on the cross valley
section where pollution calculations are  to be made,  he  should  specify  the
cross-sectional  area at the stack  to be  zero.  The  wind  speed specification at
                                     57

-------
the downwind cross section is then deactivated (i.e., is not used  in  the
model) and the plume is advected down the valley with the wind speed  specified
at the pollutant source with no plume dilution during transport.

Inversion Characteristics
     The user must specify inversion depth and inversion potential  temperature
gradient at sunrise, as well  as the rate of potential temperature  increase  at
the inversion top during the period of inversion breakup.  The user should
specify these parameters on the basis of observations of temperature  inversion
development within the valley of interest.  These observations are  critical  to
the model  results and should be based on vertical soundings taken  from  the
valley floor, if available.  The modeler should be cautioned against  using
observations taken from surface-based instruments, unless these have  been
shown to be clearly related to vertical profiles over the valley center.  A
qualified analyst may be able to determine the inversion depth from acoustic
sounder records.  The modeler must use his judgment  in specifying  a single
number for the potential temperature gradient at sunrise.  Actual  soundings
are often complicated, especially in the lower levels.  The potential
temperature gradient obtained by subtracting the surface value from the value
at the inversion top and dividing by the inversion depth may not be a
representative number for the inversion as a whole when a shallow  and very
strong inversion layer is present near the ground.   The temperature gradient
is used in the model primarily to calculate the energy deficit within the
valley represented by the presence of the temperature inversion.  The total
energy deficit is obtained under an assumption of cross valley homogeneity  by
weighting the temperature deficits at various levels by the cross-sectional
area of the valley.  Since the valley is much wider  at its top than near the
valley floor, errors made in estimating temperature  deficits (i.e.,  deter-
mining a representative potential temperature gradient) at the upper  levels
are more critical to model results than at the lower levels.  A better  esti-
mate of the valley energy deficit will  be obtained by ignoring shallow, but
very strong, temperature inversion layers present on the valley floor if this
means that the potential temperature gradient estimate is more accurate for
upper levels.

     In lieu of observations one may use the observations of Whiteman [9] to
obtain first estimates of inversion depths, strengths, and inversion  top
warming rates in Colorado valleys.

     When the model is used to estimate worst-case air pollution concentra-
tions, one should naturally attempt to arrive at worst-case estimates of
inversion characteristics parameters.

Gaussian Plume Parameters

     The model user must specify Gaussian plume parameters including  the dis-
tance to the cross section of interest, the stack distance offset  from  the
valley floor centerline, the plume-carrying wind speed, the source  emission
rate, and the atmospheric stability.
                                     58

-------
     The down-valley distance is determined from a topographic map  by
measuring the distance from the source to the cross valley section  along  the
valley floor center!ine following any curves or turns  in the valley's  course.

     The stack distance offset is determined as the shortest distance  from  the
valley floor center!ine to the pollutant source on the valley floor.   The sign
of the distance follows from the orientation of the valley coordinate  system,
as discussed in Section 3.

     The wind speeds used in VALMET are critical parameters that must  be
carefully specified from available data.  Down-valley wind speeds in Colorado
valleys have been shown to vary significantly from valley to valley [9],  so
that wind speeds in one valley are not necessarily indicative of wind  speeds
in another valley.  In short, observations are necessary for the valley being
modeled.  Furthermore, the wind speeds vary on a cross valley section  due to
vertical and horizontal position within the cross section [42].  On a  vertical
profile through the nocturnal temperature inversion, the peak winds are often
located about midway through the inversion depth with  lower wind speeds near
the ground and at the inversion top.  Near-surface wind speeds, therefore,  are
not generally indicative of wind speeds at a typical plume carrying level.
Near-surface wind direction observations, in fact, may be 180 degrees  in  error
shortly after sunrise when wind continues to flow down-valley in the elevated
stable core while the winds reverse to up-valley in the C8L which develops
over the valley floor.  Near-surface wind observations should, therefore, not
be used to drive the VALMET model.

     Two wind speeds are required inputs for the VALMET model.  Both are
required in order to allow for plume dilution due to clean air inflow  between
the source and the cross section where air pollution calculations are  made.
The user must specify the average nocturnal wind speed at plume-carrying  level
at both the source and the receptor characteristic of  the nocturnal down-
valley flow.  They may be obtained from a doppler acoustic sounder  or  from
successive tethersonde or pilot balloon ascents.

     The source emission rate may be obtained from engineering calculations
for the stack of interest.  "Worst-case" pollutant calculations should be made
on the basis of worst-case emission rates.  The model can not handle emission
rates that vary in time, so a realistic time average emission rate  is  most
appropriate for driving the model.

     The model uses Pasquill-Gifford diffusion coefficients [26,27,43] to
characterize diffusion in the nighttime, high-stability, elevated stable  core
of the valley temperature inversion.  The stability class in this stable
region should be D, E, or F following the original definitions of atmospheric
stability given in Table 3.  The user-input stability class is automatically
modified internally within VALMET to account for enhanced diffusion in complex
terrain.
                                     59

-------
Sensible Heat Flux

     The final two input parameters in the VALMET model are the two  sensible
heat flux parameters.  The parameters, which are both fractions between 0 and
1, are based on the valley surface energy budget.  Little research has yet
been done on valley energy budgets and, since the energy budget is known to
strongly influence the local diurnal meteorology of valleys, the  initiation of
such research should be considered an important priority in future studies.

     The first sensible heat flux parameter is A0, the fraction of extra-
terrestrial solar heat flux converted to sensible heat flux within the
valley.  The conversion, of course, occurs at the valley surfaces (valley
floor and sidewalls), but the bulk thermodynamic model used as the basis of
VALMET cannot distinguish between energy budgets over the different  sur-
faces.  The fraction required in the model, then, is the fraction of the
(extraterrestrial) solar heat flux coming downward across the upper  surface of
the temperature inversion which is converted to sensible heat flux within the
valley.  This fraction is assumed to be constant during the temperature
inversion breakup period.  AQ depends on the transmissivity of the earth's
atmosphere to the solar beam, the albedo of the valley surfaces and  vegetative
cover, and the presence of clouds, as well as the partitioning of the solar
beam upon striking the valley surfaces.  The long wave radiative  balance also
plays a role at these surfaces which affects sensible heat flux.  Whiteman [9]
has analyzed the individual terms in the surface energy budget to determine
that the maximum value of AQ is unlikely to be greater than 0.6.  A  valley
having a low reflectivity would approach this value under clear skies only if
the valley had no soil  moisture.  Whiteman and McKee [3] were able to simulate
temperature inversion breakup well in a dry Colorado valley in October using
an AQ value of 0.45.  Maximum values of heat flux observed in the 1968 Kansas
experiments [44], if assumed to be representative of solar noon values, corre-
spond to AQ values of about 0.25.  The value of A  may be smaller in winter
when snow cover is present in a valley and the albedo is large.   The overall;
albedo in winter may depend strongly on the presence of forest cover within
the valley or the relative proportions of other lower albedo surfaces.
Whiteman and McKee [3] were able to simulate inversion breakup in Colorado's
snow-covered Yampa Valley using AQ = 0.19.

     The modeler may estimate a value of AQ from the discussion in the para-
graph above or, if sufficient meteorological data are available for  the valley
in question, he may base his estimate on one of two alternate quantitative
methods. The first method is described below.  Then the second, and  more com-
prehensive method, which also provides a means for estimating the second
sensible heat flux parameter, k, is described.  The first method  of  determin-
ing A  involves the solution of Whiteman and McKee's [3] Equation (21), Equa-
tion (28), or a combination of the two for AQ knowing, for a particular day::

 a)  the sunrise temperature inversion potential temperature gradient,
 b)  the inversion depth at sunrise,
 c)  the valley floor width and sidewall inclination angles,
 d)  the solar parameters on the day of interest, and
 e)  the observed time of temperature inversion breakup.

                                     60

-------
     By using such meteorological information  for  a  variety  of  days  in  dif-
ferent seasons, the modeler may determine a climatology  of A0 values  that
would be useful in air pollution modeling.  For the  convenience  of the  reader,
a reprint of the Whiteman and McKee article is included  in this  report  as
Appendix C.

     The value of the second sensible heat flux parameter, k, is  the  fraction
(0-1) of the sensible heat flux used to cause  CBL  growth.  The  remaining frac-
tion (1-k) represents the fraction of sensible heat  flux used to  transport
mass up the valley sidewalls causing the inversion top to sink.   The  value of
k is 1 for flat terrain since temperature inversions  are destroyed there
entirely by CBL growth, but approaches zero in snow-covered  valleys where CBL
growth is arrested after reaching a certain height [9],  Whiteman and McKee
[3] were able to simulate inversion breakup in a dry  Colorado valley  using k =
0.14.  Inversion breakup in the snow-covered Yampa Valley was simulated well
using k = 0.

     The search for a mathematical relationship between  k and parameters that
are more easily measured is a high priority for future work.  The value of k
is expected to depend strongly on valley width and,  perhaps, on  sensible heat
flux itself.  Given the present lack of an appropriate mathematical  relation-
ship, the best approach for determining k is to simply determine  the  value of
k which best fits observations of inversion breakup  in the particular valley
being modeled.  The means by which this can be accomplished  was  described by
Whiteman and McKee [3].  They described how the value of AQ  determines  the
temperature inversion breakup time, while the  value  of k determines  the height
at which the CBL and inversion top meet to cause inversion breakup,  and have
presented plots showing the effect of varying  k and  AQ on temperature inver-
sion breakup.  Both the time and height can be observed  using an  acoustic
sounder or multiple serial tethersonde ascents in  the valley of  interest.
Observed time variations of CBL depth and inversion  top  height  can be fit with
a trial and error procedure using Whiteman and McKee's [3] Equations  (29) and
(30).  The reader should refer, again, to Appendix C.  These equations  are
programmed in VALMET module "OESCNT" where they may  be solved using  a
numerical method.  The modeler who wishes to use the  procedure  should convert
a copy of subroutine DESCNT to a main program  where  the  subroutine arguments
are explicitly specified.  The number of time  steps  to inversion  destruction,
and the CBL and inversion top heights as a function  of time  would be  required
outputs of such a program.  The effect of changing AQ and k  could then  be
easily investigated and fit to actual observations.   A time  step  of
600 seconds would be sufficient for the computations  desired.   The numerical
method requires a CBL height and inversion depth at  sunrise  as  mathematical
initial conditions.  The valley meteorologial  data necessary to  use  this
approach to determine k and AQ include:

 a)  inversion depth at sunrise,
 b)  inversion potential temperature gradient  at sunrise,
 c)  valley floor width and sidewall angles,
 d)  solar model parameters A]_ and T,
                                     61

-------
 e)  the time of inversion destruction, and
 f)  the height at which the rising C8L and descending inversion top meet
     at the time of temperature inversion destruction.

Solar parameters can be determined from model equations provided earlier in
this report.  Topographic parameters can be determined from a topographic
map.  The potential temperature gradient at sunrise requires a vertical tem-
perature sounding.  The other required parameters can also be obtained  from
such soundings, if they are made frequently through the valley depth during
the temperature inversion breakup period.  They may also be determined  from
monostatic acoustic sounder records by a qualified analyst.
ERROR MESSAGES

     The VALMET model has several built-in checking and correction  routines
which will  produce error messages when the model input parameters have  been
improperly specified by the user, or when the user attempts to use  the  model
in a way unintended by the model developers.  These messages are written  on
the screen of the user's interactive terminal.  The error messages  and  their
explanations are given below.

 1.  PARAMETER NO. nn OUT OF RANGE, PLEASE RESPECIFY

     This error message is written from Subroutine INPUT when the user  has
tried to change the default value of an input parameter (number nn=l  to 27),
but has selected a new value that is outside normal ranges of that  parameter.
The user should check that he has used the right units for the input  specifi-
cation.  Execution of the program is not terminated by this error.  The user
may respecify the input parameters as many times as he wishes before  executing
the VALMET calculations.

 2.  STACK MUST BE ON THE VALLEY FLOOR - RESPECIFY YZERO

     This error message is written from Subroutine INPUT when the user
specifies a stack location that is not on the valley floor.  VALMET was not
designed to handle this situation, so the user must respecify the stack
location.  To do this, respecify YZERO to be smaller in absolute value  than
half the valley width W.  Execution of the program is not terminated  by this
error.

 3.  DOWNVALLEY WINDSPEED AT STACK HAS BEEN SET TO 1 M/S

     This error message is written from Subroutine INPUT when the user  has
specified a very low (0-1 m/s) plume-carrying wind speed at stack height. The
nighttime VALMET simulation has, as its basis, the Gaussian plume formulation,
which fails to predict concentrations realistically as winds become calm
(us-K)).  To obviate this difficulty, VALMET automatically respecifies the wind
speed to 1 m/s and continues execution.  Mile the basis for this modification
to the wind speed is primarily mathematical, one must recognize that  perfectly
calm conditions are unusual in the valley atmosphere due to the generation of

                                     62

-------
the prevalent locally developed circulations.  The frequency of occurrence  of
calm conditions has been overestimated in the past, and is still being over-
estimated by the performance characteristics of available mechanical wind
sensors.

 4.  TOO MANY GRID ELEMENTS — PLEASE REDUCE NBOX OR RE-DIMENSION THE MODEL
     TO ALLOW MORE GRID ELEMENTS

     This error message is written from Subroutine INPUT when the input
parameters specified by the user result in the model  requiring more grid
elements than are currently dimensioned in the model  (i.e., 30).  The total
number of grid elements, NTOTI, may not exceed 30 unless the user increases
the dimensions in the model by changing the PARAMETER statements so that NB  is
the new number desired (don't forget to also change NB1, since it should be
equal  to NB+1).  If your computer memory is limited, you must decrease NBOX
and live with the poorer spatial resolution which results, as documented at
other places in this report.  Execution of the model  is not terminated by this
error.  If you reduce NBOX the program will continue to execute, but if you
decide to increase the model dimensions you must stop execution and make some
basic changes in the VALMET code.

 5.  YOU HAVE SPECIFIED A WRONG STABILITY CLASS - F WILL BE USED INSTEAD

     This error message is written from Subroutine INPUT when the user enters
a stability class other than D, E, or F.  «VALMET automatically uses F
stability when this occurs.  Execution of the program is not terminated by
this error.

 6.  THE PLUME, AFTER PLUMERISE, IS NOT WITHIN THE STABLE CORE

     This error message is written from the main program.  If the plume rises
above the stable core, or never exceeds the initial CBL height, it is not
transported down the valley in stable nighttime conditions and it makes no
physical sense to run VALMET under these conditions.  Model execution is
terminated by this error.
                                     63

-------
                                   SECTION 5

                  TECHNICAL  DESCRIPTION  OF INDIVIDUAL MODULES
VALMET-Main Program

     VALMET is the main program that controls the modules or  subroutines
which, taken-together, form the valley air quality model.  The  function of  the
main program is to provide the basic structure of the air quality model and to
call the specific modules as required.  In addition, several  computing house-
keeping functions are performed in the main program including the establish-
ment of PARAMETER statements and COMMON blocks, and the opening and  closing of
output files.  Model outputs are written from the main program, as well.

Inputs.

     Inputs to the main program come through Subroutine INPUT,  a subroutine
that is run interactively by the user and which features a default input
table.  Data coming into the main program from subroutines usually come
through arguments listed in the subroutine call statement.  Large arrays  of
data, however, are frequently transferred from the main program to the sub-
routines (and vice versa) through COMMON blocks.

Outputs

     The results of a model run are written to two output disk  files.  The
first output file, named VALMET.OUT, is a formatted file which  contains the
results of the model run.  After the model is run, the contents of this file
may be printed to obtain a summary record of the model run, including model
input parameters and outputs.  A second file, named VALMET.PLT, is also auto-
matically generated with every computer run.  This formatted  file contains
time series of pollutant concentrations, temperature inversion  depths and con-
vective boundary layer depths, as well as many of the basic parameters used in
the model run.  The file is created for the model user who wishes to develop
his own plotting programs to plot the results of a model run.   An example pro-
gram that the authors used to generate the plots used in this report is shown
in Appendix B.  Since every user's computer installation will have different
plotting software and hardware, this approach of creating a basic data file
but not including a specific plotting program in the air pollution model
seemed the most appropriate way to proceed.  Interested users can develop
their own specific plotting software using our example.
                                     64

-------
Conventions
     The following indexing conventions are used in VALMET where it has proven
convenient:
       I=grid element index

Variable Name Definitions

INPUT/  VARIABLE
OUTPUT    NAME
                                  N or MN=time step index
                                   DEFINITION
        AO

        Al
        A2
        AC

        ALPHA1
        ALPHA2
        AS

        BETA

        BOXLEN
        CCHI

        CHIOFF
-
-
I
-
I
I
-
I
-
-
-
I
-
CLCONC
DELTAT
DLH
DF
GAMMA
H
HCBLI
HZERO
I
11
IAVG
IDA
IFIX2
        I NO
        ISTAB
        ISTABY
        ISTABZ
        IT
        ITIME
                Fraction of extraterrestrial solar flux converted to
                sensible heat flux (0-1)
                Solar flux on horizontal surface at solar noon (Wm  )
                Same as AO
                Valley cross-sectional area below inversion top
                at cross-sectional distance XC (m )
                Sidewall #1 elevation angle (rad)
                Sidewall #2 elevation angle (rad)
                Valley cross-sectional area below inversion top
                at pollution source cross section (m )
                Rate of increase of potential temperature at
                inversion top (°K/s)
                Length of grid elements (m)
                Nocturnal pollutant concentration in a grid element
                Concentration offset value due to reflection off
                valley floor and sidewalls (ng/m )
                Center!ine concentration (ng/m )
                Time step size (s)
                Plume rise (m)
                Inverse of dilution factor
                Vertical potential temperature gradient  (°K/m)
                Plume center!ine height (m)
                Initial CBL height (m)
                Initial depth of inversion (m)
                Grid element index
                Intermediate variable
                Averaging period (s)
                Day of month (1-31)
                Number of grid elements on sidewall at given time
                step
                Loop index
                Stability index (1-6)
                Stability index for determination of a   (1-6)
                Stability index for determination of o^  (1-6)
                Intermediate value used in calculating time
                Time unit=LST
                                     65

-------
IYR
JULDAY
K

LAT
MN
MO
N
HAS
NB

MB1
NBOX

ND
NINDX

MS

NSTEPS

NTOT

NTOTI

NTSA
NTSR
NY
PRESS

PSTD
Q
RHO
SIGMAY
SIGMAZ
SRAD
STMP
STNDCOM
SUM1

SUM2

SVEL
Tl
T2
T3
T4
TAU
TEMP
TMAXC1
averaging intervals in NINDX
giving the maximum number of model grid
for use in dimensioning arrays
Year (00-99)
Julian date (1-366)
Fraction of sensible heat flux used to grow CBL
(0-1)
Latitude (°M)
N-l
Month (1-12)
Time step counter
Number of
Parameter
elements,
NB+1
Number of grid elements on the valley floor
half-width
Loop index
Number of time steps in simulation, including
exponential decay after breakup
Parameter giving the maximum number of model time
steps, for use in dimensioning arrays
Number of time steps required to destroy initial
temperature inversion
Total number of grid elements left in the model at
a given time step
Total number of grid elements in the model at
initiation
Number of time steps in averaging interval
Time of sunrise (LST)
Intermediate variable used to count grid elements
Atmospheric pressure at center of temperature
inversion (mb)
Standard pressure (1013 mb)
Stack emission rate (kg/s)
Air density (kg/m )
Sigma y (m)
Sigma z (m)
Stack radius (m)
Stack temperature (°K)
Factor to adjust concentration to standard conditions
Fraction of plume mass within valley cross section
(0-1)
Fraction of plume mass above valley cross section
(0-1)
Stack exit velocity (m/s)
Intermediate variable used to
Intermediate variable used to
Intermediate variable used to
Intermediate variable used to
Length of daylight period (s)
Air temperature (°K)
Ending time of the 1-hr average
(h LST)
calculate
calculate
calculate
calculate
                              time
                              time
                              time
                              time
                      concentration max
                             66

-------
        TMAXC2  Ending time of the 3-hr average concentration max
                (h LSI)
        TSTD    Standard temperature (293.16°K)
        TSTP    Time in averaging period (s)
  I     UC      Mean down-valley plume transport velocity at cross
                section distance XC (m/s)
  I     URLONG  Longitude (°W)
  I     US      Mean down-valley plume
                pollutant source cross
  I     W       Valley floor width (m)
  I     XC      Distance from stack to model cross section (km)
        Y       Y-coordinate (m)
  I     YZERO   Stack offset from valley floor centerline (m)
        Z       Z-coordinate (m)
                  transport velocity at
                  section (m/s)
Array Name Definitions

INPUT/  ARRAY
OUTPUT  NAME
               DEFINITION
        AVG(NB,2)

        CCH(2,NB)

        CHI(NB)

        CHIBAR(NB)
        CONC(NS,NB)
        HC(NS)
        HITEL(NB)

        HITEU

        HT(NS)
        NDXTIM(NB,2)
        NTS(2)
        SUM(NB)

        TIME(NS)

        V(NB1)
Maximum 1-hr and 3-hr average concentration array
Ug/m3)
Short-term storage location for concentrations
at two adjacent time steps d-ig/m )
Nocturnal  steady state concentration array
                        array (p-g/m )

                        grid element on sidewall

                        grid element on sidewall
Average pollutant concentration injected into
the top of grid elements at each time step
Ug/m3)
Pollution concentration
Height of CBL top (m)
Height of lower side of
(m)
Height of upper side of
(m)
Height of inversion top (m)
Index number of TMAXC1 and TMAXC2
Number of time steps in 1-hr and 3-hr
Summing array to accumulate pollutant
concentrations during averaging interval
Midpoint time array for averaging intervals
(h MST)
Slope flow velocity array (m/s)
Subroutines Called

   BRKUP
   DESCNT
   DILUTE
                                     67

-------
   EBDGT
   GAUSS
   INGRAT
   INPUT
   JULIAN
   PRISE
   PROFIL
   PSTPRC
   SOLAR
   VELOCY

Common Blocks

   BLK1
   BLK2   Not used in main program
   BLK3   Not used in main program
   BLK4
   BLK5
   BLK6
                                     68

-------
INPUT-Input Module
Purpose

     The VALMET program runs in an interactive mode in which the user  controls
program execution from a remote interactive terminal.  The user enters model
inputs from the remote terminal by following directions given to the user  on
the terminal screen by subroutine INPUT.  Once the input values are specified
by the user he can direct the program to begin the air pollution simulation.
The functions of the INPUT module are to obtain the proper input data  from the
model user, to convert the input data to the proper units for processing in
later modules of VALMET, to check the input data for errors, and to notify the
user of any errors.

Inputs

     Inputs to this module come entirely through user input from an inter-
active terminal, unless default values of the input parameters are selected by
the user.

Outputs

     Outputs from the module go to the main program through the subroutine
argument list.
Variable Name Definitions

INPUT/ VARIABLE
OUTPUT NAME
                                 DEFINITIONS
I
0
I

I
I
I
I
0
0
I
AO

AC
ALPHA1
ALPHA2
AS
BETA

BOXLEN
DELTAT
GAMMA

H
HCBLI
HZERO
IDA
IFIX
ISTAB
IYR
J
J6
               Fraction of solar flux converted to sensible heat
               flux (0-1)
               Valley cross-sectional area at distance XC  (m )
               Sidewall #1 elevation angle (°)
               Sidewall #2 elevation angle (°)
               Cross-sectional area at source (m )
               Rate of potential temperature increase above valley
                                          temperature gradient
               Grid element length (m)
               Time step (s)
               Inversion vertical potential
               Plume centerline height (m)
               CBL height at sunrise (m)
               Initial inversion depth (m)
               Day of month (1-31)
               Number of grid elements on sidewall
               Atmospheric stability (1-6)
               Year (00-99)
               Index
               Index
                                     69

-------
       J12     Index
       J13     Index
       J14     Index
  I    K       Fraction of sensible heat flux used to grow C8L  (0-1)
  I    LAT     Latitude (°N)
       LU      Logical unit number
  I    MO      Month (1-12)
  I    NBOX    Number of grid elements on valley floor half width
       NTOTI   Total number of grid elements
       NU      Number of input parameters to be changed from default
               values
  I    PRESS   Air Pressure (mb)
  I    Q       Stack emission rate (kg/s)
       RHO     Air density (kg/nv3)
       RNO     NBOX
  I    SRAD    Stack radius (m)
  I    STMP    Stack emission temperature (°C)
  I    SVEL    Stack emission velocity (m/s)
  I    TEMP    Air temperature (°C)
  I    UC      Plume-carrying wind speed at cross section (m/s)
  I    URLONG  Longitude (°W)
  I    US      Plume-carrying wind speed at source (m/s)
  I    w       Valley floor width (m)
  I    XC      Down-valley distance to cross section from stack  (m)
  I    YZERO   Stack offset distance from valley floor centerline
               (m)
Array Name Definitions
INPUT/
OUTPUT
_
-
-

-
-
-
-
-
ARRAY NAME
ASTAB(6)
CAT(IO)
10(27)

NAM(27)
VAL(27)
VALLO (27)
VALHI (27)
VALNEW(27)
DEFINITIONS
Atmospheric stability array
Input parameter category array
Array of identification numbers for
parameters
Input parameter name array
Default input value array
Minimum values of input parameters
Maximum values of input parameters
New values of input parameters



input






Subroutine Called

   None

Common Blocks

   None
                                     70

-------
JULIAN-Julian Day Module

Purpose

     JULIAN calculates the Julian date, given the month, day and year of the
air pollution simulation.

Inputs

     The month, day, and year are input to the subroutine from the main pro-
gram through the subroutine argument list.

Outputs

     The Julian date is sent to the main program through the subroutine
argument list.

Variable Name Definitions
INPUT/  VARIABLE
OUTPUT  NAME
    DEFINITION
  I     IDA
  I     IYR
  0     JULDAY
  I     MO
Leap year indicator
Day of month (1-31)
Year (00-99)
Julian date (1-366)
Month (1-12)
Array Name Definitions

INPUT/
OUTPUT  ARRAY NAME
             DEFINITION
        NDAY(12)
  Cumulative days of year, by month
Subroutines Called

   None

Common Blocks

   None
                                     71

-------
PRISE-Plume Rise Module

Purpose

     PRISE calculates plume rise using the Briggs plume  rise algorithms  docu-
mented in the MPTER User's Manual, taking account of both plume momentum and
buoyancy effects.

Inputs

    The inputs to PRISE come through the subroutine argument list  and  include
stack characteristics (radius, exit velocity, exit temperature), ambient atmo-
spheric characteristics (air temperature, wind speed, stability class, poten-
tial temperature gradient) and down-valley distance to the cross section of
interest.
Outputs

     The output of PRISE is the plume centerline rise above the physical  stack
height.  This value is passed to the main program through the subroutine
argument list.
FORTRAN Library Subroutines

     Name
              Function
    AMINl(xl,x2,...xn)
Determine the minimum value of a list
of real  numbers
Variable Name Definitions

INPUT/  VARIABLE
OUTPUT  NAME
     DEFINITION
        DELTT    Plume temperature excess (°K)
  0     DLH      Plume centerline rise above stack height (m)
        DLH1     Intermediate plume rise variable
        DLH2     Intermediate plume rise variable
        DTC      Crossover temperature difference (°K)
        F        Buoyancy flux (ms  )            „
        G        Acceleration due to gravity (ms  )
  I     GAMMA    Potential temperature gradient (°K/m)
  I     ISTAB    Vertical atmospheric stability index (3-6)
        S        Stability parameter (s~ )
  I     SRAD     Stack radius (m)
  I     STMP     Stack gas exit temperature (°K)
  I     SVEL     Stack gas exit velocity (m/s)
  I     TEMP     Ambient air temperature (°K)
        UCUT     Critical wind speed (m/s)
  I     US       Ambient plume-carrying wind speed (m/s)
        X        Down-valley distance (m)
                                     72

-------
  I     XC       Down-valley distance  (km)
        XF       Distance to final plume rise (m)
        XX       Intermediate distance variable

Array Name Definitions

   None

Subroutines Called

   None

Common Blocks

   None

Special Note

The user might wish to specify an "effective stack height" directly,  rather
than specifying a physical stack height and allowing this subroutine  to
calculate the additive plume rise.  This can be done by inputting the effec-
tive stack height in place of the physical stack height, and inputting a zero
value for either stack radius or stack gas exit velocity.  PRISE returns a
zero value for plume rise when a zero  value is input for either of these
parameters.
                                     73

-------
DILUTE-Plume Dilution Module

Purpose

     DILUTE calculates the dilution factor necessary to account  for  clean  air
dilution of the nocturnal pollutant plume during its transport down  the  valley
due to clean air inflow from valley tributaries, slope flows, or entrainment
at the top of the down-valley flow layer.

Inputs

     The inputs to DILUTE come from the main program through the subroutine
argument list.  They include the cross-sectional areas and wind  speeds at  the
pollutant source and the receptor cro'ss sections.

Outputs

     The output from DILUTE is the dilution factor calculated from the ratio
of volume fluxes at the two cross sections.  It is passed to the main program
through the subroutine argument list.

Variable Name Definitions

INPUT/  VARIABLE
OUTPUT  NAME	DEFINITION	
                                                            O
  I     AC     Area of valley cross section at distance x (nr)
  I     AS     Area of valley cross section at pollutant source  (m2)
  0     OF     Inverse of dilution factor
  I     UC     Wind speed at cross section at distance x (m/s)
  I     US     Wind speed at pollutant source cross section (m/s)


Array Name Definitions

   None

Subroutines Called

   None

Common Blocks

   None
                                     74

-------
INGRAT-Valley Plume Reflection Module

Purpose

     The purpose of Subroutine INGRAT is to calculate Gaussian diffusion
coefficients and to accomplish cross wind integrations of plume mass on a
valley cross section in order to determine the amount of plume mass within the
valley cross section and the amount of plume mass which has escaped out the
top of the valley inversion.  These mass calculations are necessary to simu-
late valley plume channeling so that nocturnal plume reflection off the valley
floor and sidewalls can be handled in the main program.

Inputs

     Inputs to Subroutine INGRAT come into the subroutine through the sub-
routine argument list.  Outputs include the fraction of plume mass within the
valley inversion, the fraction which has diffused out the top of the inver-
sion, and the two dispersion coefficients, sigma y and sigma z.

Variable Name Definitions
INPUT/  VARIABLE
OUTPUT  NAME
DEFINITION
  I     ALPHA1   Sidewall #1 elevation angle (rad)
  I     ALPHA2   Sidewall #2 elevation angle (rad)
        F        Analytical  formula for Gaussian distribution
  I     H        Plume center!ine height (m)
  I     HZERO    Initial inversion height (m)
        I        Height increment counter
  I     ISTABY   Horizontal  atmospheric stability index (1-6)
  I     ISTABZ   Vertical atmospheric stability index (1-6)
        NP       Number of height increments to inversion top
        P        Number of height increments to inversion top
        PHIY     Area under Gaussian curve from yl to y2
        PHIY1    Area under Gaussian curve from minus infinity to yl
        PHIY2    Area under Gaussian curve from minus infinity to y2
        PI       Trigonometric constant
        R        Height increment counter
  0     SIGMAY   Standard deviation of plume concentration in y-direction (m)
  0     SIGMAZ   Standard deviation of plume concentration in z-direction (m)
  0     SUM1     Fraction of pollution within valley inversion (0-1)
  0     SUM2     Fraction of pollution diffusing out inversion top  (0-1)
  I     W        Width of valley floor (m)
  I     XC       Down-valley distance from stack to cross section (km)
  I     Yl       Y-coordinate of sidewall #1 at height z (m)
        Y2       Y-coordinate of sidewall #2 at height z (m)
  I     YZERO    Off-centerline displacement of stack (m)
        I        Vertical coordinate (m)
        ZINC     Height increment for integration (m)
                                     75

-------
Array Name Definitions

INPUT/
OUTPUT  ARRAY NAME	DEFINITION	

        SIGGY(18)   McMullen's coefficients for P-G a
        SIGGZ(18)   McMullen's coefficients for P-G a
                                                     z
Subroutines Called

   NORMAL

Common Blocks

   None
                                     76

-------
NORMAL-Normal or Gaussian Curve Integration Module

Purpose

     Subroutine NORMAL is used to calculate the area under a Gaussian or
Normal distribution curve from minus infinity to X standard deviations.  The
polynomial approximation technique used in this subroutine comes from
Abramowitz and Stegun [33, p. 932] following a formulation attributed to
Hastings [34].

Inputs

     The sole input is the standard deviation X where X is between 0 and
positive infinity.  This input comes into the subroutine through the
subroutine argument list.

Outputs

     The sole output is the area under the Gaussian curve, PHI.  This output
is carried back to the calling program (Subroutine INGRAT) through the
argument list.

Variable Name Definitions

INPUT   VARIABLE
OUTPUT  NAME         DEFINITION
0
-
-
-
I
PHI
PI
SIGH
T
X
Area under curve (0-1)
Trigonometric constant
Intermediate variable
Intermediate variable
Standard deviation
Array Name Definitions

INPUT/  VARIABLE
OUTPUT  NAME	DEFINITION	

        C(6)     Coefficients of polynomial


Subroutines Called

   None

Common Blocks

   None
                                     77

-------
GAUSS-Gaussian Plume Module

Purpose
     This subroutine uses a Gaussian plume algorithm to calculate pollutant
concentration at an arbitrary receptor location P(x,y,z).  The present version
uses Pasquill-Gifford dispersion coefficients, sigma y and sigma z, calculated
using McMullen's [28] method.

Inputs

     The subroutine requires information on the coordinates of the receptor
and source, the plume centerline height, source strength, wind velocity, dis-
persion coefficients, atmospheric temperature, and atmospheric pressure.  The
inputs come into the subroutine through the subroutine argument list.

Outputs
         /
     The subroutine output is the pollutant concentration at the receptor in
micrograms per cubic meter.  The output is sent to the calling program (MAIN
or PROFIL) through the subroutine argument list.

Variable Name Definitions

INPUT/  VARIABLE
OUTPUT   NAME                 DEFINITION
0
-
I
-
-
I
I
I
I
I
I
I
I
CHI
CHIOQ
OF
Gl
G4
H
Q
SIGMAY
SIGMAZ
US
Y
YZERO
Z
Pollutant concentration (ng/m )
Concentration/source strength (ng s m" kg" )
Inverse of plume dilution factor
Of f-centerl ine term in y direction
Off-centerline term in z direction
Plume centerline height (m)
Source strength (kg/s)
Sigma y (m)
Sigma z (m)
Wind speed (m/s)
Receptor coordinate (m)
Stack offset distance from valley floor centerline
Receptor coordinate (m)
                                                                    (m)
Array Name Definitions

   None

Subroutines Called

   None

Common Blocks
   None
                                     78

-------
SOLAR-Extraterrestrial Solar Radiation Module
Purpose

     The purpose of Subroutine SOLAR is to calculate the time of sunrise,
length of day, and extraterrestrial solar flux on a horizontal surface at
solar noon given the day of year and the latitude and longitude of the site.
The outputs of SOLAR are necessary to drive the daytime portion of VALMET.

Inputs

     Necessary subroutine inputs include the latitude and longitude of the
site and the Julian date of the simulation.  These three inputs come into
SOLAR through the subroutine argument list.

Outputs

     Outputs from Subroutine SOLAR include the local standard time of sunrise,
the length of the daytime period, and the extraterrestrial  solar flux on a
horizontal surface at solar noon.

Variable Name Definitions
INPUT/ VARIABLE
OUTPUT NAME
0 Al Solar
DEFINITION
flux on horizontal surface

at solar noon
         Conversion factor-degrees to radians (rad/deg)
         Cosine of zenith angle
         Julian date (1-366)
         Date interpolation variable for equation of time
         Declination (rad)
         Maximum declination (rad)
         Julian date of vernal  equinox
         Eccentricity of earth's orbit
         Date interpolation variable for equation of time
         Time unit conversion variable
         Julian date (1-365)
         Latitude (°N)
         Longitude correction (h)
         Longitude of earth in its orbit around sun (rad)
         Time of sunrise (hhmm)
         Earth's position in orbit around sun on date of
         interest (rad)
OMDZRO   Earth's position in orbit around sun at date of
         vernal  equinox (rad)
OMEGA    Revolution rate of earth (rad/day)
ONEHR    Conversion factor (rad/h)
PI       Trigonometric constant
       CONV
       COSZ
       D
       02
       DECLIN
       DECMAX
       DZERO
       ECCENT
       ID
       IT
       JULDAY
       LAT
       LONCOR
       LONG
       NTSR
       OMD
                                     79

-------
       RDVCSQ   Earth-sun distance factor
       SC       Solar constant (W/m2)
       SR       Hour angle of sunrise (rad)
       STOLON   Longitude of standard meridian of time zone  (°W)
       Tl       Time unit conversion variable
       T2       Time unit conversion variable
       TAU      Length of daylight period  (h)
       TIMCOR   Correction due to equation of time (h)
       TMNOON   Local standard time of solar noon (h)
       URLONG   Longitude of site (°W)
Array Name Definitions

INPUT/
OUTPUT  ARRAY NAME	     DEFINITION	

        EQNTIM(25)    Equation of time correction at 15-day intervals
                      (h)
Subroutines Called

   None

Common Blocks

   None
                                     80

-------
EBDGT-Surface Energy Budget Module

Purpose

     The purpose of EBDGT is to determine the fraction of extraterrestrial
solar heat flux that is converted to sensible heat flux in the valley of
interest.  It is this sensible heat flux that drives the post-sunrise inver-
sion breakup that leads to air pollution fumigations on the valley floor and
sidewalls.  The fraction depends on factors associated with the surface energy
budget, including surface albedo, soil moisture, surface cover, cloud cover,
atmospheric transmissivity, and long-wave radiative transfer within the val-
ley.  The present version of the subroutine is not fully developed and does
not yet explicitly include these factors.  The present skeletal version of the
subroutine merely specifies a fraction that the user has previously input
using guidance given in earlier sections of this report.  Further development
of this subroutine is a suggested priority for future work.

Inputs

     The fraction of extraterrestrial solar flux converted to sensible heat
flux in the valley is the sole input to the present version of EBDGT, and is
input to the subroutine through the subroutine argument list.

Outputs

     The present version of EBDGT simply outputs to the main program the user-
specified input value.  The output is transferred to the main program through
the subroutine argument list.

Variable Name Definitions

INPUT/   VARIABLE
OUTPUT     NAME	DEFINITIONS	

  I     AO       Fraction of extraterrestrial solar flux converted
                 to sensible heat flux (0-1)
  0     A2       Same as AO (0-1)


Array Name Definitions

   None

Subroutines Called

   None

Common Blocks

  None
                                     81

-------
DESCNT-Inversion Descent and CBL Growth Module

Purpose

     DESCNT uses Whiteman and McKee's [3] bulk thermodynamic model of tem-
perature inversion breakup to calculate post-sunrise changes in valley inver-
sion and CBL depths.  The method considers that the valley temperature
inversion represents an energy deficit relative to the warmer air above the
inversion.  The energy deficit is destroyed after sunrise by solar energy
input into the valley as it is converted to sensible heat flux at valley sur-
faces.  The energy is used in two ways:  (1) it is used to grow CBLs over
valley surfaces and (2) it is used to cause air to flow upslope in the valley
sidewall CBLs.  Air flowing up the sidewalls in the CBLs causes corresponding
descending motions over the valley center, resulting in a descending inversion
top after sunrise.  The module uses a numerical method for the calculations,,
and incorporates the effects of warming above the valley in retarding the
inversion breakup.

Inputs

     The inputs to DESCNT include solar flux parameters, valley characteris-
tics parameters, model time step, initial CBL and inversion top heights, an
above valley warming rate parameter, inversion potential temperature gradient,
and mean valley air temperature and pressure.  These inputs are passed to the
subroutine through the subroutine argument list.

Outputs

     The main outputs of DESCNT are CBL heights and inversion top heights as a
function of time (actually, time step).  These outputs are passed back to the
main program in arrays located in common block BLK1.  The number of time steps
required to destroy the valley inversion is also an output of the subroutine,
passed to the main program through the subroutine argument list.

Variable Name Definitions

INPUT/  VARIABLE
OUTPUT   NAME                      DEFINITIONS
I
I
I
I
I
_
-
I
-
-
-
AO
Al
ALPHA1
ALPHA2
BETA
C
CP
DELTAT
DHC
DHDT
DHT
Fraction of Al converted to sensible heat flux (0-1)
Solar flux on horizontal surface at solar noon (Wm )
Sidewall #1 elevation angle (rad)
Sidewall #2 elevation angle (rad)
Rate of temperature change at inversion top (°K/s)
Intermediate factor ' . ,
Specific heat at constant pressure (J kg" K~ )
Time step (s)
Change in CBL height (m)
Change of CBL height with time (m/s)
Change in inversion top height (m)
                                     82

-------
       DHTDT   Change in inversion top height with time (m/s)
       FACT1   Multiplicative factor
       FACT2   Multiplicative factor
       FDENOM  Intermediate variable-denominator
       FNUM    Intermediate variable-numerator
  I    GAMMA   Vertical  potential temperature gradient (°K/m)
       HCBL    Height of CBL top (m)
  I    HCBLI   Initial CBL height (m)
       HTOP    Height of inversion top (m)
  I    HZERO   Initial height of inversion top (m)
  I    K       Fraction of sensible heat flux going into CBL growth
               (0-1)
       N       Time step counter
  0    NSTEPS  Number of time steps required to destroy inversion
       PI      Trigonometric constant
  I    PRESS   Average pressure at inversion center at sunrise (mb)
  I    RHO     Air density (kg/nr)
       T       Elapsed time since sunrise (s)
  I    TAD     Length of daylight period (s)
       THTOT   Potential temperature/ambient temperature
       TI      Time of sunrise (s)
  I    W       Valley floor width (m)
Array Name Definitions

INPUT/
OUTPUT  ARRAY NAME	DEFINITIONS	

  0     HC(NS)      Array of CBL heights (m)
  0     HT(NS)      Array of inversion top heights (m)
Subroutines Called

   None

Common Blocks

   BLK1
                                     83

-------
PROFIL-Concentration Profile Module

Purpose

     The purpose of Subroutine PROFIL is to calculate the pollutant concentra-
tion injected into the top of each of the model grid elements at each time
step.

Inputs

     Subroutine PROFIL requires a number of inputs, dealing primarily with the
valley characteristics, grid element locations and specifications, and
Gaussian plume characteristics.  Subroutine PROFIL is called for each model
time step and does calculations for each of the grid elements.  All inputs
come into the subroutine through the subroutine argument list.

Outputs

     For each model time step, Subroutine PROFIL calculates the pollutant
concentration injected into the top of each of the growing grid elements.  To
maintain accuracy in the calculations, the concentration for each grid element
is calculated as the average of four concentration determinations made at the
various sides of the quadrilateral Gaussian plume element which sinks into the
grid element.

Variable Name Definitions
INPUT/  VARIABLE
OUTPUT  NAME
DEFINITIONS
_
I
I
I
I
mf
~
-
-
I
I
I
I
I
I
I
-
I
I
ALPHA
ALPHA1
ALPHA2
BOXLEN
CHIOFF
CHIY1
CHIY2
CHIZ1
CHIZ2
DF
H
HCL
HCU
HTL
HTU
HZERO
I
NBOX
NTOT
Average of ALPHA1 and ALPHA2
Sidewall #1 elevation angle (rad)
Sidewall #2 elevation angle (rad)
Grid element length (m)
Offset air pollution concentration
Concentration at P(X,Y1,Z) (ng/m;:
Concentration at P(X,Y2,Z) (p-g/m^
Concentration at P(X,Y,Z1) (ng/m;r
Concentration at P(X,Y,Z2) (M-g/m'3
Inverse of plume dilution factor
Plume centerline height (m)
CBL height at lower time step (m)
CBL height at upper time step (m)
Inversion top height at lower time
Inversion top height at upper time
Inversion depth at sunrise (m)
Grid element index
Number of grid elements on valley
Number of model grid elements at a




due to reflections
}
)
)
)




step (m)
step (m)


floor half-width
given time step
                                     84

-------
I
I
I
I
I
I
I
-
-
-
I
-
-
NTOTI
PRESS
Q
SIGMAY
SIGMAZ
TEMP
US
Y
Yl
Y2
YZERO
Z
Zl
       Z2
Initial number of model grid elements
Ambient pressure (mb)
Source strength (kg/s)
Sigma y (m)
Sigma z (m)
Ambient temperature (°K)
Down-valley wind speed at stack cross section  (m/s)
Y-coordinate of receptor (m)
Y-coordinate at left side of sinking mass element  (m)
Y-coordinate at right side of sinking mass element (m)
Y-distance of stack from valley floor centerline  (m)
Z-coordinate of receptor (m)
Effective z-coordinate at bottom of sinking mass
element (m)
Effective z-coordinate at top of sinking mass  element
(m)
Array Name Definitions

INPUT/
OUTPUT  ARRAY NAME
                       DEFINITIONS
  0     CHIBAR(NB)    Average concentration injected into top of each
                      grid element
Subroutines Called

   GAUSS

Common Blocks

   BLK2
                                     85

-------
VELOCY-Upslope Flow Velocity Module

Purpose

     Subroutine VELOCY uses the mass continuity equation for air to calculate
the slope flow velocity profile, under the assumption that the convective
boundary layer depth does not vary as a function of distance up the slope.   An
implicit assumption is that all mass lost from the valley inversion cross sec-
tion as the inversion top sinks is transported from the valley in the upslope
flows which develop in the convective boundary layers over the slopes.  The
upslope velocities are calculated at the boundaries between grid elements.

Inputs

     The inputs to this subroutine come through the subroutine argument list
and through the common block labeled BLK1.  BLK1 provides the inversion top
and CBL heights for every time step.  Inputs from the argument list include
the time step index number, the time step length, and information about the
size and location of the grid elements.

Output

     The output of VELOCY is an array of upslope wind velocities calculated  at
the NTOTI+1 boundaries between the NTOT grid elements. The subroutine is
called from the main program at each model time step, and the calculated
velocities are passed back to the main program through the common block
labeled BLK3.

Variable Name Definitions
INPUT/  VARIABLE
OUTPUT    NAME
DEFINITIONS
  I     BOXLEN    Length of grid elements (m)
        CIH       Inversion top displacement during time step (m)
  I     OELTAT    Time step (s)
        I         Grid element index
  I     MN        N-l
  I     N         Time step index
  I     NBOX      Number of grid elements on valley floor half-width
  I     NTOT      Total number of grid elements in simulation at a
                  given time step
  I     NTOTI     Number of grid elements in simulation at beginning
                  of simulation
                                     86

-------
Array Name Definitions

INPUT/
OUTPUT  ARRAY NAME         DEFINITIONS
I
I
0
HC(NS)
HT(NS)
V(NB1)
CBL height array (m)
Inversion top height array (m)
Upslope wind velocity array (m/s)
Subroutines Called

   None

Common Blocks

   BLK1
   BLK3
                                     87

-------
BRKUP-Pollutant Mass Budget Module

Purpose

     The purpose of Subroutine BRKUP is to calculate pollutant concentrations
in each of the grid elements using a pollutant mass balance.  The calculations
are made for each model time step, taking account of pollution sinking  into
the top of each grid element, CBL growth, upslope advection of pollution,
mixing within each grid element and carryover of pollution in each grid ele-
ment from the previous time step.  A currently unused feature of the pollution
mass budget in each grid element is the provision for sources and sinks of
pollution within the element.  A special feature of BRKUP is the exponential
decay of air pollution concentrations in grid elements that are dropped from
the simulation as the top of the inversion sinks below the grid element.

Inputs

     Inputs to the subroutine coming through the subroutine argument list
include time step index, number of grid elements on the valley floor half-
width, total number of grid elements at model initiation, the grid element
length, and model time step length.  Other inputs are available through
labeled common blocks BLK1, BLK2, and BLK3.  BLK1 provides information  on CBL
and inversion depths for each time step.  BLK2 provides information generated
in Subroutine PROFIL on the amount of air pollution mass sinking into the top
of each grid element at the given time step.  BLK3 provides the upslope wind
velocities used to advect pollutant mass up the valley sidewalls at the given
time step.  These velocities were generated in Subroutine VELOCY.

Outputs

     BRKUP calculates air pollution concentrations in the model grid elements
for each time step.  These calculations constitute the main output of the
VALMET model, and are output to the main program through the subroutine argu-
ment list.  The main program performs further calculations on these concentra-
tions to determine 5-min-average concentrations which are stored in the main
program for further processing into maximum 1- and 3-h averages.

Variable Name Definitions

INPUT/   VARIABLE
OUTPUT     NAME	DEFINITIONS	

         A        Pollutant mass within the grid element at previous
                  time step (ng)
         B        Pollutant mass coming from adjacent downhill grid
                  element (|ig)
  I      BOXLEN   Grid element length (m)
         C        Pollutant mass sinking through growing top of CBL
                  (ng)
         D        Pollutant mass transported into adjacent uphill grid
                  element (,ug)

-------
      DELTAH   Entrainment at top of grid element  (m)
  I   DELTA!   Time step (s)
      I        Grid element index
  I   MN       Index of previous time step
  I   N        Current time step index
  I   NBOX     Number of grid elements on valley half-width
  I   NTOTI    Total number of grid elements at model  initiation
      S        Exponential decay time constant (s"1)
Array Name Definitions

INPUT/
OUTPUT  ARRAY NAME	DEFINITIONS	

  0     CCH(2,N8)     Grid element concentration array  (ng/m)
        CHIBAR(NB)    Average pollutant concentration injected into
                      top of grid element  (ug/m )
        HC(MS)        CBL height array (m)
        HT(NS)        Inversion top height array (m)
        V(NB1)        Upslope velocity array (m/s)
Subroutines Called

   None

Common Blocks

   BLK1
   BLK2
   BLK3
                                     89

-------
PSTPRC-Post Processor Module

Purpose

     This subroutine is a post-processor.  It processes the time series of
calculated 5-min-average concentrations for each grid element to obtain
maximum short-term (1- and 3-hr) average concentrations which can be compared
to regulatory standards.  The maximum 1- and 3-hr-average concentrations are
determined by means of a moving average of the 5-min-average concentration
time series, determining maximum values for each of the grid elements.  A time
index value is saved for each of the maximum values so that the time of
occurrence of the maximum can be calculated in the main program.  The
subroutine is written so that the averaging intervals can be easily changed,
if necessary.

Procedure

     The moving average is calculated by stepping backwards through the
5-min-average concentration array using a 1- and 3-hr averaging "window",
beginning at the last 5-min average calculated for each grid element.  As they
travel  backwards through the 5-min concentration array the 1- and 3-hr
averaging intervals will eventually pass the time of sunrise.  The nocturnal
steady state concentrations are added into the average at that point.

Inputs

     The inputs to PSTPRC include the series of 5-min-average concentrations
for each of the grid elements, the steady-state nocturnal concentrations in
each grid element, the averaging interval (300 s), total number of grid ele-
ments and index number of the last 5-min concentration values.  The time
series of 5-min averaged values is input through common block BLK2.  The
nocturnal concentrations are input through common block BLK.  The other
parameters are input through the subroutine argument list.

Outputs

     The outputs of Subroutine PSTPRC are passed to the main program through
common block BLK6.  They include the maximum 1- and 3-hr average concentra-
tions for each grid element, the time index numbers for each of these
concentrations, and the number of 5-min time steps in the regulatory averaging
intervals (1- and 3-hr).

Variable Name Definitions

INPUT/  VARIABLE
OUTPUT    NAME	DEFINITIONS	

        I          Index for grid elements
        K          Index for averaging intervals
                   =1 for 1-hr averages
                   =2 for 3-hr averages
                                     90

-------
        N          Index for 5-min-average time series
        MAS        Number of elements in 5-min series
        NSTART     NAS+1
        NTIMES     Time index intermediate variable
        NTOTI      Total number of grid elements in model at sunrise
        ONEHR      Number of seconds in 1 hr (s)
        Ql         Intermediate summation variable
        SUM        Summation variable for concentrations
        THREE      Number of seconds in 3 hr (s)
        TS         Number of 5-min-averages in averaging interval
        TSTP       Averaging interval of basic model concentration
                   array =5 min= 300 s
Array Name Definitions

INPUT/
OUTPUT  ARRAY'NAME	DEFINITIONS	

  0     AVG(NB,2)    Maximum 1- and 3-hr-averages for each grid
                     element (|j.g/m )
  I     CHI(MB)      Nocturnal  concentrations in grid elements
                     Ug/rr)
  I     CONC(NA,NB)  Array,,of 5-min-average concentrations
                     (ug/m-3)
  0     NDXTIM(NB,2) Array of time index values for maximum 1- and
                     3-hr averages
  0     NTS(2)        Number of 5-min intervals in 1 and 3 hr
        SUMM(NB,2)   Summing variable for 5-min-average
                     concentration
Subroutines Called

  None

Common Blocks

   BLK4
   BLK5
   BLK6
                                     91

-------
                                   SECTION 6

                               SAMPLE  MODEL RUNS


     In this section we present sample model runs to illustrate  model  perfor-
mance and model outputs.  In the first model run we present the model  results
for a run using default input parameters.  The  first-time model user may
compare his results with the results  from  this  run to see that the model  is
operating properly at his installation.   In the second  sample run we make
calculations for a downwind distance  of 30 km rather than the default  10  km,
but maintain all other default values for  model inputs.

SIMULATION NUMBER 1

     The first simulation to be illustrated was obtained using default input
values obtained from Table 5.  A full listing of the summary output file
(VALMET.OUT) generated from this simulation is  given below in Figure 21.   This
output file would be printed by the model  user  at the completion of the model
run and consists of three pages of printer output.  Outputs shown in the
figure include:

 1.  a model input table,
 2.  a listing of numerical simulation parameters,
 3.  a summary listing of nocturnal model  output values,
 4.  a listing of the output values from  the solar model,
 5.  an output table of maximum 1- and 3-hr-average concentrations in  each
     grid element, and
 6.  a table of 5-min-average (300 s) concentrations as a function of  time for
     each model grid element.

The user who wishes to get a graphical depiction of model results may  manually
plot the model results from the output listing.  Alternatively, he may develop
a plotting program using as input the special file of model output data,
VALMET.PLT, that is generated with each model run.  Figures 22 and 23
illustrate the types of plots that may be  generated using such a plotting
program.  A listing of the plotting program that generated these figures  is
given in Appendix B.
                                     92

-------
I
»3


U

S

g

H
              u     a  a  u


                       a
i~t m    rjoc4oo

o       • in    o    fe
        3 O    *n
                        .
   i.    t  -3 < U W < S
   _;    <  =; - a: « e- i- '/j
   <    'J— pH&i>i05U5<
           •* —« ^ ** p* EN r«
^  .en  > O ff) O  • S Q O
CM -3    S> 3   OO^OO
 . c    :>  •     • o o  •  •
o«    in o   ooooo
            t .^ r* ^ fM C4 *N
<
S
z

H
X
0*

=
X
3
Eu
S
U

U
2

yj
z
v
'•&
a
u
ARACT
=
U
>*
u
_:
1
u
-•§
s§
2s
as
< z
S2
yj
j s
w u
a >
i5

z
a
S
<
8
>J

U
u
e
u
X U
a. s
co a
i*

^S

U U]
Id- U
s- C-* « —
w « <
« 3 > a
CB
RACTE
<
u

a«
U
<
JH
U





a
'ARAHE'J'tH

^
<
O
at
i
r*
an
W
fH
V
1
5
1
in
3
Z
O
i
»•»
as
u
-£
r*i <
-« >•
S
zg
Sj
§S ~
«z in
6-1 U
M z
S2 1
z282
-« WO «
W £d O -3
Sous
a tax
Z U O «!
M z O
1-1 m u
l*;.g
M u o a z
u to **<'•«
C SS S •« OS W
Off* 3 36- <
CQ fS U CC £ Z >
a c -H ^ o:
z u a to u
o too o .« fr«
U Z O fa f» IM H.
ua w o ^
3C *-* Cfi CQ < :?
•H < M W < <
a u u _ a
U 7-
U ^ -3 H
,&* as w ae ts, z cs
03 CJ £d U U 03
Siilill
an
j§
S8
U2
uu
M a; ^
§gj
as> to
<~a3
sSg£
o3&* Z
irt < O O *
M > -* «
a* 03 tii
Mp a: e-
2PSSS
3 M Z
asp-si
^g^-ss
a- o c-o •
SsS*«*
u O C 3 rd IN
£ U U Q & ^
ZJJ =
23£B2
ca H &< a < w
3Z« K BW
uaava
u &* o w
an ens
S2S5yU
gssie*
< = = u.^o
x u ^ « a «
u Q* a* 2 •
C 3 O fci
X U U O
u ssa
-i £-• &» cd &• ua
*i.u.5°i
OOO X Z <
!N O W
M . of •• Cfl — 3
d-^-i-Si
> in o W a e-t
< ot ai &• o
K A* < S Z O
r- oir 2 u^o
ni M U
>j z u yi z *-*
< o o o * q o
K &4.9Q U .-
u f* cd o w
O S £«. i* Z fa. Z
;J O tO pt
lsiz°i§£
r? tu^t t» ec ia z z
Z -4 Q Q £ 5" c-
o a o < —
3 z > > •-» £ a
Ci. U < < Z p —
_ ji§ 5 ^ w S
a ca z z c* <
I ca M a o <
M O Z 3 CS M —
l
ca
S: M S W
< S — -3 I 1 1 t
3£u~33SS
SxstU ^^
Ji-S
H GO ooor-
u3£'~
U £ ( I t l
u 3 H
1 §*
a u a»
3£5S~^,.


1 1 1 1 1 1 I 1 ( l 1 1 1 I i l

00030090000<=0<303
t 1 , 1 1 ' 1 1 1 1 1 I ' I L 1


JgS*"^£'5'°S^5'n2a"n
t i i i i j i i i i ! : , i i
r--»o^f-r--.-,«^-.c=^
                                                                                                                                                              0)
                                                                                                                                                              01
                                                                                                                                                              c
                                                                                                                                                              01
                                                                              93

-------
             i-*^nr4(N»«M'v>.3p*r-aio>ff*a>oor-^ir4f«N-^ooooQoaaooocjooooooooooi3OOQOooooooooo
    •-»  CQ   0000000000000000000000000000000000000000000000000020000=90  o
III  *   °IIIII°ir°°IIo!IIIIIIoIII°°II°III^^                                                         *







£O          CH^r*fflflOCDflOflQWOOWff»W
-------
                              600
                           o
                              200
                                                                          b)
                                                              \
                                                               \

     0       1       2
       CONC (ug/m3)
789
    TIME (LSI)
10
11
                           n
                           E
                           o>
                           O
                           z
                           o
                           o
                                                                          c)
                              0-4-3-2-10    1     2     3    4
                                            CROSS-VAUH DISTANCE (km)


 Figure 22.  Plots of Sample Simulation  1:   (a)  nocturnal  vertical  concentra-
             tion profile through plume  centerline,  (b)  CBL  height  (I°n9
             dashes) and inversion height  (short dashes)  as  a function of
             time, and (c) nocturnal cross-valley  concentration  profile
             through plume centerline.


Figure 22 illustrates some of the characteristics  of the  simulation.   It
presents plots of the nocturnal vertical  air pollution concentration  profile
at the valley center (Figure 22a), the  nocturnal  cross-valley concentration
profile through the plume centerline (Figure 22c),  and plots  of  CBL height
(long dashes) and inversion top height  (short  dashes)  as  a function of time
(Figure 22b).

     Figure 23 presents plots of the 5-min-average pollution  concentrations
for selected model grid elements.  Twenty  grid elements are  necessary for the
                                     95

-------
     1.0
    0.8
X= 10km

CL CONC=  1.20jjg/m3

%  OUT TOP=   0.1 %
 E
^
O
o
    0.4
    0.2
    0.0
                                          8
                                                 10
11
                                     TIME  (1ST)
  Figure 23.  Pollutant conentration  versus  time  for  selected  grid  elements  for
              Sample  Simulation  1.  The  down-valley distance,  centerline con-
              centration and  fraction  of mass  diffusing  out  the  top of  the
              valley  inversion during  the plume's  nocturnal  transport are
              indicated on  the figure.
 default simulation—three  on  the  valley  floor  and  17  on  the  valley  sidewall.
 Plots are shown  for grid elements  3,  6,  9,  12,  and  15.   Sunrise on  the date of
 the simulation was  at  0550 am LSI.   Low  steady-state  concentrations  are  seen
 in the figure before sunrise, with  time-varying  concentrations arising from
 fumigations and  upslope  advection  after  sunrise.   Nocturnal  concentrations on
 the valley floor and sidewalls are  low since the plume,  traveling down the
 valley centerline at 250 m height,  has diffused  insufficiently both  vertically
 and horizontally to produce high  concentrations  on  the valley surfaces.  The
 predominant feature of the concentration curves  at  the 10  km travel  distance
 is the post-sunrise fumigation of  the pollutant  plume on the valley  floor and
 sidewalls.  The  highest  concentrations occur on  the valley floor as  the  high
 concentrations at plume  centerline  subside  into  the growing  CBL.  The highest
 concentration shown occurred  in grid  element number 3 and  is 0.77 u.g/m .  This
 may be compared  to  the nocturnal  plume centerline  concentration of  1.20  ng/m  .
 Grid elements on the valley sidewalls are strongly  affected  by upslope advec-
 tion.  For example, the  concentration in grid  element 15 increases  rapidly
 just following sunrise as  the higher  concentrations in downslope grid elements
                                     96

-------
(e.g., see grid element 12) are advected up the sidewall.  Another  feature  of
the simulations, apparent in the figure, is the exponential decay of  concen-
trations in individual grid elements as the temperature  inversion top subsides
below their elevation.  This occurs earliest at the  highest grid elements.
The exponential decay in concentrations simulates  the effect of a sudden
increase in mixing depth that occurs when the temperature  inversion descends
below the grid element.  The concentration in grid element 3 begins to  drop
rapidly to zero at the time of temperature inversion destruction (0933  1ST).

SIMULATION NUMBER 2

     The results of sample simulation number 2 are shown below in Figures 24
and 25, in the same way that the results of simulation number 1 were  illus-
trated.  Input parameters for this simulation, 30-km down-valley from the
pollutant source, differ from the previous simulation only in down-valley
distance.

     At the 30 km cross section, the nocturnal plume had diffused sufficiently
during travel  that substantial reflections had occurred  from the valley sur-
faces.  The vertical and horizontal concentration  profiles through  the  plume
center!ine (Figure 24) show that the concentrations  were more uniform within
the valley than at the 10-km section.  At the 30-km  section, 2.7% of  the plume
had diffused out the top of.,the valley inversion,  and the  plume centerline
concentration was 0.39 ng/m .  The CBL growth, inversion top descent,  and
inversion breakup occur exactly as shown in the first simulation.   Concentra-
tions (Figure 25) differ during both the pre-sunrise and post-sunrise periods,
however, due to the different nocturnal concentration profiles within the
cross section at the longer travel distance.  Nocturnal  concentrations  are
higher at the valley surfaces due to the broader distribution of the  plume
about its centerline.  The highest concentration in  sidewall grid element 12
occurs during the night.  Since the pollutant is more evenly distributed
across the cross section, the effects of fumigations are less pronounced than
at the 10-km cross section.  Again, as at the 10-km  cross  section,  the
greatest effect of fumigation is on the valley floor.

     The example shown here as simulation number 2 is used merely to  illu-
strate the effect of changing down-valley distance,  other  parameters  remaining
constant.  Other parameters could have been changed  along  with the  down-valley
distance (such as valley width, for example) to simulate the physical charac-
teristics of a real  valley more accurately.
                                     97

-------


0 400
<
£
§200
X
0
a)
\

r\
\
f 1
i.i.
600

/•^
< ***
J,
O
| 200

b)

h
"

h
L 3
—••**"*'
1 i 1 1 1
0       1       2
  CONC (ug/m3)
789
    T1ME(USIO
10
11
/*"V A
lO
E
3
^ 1
0
z
O
0
0
•



. _ _ a
_^ — »-^""— "~~"'^™*>—- ^.
"T^ 	 	 T . .
                           0-4-3-2-10     1     2
                                         CROSS-VALLEY DISTANCE (km)

        Figure  24.   Same  as  Figure 22, for Sample Simulation  2.
                                   98

-------



,-N
fO
E
1?
V— X
o
z
o
o






1.0

0.8



0.6

0.4




0 7
u.z.

/N A\

. X=30km
. CL CONC= 0.39jjg/m3
. % OUT TOP= 2.7%




i
i
i

	 ^1 -^^=3«~, 	 in — . —a -»*>3f*^-. __ "*^^>^ ft"I
	 	 "" ' 	 ^""Z"*^^? \ ^ \ \
"~ — — •" **^^«-^^*^' \ « ^nQ*rtc
	 	 - — \15 \'2 \ , "\
\ \ \% \\
1 , L , i ^-^v X>— 1 ^ — ^--S^a—
                 7          8
                       TIME (1ST)
10
11
Figure  25.   Same as Figure 23,  for  Sample Simulation 2.
                        99

-------
                                  SECTION 7

                                 FURTHER  WORK
     In this section, we will discuss the need for the further development
and testing of the VALMET model.  VALMET, as described in this report,  is a
working model able to simulate many of the observed features of the
meteorology of Colorado's deep valleys.  The model has been developed  in a
modular fashion so that the modules can be upgraded as we learn more  about
valley meteorology.  The model, while incorporating what are thought  to be
the relevant physical processes that lead to air  pollution transport  and
dispersion, has not yet been fully tested against actual air quality  data.
Nevertheless, some initial results [52,53,54] are now available from
meteorological  and tracer experiments conducted in the Brush Creek Valley of
Colorado in 1982.  The tracer experiments, supported by the EPA in coopera-
tion with the U.S. Department of Energy's (DOE) Atmospheric Studies in
Complex Terrain (ASCOT) Program, were designed to provide the data necessary
to perform an initial evaluation of the VALMET model.  The evaluation  is not
yet complete, but is sufficient to provide us with initial guidance on  the
further development of VALMET.  In addition to the 1982 tracer experiments,
a more comprehensive set of tracer and meteorological experiments was
conducted in the same valley in September and October of 1984 as part  of
DOE's ASCOT program.  The 1984 ASCOT data are not yet available, but  they
will constitute an important resource to further  test the model.  A portion
of the tracer experiment, supported by the EPA, was designed specifically to
test the VALMET and MELSAR [55] air quality models developed in its GRAMA
program.  In the following, we will briefly summarize the results from  the
1982 experiments, pointing out the implications for future modifications to
the VALMET model.  Suggested modifications resulting from these experiments
and from other sources will  then be individually  discussed.

GUIDANCE FROM THE 1982 EXPERIMENTS

     The 1982 tracer experiments were designed to provide the data required
to evaluate the initial version of VALMET [52].   We did not consider  it
sufficient to simply collect tracer concentration data on a cross-valley arc
and compare this with model  calculations.  Rather, the approach taken  was to
collect meteorological and tracer data to test the full range of
meteorological  assumptions and parameterizations  used in modules within the
model.  For example, the model predicts that convective boundary layers will
grow over heated surfaces after sunrise, that upslope flows will develop
within these boundary layers, that pollutants from the elevated nocturnal
plume will fumigate into the convective boundary  layers, and that they  will
be transported out of the valley by the upslope flows.  Thus, within  the

                                     100

-------
restraints of the resources available, it was necessary to observe the
development of convective boundary layers over the slopes, the upslope wind
systems, fumigation of pollutants, and transport of pollutants up the slope.
In addition, it was necessary to simulate an elevated release of pollutants
and to observe the characteristics of the nocturnal plume.

     The EPA tracer experiments were conducted in a valley chosen by DOE
using criteria unrelated to the testing of the VALMET model.  The Brush
Creek Valley was a useful "target of opportunity" for the initial evaluation
of VALMET, but, as is usual with such opportunities, there were advantages
and disadvantages to the choice of this particular valley.

     There were several advantages to choosing the Brush Creek Valley for
the initial evaluation of VALMET.  First, the valley has a rather simple
topography.  The narrow, 25-km-long valley has no major changes in valley
orientation along its length.  It has nearly equal sidewall inclinations.
The valley drains a plateau, so that the ridges are at a constant altitude
regardless of location along the valley axis.  The valley has no major
tributaries.  Second, the valley axis is oriented from NW to SE so that the
sidewalls would be exposed to quite different insolation during the post-
sunrise temperature inversion breakup period.  The effect of this unequal
heating was a major uncertainty in the model formulation.  On the basis of
meteorological data collected in wider Colorado valleys, and numerical model
results, the VALMET model was developed under an assumption of horizontal
homogeneity of atmospheric structure on a valley cross section.  This
assumption could be readily tested in the Brush Creek Valley, where the
narrowness of the valley and the NW-SE orientation of the valley would
clearly maximize any horizontal gradients in atmospheric structure between
the sidewalls.  Third, the Brush Creek Valley would be heavily instrumented
with meteorological sensors by the ASCOT program.  Access to their
meteorological data would be a great benefit to the model evaluation effort.

     Along with the above advantages, there was a major disadvantage to
conducting an initial evaluation of VALMET using data from the Brush Creek
Valley.  This disadvantage was related to the short segment of the valley
that was accessible for tracer instrumentation.  VALMET is a two-dimensional
model, predicting concentrations on a cross section oriented perpendicular
to the valley axis some distance down-valley from a source.  Restrictive
assumptions are present in VALMET regarding a required homogeneity of the
temperature and wind structure in the along-valley direction.  The Brush
Creek Valley, however, is a short tributary valley that flows into the Roan
Valley a few kilometers below the valley cross section where most measure-
ments were made.  Consequently, tracer plume carried down the Brush Creek
Valley during the night would be carried into Roan Creek.  Reversal of the
down-valley winds (to up-valley) after sunrise would result in a large part
of the tracer plume being carried up the Roan Creek Valley, rather than
being carried back up the Brush Creek Valley as assumed in the model.
Evaluation of VALMET would be complicated by this violation of a major
assumption in the model, which had been designed for longer valleys.
                                     101

-------
     The full evaluation of the VALMET model will be the subject of  future
work.  It is appropriate here, however, to make some initial qualitative
statements concerning the evaluation of the model.  First, with respect to
the nocturnal portion of the model, the nocturnal plume was carried  down the
valley, as expected.  The nocturnal plume, although released above the
valley center, was found to be displaced somewhat towards one sidewall as it
was transported down the valley.  No completely satisfactory physical
explanation for this phenomenon has yet been offered, and no mechanism is
available in the VALMET model to model it.  The valley is not strictly
linear, but turns slightly with down-valley distance.  Because the plume was
displaced towards the "outside" of the turn, it is conceivable that  inertia!
effects are responsible for the displacement of the plume from the valley
centerline.  Further information on this phenomenon will be available from
the 1984 experiments.

     The nocturnal plume was carried down the valley in a rather strong
"jet" of down-valley winds, with the level of maximum winds at about release
height.  The nocturnal model, based on the Gaussian formulation, is
incapable of treating vertical shears in transport winds but, when winds at
release height are used for transport, the model approximates transport and
diffusion along the valley direction fairly well.

     Assumptions in the daytime portion of the model were verified with
actual meteorological and tracer data [53,54].  The post-sunrise period was
characterized by the growth of convective boundary layers over the sunlit
valley surfaces.  The tracer plume fumigated the valley sidewalls as
convective boundary layers grew upwards into the remnants of the nocturnal
temperature inversion containing the elevated tracer plume.  Tracer  was
carried from the valley by upslope flows, which developed within the growing
convective boundary layers.  Corresponding subsiding motions over the valley
center were noted in the temperature profiles at several of the tethered
balloon sites, but the limited vertical resolution of the tracer plume did
not allow this feature to be seen in the tracer concentration analyses.

     Due to the northwest-southeast orientation of the deep, steep-walled
valley, very significant differences occurred in the timing and rates of
convective boundary layer growth on the opposing sidewalls following sun-
rise.  As a result of the unequal heating of the different sidewalls, a
cross-valley flow developed, carrying the elevated plume towards the warmer
sidewall [53].  Due to the cross valley advection, tracer concentrations
were higher on this sidewall than predicted by the model.  A future
modification of the VALMET model will be required to handle this situation
properly in narrow valleys where post-sunrise insolation on the opposing
sidewalls is quite different.  The Brush Creek tracer experiments were the
first direct experimental confirmation of the importance of this physical
effect on tracer plume dispersion.

     The short length of the Brush Creek Valley, as expected, affected the
results of the tracer experiments.  The primary effect, from initial
analyses, seems to be that the tracer concentrations in the valley fell more
rapidly than expected after the post-sunrise wind reversal.  This is thought


                                     102

-------
to be due to the nocturnal plume being carried largely up Roan Creek  after
the wind reversal rather than reversing direction to come back up Brush
Creek.

SUGGESTED MODIFICATIONS TO THE VALMET MODEL

     The basic modeling approach in VALMET is to explicitly parameterize  a
number of physical processes acting to disperse pollutants in the valley
atmosphere.  Based on the parameter!zations now included in the model  and
initial results from the 1982 tracer experiments, a number of suggestions
can be made for model improvement.  Some of these suggestions are presented
below.  The simpler modifications are listed first.

Deposition

     Deposition of pollutants during transport could be incorporated  into
the model at two points.  Deposition during along-valley transport of  the
plume could be included in modules INGRAT and NORMAL, while deposition
occurring in the slope flows could be rather easily incorporated into  module
BRKUP using a surface sink term in the pollutant mass budget calculations.

Emission Above or Below Stable Core

     The present model assumes implicitly that pollutants are emitted  into
the stable core, which is then advected down-valley.  Concentrations  are
produced by fumigation of pollutants from this elevated plume as th*e  CBL
grows upward into it.  Modifications to the model will be required if  the
emissions fail to enter the stable core.  This will happen at the stack
location (x=0) once the CBL grows to the effective stack height.  Pollutants
would then be emitted into the CBL.  Similarly, once the inversion top sinks
below the effective stack height, the pollutants are dispersed above  the
valley inversion and do not affect the stable core.

     Calculations of air pollution concentrations down-valley of the  stack
should take account of these possibilities by modifying pollutant
calculations when changes in the stable core concentration profiles are
advected over the cross section of interest.  The advection time depends  on
the distance x down-valley of the stack and the wind speed in the stable
core, such that the advection time is

                                tx  -  t0 -  x/us

where t~ is the time of sunrise and t, is the time when the inversion  top
descends below the effective stack height or the CBL grows above the
effective stack height.  When calculations are made far downwind of the
stack (x»0), or down-valley flows are weak (ug small) the travel time will
be so long that the inversion will be broken before any change in
concentration calculations is necessary.  It may be necessary, however, to
modify the model calculations in short, well-drained valleys.  Similarly,
the model, when applied to ground level sources, will need modification to
account for the post-sunrise emission directly into the CBL.


                                     103

-------
 Energy Budget

      The energy budget model needs a great deal of further work.   It
 presently includes a simple bulk parameterization of the fraction  of
 extraterrestrial solar radiation converted to sensible heat flux at the
 valley surfaces.  This fraction, in general, should vary throughout the day
 and should depend on cloudiness, soil moisture, albedo, ground heat flux,
 vegetation, atmospheric absorption of the solar beam, etc.  It may be
 worthwhile to adapt existing surface energy budget parameterization schemes
 for use in the energy budget module.  This will, of course, increase the
 input requirements of the overall model.  A different approach is
 possible.  One could look at existing data sets and try to determine the
 value of this fraction leading to the "worst-case" pollution episodes  in
 different parts of the country.  The model could then be used with these
 fractions to simulate such conditions.  Such an empirical approach should
 also be focused on determining the other free parameter in the model,  the
 fraction k of sensible heat flux causing CBLs to grow.  A major effort is
 required to do a good job on either approach.

 Cross Valley Flows

      The model should be modified to incorporate cross valley flows in the
 stable core.  Such flows have been postulated by others [45] and were
 clearly observed in the 1982 Brush Creek tracer experiments.  They are
 expected to be strongest in valleys where differential heating occurs  on the
 sidewalls in the morning.  An example would be a north-south valley where
 the morning sun would shine on one sidewall while the other was still  in the
 shade.  The existence of such flows may strongly affect post-sunrise concen-
 trations on the sidewall since the plume, containing high concentrations, is
 advected towards the most-strongly sunlit sidewall.  Several methods seem
 possible for incorporating the modification into VALMET, but little
 observational information is available to test the different methods.
 Guidance may be available from the more-sophisticated primitive equation
 models [46-49].

 Turbulent Erosion of the Valley Inversion by Overlying Flows

      Too little is presently known about postulated turbulent erosion
 [15,50] of the top of a valley temperature inversion to include it in  the
 present version of the model.  It is not clear that it should be
 incorporated anyway, since the model is more conservative without  it.
 Lenschow et al. [50] have developed an approach that could be rather easily
 incorporated into the present model if a means could be found to relate a
 local  Richardson number at the top of the inversion to the rate of turbulent
 erosion.  No way is yet available to do this.  Additionally, recent
 primitive equations modeling indicates that turbulent erosion is not a major
 factor affecting temperature inversion breakup under conditions of light to
 moderate winds aloft when inversions are strong.^ '  Fluid modeling
 experiments may be useful here in the future.
(a)   Personal  communication with T. B. McKee, Colorado State University,
     Fort Collins,  Colorado, 1983.

                                      104

-------
Effect of Tributary Flows on the Enhancement  of  Diffusion

     The present version of the model  incorporates  a  simple  means  of
handling dilution due to valley tributary  flows.  The model  does  not treat
the effect of these tributary flows on enhancing  diffusion within  the main
valley.  At present there seems to be  too  few observational  data  to
incorporate this effect or the related thermodynamic  effects  caused  by
converging air masses of different temperature or stratification.  The DOE
ASCOT program may look into these effects  in  their  future Colorado field
programs.

Piffusion Coefficients

     The diffusion coefficients in the model  need improvement.  At present
we are modifying Pasquill-Gifford [43] flat terrain coefficients  on  the
basis of the Huntington canyon data of Start  et  al. [20] and  other data
summarized by others [29-31].  We need to  look at more  data  or  conduct
specific diffusion experiments to determine the  most  appropriate  values of
these coefficients.  Alternatives to the Gaussian plume equation  should also
be seriously considered.

Time-Varying Wind Speeds in the Stable Core

     The model could be modified to handle time-varying wind  speeds  in the
stable core.  There is no doubt that the down-valley  wind speeds  in  the
stable core decrease after sunrise during  the temperature inversion
destruction period, since this has been frequently  observed  [9].   It is not
entirely clear what the effect of this decrease will  be on calculations,
since the wind speed, in addition to modifying plume  concentration
calculations, will  also affect plume rise  calculations.

Temperature Inversion Buildup

     The present version of the model  simulates  the nocturnal steady-state
period and the inversion breakup or morning transition  period.  The  model
could be extended to simulate the temperature inversion formation  and growth
period, when down-valley winds first become established within  a  valley.  A
large quantity of data on this period was  collected in  Whiteman's  National
Science Foundation-supported dissertation  program in  1977 and 1978.   The
data have been processed, but not analyzed.  A case study [6] indicated some
promising research approaches that could lead to  an air pollution  model for
the inversion buildup period.

Differential Heating

     One might expect CBLs to grow at different  rates over the  different
sidewalls of a valley, especially when the energy budgets of  the  sidewalls
are substantially different.  This might occur frequently in  north-south
oriented valleys where the differential insolation  on the sidewalls  varies
strongly with time (see Figure 26 for  an example).  The present version of
VALMET has as its basis a bulk energy budget model  of the valley  inversion
that does not treat the sidewall  energy budgets  separately.   A  major

                                     105

-------
modification to the model would be required to simulate CBLs growing  at
different rates on the individual  sidewalls.  This would require an addi-
tional free parameter in the model, specifying the fraction of  sensible  heat.
used for growing CBLs over one sidewall, as opposed to the other.  At  pre-
sent there seem to be no vertical  structure data available taken on a  single
day on two opposite sidewalls.  Without this kind of data it is difficult to
come up with accurate parameter!zations of the differential CBL growth.
Brehm [51], based on the Innsbruck, Austria, Slopewind Experiment of  1978,
has presented a conceptual  model snowing differential CBL growth on opposite
sidewalls (Figure 27).  His conceptual model also shows subsidence in  the
stable core.  Bader and McKee [47], however, in their two-dimensional  simu-
lation of valley inversion breakup, noted no significant differential
boundary layer growth over opposite valley sidewalls even when  substantially
different sensible heat flux functions were used on the different valley
sidewalls.  They explained this as due to the effects of horizontally
propagating gravity waves that act to keep isotherms horizontal within the
stable core.  They also noted the existence of multiple closed  cross  valley
circulations within the stable core early in the post-sunrise period  that
acted to warm the valley temperature  inversion differentially with height,
destabilizing the valley atmosphere.  This process acts in consort with
stable core subsidence to warm the valley atmosphere.  As shown by the
different conclusions reached in these two studies, it seems appropriate to
verify differential CBL growth using  specially designed field studies.

Stacks Located on Sidewalls
     The model could be modified to handle emissions  from  stacks  located
above the valley floor on the sidewalls.  In view of  the results  from  the
1982 tracer experiments, this modification should be  combined with  a new
treatment of differential heating and cross valley flows,  since the
development of cross valley flows due to differential  heating may  strongly
affect concentrations on the sidewalls in the vicinity  of  the elevated
stack.
                                     106

-------
                          BRUSH CR.-AUGUST 1, 1982
      1400 r
E-FAC1NG SIDEWALL
W-FACING SIDEWALL
                            10      12      14      16      18      20
                                   Time (MDT)
Figure 26.  Illustration of differential solar flux on opposing sidewalls
            for Brush Creek Colorado on August 4, 1982.  The Brush Creek
            Valley drains from NW to SE.  The curves represent extra-
            terrestrial  solar flux (i.e., they assume no atmosphere) and
            were determined using a solar radiation model.
                                   107

-------
            ®
                                         MOUNTAIN
                                         WIND
                                         VALLEY
                                         WIND

                                         SLOPE
                                         WIND
STABLE
LAYER
Figure 27.  Brehm's [51] conceptual  model  of temperature inversion
            destruction in Austria's Inn  Valley,  showing
            differential CBL  growth  over  the opposite sidewalls and
            continued down-valley  flow in the elevated stable core.
                                108

-------
                                  REFERENCES
1.  Hewson, E. W., and G. C. Gill.  1944.
               ver Valley Near Trail ,  B.C.
Columbia Ri
of Interior,
                 453, 23-228.
                                       "Meteorological  Investigations in
                                       " Bur. Mines Bull., U.S. Department
2.  Bierly, E. W., and E. W. Hewson.  1962.  "Some Restrictive Meteorological
    Conditions to be Considered in the Design of Stacks."  J. Applied Meteor.,
    1, 383-390.

3.  Whiteman, C. D., and T. B. McKee.  1982.  "Breakup of Temperature
    Inversions in Deep Mountain Valleys:  Part II. Thermodynamic Model."  j_._
    Applied Meteor., 21, 290-302.
4.  U.S. Environmental
    Office of Research
                   Protection Agency.  1978.  Handbook for Preparing
                   and Development Reports, Revised.
    EPA-600/9-78-032,
    Cincinnati, Ohio.
                  December 1978, U.S. Environmental
                                                      Report
                                                    Protection
                                              Agency,
5.  Peterson, W.
    Handbook for
             B., J. S. Irwin,
             Preparing Users'
                              D. B. Turner and J. A.
                              Guides for Air Quality
    U.S. Environmental
    North Carolina.
                   Protection Agency, Research Triangle Park,
                                    Catalano.  1982.
                                   _Mod_e_ls.  August 1982,
6.
7.
8.
Whiteman, C. D.
Rocky Mountains.
Nov. 9-12, 1981,
                 1981.  "Temperature Inversion Buildup in Valleys of the
                1   Preprints, Second Conference on Mountain Meteorology.
                 Steamboat Springs, Colorado, pp. 276-272.
Whiteman, C. D. and T. B. McKee.  1977.  "Observations of Vertical
Atmospheric Structure in a Deep Mountain Valley."  Arch. Meteor. Geophys.
Bioklimat., Ser. A, 26,  29-50.
Whiteman, C. D., and T. B. McKee.  1979.  Temperature Inversion
Destruction in Mountain Valleys - Implications for Air Pollution
Dispersion.  Proceedings of the Tenth International  Technical  Meeting
Air Pollution Modeling and Its Application, NATO/CCMS Pilot Study,
Oct. 23-26, 1979, Rome, Italy, 217-224.
                                                                          on
9.
Whiteman
Mountain
          C. D.
         Valleys.
                 ey;
                 Til
1980.  Breakup
  Ph.D.
                                of Temperature Inversions in Colorado
                         Dissertation, Colorado State University,
pp. 250.Available from University Microfilms, Ann Arbor, Michigan, or as
Atmospheric Science Paper No. 328 from Department of Atmospheric Science,
Colorado State University, Fort Collins, Colorado.
                                     109

-------
10.  Whiteman, C. D.   1982.   "Breakup  of Temperature  Inversions  in  Deep
     Mountain Valleys:   Part  I.  Observations."   J.  Applied  Meteorology,  21,
     270-289.

11.  Whiteman, C. 0.,  and T.  8.  McKee.   1978.   "Air Pollution  Implications  of
     Inversion Descent  in Mountain  Valleys."   Atmospheric Environment,  12,
     2151-2158.

12.  Whiteman, C. D.,  and T.  8.  McKee.   1981.   "Valley  Wind Systems Under
     Temperature Inversion Conditions."   Proceedings, Symposium  on  Intermediate
     Range Atmospheric  Transport Processes  and  Technology Assessment.
     October 1-3, 1980,  Gatlingburg, Tennessee,  pp. 351-358.

13.  Whiteman, C. D.,  and R.  L.  Drake.   1981.   The  Green River Ambient  Model
     Assessment Program  September 1981  Progress  Report  on Local  Scale Modeling
     Activities.  PNL-4058.   September 1981.   Pacific Northwest  Laboratory,
     Richland, Washington, 249 pp.

14.  Whiteman, C. D.,  and K.  J.  Allwine.  1982.   Green  River Ambient Model
     Assessment Program  FY-1982  Progress Report.  PNL-4520. October 1982.
     Pacific Northwest  Laboratory,  Richland,  Washington, 187 pp.

15.  Davidson, 8., and  P. K..  Rao.  1958.  Preliminary Report on  Valley  Wind
     Studies in Vermont, 1957.   Final  Report  AFCRC-RT-58-29,  Contract  AF
     19(604)-1971, College of Engineering,  New  York University,  54  pp.

16.  Beuttner, J. K.,  and N.  Thyer. 1966.   Valley  Winds in the  Mount Rainier
     Area.  Arch. Meteor. Geophys.  Bioklimat.,  Ser. B., 14(2):125-147.

17.  Tyson, P. 0., and  R. A.  Preston-Whyte.   1972.   "Observations  of Regional
     Topographically-Induced  Wind Systems in  Natal."  J. Applied  Meteor., 11,
     643-650.

18.  Sterten, A. K.  1963.  A Further  Investigation of  the  Mountain and  Valley
     Wind System in South-Eastern Norway.  Intern Rapport K-254,  Forsvarets
     Forskningsinstitutt, Norwegian Defence  Research  Establishment, Kjeller,
     Norway, 51 pp.

19.  Urfer-Henneberger,  C.  1970.  Neuere Beobachtungen uber die  Entwicklung
     des Schonwetterwindsystems  in  einem V-formigen  Alpental  (Dischmatal bei
     Davos).  Arch. Meteor.  Geophys. Bioklimat.,  Ser. B, 18, 21-42.

20.  Start, G. E., C.  R. Dickson and L.  L.  Wendell.,  1975.   "Diffusion  in a
     Canyon Within Rough Mountainous Terrain."   J.  Applied  Meteor., 14(3):333-
     346.

21.  Briggs, G. A.  1969.  Plume Rise.   AEC  Critical  Review Series, TID-25075.

22.  Briggs, G. A.  1971.  "Some Recent  Analyses  of Plume Rise Observations."
     In:  Proceedings  of the  Second International Clean Air Congress,
     H. M. Englund and  W. T.  Beery, eds.  Academic  Press, New  York,
     pp. 1029-1032.


                                      110

-------
23.  Briggs, G. A.  1973.  "Diffusion Estimation for Small  Emissions."  Atmos.
     Turbulence and Diffusion Laboratory, Contribution File No. (Draft) 79.
     Oak Ridge, Tennessee, 59 pp.

24.  Briggs, G. A.  1975.  "Plume Rise Predictions."  Lectures on Air Pollution
     and Environmental  Impact Analysis, D. A. Haugen, ed.  American
     Meteorological Society, Boston, Massachusetts, pp. 59-111.

25.  Pierce, T. E., and D. B. Turner.  1980.  User's Guide^for MPTER:  A
     Multiple Point Gaussian Dispersion AlgoriThm with Optional Terrain
     Adjustment.  EPA-600/8-80-016.  April 1980, U.S. Environmental Protection
     Agency, Research Triangle Park, North Carolina.

26.  Turner, D. B.  1969.  Workbook on Atmospheric Dispersion Estimates.
     Public Health Service Publications No. 999-AP-26.

27.  Slade, D. H.  (ed.).   1968.  Meteorology and Atomic Energy 1968.  U.S.
     Atomic Energy Commission, Division of TechnicalInformation, Oak Ridge,
     Tennessee, 445 pp.

28.  McMullen, R.  W.  1975.  "The Change of Concentration Standard Deviations
     with Distance."  J.  Air Poll. Control Assoc., 25(10):1057-1058.

29.  Minnott, D. H., D. L. Shearer and Y. Gotaas.  1979.  "Measurements of the
     Vertical Dispersion  Rate in Deep-Valley Terrain."  Preprint Volume:
     Fourth Symposium on  Turbulence, Diffusion and Air Pollution.  January 15-
     18, 1979, Reno, Nevada.  Published by American Meteorological Society,
     Boston, Massachusetts.

30.  Minnott, D. H., D. L. Shearer and and R. S. Marker.  1977.  Development of
     Vertical Dispersion  Coefficients for Deep-Valley Terrain.  Report
     COO-4026-3 prepared  for U.S. Energy Research and Development Administra-
     tion by TRC Environmental Consultants, Inc.

31.  Shearer, D. L., D. H. Minnott and G. R. Hilst.  1977.   Development of
     Vertical Dispersion  Coefficients for Rolling Terrain Environments.  The
     Research Corporation of New England (TRC) Report COO-4026-1, January 1977.

32.  Gotaas, Y.  1972.   A Model of Diffusion in a Valley from a Continuous
     Point Source."  Arch. Meteor. Geophys. Bioklimat., Ser. A, 21. 13-26.

33.  Abramowitz, M., and  I. A. Stegun.  1965.  Handbook of Mathematical
     Functions with Formulas, Graphs, and Mathematical Tables.  Dover
     Publications, Inc.,  New York, 1046 pp.

34.  Hastings, C., Jr.   1955.  Approximations for Digital Computers.  Princeton
     University Press,  Princeton, New Jersey.

35.  Sellers, W. D.  1965.  Physical Climatology.  The University of Chicago
     Press. Chicago, Illinois, 272 pp.
                                      Ill

-------
36.  Kreith, F., and J. F. Kreider.   1978.   Principles  of Solar Engineering.
     Hemisphere Publishing Corp.,  Washington,  D.C.,  778 pp.

37.  McCullough, E. C.  1968.   Total  Daily  Radiant  Energy Available Extrater-
     restrial ly as a Harmonic  Series  in the Day  of  the  Year.   Arch. Meteor.
     Geophys.  Bioklimat..  Ser. B.  16. 129-243.

38.  Waugh, A. E.  1973.  Sundials:   Their  Theory and  Construction.  New York,
     Dover Publications, Inc., 228 pp.

39.  Thompson, P. D.  1961.  Numerical  Weather Analysis and  Predictions.  The
     MacMillan Co., 170 pp.

40.  Haltiner, G. J.  1971.  Numerical  Weather Prediction.  John Wiley and
     Sons,  Inc., New York, 317 pp.

41.  Friedman, F. L., and  E.  B. Koffman.  1981.   Problem Solving and Structured
     Programming in FORTRAN,  Second  Edition.  Addison-Wesley  Publishing Co.,
     Reading,  Massachusetts.

42.  MUller, H., R. Reiter, R. Sladkovic,  and  K. Munzert.  1982.  "Aerologische
     Untersuchungen des Tagesperiodischen  Windsystems  im Loisachtal."   XVII
     Internationale Tagung fur Alpine Meteorologie.  Berchesgaden, 21-25
     September 1982.  Annalen  der  Meteorologie,  (Neue  Folge), Nr. 19,  186-188.

43..  Hanna, S. R., G. A. Briggs and  R.  P.  Hosker, Jr.   1982.   Handbook on
     Atmospheric Diffusion.  DOE/TIC-11223. Technical  Information Center, U.S.
     Department of Energy. 102 pp.

44.  Sundararajan, A., and S.  A. Macklin.   1976.  "Comments  on the Heat Flux
     and Friction Velocity in  Free Convection  Near  the  Ground."  J. Applied
     Meteor.,  33, 715-718.

45.  Gleeson,  T. A.  1951.  "On the  Theory  of  Cross-Valley Winds Arising From
     Differential Heating  of  the Slopes."   J.  Meteor.,  8(6),  398-405.

46.  Bader, D. C.  1981.  Simulation  the Daytime Boundary Layer Evolution in
     Deep Mountain Valleys. M.S. Thesis, Atmospheric Science Department,
     Colorado  State University, Fort  Collins,  Colorado.

47.  Bader, D. and T. B. McKee.  1983.   "Dynamical  Model Simulation of the
     Morning Boundary Layer Development in  Deep  Mountain Valleys."  J. Climate
     and Appl. Meteor, 22(3),  341-351.

48.  McNider,  R. T.  1981.  "Investigation  of  the  Impact of  Topographic
     Circulations on the Transport and Dispersion  of Air Pollutants."   Ph.D.
     Thesis.  Department of Environmental  Science,  University of Virginia,
     Charlottesville, Virginia (available from University Microfilm, Ann Arbor,
     Michigan).
                                      112

-------
49.  McNider, R. T., and R. A. Pielke.  1981.  "Diurnal  Boundary-Layer
     Development Over Sloping Terrain."  J.  Atmos.  Sci..  38(10),  2198-2212.

50.  Lenschow, D. H., and B. B. Stankov.  1979.  "The Rapid Morning Boundary-
     Layer Transition."  J. Atmos.  Sci., 36, 2108-2124.

51.  Brehm, M.  1981.  Hangwindexperiment Innsbruck 1978  - Gebirgswindsystem
     und Inversionsauflosung.Ph.D. Dissertation,  Ludwig-Maximinans-
     Universitat, Munich, West Germany.

52.  Whiteman, C. D., R. N. Lee, M. M. Orgill and B. D.  Zak.  1984.  Green
     River Air Quality Model Development.  Meteorological  and Tracer Data -
     July/August 1982 Field Study in Brush Valley,  Colorado.  PNL-5163.
     Pacific Northwest Laboratory,  Richland, Washington,  139 pp.

53.  Whiteman, C. D., A. H. Huber,  R. W. Fisher, and B.  D. Zak.   1984.
     "Atmospheric Tracer Experiments in a Deep Narrow Valley."   Conference
     Volumes:  (Joint Session) Fourth Joint  Conference on Applications of Air
     Pollution Meteorology and Third Conference on  Mountain Meteorology,
     October 16-19, 1984, Portland, Oregon,  pp. J1-J4.

54.  Orgill, M. M., R. N. Lee, R. I. Schreck, K. J. Allwine and  C.  D.  Whiteman.
     1984.  "Early Morning Ventilation of an SF§ Tracer  from a Mountain
     Valley."  Conference Volumes:   (Joint Session) Fourth Joint  Conference  on
     Applications of Air Pollution  Meteorology and  Third  Conference on Mountain
     Meteorology, October 16-19, 1984, Portland, Oregon,  pp. J36  J-39.

55.  Allwine, K. J., and C. D. Whiteman.  1984.  Technical Description of
     MELSAR:  A Mesoscale Air Quality Model  for Complex  Terrain.   PNL-5048,
     Pacific Northwest Laboratory,  Richland, Washington.
                                      113

-------
       APPENDIX A



FORTRAN LISTING OF  VALMET
           115

-------
 1          PROGRAM VALHET
 2    c****************************************************************************
 3    C**** THIS COMPUTER MODEL  PREDICTS  AIR  POLLUTION CONCENTRATIONS ON A        *
 4    c**** VALLEY FLOOR AND SIDEWALLS  ARISING  FROM AN ELEVATED CONTINUOUS        *
 5    C**** POINT SOURCE OF POLLUTION DURING  NOCTURNAL STEADY-STATE               *
 g    c**** DOWN-VALLEY DRAINAGE FLOWS  AND  DURING  THE POST-SUNRISE                *
 7    C**** TEMPERATURE INVERSION BREAKUP PERIOD.   A MODIFIED GAUSSIAN PLUME      *
 8    C**** ALGORITHM IS USED DURING THE  NOCTURNAL PERIOD.   A NUMERICAL SCHEME    *
 9    C**** IS USED FOR THE CALCULATIONS  AFTER  SUNRISE WHEN FUMIGATIONS OCCUR.    *
10    C**** SIMPLE PAP.AMETERIZATIONS ARK  USED IN THE POST-SUNRISE SIMULATIONS     +
11    C**«* TO ACCOUNT FOR SOLAR FORCING,  SENSIBLE HEAT FLUX, CBL GROWTH,
12    C**** INVERSION DESCENT, AND UPSLOPE  TRANSPORT AND DIFFUSION.  A USER'S
13    C**** MANUAL IS NOW AVAILABLE FOR THIS  MODEL.  CONTACT C.D. WHITEMAN
14    c**** OR K.J. ALLWINE, GEOSCIENCES  RESEARCH  AND ENGINEERING DEPARTMENT,
15    c**** BATTELLE PACIFIC NORTHWEST  LABORATORIES,  RICHLAND.WA 99352.
16    C****        VERSION 1.1      DECEMBER  1, 1984
^7    £****************************************************************************
18
19          PARAMETER(NS=6000,NA=100,NB=30,NB1-31)
20
21          COMMON/BLK1/HC(NS),HT(NS)
22          COMMON/BLK2/CHIBAR(NB)
23          COMMON/BLK3/V(NB1)
24          COMMON/BLK4/CONC(NA,NB)
25          COMMON/BLK5/CHKNB)
26          COMMON/BLK6/AVG(NB,2),NDXTIM(NB,2),NTS(2)
27
28          REAL HITEL(NB),HITEU(NB),TIME(NS)
29          REAL K,LAT
30          REAL SUM(NB),CCH(2,NB)
31          CIIARACTER*3 ITIME
32
33    c**** OPEN FILES
34          OPEK(UNIT=2,NAME='VALMET.OUT',TYPE='NEW',FORM='FORMATTED1)
35          OPEN(UNIT=3,NAME='VALMET.PLT',TYPE='NEW',FORM='FORMATTED')
36
37    c**** INITIALIZE PARAMETERS
38          HCBLI=25.
39          ITIME='LST'
40    C***« STANDARD TEMPERATURE AND PRESSURE
41          PSTD=1013.
42          TSTD=20.+273.16
43
44    c**** READ INPUT DATA FROM SUBROUTINE
45          CALL INPUT(HCBLI,H,W,ALPHA1,ALPHA2,DELTAT,NBOX,HZERO,
46         +           GAMMA,BETA,LAT,URLONG,MO,IDA,IYR,K,RHO,PRESS,
47         +           Q,ISTAB,XC,UC,STMP,SRAD,SVEL,TEMP,AO,YZERO,US,
48         +           AS,BOXLEN,NTOTI,AC)
49    C     NUMBER OF TIMESTEPS  IN THE  CONCENTARTION AVERAGING PERIOD
50          NTSA=30
51    C     CONCENTRATION AVERAGING  INTERVAL  IN SECONDS
52          IAVG=NTSA*INT(DELTAT)
53    C     FACTOR TO CONVERT TO STANDARD CONDITIONS
54          STNDCON=PSTD/PRESS*TEMP/TSTD
55
56
57    c**** CALCULATE JULIAN DATE
58          CALL JULIAN(MO,IDA,IYR,JULDAY)
59
60    c**** PLUME RISE
61          CALL PRISE(TEMP,SVEL,SRAD,STMP,XC,US,ISTAB,GAMMA,DLH)
62          H=H+DLH
63          IF (H. GT. HZERO. OR. H. I.E. HCBLI) THEN
64          WRITE(6,115)
                                             117

-------
 65      115 FORMAT(1HO, "THE PLUME, AFTER PLOMERISE, IS NOT WITHIN',
 66         +' THE STABLE CORE1)
 67          STOP 4
 68          ENDIF
 69
 70    c**** ADJUST STABILITY CATEGORIES TO ACCOUNT FOR ENHANCED
 71    C     DIFFUSION IN COMPLEX TERRAIN, AFTER START ET AL. (1975) .
 72          ISTABY=ISTAB-2
 73          ISTABZ-ISTAB-1
 74
 75    C**** VOLUMETRIC DILUTION DUE TO SLOPE FLOW CONVERGENCE, TRIBUTARY
 76    C**** FLOWS, OR ENTRAINMENT.
 77          CALL DILUTE(US,UC,AS,AC,DF)
 78
 79    C**** DETERMINE DIFFUSION COEFFICIENTS AND CALCULATE THE FRACTION
 80    c**** OF POLLUTANT MASS IN THE VALLEY CROSS-SECTION THAT HAS
 81    c**** DIFFUSED "BEYOND" THE VALLEY WALLS AND "BELOW" THE GROUND.
 82          CALL INGRATJXC,YZERO,ISTABY,ISTABZ,W,ALPHA1,ALPHA2,
 83         +            HZERO,H,SUM1,SUM2,SIGMAY,SIGMAZ)
 84
 85    c**** DISTRIBUTE THE LOST MASS FRACTION BACK INTO THE  CROSS-
 86    c**** SECTION. USE WELL-MIXED ASSUMPTION.
 87          CHIOFF=(1.-SUM1-SUM2)*Q*1.E09/(US*DF*AC)
 88
 89    c**** USE GAUSSIAN PLUME EQUATION TO CALCULATE THE MAXIMUM
 90    c**** CENTERLINE CONCENTRATION WITHIN THE VALLEY CROSS-SECTION
 91          CALL GAUSS(YZERO,H,YZERO,H,Q,US,DF,SIGMAY,SIGMAZ,
 92         +           CLCONC)
 93          CLCONC=(CLCONC+CHIOFF)*STNDCON
 94
 95    c**** USING GAUSSIAN PLUME EQUATION, CALCULATE STEADY-STATE
 96    c**** NOCTURNAL CONCENTRATIONS AT POINTS ON THE VALLEY FLOOR AND
 97    c**** SIDEWALLS CORRESPONDING TO THE GRID ELEMENTS USED FOR THE
 98    c**** AFTER-SUNRISE CALCULATIONS.
 99          DO 1 I=1,NTOTI
100          IF(I.LE.NBOX) THEN
101    C  VALLEY FLOOR
102             Z»0.
103             Y=(BOXLEN/2.)+BOXLEN*(I-D
104             CALL GAUSS(Y,Z,YZERO,H,Q,US,DF,SIGMAY,SIGMAZ,CCHI)
105             CCHI=CCHI+CHIOFF
106    C  SIDEWALLS
107          ELSE
108             Z=(BOXLEN*TAN(ALPHA1))/2.+(I-NBOX-1)*BOXLEN*TAN(ALPHA1)
109             Y=W/2.+Z/TAN(ALPHAl)
110             CALL GAUSS(Y,Z,YZERO,H,Q,US,DF,SIGMAY,SIGMAZ,CCHI)
111             CCHI=CCHI+CHIOFF
112          ENDIF
113          CHI(I)=CCHI
114        1 CONTINUE
115
116    C**** THE FIRST ELEMENT IN THE MODEL OUTPUT CONCENTRATION ARRAY IS
117    C**** DEFINED AS THE STEADY-STATE NOCTURNAL CONCENTRATION.
118          DO 2 I=1,NTOTI
119          CONC(1,I)=CHI(I)
120          CCH(1,I)=CHI(I)
121        2 CONTINUE
122
123    C**** CALCULATE SOLAR PARAMETERS FOR THE SITE AND DATE.
124          CALL SOLAR(LAT,URLONG,JULDAY,NTSR,TAU,A1)
125          TAU=TAU*3600.
126          T4=NTSR
127          IT=NTSR/100
128          T2=IT
                                              118

-------
129          T1=T4/100
130          T3=(T1-T2)*100./60.+T2
131
132    c**** ENERGY BUDGET
133          CALL EBDGT(AO,A2)
134          AO=A2
135
136    C**** USE BULK THERMODYNAMIC MODEL TO DETERMINE  INVERSION TOP
137    c**** DESCENT AND CBL RISE AS FUNCTION OF TIME.  THE NUMBER OF
138    C**** TIME STEPS REQUIRED TO DESTROY SHE INVERSION IS ONE OF
139    c**** THE OUTPUTS OF THE THERMODYNAMIC MODEL.
140          CALL DESCNT(AO,A1,ALPHA1,ALPHA2,BETA,DELTAT,GAMMA,
141         +            HCBLI,HZERO,K,W,PRESS,RHO,TAU,NSTEPS)
142          NINDX=NSTEPS+NINT(3600./DELTAT)
143          NAS=NINDX/NTSA-H
144
145    C**** CALCULATE AIR POLLUTION CONCENTRATION AS A FUNCTION OF TIME
146    C**** AFTER SUNRISE.
147          N=l
148          DO 9 IND=1,NAS
149
150    C     INITIALIZE CONCENTRATION SOMMING ARRAY TO  ZERO
151          DO 3 I=1,NTOTI
152          SUM(I)=0.
153        3 CONTINUE
154
155    C     LOOP ON NUMBER OF TIMESTEPS PER AVERAGING  PERIOD
156          DO 7 ND«1,NTSA
157          N-N+1
158          MN«N-1
159
160          IF(N.GT.NSTEPS)THEN
161          DO 4 I=1,NTOTI
162          V(I)=999.9
163        4 CONTINUE
164          GO TO 5
165          ENDIF
166
167          IFIX2=(HT(MN)-HC(MN))/(BOXLEN*TAN(ALPHA1))
168          NTOT=NBOX+IFIX2
169
170    C**** CALCULATE POLLUTANT MASS COMING INTO GROWING BOX  AT ITS  TOP.
171          CALL PROFIL(MBOX,NTOTI,NTOT,BOXLEN,HC(MN),ALPHA1,
172         +            ALPHA2,HZERO,H,HC(N),HT(MN),HT(N),YZERO,Qr
173         +            US,DF,SIGMAY,SIGMAZ,CHIOFF)
174
175    c**** CALCULATE VELOCITIES AT BOX SEPTA.
176          CALL VELOCY(N,MN,NTOTI,NBOX,NTOT,BOXLEN,DELTAT)
177
178    c**** CALCULATE POLLUTANT MASS BUDGET FOR EACH BOX.
179        5 CALL BRKUP(N,MN,NTOTIJ.BOXLEN,DELTAT,CCH)
180
181          DO 6 I»1,NTOTI
182          SUM(I)=CCH(2,I)+SUM(I)
183          CCH(1,I)=CCH(2,I)
184        6 CONTINUE
185
186        7 CONTINUE
187
188          DO 8 I=1,NTOTI
189          CONC(IND,I)=SUM(I)/REAL(NTSA)*STNDCON
190        8 CONTINUE
191
                                             119

-------
192          IF(IND.EQ.1)THEN
193          TIME(IND)=T3 +REAL(NTSA)*DELTAT/2./3600.
194          ELSE
195          TIME(IND)=TIME(IND-1)+REAL(NTSA)*DELTAT/36 00.
196          ENDIF
197
198        9 CONTINUE
199
200    C**** EMPLOY THE POST-PROCESSOR  TO  CALCULATE  MAXIMUM 1- AND 3-HOUR
201    C**** AVERAGE CONCENTRATIONS  FOR REGULATORY APPLICATIONS.
202          TSTP=REAL(NTSA)*DELTAT
203          CALL PSTPRC(TSTP,NTOTI,NAS)
204
205    c**** WRITE OUT RESULTS TO UNIT  2.
206          WRITE(2,100) DELTAT,NTOTI,BOXLEN,NSTEPS,NINDX
207      100 FORMAT(1HO,'NUMERICAL SIMULATION  PARAMETERS:',/,IX,
208         +'TIMESTEPS ARE',F5.1,'  SECONDS  LONG',/,IX,'NUMBER OF GRID',
209         +'  ELEMENTS"1,13,/,IX,'GRID ELEMENTS ARE',F6.1,' METERS',
210         -I-1  LONG',/, IX,'NUMBER OF TIMESTEPS  FROM  SUNRISE TO INVERSION',
211         +'  DESTRUCTION"1,I6,/,IX,'NUMBER OF TIMESTEPS IN SIMULATION1,
212         +'  (INCLUDES  EXPONENTIAL DECAY AFTER INVERSION DESTRUCTION)=',
213         +16)
214          WRITE(2,101) IAVG,NAS
215      101 FORMATdH ,'CONCENTRATION  AVERAGING INTERVAL" ' , 15 , ' SECONDS',
216         +/flH  ,'NUMBER  OF AVERAGING INTERVALS IN THE SIMULATION=',14)
217          WRITE(2,102) CLCONC,H,XC,(SUM1*100.)
218      102 FORMAT(1HO,'PLUME DIFFUSION  DURING NOCTURNAL TRAVEL:',/,lH ,
219         +  'A  PLUME CENTERLINE CONCENTRATION OF',F7.2,
220         +  ' MICROGRAMS  PER CUBIC METER OCCURS',F6.0,' M ABOVE ',
221         •*•  'THE VALLEY FLOOR  CENTER. ' ,/,lH , ' AT THE TRAVEL DISTANCE OF1
222         +  ,F6.1,' KM',F5.1,'% OF THE  PLUME  IS CONTAINED WITHIN THE'
223         +  ' VALLEY INVERSION CROSS  SECTION.')
224          WRITE  (2,103)  XC,(SUM2*100.),((1.-SUM1-SUM2)*100.),CHIOFF
225      103 FORMATdH , ' AT THE  TRAVEL  ',
226         +  'DISTANCE OF',F6.1,' KM',F5.1,'%  OF THE PLUME HAS DIFFUSED',
227         +  ' OUT THE TOP OF THE VALLEY.',/,lH ,'THE REMAINING',F5.1,
228         +  '%  OF THE PLUME  MASS  IS MIXED  UNIFORMLY THROUGHOUT THE ',
229         +  'INVERSION  CROSS',/,1H ,'SECTION  TO PRODUCE AN OFFSET  COHC',
230         +  'ENTRATION  OF',F7.3,'  MICROGRAMS  PER CUBIC METER.1)
231          WRITE  (2,104)  (AC/1000.),(1./DF)
232      104 FORMATdH ,'AREA  AT SIMULATION  CROSS SECTION IN THOUSANDS OF  ',
233         +  'SQ METERS  IS',F8.0,/,lH  ,'CLEAN  AIR DILUTION FACTOR IS',F8.3)
234          WRITE(2,105)  NTSR,ITIME,(TAU/3600.),Al
235      105 FORMATdHO,'SOLAR  MODEL RESULTS:',/,lH  ,'TIME OF SUNRISE=',
236         +  I6,1X,A3,/,IX,'LENGTH  OF  DAYLIGHT PERIOD=',F6.2,' HRS',/,
237         +  1H  ,'EXTRATERRESTRIAL  SOLAR  FLUX  ON HORIZONTAL SURFACE AT',
238         +  ' SOLAR NOON=',F7.1,'  WATTS  PER SQUARE METER')
239          WRITE(2,106)
240      106 FORMAT(1HO,'MODEL OUTPUT:  CONCENTRATIONS IN MICROGRAMS/CUBIC',
241         +  ' METER:1,/,1H  ,2X,'GRID',5X,'HEIGHT ABOVE',3X,'MAXIMUM1,4X,
242         +  'TIME OF',5X,'MAXIMUM1,4X,'TIME OF',/,1H ,IX,'ELEMENT1,
243         +  3X,'VALLEY  FLOOR',4X,'1-HOUR',2X,'OCCURRENCE1,5X,'3-HOUR',
244         -I-  2X,'OCCURRENCE',/,lH  ,IX, ' NUMBER' ,9X, ' (M) ' ,7X,'AVERAGE',
245         +  5X, ' (LST) ' ,6X,'AVERAGE',5X, ' (LET)')
246
247          DO 10  I=1,NTOTI
248          IF(I.LE.NBOX)THEN
249             HITEL(I)=0.
250             HITEU(I)=0.
251          ELSE
252             HITEL(I) = (I-NBOX-1) *BOXLEN*TA£J(ALPHA1)
253             HITEU(I)=(I-NBOX)*BOXLEN*TAN(ALPHA1)
254          ENDIF
255        10 CONTINUE
                                               120

-------
256
257          DO 11 1=1,NTOTI
258          TMAXC1=TIME(NDXTIM(I,1))
259          TMAXC2=TIME(NDXTIH(I,2))
260          WRITE(2,107) I,HITEL(I),HITEU(I),AVC{I,1),TMAXC1-1.,
261         + TMAXC1,AVG(I,2),T«AXC2-3.,TMAXC2
262      107 FORMATdH , 14 ,9X,F4 .0 , ' - ' ,F4 .0 ,2 (2X,F8.2 ,2X, F5 .2 ,'-' , F5 .2 ) )
263       11 CONTINUE
264
265          WRITE(2,108) IAVG,IAVG
266      108 FORMAT(1H1,//,15X,I5,'  SECOND  AVERAGE POLLUTANT CONCENTRATIONS',
267         + ' (ug/m3) IN MODEL GRID  ELEMENTS  AS  A FUNCTION OF TIME',//
268         + 1H  ,32X,'INDICATED TIME  IS AT THE MIDPOINT OF',15,'  SECOND',
269         + ' AVERAGING PERIOD',/,1H ,32X,'PRE-SUNRISE STEADY STATE',
270         + ' CONCENTRATIONS  ARE GIVEN  IN FIRST  LINE')
271          WRITE(2,109)
272      109 FORMAT (1 HO, 2X, 'TIME' ,2X, (20CBOX1 ,3X) ) ,/,lH ,2X, ' LST',4X,
273         +'01',4X,'02',4X,'03',4X,'04',4X,'05',4X,'06',4X,'07',4X,
274         +108',4X,'09',4X,110I,4X,'11I,4X,'121,4X,'13',4X,'14',4X,
275         +'15',4X,'16',4X,'17',4X,'18',4X,'19',4X,'20')
276
277          IF (NTOTI. LE. 20)  THEN
278          NY=NTOTI
279          ELSE
280          NY=20
281          ENDIF
282
283          WRITE(2,110) T3,(CHI(I),1=1,NY)
284      110 FORMATdH ,1X, ' < ' ,F4 .2 ,20F6 .2)
285          DO 12 MN-1,NAS
286          WRITE(2,111) TIME(MN),(CONC(MN,I),1=1,NY)
287      111 FORHAT(1H ,21F6.2)
288       12 CONTINUE
289
290          IF(NTOTI.GE.21)THEN
291          WRITE(2,112)
292      112 FORMAT(1H1,//////,2X,'TIME',2X, (20CBOX1 ,3X)) ,/,lH ,2X,
293         +  'LST',4X,'21r,4X,'22',4X,'23',4X,'24',4X,'25',4X,'26',
294         +4X,'27',4X,'28',4X,'29',4X,'30',4X,'31',4X,'32',4X,'33',
295         +4X,'34',4X,'35',4X,'36',4X,'37',4X,'38',4X,'39',4X,'40')
296          WRITE(2,113) T3,(CHI(I),1=21,NTOTI)
297      113 FORMATdH ,1X, ' < ' , F4 .2 ,20F6 .2)
298          DO 13 MN=1,NAS
299          WRITE(2,114) TIME(MN),(CONCfMN,I),1=21,NTOTI)
300      114 FORMATdH ,21F6.2)
301       13 CONTINUE
302          ENDIF
303
304          CLOSE(UNIT=2)
305
306    C**** WRITE OUT RESULTS  TO  UNIT 3  FOR PLOTTING.
307          11=1
308          TIME(1)=T3
309          WRITE(3,201) NSTEPS,NTSR,NINDX,NTOTI,NAS,ITIME
310      201 FOSMATdH ,518,A3)
311          WRITE(3,202) CLCONC,DELTAT,SUM1,SUM2,XC,CHIOFF,
312         + SIGMAZ,SIGMAY,ALPHA1,ALPHA2,W,HZERO,H
313      202 FORMATdH ,13F10.5)
314          DO 14 1=1,NTOTI
315          WRITE(3,203) I,II,T2,CHI(I)
316          DO 14 IKD=1,NAS
317          WRITE(3,203) I,IND+1,TIME(IND),CONC(IND,I)
318      203 FORMATdH  , 12 ,15 , F12 .7 , F12 .7)
319       14 CONTINUE
                                             121

-------
320          DO 15 I=1,NSTEPS
321          WRITE(3,204) HC(I),HT(I)
322      204 FORMAT(1H ,6(2F9.3))
323       15 CONTINUE
324          CLOSE(UNIT=3)
325
326          STOP 'NORMAL EXIT1
327          END
                                            122

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
      SUBROUTINE INFUT(HCBLI, H, W, ALPHA1 , ALPHA2 ,DELTAT,
     1   NBOX, HZ ERO, GAMMA, BETA, LAT, URLONG, MO, IDA, IYR, K, RHO,
     2   PRESS, Q, ISTAB, XC, UC, STMP, SRAD, SVEL, TEMP, AO , YZERO, US,
     3   AS,BOXLEN,NTOTI, AC)

      PARAMETER  (NS=6000 , NA=100 , NB=30 ,NB1=31 )

C**** THIS SUBROUTINE CONTROLS AN  INTERACTIVE TERMINAL THROUGH WHICH
C**** INPUT DATA ARE RECEIVED BY THE MODEL.

      REAL K, LAT, VAL(27),  VALNEW(27),  VALLO(27),  VALHI(27)
      REAL ASTAB(6)
      INTEGER ID (27)
      CHARACTER*? NAM (27)
      CHARACTER*20 CAT (10)
            *********** Jjjpprp  pgfJfljrpjQjqg***** ****************************
             VALLEY SIDEWALL  SLOPE (DEC ELEVATION)                        *
             OTHER VALLEY  SIDEWALL  SLOPE (DEG)                           *
             CROSS SECTIONAL  AREA AT SOURCE (THOUSANDS OF M2)           *
             RATE OF TEMP  RISE ABOVE VALLEY (DEG K/SEC)
             TIME STEP (SEC)
             VERTICAL POTENTIAL TEMPERATURE GRADIENT (DEG K/M)
             INITIAL HT  OF POLLUTION MAX CONCENTRATION(M) .
             INITIAL INVERSION HEIGHT(M)
             DAY OF MONTH  (1-31)
             ATMOSPHERIC STABILITY  CLASS (1-6),  E.G., 6=F
             YEAR  (00-99)
             FRACTION OF SENSIBLE HEAT GOING INTO CBL GROWTH  (0-1)
             LATITUDE  (DEG N)
             MONTH OF YEAR (1-12)
             NUMBER OF BOXES  ON HALF OF THE VALLEY FLOOR.
             PRESSURE  (MB)
             POLLUTANT SOURCE STRENGTH (KG/SEC)                          *
             DENSITY  (KG/M**3)                                           *
             STACK RADIUS  («)                                            *
             STACK EFFLUENT TEMPERATURE (DEG C)                          *
             STACK EXIT  VELOCITY (M/S)                                  *
             AMBIENT AIR TEMPERATURE (DEG C)                            *
             DOWNVALLEY  WIND  VELOCITY AT CROSS SECTION (M/SEC)          *
             LONGITUDE (DEG W)                                           *
             PLUME-CARRYING DOWNVALLEY WINDSPEED AT STACK (M/S)         *
             FULL WIDTH  OF VALLEY FLOOR (M)                             *
             OFF-CRNTERLINE DISPLACEMENT OF STACK (M)                   *
             DOWNVALLEY  DISTANCE FROM SOURCE TO CROSS SECTION  (M)       *
C************************************************************************

C**** SPECIFY NAMES OF CATEGORIES AND PARAMETERS.
c***********
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
ALPHA1
ALPHA2
AS
BETA
DELTAT
GAMMA
H
HZ ERO
IDA
ISTAB
IYR
K
LAT
MO
NBOX
PRESS
Q
RHO
SRAD
STMP
SVEL
TEMP
UC
URLONG
US
W
YZERO
XC
DATA
1
2
3
4
DATA
1
2
3
4
5
6
CAT /'SENSIBLE HEAT FLUX
'MODEL CHARACTER.
'SITE LOCATION
'VALLEY ATMOSPHERE
•STACK CHARACTER.
NAM /'AO
'ALPHA2=
1 BETA =
1 IYR
'UC
'Q
1 SRAD =
'K
'NBOX =
'LAT
1 TEMP =

' STAB =
1 SVEL =
'VALLEY CHARACTER.
'INVERSION CHARACTER.
'DATE
'GAUSSIAN PLUME
'
•WIDTH =' , 'ALPHA1=' ,
•HZERO =' , 'GAMMA =' ,
1 URLONG= ' , ' MO = ' , ' IDA!
•PRESS =' , 'XC =', 'YZEF

'US = ' , ' H = ' , ' STMI
'AS ='/
                                              123

-------
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
^78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
C**** SET DEFAULT VALUES.
DATA VAL /. 24, .15, 600

.,15. ,15. ,3. ,500. ,.025,0.,
1 40. 00, 105. 00, 9., 21. ,82., 10., 7 50. ,10000. ,
2 0.,4.,.001,

'F',4.,250.,100.,3.,0.,0./

C**** SET LOWER LIMITS OF PARAMETERS
DATA VALLO /5*0.,1. ,3
1 600. ,100.

*0.,-90.,-180.,1.,1.,0.,-50.,
,-50000., 2*0. ,'D' ,6*0./

c**** SET UPPER LIMITS OF PARAMETERS
DATA VALHI /I . ,1 . ,50000 . ,90 . ,90 . ,30 . ,2000 . , .050 , .0005 ,
1 90. ,180.,
12., 31., 99., 50., 1050., 100000., 50000.,
2 15., 1.,'P1, 15., 1000., 600., 5., 50., 150000. /


C**** NAME STABILITY CATEGORIES.
DATA ASTAB /'A' , 'B' , '

C','D' ,'E' ,'F'/

C**** WRITE OUT TABLE OF INPUT VALUES.
LU=6
10 WRITE(LU,1CO)
WRITE(LU,110) CAT(l),
WRITE(LU,120) CAT(2),
J6=6
NBOX=INT(VAL(6))
WRITE(LU,130) CAT(3),
WRITE(LU,140) CAT(4),
WRITE (LU, 150) CAT (5),
J12=12
J13=13
J14=14
MO=INT(VAL(12) )
IDA=INT(VAL(13))
IYR=INT(VAL(14»


(J,NAM(J) ,VAL(J) ,J=1,2)
(J,NAM(J) ,VAL(J) ,J=3,5)


J6,NAM(6) ,NBOX
(J,NAM(J) ,VAL(J) ,J=7,9)
(J,NAM(J) ,VAL(J) , J=10,ll)






WRITE(LU,160)CAT(6) ,J12,NAM(12) , MO, J13 ,NAM(13) , IDA, J14 ,NAM(14) , IYR
WRITE(LU,170) CAT(7)
WRITE(LU,180) CAT(8)
WRITE (LU, 190)
WRITE(LU,192) CAT(9),
WRITE(LU,194)
IF(LU.EQ.2) GO TO 70

(J,NAM(J) ,VAL(J) , J=15,16)
(J,NAM(J) ,VAL(J) ,J=17,19)
(J,NPM(J) ,VAL(J) ,J=20,22)
(J,NAM(J) ,VAL(J) , J=23,25)
(0,HAM(,1) ,VAL(J) ,J=26,27)


100 FORMAT(/,1H , ' THE PROGRAM INITIALIZATION PARAMETERS ARE SET TO THE
1 FOLLOWING VALUES:',/)
110 FORMAT(1H ,A20,2(2X,
120 FORMATflH ,A20,3(2X,
130 FORMATdH ,A20,2X,'(
140 FORMATdH ,A20,2X,'(
1 F6.3))
150 FORMATdH ,A20,2(2X,
160 FORMATdH ,A20,3(2X,
170 FORMATdH ,A20,2(2X,
180 FORMATdH ,A20,2X,'(
1 2X,'(
2 2X,'(
190 FORMATdH ,20X,2X,'(
1 2X,'(
2 2X,'(
192 FORMATdH ,A20,3(2X,
194 FORMATdH ,20X,2X,'(
1 2X,'(

(',12, ) ',1X,A7,F6.2))
(',12, ) MX,A7,F6.0))
,12,') ,1X,A7,I6)
,12,') ,1X,A7,F6.0,2(2X,'(',I2,') ',1X,A7,

(| ,12, ) |,1X,A7,F6.2))
\ t i2> t } fJ-XfA/fXa)/
(',12, ) ',1X,A7,F6.2))
,12,') ,1X,A7,F6.0,
,12,') ,1X,A7,F6.0,
,12,') ,1X,A7,F6.2)
,12,') ,1X,A7,F6.4,
,12,') ,1X,A7,A6,
,12,') ,1X,A7,F6.2)
(',12, ) ',1X,A7,F6.2)>
,12,') ,1X,A7,F6.2,
,12,') ,1X,A7,F6.0)

124

-------
124    C**** DETERMINE  IF CHANGES  TO  INPUT VALUES ARE TO BE HADE.
125          WRITE(6,200)
126      200 FORMAT)/,1H  ,'IF  NO CHANGES  ARE TO BE MADE IN VALUES IN TABLE',
127         1  ' 	 ENTER  99',/,lH  ,'IF  "STAB" IS TO BE CHANGED 	 ENTER1,
128         2  ' 98',/rlH  ,'IF  CHANGES  OTHER THAN TO "STAB" ARE TO BE MADE',
129         3  ' 	 ENTER TOT.  NUMBER  OF  CHANGES')
130          READ(5,*)  NU
131          IF(NU.EQ.99) GO TO 40
132          IF(NU.EQ.98) GO TO 30
133
134    C**** CHANGE VALUES  IN  TABLE.
135          WRITE(6,210)
136      210 FORMAT(/,1H  ,'SPECIFY ID.  NUMBER OF PARAMETER TO BE CHANGED1,
137         1  ' FOLLOWED BY THE NEW VALUE.',/,lH ,'ALL CHANGES ARE ENTERED',
138         2  ' SEPARATED BY COMMAS  (E.G.,  1,200.,2,50.,4,600., ETC).1)
139          READ(5,*)  (ID( J) , VALNEW( J) , J=l, NU)
140
141          DO 20 J=1,NU
142
143          IF(VALNEW(J).LT.VALLO(ID(J)).OR.VALNEW(J).GT.VALHI(ID(J)))THEN
144          WRITE(6,240) ID(J)
145      240 FORMAT(/,1H  ,'PARAMETER  NO.  ',12,' OUT OF RANGE, PLEASE',
146         1  ' RESPECIFY')
147          GO TO 20
148          ENDIF
149
150          VAL(ID(J))=VALNEW(J)
151
152          IF(ID(J).EQ.18)THEN
153            IF(ABS(VAL(18)>.GT.VAL(3)/2.)THEN
154            WRITE(6,250)
155      250   FORMATf/,1H  ,'STACK MUST BE ON VALLEY FLOOR-RESPECIFY YZERO1)
156            ENDIF
157          ENDIF
158
159          IF(ID(J).EQ.22)THEH
160            IF(VAL(22).LE.1.1THEN
161            VAL(22)=1.
162            WRITE(6,260)
163      260   FORMAT(/,1H  ,'DOWNVALLEY WINDSPEED AT STACK HAS BEEN SET',
164         +  ' TO 1 M/S')
165            ENDIF
166          ENDIF
167
168       20 CONTINUE
169          GO TO 10
170
171    C**** CHANGE STABILITY  DESIGNATION.
172       30 CONTINUE
173          WRITE(6,220)
174      220 FORMAT(/,1H  ,'SPECIFY NEW STABILITY: D, E OR F  ')
175          READ(5,230)  VALI21)
176      230 FORMAT(A4)
177          GO TO 10
178
179    C**** SET VALUES FOR EACH  PARAMETER.
180       40 CONTINUE
181    C SENSIBLE HEAT  FLUX
182          AO=VAL(1)
183          K=VAL(2)
184    C VALLEY PHYSICAL  CHARACTERISTICS
185          W=VAL(3)
186          ALPHA1=(VAL(4)/J60.)*6.28
187          ALPHA2=(VAL(5)/360.)*6.28
                                              125

-------
188    C MODEL CHARACTERISTICS
189          NBOX=INT(VAL(6))
190          RNO=NBOX
191    C LENGTH OF BOXES.
192          BOXLEN=.5*W/RNO
193    C TIME STEP
194          DELTAT=((W/2.)/RNO)/10.
195    C INVERSION CHARACTERISTICS
196          HZERO=VAL(7)
197          GAMMA=VAL(8)
198          BETA=VAL(9)
199    C CROSS-SECTIONAL AREA OF VALLEY
200          AC= HZERO*W+(.5*(HZERO)**2)*(1-/TAN(ALPHAl)+
201         + l./TAN(ALPHA2))
202    C TOTAL NUMBER OF WHOLE BOXES  IN  MODEL  AT INITIATION.
203          IFIX=(HZERO-HCBLI)/(BOXLEN*TAN(ALPHA1))
204          NTCTI=NBOX+IFIX
205          IF(NTOTI.GT.NB)THEN
206          WR1TE(6,280)
207      280 FORMAT(/,1H ,'TOO MANY GRID ELEMENTS—  PLEASE REDUCE NBOX1
208         + ' OR RE-DIMENSION THE MODEL  TO ALLOW MORE GRID  ELEMENTS')
209          GO TO 10
210          ENDIF
211    C SITE LOCATION
212          LAT=VAL(10)
213          URLONG=VAL(11)
214    C DATE
215          MO»INT(VAL(12))
216          IDA=INT(VAL(13))
217          IYR=INT(VAL(14))
218    C VALLEY ATMOSPHERE
219          TEMP=VAL(15)
220          PRESS=VAL(16)
221          RHO=PRESS/(TEMP+273.16)/2.8704
222    C GAUSSIAN PLUME
223          XC=VAL(17)
224          YZERO=VAI,(18)
225          UOVAL(19)
226          Q=VAL(20)
227          DO 50 J=4,6
228          IF(VAL(21).EQ.ASTAB(J))  GO  TO 60
229       50 CONTINUE
230          WRITE(6,290)
231      290 FORMAT(/,1H ,'YOU HAVE SPECIFIED  A WRONG STABILITY CLASS-
232         1 F WILL BE USED  INSTEAD')
233          J=6
234       60 ISTAB=J
235    C STACK CHARACTERISTICS
236          US=VAL(22)
237          H=VAL(23)
238          STMP=VAL(24)
239          SRAD=VAL(25)
240          SVEL=VAL(26)
241          AS=VAL(27)*1000.
242          LU=2
243          GO TO 10
244
245    c**** CONVERT DOWN-VALLEY DISTANCES TO  KILOMETERS
246       70 XC=XC/1000.
247
                                             126

-------
248    c**** CONVERT TEMPERATURES  TO  DEGREES
249          TEMP"=TEMP+273.16
250          STMP=STMP+273.16
251
252          RETURN
253          END
                                              127

-------
 2          SUBROUTINE JULIAN(MO,IDA,IYR,JULDAY)
 3    C     THIS SUBROUTINE CALCULATES THE JULIAN DATE GIVEN THE
 4    C     MONTH,  DAY, AND YEAR.
 5
 6          DIMENSION NDAY(12)
 7
 8          DATA NDAY/0,31,59,90,120,151,181,212,243,273,304,334/
 9
10          JULDAY=IDA+NDAY(MO)
11    C     ADJUST FOR LEAP YEAP
12          A=FLOAT(IYR)/4-IYR/4
13          IF(A.EQ.O. .AND. MO.GE.3) JULDAY=JULDAY+1
14
15          RETURN
16          END
                                            128

-------
 1
 2          SUBROUTINE PRISE(TEMP,SVEL,SPAD,STMP,XC,US,ISTAB,GAMMA,DLH)
 3    c**** THIS SUBROUTINE IS USED TO CALCULATE PLUME  RISE.   IT  IS
 4    C**** DERIVED FROM THE PLUME RISE ALGORITHM  IN THE  E.P.A. MODEL
 5    c**** "MPTER".
 6
 7    C**** DEFINITIONS **********************************************
 8    C  THE INPUT VARIABLES ARE:                                    *
 9    C      GAMMA  - POTENTIAL TEMPERATURE GRADIENT  (DEC K/M).      *
10    C      ISTAB  - P-G STABILITY (1,2,3,4,5 OR  C).                *
11    C      SVEL   - STACK EXIT VELOCITY  (M/S).                     *
12    C      SRAD   - INSIDE STACK RADIUS  (M).                       *
13    C      STMP   - STACK EXIT TEMPERATURE  (DEC  K).                *
14    C      TEMP   - AMBIENT AIR TEMPERATURE  (DEG K).               *
15    C      US     - AMBIENT WIND SPEED  (M/S).                      *
16    C      XC     - DISTANCE TO DOWNVALLEY CROSS SECTION  (KM).     *
17    C  THE OUTPUT IS:                                              *
18    C      DLH    - PLUME CENTER-LINE RISE ABOVE STACK  HEIGHT  (M). *
19    C***************************************************************
20
21          DATA G/9.8/
22
23          IF(SVEL.EQ.O..OR.SRAD.EQ.O.) THEN
24            DLH=0.
25            RETURN
26          ENDIF
27
28          X=XC*1000.
29          DELTT=STMP-TEMP
30          F«G*SVEL*SRAD*SRAD*DELTT/STMP
31
32    C            **COMPUTE FOR NEUTRAL-UNSTABLE
33          IF(ISTAB.LE.4) THEN
34            IF(F.GE.55.) THEN
35              DTC=0.00575*STMP*SVEL**.6667/((2.*SRAD)**.3333)
36            ELSE
37              DTC=0.0297*STMF*SVEL**.3333/((2.*SRAD)**.6667)
38            ENDIF
39            IF(DELTT.LE.DTC.OR.STMP.LE.TEMP) THEN
40    C            *MOMENTUM DOMINATED
41              DLH=6.*SRAD*SVEL/US
42            ELSE
43    C            *BCUYANCY DOMINATED
44              IF(F.GE,55.) THEN
45                XF=3.5*34.*F**(.4)
46              ELSE
47                XF=3.5*14.*F**(.625)
48              ENDIF
49              XX=AMIN1(X,XF)
50              DLH=1.6*(F**.3333)*(XX**.6667)/US
51            ENDIF
52
53    C            "COMPUTE FOR STABLE CONDITIONS
54          ELSE
55            S=G*GAMKA/TEMP
56            DTC=0.01958*TEMP*SVEL*SQRT(S)
57            IF(DELTT.LE.DTC.OR.STMP.LE.TEMP) THEN
58    C            *MOMENTUM DOMINATED
59              DLH1=6.*SRAD*SVEL/US
60              DLH2=(1.5*((SVEL**2*SRAD**2*TEMP)/(STMP*US))**,3333)
61         +         /(S**.1667)
62              DLH=AMIN1(DLH1,DLH2)
                                            129

-------
63            ELSE
64    C            *BOOYANCY DOMINATED
65              UCUT=((2.6/4.)**3)*(F**.25)*(S**.125)
66              IF(OS.GT.UCUT)  THEN
67                XF=((2.6/1.6)**1.5)*US/SQRT(S)
68              ELSE
69                XF-((4./1.6)**1.5)*(US**1.5)/((S**(9./16.))*(F**.125)]
70              ENDIF
71              XX=AMIN1(X,XF)
72              DLH-1.6*(F**.3333)*(XX**.6667)/US
73            ENDIF
74          ENDIF
75
76          RETURN
77          END
                                            130

-------
 1    C+++++-H-+++++++++++++++++++-M
 2          SUBROUTINE DILUTE(US,UC,AS,AC,DF)
 3    C**** CALCULATE DILUTION FACTOR TO ACCOUNT  FOR  CLEAN AIR DILUTION OF THE
 4    c**** NOCTURNAL POLLUTANT PLUME DURING  ITS  TRANSPORT DOWN THE VALLEY AS
 5    c**** A RESULT OF CLEAN AIR FLUX INTO THE VALLEY FROM TRIBUTARIES,  SLOPE
 6    C**** FLOWS, AND/OR ENTRAINMEST AT THE  TOP  OF THE DOWN VALLEY FLOW
 7    C**** LAYER.
 8
 9          IF(AS.EQ.O.)THEN
10          DF=1.
11          ELSE
12          DF=(UC*AC)/(US*AS)
13          IF(DF.LT.1.)DF=1.
14          ENDIF
15          RETURN
16          END
                                             131

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
DATA SIGGS/ 6.035, 2.1097,
      SUBROUTINE INGRAT(XC,YZ ERO,ISTABY,ISTABZ,W,ALPHA1,
     1           ALPHA2,HZERO,H,SUM1,SUM2,SIGMAY,SIGMAZ)
C**** CALCULATE GAUSSIAN DIFFUSION COEFFICIENTS AND DO  A
C**** CROSSWIND INTEGRATION WITHIN THE VALLEY  CROSS-SECTION
C**** TO CALCULATE THE  MASS OF POLLUTION  FLOWING THROUGH
C**** THE SECTION (KG/SEC).  SIMILARLY, DO A CROSS-WIND INTEGRATION
C**** ABOVE THE VALLEY  TO CALCULATE  THE MASS OF POLLUTION  FLOWING
C**** ABOVE THE VALLEY.  THE REMAINING MASS  (CALCULATED FROM  THE
C**** EQUATION OF CONTINUITY AS Q MINUS THESE  TWO QUANTITIES)  CAN
C**** BE FOLDED BACK INTO THE CROSS-SECTION.   THIS IS DONE IN THE
C**** MAIN PROGRAM UNDER AN INSTANTANEOUS PERFECT-MIXING
C**** ASSUMPTION.

      REAL SIGGY(18),SIGGZ(18)

      DATA SIGGY/ 5.357, 0.8828, -0.0076, 5.058, 0.9024, -0.0096,
     1            4.651, 0.9181, -0.0076, 4.230, 0.9222, -0.0087,
                  3.922, 0.9222, -0.0064, 3.533, 0.9181, -0.0070/
                  6.035, 2.1097,  0.2770, 4.694, 1.0629,   0.0136,
     1            4.110, 0.9201, -0.0020, 3.414, 0.7371, -0.0316,
     2            3.057, 0.6794, -0.0450, 2.621, 0.6564, -0.0540/

C**** DEFINITIONS *************************************************
C     H        PLUME CENTERLINE HEIGHT  (M)                         *
C     HZERO    INITIAL  INVERSION HEIGHT  (M)                        *
C     ISTABY   HORIZONTAL STABILITY  INDEX (1-6)                    *
C     ISTABZ   VERTICAL STABILITY INDEX  (1-6)                      *
C     SIGMAY   STD DEVIATION OF PLUME CONC IN  Y DIRECTION  (M)      *
C     SIGMAZ   STD DEVIATION OF PLUME CONC IN  Z DIRECTION  (M)      *
C     ALPHA1   VALLEY SIDEWALL SLOPE (RAD)                         *
C     ALPHA2   OTHER VALLEY SIDEWALL SLOPE  (RAD)                   *
C     W        FULL WIDTH OF VALLEY  FLOOR (M)                      *
C     XC       DOWNVALLEY DISTANCE OF CROSS  SECTION  (KM)           *
C     YZERO    OFF-CENTERLINE DISPLACEMENT OF  STACK  (M)            *
C     Y,Z      COORDINATE AXES  (M)                                 *
C     ZINC     VERTICAL HEIGHT  INCREMENT  FOP INTEGRATION  (M)       *
£******************************************************************


C**** CALCULATE THE SIGMAS USING MCMULLEN'S  METHOD  (.1975) **********
      SIGMAY-EXP(SIGGY(3*ISTABY-2)+SIGGY(3*ISTABY-l)*LOG(XC)+
     1       SIGGY(3*ISTABY)*(LOG(XC))**2)
      SIGMAZ = EXP(SIGGZ(3 *ISTABZ-2)+SIGGZ(3*ISTABZ-1)*LOG(XC) +
     1       SIGGZ(3*ISTABZ)*(LOG(XC))**2)

C**** INITIAL VALUES
      PI=3.14159
      P=100.
      NP=P
      ZINC=HZERO/P
      SUM1=0.

C**** RATE OF MASS FLOW ACROSS THE VALLEY SECTION
      DO 1 1=1,NP
      R=I
      Z=(R-1.)*ZINC
      F=(1./(SQRT(2.*PI)*SIGMAZ))*EXP(-.5*((Z-H)/SIGMAZ)**2)
      Yl=(W/2.+Z/TAN(ALPHAl)-YZERO)/SIGMAY
      Y2=(-W/2.-Z/TAN(ALPHA2)-YZ ERO)/SIGMAY
                                             132

-------
 60
 61           IF(Y1.LT.O.)THEN
 62           Y1=-Y1
 63           CALL NORMAL(Y1.PHIY1)
 64           PHIY1-1.-PHIY1
 65           ELSE
 66           CALL NORMAL(Y1,PHIY1)
 67           ENDIF
 68
 69           IF(Y2.LT.O.JTHEN
 70           Y2=-Y2
 71           CALL NORMAL(Y2,PHIY2)
 72           PHIY2=1.-PHIY2
 73           ELSE
 74           CALL NORMAL(Y2,PHIY2)
 75           ENDIF
 76
 77           PHIY=PHIY1-PHIY2
 78           SUM1=SUM1+PHIY*F*ZINC
 79         1 CONTINUE
 80
 81     C**** RATE OF MASS FLOW ABOVE THE CROSS SECTION.
 82           SUM2=0.
 83           DO 2 1=1,NP
 84           R=I
 85           Z=(R-1.)*ZINC*2.+HZERO
 86           ARG=-.5*((Z-H)/SIGMAZ)**2
 87           IFfARG.LT.-80.JTHEN
 88           F=0.
 89           ELSE
 90           F=(1./(SQRT(2.*PI)*SIGMAZ))*EXP(ARG)
 91           ENDIF
 92           Yl=5.
 93           Y2=-5.
 94
 95           IF(Yl.LT.O.JTHEN
 96           Y1=-Y1
 97           CALL NORMAL)Yl,PHIYl)
 98           PHIY1=1.-PHIY1
 99           ELSE
100           CALL NORMAL(Y1,PHIY1)
101           ENDIF
102
103           IF(Y2.LT.O.JTHEN
104           Y2=-Y2
105           CALL NORMAL(Y2,PHIY2)
106           PHIY2=1.-PHIY2
107           ELSE
108           CALL NORMAL(Y2,PHIY2)
109           ENDIF
110
111           PHIY=PHIY1-PHIY2
112           SUM2=SUM2+PHIY*F*ZINC*2.
113         2 CONTINUE
114
115           RETURN
116           END
                                            133

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
      SUBROUTINE NORMALfX,PHI)
C**** CALCULATE THE INTEGRAL OF THE AREA UNDER THE GAUSSIAN CURVE
C**** FROM MINUS INFINITY TO X.  THE POLYNOMIAL APPROXIMATION
C**** USED HERE COMES FROM ABRAMOWITZ AND STEGUN, 1965,
C**** HANDBOOK OF MATHEMATICAL FUNCTIONS, DOVER PRESS, NINTH
C**** PRINTING, P. 932.  THEY ATTRIBUTE THE FORMULA TO HASTINGS  (1975)

      REAL C(6)

      DATA C/0.2316419, 0.31938153, -0.356563782,
     1    1.781477937, -1.821255978, 1.330274429/

      PI=3.14159
      T-1./(1.+C(1)*X)
      SIGH-(l./SQRT(2.*PI))*EXP(-X**2/2.)
      PHI=1.-SIGH*(C(2)*T+C(3)*T**2+C(4)*T**3+C(5)*T**4+C(6)*T**5)

      RETURN
      END
                                            134

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
      SUBROUTINE GAUSS ( Y, Z,YZERO,H,Q, OS, DF,SIGMAY,SIGMAZ , CHI)
C**** THIS PROGRAM USES THE WELL-KNOWN GAUSSIAN PLUME EQUATIONS
C**** (TURNER, 1969) TO CALCULATE VALLEY POLLUTION CONCENTRATION
C**** DOWNVALLEY OF A POLLUTANT SOURCE.

      DEFINITIONS *************************************************
      CHI      POLLUTANT CONCENTRATION (MICROGRAMS/M**3)          *
      CHIOQ    CHI OVER Q (S/M**3)                                *
      CHIOQN   NORMALIZED AND STP CORRECTED .CONCENTRATION  (S/M**3)*
      DF       DILUTION FACTOR                                    *
      H        PLDME CENTERLINE HEIGHT (M)                        *
      Q        SOURCE STRENGTH (KG/S)                             *
      SIGMAY     SIGMA Y (M)                                      *
      SIGMAZ     SIGMA Z (M)                                      *
      US       WINDSPEED AT SOURCE (M/SEC)                        *
      YZERO    OFF-CENTERLINE DISPLACEMENT OF STACK  (M)           *
      XC,Y,Z   COORDINATES OF RECEPTOR (KMFM,M)                   *
C
C
C
C
C
C
C
C
C
C
C
C**** CALCULATE INDIVIDUAL TERMS IN THE GAUSSIAN PLUME EQUATION ***
C        Gl : OFF CENTERLINE IN Y DIRECTION                       *
C        G4 : OFF CENTERLINE IN Z DIRECTION                       *
£******************************************************************

C**** GAUSSIAN PLUME EQUATION
      G1=EXP(-0.5*((Y-YZERO)/SIGMAY)**2)
      G4=EXP(-0.5*((Z-H)/SIGMAZ)**2)
      CHIOQ=1./(2.*3.14*SIGMAY*SIGMAZ*US*DF)*(G1)*(G4)

C**** CONCENTRATION IN MICROGRAMS PER CUBIC METER
      CHI-CHIOQ*Q*1.E09

      RETURN
      END
                                            135

-------
 2          SUBROUTINE SOLAR(LAT,URLONG,JULDAY,NTSR,TAU,Al)
 3    C**** CALCULATE TIME OF SUNRISE, LENGTH OF DAY, AND  SOLAR  FLUX ON
 4    c**** HORIZONTAL SURFACE AT SOLAR NOON FOR ANY SITE  GIVEN  LATITUDE,
 5    C**** LONGITUDE, AND JULIAN DATE. THIS SUBROUTINE USES  KCCULLOUGH'S  1968
 6    C**** APPROXIMATIONS AS GIVEN IN ARCH. METEOR. GEOPHYS. BIOKLIMAT.,
 7    C**** SER. B, 16, 129-143.  REFER TO WHITEMAN, 1980, ATMOSPHERIC
 8    C**** SCIENCE PAPER NO. 328, COLORADO STATE UNIVERSITY, FORT
 9    C**** COLLINS, COLORADO, 250 PP.
10
11          REAL EQNTIM(25)
12          REAL LAT,LONG,LONCOR
13
14    C**** EQUATION OF TIME IN HRS FAST OR SLOW.   .HHHH VALUES  GIVEN FOR
15    c**** EVERY 15 JULIAN DAYS, FOR EXAMPLE, 1,16,31,46	..360.
16          DATA EQNTIM/-0.0533,-0.1589,-0.2233,-0.2378,-0.2064,-0.1442,
17         1             -0.0689,+0.0003,+0.0475,+0.0622,+0.0425,-0.0028,
18         2             -0.0558,-0.0961,-0.1058,-0.0789,-0.0186,+0.0639,
19         3             +0.1517,+0.2258,+0.2683,+0.2647,+0.2094,+0.1094,
20         4             -0.0131/
21
22    C**** CONSTANTS
23    C JULIAN DATE OF VERNAL EQUINOX.
24          DZERO=80.
25          PI-3.14159
26    C CONVERSION-DEGREES TO RADIANS.
27          CONV=PI/180.
28    C ECCENTRICITY OF EARTHS ORBIT
29          ECCENT=.0167
30    C MAX DECLINATION.
31          DECMAX=(23.+27./60.)*PI/180.
32    C REVOLUTION RATE OF EARTH.
33          OMEGA=2.*PI/365.
34    C SOLAR CONSTANT IN WATTS/M**2
35          SC=1367.
36    C 15 DEGREES OF LONGITUDE  IS EQUIVALENT TO 1  HOUR.
37          ONEHR=15.*PI/180.
38    C LONGITUDE OF STANDARD MERIDIAN FOR YOUR TIME ZONE.
39          STDLON=105.
40          LAT=LAT*CONV
41
42    C JULIAN DATE
43          D=JULDAY
44
45    c**** MAKE COMPUTATIONS.
46          OMD=OHEGA*D
47          OMDZRO=OMEGA*DZERO
48    C RATIO OF RADIUS VECTORS  SQUARED. ACCOUNTS FOR VARYING DISTANCE
49    C BETWEEN EARTH AND SUN.
50          RDVCSQ=1./(1.-ECCENT*COS(OMD))**2
51          LONG=OMEGA*(D-DZERO)+2.*ECCENT*(SIN(OMD)-SIN(OMDZRO))
52          DECLIN=A  SIN(SIN(DECMAX)*SIN(LONG))
53          SR=-ABS(ACOS(-TAN(LAT)*TAN(DECLIN)))
54    C LENGTH OF DAYLIGHT PERIOD
55          TAU=-SR*2./ONEHR
56
57    c**** CALCULATE TIME OF SOLAR NOON AT SITE AS A PRELIMINARY TO  CALCULATING
58    c**** THE LOCAL TIME FOR ANY GIVEN HOUR ANGLE.
59          LONCOR=(STOLON-URLONG)*1./15.
60          IF(D.GT.361.) GO TO  1
61          ID=((D-1.)/15.)+1.
62          D2=(ID-1)*15+1
63          TIMCOR=(EQNTIM(ID+1)-EQNTIM(ID))*(D-D2)/15.+EQNTIM(ID)
                                            136

-------
64        1 IF(D.EQ.362.) TIMCOR=-0.0211
65          IF(D.EQ.363.) TIMCOR=-0.0292
66          IF(D.EQ.364.) TIMCOR=-0.0372
67          IF(D.EQ.365.) TIMCOR=-0.0453
68          TMNOON-12.00-LONCOR-TIMCOR
69
70    C**** CALCULATE SOLAR FLUX AT SOLAR NOON
71          COSZ=SIN(LAT)*SIN(DECLIN)+COS(LAT)*COS(DECLIN)*1
72    C SOLAR FLUX AT SOLAR NOON
73          A1»SC*RDVCSQ*COSZ
74          IF(A1.LT.0.) Al=0.0
75          T1-(12.+(SR/ONEHR)+(TMNOON-12.00))
76          IT=T1
77          T2 = IT
78    C TIME OF SUNRISE.
79          NTSR-(T1-T2)*60.+(T2*100.)
80
81          RETURN
82          END
                                           137

-------
 1
 2          SUBROUTINE EBDGT(AO,A2)
 3    c**** DETERMINE FRACTION OF EXTRATERRESTRIAL  SOLAR  FLUX  THAT PRODUCES
 4    C**** SURFACE SENSIBLE HEAT FLUX.
 5    c**** AT PRESENT, THIS SUBROUTINE IS NOT BEING  USED,  SINCE  THE USER NOW
 6    c**** SIMPLY INPUTS FRACTION AO.  IN FUTURE,  THIS SUBROUTINE WOULD BE
 7    c**** DEVELOPED TO EXPLICITLY INCORPORATE  FEATURES  OF THE SURFACE ENERGY
 8    C**** BUDGET SUCH AS ALBEDO, SOIL MOISTURE AND  CLOUD  COVER.
 9
10          A2=AO
11
12          RETURN
13          END
                                             138

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
j t\
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
C++-f++++-H"t-+++++++++ +++++++++++++++++-f +++++++++++ +++++++++++++++++
SUBROUTINE DBS CUT (AO , Al , ALPHA1 , ALPHA2 , BETA, DELTAT, GAMMA,
1 HCBLI,HZERO,K,W, PRESS, RHO, TAU, NSTEPS)
C**** THIS PROGRAM CALCULATES THE HEIGHTS OF THE INVERSION TOP
C**** AND CBL TOP AS A FUNCTION OF TIME AFTER SUNRISE AS VALLEY
C**** INVERSIONS ARE DESTROYED, USING WHITEMAN'S (1980)
C**** NUMERICAL METHOD AS GIVEN IN HIS EQUATIONS 41-50.

PARAMETER (NS=6000)

COMMON/BLK1/HC(NS) ,HT(NS)

REAL K

C AO FRACTION (0-1) OF Al CONVERTED TO SENSIBLE HEAT
C Al SOLAR FLUX ON HORIZ SFC AT SOLAR NOON (W/M2)
C ALPHA1 SIDEWALL NO. 1 INCLINATION ANGLE (RAD)
C ALPHA2 SIDEWALL NO. 2 INCLINATION ANGLE (RAD)
C BETA RATE OF TEMP CHANGE AT INVERSION TOP (K/SEC)
C CP SPECIFIC HEAT AT CONSTANT PRESSURE (J/KG/K)
C DELTAT TIME STEP (SEC)
C GAMMA VERT POT TEMPERATURE GRADIENT IN INVERSION (K/M)
C HC CBL HEIGHT (M)
C HCBL HEIGHT OF CBL TOP (M)
C HCBLI INITIAL CBL HEIGHT (M)
C HT INVERSION TOP HEIGHT (M)
C HTOP HEIGHT OF INVERSION TOP (M)
C HZERO INITIAL HEIGHT OF INVERSION TOP (M)
•++++•













FLUX
*
*
*
Ik
*
*
*
*
*
It
*
*
*
C K FRACTION (0-1) OF SENS HT FLUX GOING INTO CBL GROWTH
C N TIMESTEP COUNTER
C NSTEPS NUMBER OF TIME STEPS RQD TO DESTROY INVERSION
C PRESS AVG PRESSURE (MB) AT INV CENTER AT SUNRISE
C PI TRIGONOMETRIC CONSTANT
C RHO AIR DENSITY (KG/M3)
C TAU LENGTH OF DAYLIGHT PERIOD (SEC). 43200SEC=12HR
C T ELAPSED TIME SINCE SUNRISE (SEC)
C TT TIME OF SUNRISE (SEC)
*
*
it
*
*
*
*
*
C W VALLEY FLOOR WIDTH (M) *
°
C**** INPUT PARAMETERS
CP=1005.
PI=3.14
N=0
TI=0.
T=0.

C**** CALCULATE SECONDARY PARAMETERS
FACT1=AO*A1/(RHO*CP)
THTOT=(1000./PRESS)**.286
C=(1./TAN(ALPHA1))+(1./TAN(ALPHA2))

c**** D0 CALCULATIONS
HTOP= HZERO
HCBL=HCBLI
GO TO 2
1 CONTINUE
T=T+ DELTAT
2 CONTINUE
N=N+1
FACT2=FACT1*SIN(PI*(T-TI)/TAU)























139

-------
64          IF(K.NE.O.) THEN
65             DHDT=FACT2*THTOT*K*(W+HCBL*C)/
66         1   (GAMMA*HCBL*(W+(HCBL*C)/2.))
67          ELSE
68             DHDT=0.
69          ENDIF
70
71          FNUM=(W+HTOP*C-K*(W+HCBL*C))*FACT2-((BETA*
72         l(HZERO-HTOP)*(W+( (HZERO+HTOP)/2.)*C) )/{2.*THTOT) )
73          FDENOM=HTOP*GAMMA*(W+(HTOP*C)/2.)
74         l+(BETA*(T-TI)*(W+HTOP*C))/2.
75          DHTDT=-THTOT*FNUM/FDENOM
76          DHT=DHTDT*DELTAT
77          HTOP=HTOP+DHT
78          HT(N)=HTOP
79          DHC=DHDT*DELTAT
80          HCBL=HCBL+DHC
81          HC(N)=HCBL
82
83    c**** STOP CALCULATIONS WHEN INVERSION TOP SINKS BELOW CBL  TOP
84          IF(HTOP.LE.HCBL) THEN
85             NSTEPS=N
86             RETURN
87          ELSE
88             GO TO 1
89          ENDIF
90
91          END
                                             140

-------
 1
 2          SUBROUTINE PROFIL(NBOX,NTOTI,NTOT,BOXLEN,HCL,ALPHA1,
 3         1ALPHA2,HZERO,H,HCU,HTL,HTU,YZ ERO,Q,OS,DF,
 4         2SIGMAY,SIGMAZ,CHIOFF)
 5    C**** THIS SUBROUTINE DETERMINES THE AVERAGE POLLUTANT  CONCENTRATION
 g    c**** INJECTED INTO THE TOP OF EACH BOX.
 7
 8          PARAMETER (NB=30)
 9
10          COMMON/BLK2/CHIBAR(NB)
11
12    c**** DEFINITIONS ***************************************************
13    C     HCL    CBL HEIGHT AT LOWER TIME STEP (M)
14    C     HCU    CBL HEIGHT AT UPPER TIME STEP (M)
15    C     HTL    INVERSION TOP HEIGHT AT LOWER TIME STEP  (M)
16    C     HTU    INVERSION TOP HEIGHT AT UPPER TIME STEP  (M)
17    C     Yl     Y COORD AT LEFT SIDE OF SINKING MASS ELEMENT  (M)
18    C     Y2     Y COORD AT RIGHT SIDE OF SINKING MASS ELEMENT  (M)
19    C     Zl     EFFECTIVE Z COORD AT BOTTOM OP SINKING MASS ELEMENT  (M)
20    C     Z2     EFFECTIVE Z COORD AT TOP OF SINKING MASS ELEMENT  (M)
2i    £********************************************************************
22
23          ALPHA=(ALPHAl+ALPHA2)/2.
24
25    C**** CALCULATE CHIBAR FOR THE BOXES ON THE VALLEY FLOOR.
26          DO 1 I=1,NTOTI
27          Y1=(I-1)*BOXLEN
28          Y2-I*BOXLEN
29          Y»(Yl+Y2)/2.
30
31          IF(I.GT.NTOT) THEN
32             CHIBAR(I)-999.9
33             GO TO 1
34          ELSE IFd.GT.NBOX .AND. I.LE.NTOT) THEN
35             Zl=HCL-l"HZERO-nTL-l-(FLOAT(I-(NBOX+l))+.5)*BOXLEN*TAN(ALPHA)
36             Z2=HCU+HZERO-HTU+(FLOAT(I-(NBOX+1))+ .5)*BOXLEN*TAN(ALPHA)
37          ELSE
38             Z1=HZERO-HTL+HCL
39             Z2=HZERO-HTU+HCU
40          ENDIF
41
42          Z=(Zl+Z2)/2.
43          CALL GAUSS(Y,Z1,YZERO,H,Q,US,DF,SIGMAY,SIGMAZ,CHIZ1)
44          CALL GAUSS(Y,Z2,YZERO,H,Q,US,DF,SIGMAY,SIGMAZ,CHIZ2)
45          CALL GAUSS(Y1,Z,YZERO,H,Q,OS,DF,SIGMAY,SIGMAZ,CHIY1)
46          CALL GAUSS(Y2,Z,YZERO,H,Q,US,DF,SIGMAY,SIGMAZ,CHIY2)
47          CHIBAR(I)=(CHIZH-CHIZ2+CHIYl+CHIY2)/4. + CHIOFF
48
49        1 CONTINUE
50          RETURN
51          END
                                            141

-------
 1
 2          SUBROUTINE VELOCY(N,MN,NTOTI,NBOX,NTOT,BOXLEN,DELTAT)
 3    c**** THIS SUBROUTINE CALCULATES VELOCITY  AT  EACH  SEPTUM BASED ON
 4    C****  MASS CONTINUITY.
 5
 6          PARAMETER  (NS=6000,NB1=31)
 7
 8          COMKON/BLK1/HC(NS),HT(NS)
 9          COMMON/BLK3/V(NB1)
10
11          V(l)=0.
12          CIH=HT(MN)-HT(N)
13          IF(NTOT.EQ. NBOX. AND. HT(N) .LE.HC(N))  GO  TO 3
14
15          DO 2 I=1,NTOTI
16          IF(I.GT. KTOT) GO  TO 1
17          V( I-H) = (I *BOXLEN*CIH/DELTAT) /HC (N)
18          GO TO 2
19        1 V(I+1)=999.9
20        2 CONTINUE
21          GO TO 5
22
23        3 DO 4 I=1,NTOTI
24          V(I+1)=999.9
25        4 CONTINUE
26
27        5 RETURN
28          END
                                            142

-------
 1
 2          SUBROUTINE BRKUP(N,UN,NTOTI,BOXLEN,DELTAT,CCH)
 3    c**** THIS SUBROUTINE CALCULATES  POLLUTANT  CONCENTRATIONS IN EACH
 4    c**»* op THE GRID ELEMENTS  (BOXES) USING  A  POLLUTANT  MASS BALANCE.
 5
 6          PARAMETER  (NS=6000,NB=30,NB1=J1)
 7
 8          REAL CCH(2,NB)
 9
10          COMHON/BLK1/HC(MS),HT(NS)
11          COHMON/BLK2/CHIBAR(NB)
12          COMMON/BLK3/V(NB1)
13
14    C*DEFINITIONS **********************************************************
15    C     A IS THE POLLUTANT MASS WITHIN  THE  BOX  AT  THE PREVIOUS TIME STEP.
16    C     B IS THE MASS COMING  FROM THE ADJACENT  DOWNHILL BOX.
17    C     C IS THE MASS COMING  INTO TOP OF BOX  DUE TO SINKING INVERSION
18    C       AND/OR GROWING CBL
19    C     D IS THE MASS GOING OUT  UPHILL  SIDE OF  THE BOX.
20    C     E IS THE MASS SOURCE  OR  SINK FROM SURFACE  OF BOX.                  *
21    c************************************************************************
22
23          E=0.0
24          DELTAH=HC(N)-HC(MN)+HT(MN)-HT(N)
25          DO 4 1=1,NTOTI
26          IF(V(I+1).EQ.999.9) GO TO 3
27          A=CCH(1,I)*BOXLEN*HC(MN)
28          IF(I.EQ.l)  GO TO 1
29          B=D
30          GO TO 2
31        1 B=0.0
32        2 C=CHIBAR(I)*DELTAH*BOXLEN
33          D=(DELTAT/BOXLEN*V(I+1)+(HC(N)-HC(MN))/HC(N))*A
34         1 -HDELTAT/BOXLEN*V(I+1))*(B+C)
35          CCH(2,I)=(A+B+C+E-D)/(BOXLEN*HC(N))
36          GO TO 4
37    C**** EXPONENTIAL DECAY
38        3 S=ALOG(.90)
39          CCH(2,I)=CCH(1,I)*EXP(S*DELTAT/10.)
40        4 CONTINUE
41
42          RETURN
43          END
                                             143

-------
 1
 2          SUBROUTINE PSTPRC{TSTP,NTOTI,NAS)
 3    c**** THIS SUBROUTINE PROCESSES THE TIME  SERIES  OF CALCULATED
 4    c**** 5-MINUTE (300 S) AVERAGE CONCENTRATIONS  TO OBTAIN MAXIMUM
 5    c**** SHORT TERM (1 AND 3 HOUR) AVERAGE CONCENTRATIONS WHICH CAN
 g    c**** BE COMPARED TO REGULATORY STANDARDS.
 7
 8          PARAMETER (NA=100,NB=30,NB1=31)
 9
10          REAL SUMM(NB,2)
11
12          COMMON/BLK4/CONC(NA,NB)
13          COMMON/BLK5/CHI(NB)
14          COMMON/BLK6/AVG(NB,2),NDXTIM(NB,2),NTS(2)
15
16    c**** TIME IN SECONDS
17          ONEHR-3600.
18          THREE«10800.
19
20    c**** NUMBER OF TIMESTEPS
21          NTS(1)=OHEHR/TSTP
22          NTS(2)=THREE/TSTP
23          NSTART=NAS+1
24
25    c**** CALCULATE MAX 1-HR AND 3-HR  AVERAGE CONCENTRATIONS
26          DO 7 K=l,2
27          TS=NTS(K)
28          DO 7 1=1,NTOTI
29          SUM=0.
30
31    c**** AVERAGING INTERVAL LESS THAN  LENGTH OF DAYTIME CONC SERIES.
32          IFfNSTART.GT.NTS(K))  THEN
33             DO 1 N=HAS,NSTART-NTS(K),-1
34             SUM=SUM+CONC(N,I)
35        1    CONTINUE
36             SUMM(I,K)=SDM
37             Q1=SUM
38             NTIMES=NAS
39             DO 2 N=NAS,NTS(K)-H,-1
40             SUMM(I,K)=SUMM(I,K)+CONC(N-NTS(K),I)-CONC(N,I)
41             IF(SUMM(I,K).GT.Ql)THEN
42             Q1=SOMM(I,K)
43             NTIMES=N-1
44             ENDIF
45        2    CONTINUE
46             DO 3 N=NTS
-------
64             Q1=SOMM(I,K)
65             NTI«ES=N-1
66             ENDIF
67        6    CONTINUE
68          ENDIF
69
70          NDXTIM(I,K)=NTIHES
71          AVG(I,K)=Q1/TS
72        7 CONTINUE
73
74          HETURN
75          END
                                            145

-------
                   APPENDIX  B



FORTRAN LISTING OF VALMET OUTPUT PLOTTING PROGRAM
                        147

-------
 1     C      PROGRAM PLOTT.FOR
 2
 3     C     THIS PR06RAM IS DESIGNED  TO  PRODUCE PLOTS OF CONCENTRATION
 4     C     VERSUS TIME FOR ARBITRARY 6RID  ELEMENTS  IN MODEL  VALMET.
 5     C     THIS PROSRAfl ALSO PRODUCES A SINGLE PA6E CONTAININ6 THREE OTHER PLOTS:
 6     C     CONC VS HEIBHT, CONC VS X-VALLEY  DISTANCE, AND HEI6HT VS TIME.
 7
 3           PARAMETER ,CONCEN(NA),)C80RD(5),
11          1  YBORD(5),NBOX(10),l(MAn(10),YMAn!10)
12           DIMENSION X1(4),K2(4),Y1(4),Y2(4),ICNT!3,2)
13          1  ,T(3,4!,Z(1000),TIC(3,2},
14          2  YU000),CHIZUOOO>,CHIY(1000>,HC(NS),HT(NS),
15          3  TIME!NS),SZ1(3},SZ2(3),TIK(3,2)
16
17           CHARACTER120 INFILE
18           CHARACTERS ITIME
19           CHARACTERI2 ICHARI30)
20           CHARACTER126 ILBL(3,2)
21           CHARACTER}! ANSi,ANS2
22           CHARACTERI2 JLBL1
23           CHARACTERJ5 JLBL2
24           CHARACTERI5 JLBL3
25
26           DATA Xl/3.0,2.5,5.5,5.5/
27           DATA X2/10.0,4.0,10.0,10.0/
28           DATA Yl/3.0,5.0,5.0,2.0/
29           DATA Y2/7.5,8.0,8.0,3.5/
30
31           DATA ICHAR/'Or,'02','03','04','05','06','07','08',
32          1           '09','10','11','12','13','14','15','16',
33          2           '17','18','19','20','21','22','23','24',
34          3           '25','26','27','28','29','30'/
35
36           SZH1K35
37           SZ1(2)=.35
38           SZ1(3)=.70
39           SZ2(1)=.40
40           SZ2(2)=.40
41           SZ2(3)=.80
42
43           1CNT(1,1)=12
44           ICNT(2,1)=10
45           ICNT(3,l)=2i
46           ICNT(1,2)=14
47           ICNT!2,2!=14
48           ICNT!3,2)=12
49
50           ILBLU,1!='CONC (ug/»3)'
51           ILBL!2,l)='TIt1E (LOT)'
52           ILBL(3,1)='CROSS-VALLEY DISTANCE  !k«)'
53           ILBL(1,2)='HEIGHT  !• A6L)'

                                        149

-------
 54           ILBL(2,2)='H£IBHT  !a  A6L)'
 55           ILBL(3,2!='CQNC  (ug/§3)'
 54
 57           TIC(l,l)=i.O
 58           TIC!1,2)=200.
 59           TIC(2,1)=1.0
 60           TIC(2,2)=200.
 61           TIC(3,l)=i.O
 62           TIC<3,2)=1.0
 63
 44           TIK(1,1)=0.5
 65           TIKU,2)=100.
 66           TIK(2,1)=1.0
 67           TIK(2,2)=iOO.
 68           TIK(3,1)=0.5
 69           TIK(3,2)=0.5
 70
 7!           HR!TE!6,1QO)
 72       100  FORHATtlH  ,'UHAT IS THE NAHE OF THE DATA FILE?')
 73           READ(5,'(A20)')INFILE
 74           QPEN
-------
107       501  FORMAT(Al)
108           IF(ANS2.NE.'Y')  60  TO  502
109           WR!TE(6,1Q3>
110       103  FORMAT(1H ,'THE  FIRST  PLOT  IS CONCENTRATION VS TIME')
111           XMAX=TTIME(1)+NINDXtDELTAT/3600.
112           HRITE!6,104)  NTSR,XHAX,CLCONC
113       104  FORMATUH ,'NIN  AND MAX VALUES ARE:  TIME  ',I5,2X,
114          1F8.2,5K,'CONCENTRATION 0.0',2X,F4.2)
115           WRITE(6,105)
Hi       105  FORMATdH ,'ENTER T«IN,TMAX,CONCMIN,CONCMAX')
117           READ(5,I> XMIN,XMAX,YHIN,YHAX
118           XBQRD!1)=XHIN
119           XBQRD(2)=XNAX
120           IBORDttMHAX
121           XBORD(4>=XNIN
122           XBORD(5)=X«IN
123           YBORD(1)=YMIN
124           YBORD(2)=YMIN
125           YBORD(3)=YMAX
126           YBORD!4)=YMAX
127           YBORD(5)=YMIH
128           WRITE(6,106)  HTOTI
129       106  FQRMATilH ,'THERE ARE  ',12,' GRID  ELEMENTS')
130           WRITE (6,107)
131       107  FQRHATUH ,'HO«  MANY CURVES DO YOU HAMT TO PLOT?')
132           READ(5,t) NC
133           WRITE!6,108)
134       108  FORHATIJH ,'CURVES  ARE TO BE DRAWN FOR WHICH 6RID ELEMENT
135          1 NUMBERS?  EX:3,6,9')
136           READ(5,I) (NBOX(N),N=1,NC)
137
138     Cttll  DRAW THE PLOTS
139           CALL KTERH(IMODE)
140           CALL KBE6IN
141           CALL KPA6E(X1(1),X2(1),Y1(1),Y2(1),0)
142           CALL KSCALE(XMIN,XMAX,0,YMIN,YHAX,0)
143           CALL KAXIS(XMIN,YMIN,XMAX,1.0,0.5,0,0.0,1,0.35,0.,
144          1ILBLI2,!),10,1,0.40,0.)
145           CALL KAXISUMIN,YNIN,YHAX,0.2,0.1,1,0., 1,0.35,0.,
146          1ILBL(3,2),12,1,0.40,0.)
147           CALL KCURVEC(BDRD,Y80RD,5)
148           IMAXI(N)=0.
149           YMAXI(N)=0.
150           DO 3 N=1,NC
151           DO 2 MN=1,NAS+1
152           CONCEN(MN)=CONC(«N,NBQX!N»
153           IF'MN.EQ.l)  SO TO 14
154           IF(CONCEN(«N).ST.CONCEN(MN-D) THEN
155             XMAXI!N)=TTIME!MN)
156             YMAXI!N)=CONCEN(MN)
157           ENDIF
!53        14  CONTINUE
15"         2  CONTINUE
                                         151

-------
160           HRITE(6,109)  XHAXI(N),YMAXI!N)
161       109 FORMAT UH ,2F8.2)
162           CALL KLINE(N,0,0.,0.)
163           CALL KCURVE(TTIME,CONCEN,NAS+i)
164         3 CONTINUE
165           DO 4 N=1,NC
166           YNAXI(N}=YNAmN}+.10l
167           HRITE«6,HO)  Y«AXI(N)
168       110 FORMAT(F8.2)
169         4 CONTINUE
170           DO 5 N=1,NC
171           !=N80X(N)
172           »RITE(6,111)  I.ICHARd)
173       111 FORMAT(1H ,I3,A2)
174           CALL KUALFA(XHAXI(N),Y?1AXI(N),ICHAR(I),
175          12,1,0.35,0.,0.)
176         5 CONTINUE
177           IX=X
178           XA=XH!N+.74
179           XB=XMIN+1.45
180           XC=XHIN+l.i5
181           XPOS1=XNIN+0.4
182           YPOSl=.9tYf!AX
183           YPOS2=.8»YMAK
184           YPQS3=.7tYHAX
185           HTL=0.4
186           HRITE(JLBL1,300)  IX
187       300 FORHAT(I2)
138           WITE(JLBL2,30i)  CLCONC
139       301 FOR«AT(F5.2)
190           URITE
-------
213           TINE(HN)=TTI«E(1)
214           ELSE IF!!1N.E9.2)THEN
215           TIME(nN)=TTIKE(2!
216           ELSE
2i7           TINE(HN)=TTIME(2)HMN-2)tDELTAT/3600.
218           EHDIF
219         i CONTINUE
220
221     C     WRITE
-------
266           CALL K8E6IN
267           DO 12 H=l,3
268           CALL KPASEai(Htl),X2(«tl),YHH+l),Y2«N+l)tO)
269           CALL KSCALE(T(H,l),T(n,2),0,T(l1,3),T!N,4},0)
270           mjRD(l)=T(N,l)
271           XBOR»(2)=T(«,2)
272           JBORD(3)=T(«,2)
273           HBORD(4)=T(H,1)
274           XBQRD(5)=T
230           CALL KLINE(1,0,0.0,0.0)
281           CALL KCURVEUBORD,YBORD,5)
282           CALL KAXIS,0.)
286           IF(n.EQ.l) THEN
287             CALL KLINEI4,0,0.0,0.0)
288             CALL KCURVE!CHIZ,Z,K)
289           ELSE IF(H.E9.2)  THEN
290             CALL KLINEI4,0,0.0,0.0)
291             CALL KCURVE(TIHE,HC,NSTEPS)
292             CALL KLINE(2,Q,0.0,0.0)
293             CALL KCURVE(TIHE,HT,NSTEPS)
294           ELSE IFIN.EQ.3)  THEN
295             CALL KLINE!4,0,0.0,0.0)
296             CALL KCURVE!Y,CHIY,J>
297           ELSE
298             STOP 4
299           ENDIF
300        12 CONTINUE
301        13 CONTINUE
302           STOP 1
303           END
                                         154

-------
                   APPENDIX  C

RESEARCH PAPER ENTITLED "BREAKUP OF TEMPERATURE
     INVERSIONS  IN  DEEP  MOUNTAIN  VALLEYS:
         PART  II.   THERMODYNAMIC  MODEL"
                      155

-------
   Reprinted from JOURNAL OF APPLIED METEOROLOGY, Vol. 21, No. 3, March 1982
                      Amman Mcuorotofical Society
                          Pnmod m U. S. A.
Breakup of Temperature Inversions in Deep Mountain Valleys:
               Part II. Thermodynamic Model
                     C. DAVID WHITEMAN
                      THOMAS B. McKEE
                           157

-------
290
JOURNAL OF APPLIED  METEOROLOGY
                                                                   VOLUME 21
                Breakup of Temperature Inversions in Deep Mountain Valleys:

                                  Part II. Thermodynamic  Model


                                         C. DAVID WHITEMAN

                                Pacific Northwest Laboratory, Richland, WA 99352


                                          THOMAS B. McKEE

                   Department of Atmospheric Science. Colorado State University. Fort Collins 80523

                        (Manuscript received 31 March  1981, in final form 7 December 1981)

                                               ABSTRACT

             A thermodynamic model is developed to simulate the evolution of vertical temperature structure during
           the breakup of nocturnal temperature inversions in mountain valleys. The primary inputs to the model are
           the valley floor width, sidewail inclination angles, characteristics of the valley inversion at sunrise, and an
           estimate of sensible heat flux obtained from solar radiation calculations. The outputs, obtained by a numerical
           integration of the model equations, are the time-dependent height of a convective boundary layer that grows
           upward from the valley floor after sunrise, the height of the inversion top, and vertical potential temperature
           profiles of the valley atmosphere. The model can simulate the three patterns of temperature structure
           evolution observed in deep valleys of western Colorado. The well-known inversion breakup over flat terrain
           is a special case of the model, for which valley floor width becomes infinite. The characteristics of the model
           equations are investigated for several limiting conditions using the topography of a reference valley and
           typical inversion and solar radiation characteristics. The model is applied to simulate observations of inversion
           breakup taken in Colorado's Eagle and Yampa Valleys in different seasons. Simulations are obtained by
           fitting two constants in the model, relating to the surface energy budget  and energy partitioning, to the
           data. The model accurately simulates the evolution of vertical potential temperature profiles and predicts
           the time of inversion destruction.
1. Introduction

  In Part I (Whiteman, 1982) observations of ver-
tical temperature structure and wind evolution were
summarized for 21 case studies of nocturnal tem-
perature inversion breakup in deep Colorado moun-
tain valleys. Breakup occurred following one of three
patterns of vertical temperature structure evolution.
A hypothesis was offered to explain the observations
in which sensible heat flux was the driving force and
one of the  three patterns occurred, depending  on
whether the sensible heat flux was used primarily to
cause  convective boundary layers (CBLs) to grow
over the valley floor  and sidewalls  (Pattern 1), to
remove mass from the inversion in the upslope flows
(Pattern 2), or to accomplish both  (Pattern 3).  In
Part II, a bulk thermodynamic model of temperature
inversion destruction is  developed based  on the  hy-
pothesis, using several simplifying assumptions.

2. Mathematical model of inversion destruction

a. General  equations

  Two approaches can  be used to develop a math-
ematical model able to simulate temperature changes
                          in the valley atmosphere.  In the first  approach de-
                          tailed mathematical equations can be developed for
                          the individual components of the overall system, in-
                          cluding the various boundary layers and stable core.
                          However, the valley atmosphere consists of many
                          interrelated layers, and the coupling of the equations
                          for the different layers to simulate potential tem-
                          perature changes in the valley atmosphere as a whole
                          would be difficult  due to geometrical considerations
                          and lack of detailed information on physical char-
                          acteristics of the various layers. Consequently, a sec-
                          ond approach is taken in which a bulk  thermody-
                          namic model  is developed for  the valley inversion.
                          As more  is learned about the individual components
                          of the system, the  model can be refined to introduce
                          greater detail into the simulation of the individual
                          components.
                            The thermodynamic model of valley temperature
                          structure evolution developed here is  based on the
                          hypothesis of  Part I.  Fig.  1 shows a  unit thick cross
                          section of a mountain valley having  sidewalls of in-
                          clination or, and a2 and valley floor width /. At sunrise
                          (t,)  the  valley is  assumed to  have an inversion  of
                          depth hi and constant vertical potential temperature
                          gradient  7. The variable width  of the valley at the
0021-8952/82/030290-I3S07 25
9 1982 American Meteorological Society
                                                   159

-------
MARCH 1982
                      C. DAVID  WHITEMAN AND THOMAS  B.  McKEE
                                                                                               291
          INVERSION TOP AT t«ti
                                                                     H  ••
     FIG. 1. Valley geometry and potential temperature profiles used to formulate a mathematical model of inversion destruction.
top of the inversion is designated by L, and the origin
of a y-z coordinate system is placed at the center of
the valley floor at point  0. After sunrise  a  typical
Pattern 3 potential temperature evolution ensues, in
which  a CBL develops  over the  valley floor and
sidewalk. Removal of mass from the valley in these
CBLs allows the stable core to sink so that a sounding
taken at an arbitrary later time r will  show a lower
inversion top height h(t ) and a shallow CBL of height
H(t) near the ground. A later sounding, taken at a
time t0 when the inversion has just been destroyed,
will show a neutral  atmosphere having a  potential
temperature 8 = 9h.
  From the first law of thermodynamics, the incre-
ment  of energy  required to increase  the  potential
temperature of a mass of air m by the potential tem-
perature increment A# is
                                             (1)
where T/9 = (P/IOOQ)*'' =  1, p is the density (as-
sumed constant) and V  is volume. Using (1),  the
energy required to change the valley potential tem-
perature profile at time t, to the profile at time t can
be obtained by an integration  over the valley volume
below the height of the inversion top. The total en-
ergy requirement is composed of two pans: the en-
ergy increment g2 that removes mass from the valley
and allows the top of the inversion to sink, and the
energy increment Q3  that causes a  CBL  to  grow.
These energies  are  represented by the  areas  desig-
nated in Fig. 1  and are given by
         T r c*i f?*  f>
         -            MJxdydz
         9 [_Jo JV£ Jo
   f  (•>•<<  /-i
-           &6tdxdydz
  Jo Jxt. Jo
where
                                             (2)
                                             (3)
                                             (4)
               yL = -  - H	
                      v2   tana2/'
               Vt> =	1" 	 .
               y    2   tana,

                ~     l   . + .  l
                    tana,   tana2

                  = f(h - Z),
and
                 J  fH «-«  (M
                 -           Mydxdydz,
                 0 JO Jyi JO
where
                      y(H - z).
                                                                              (5)


                                                                              (6)


                                                                              (7)

                                                                              (8)


                                                                              (9)


                                                                            (10)


                                                                            (11)
In order to simplify the integration, it was assumed
that the valley temperature structure is horizontally
homogeneous across the valley, that mass is removed
from the valley in such a way that the potential tem-
perature  gradient in  the  inversion  layer  does not
change with time, and that p and cf are constant. By
differentiating the individual energies Q2 and Q3 with
respect to time, the rates of change of the height of
the top of the inversion and the height of the CBL
are obtained, such that
         dt
                                                           dt
                                                        . dH,
                                 HC\
                                                                                              (13)
  The total rate of energy input into the valley to
accomplish these changes is the fraction A0 of solar
irradiance F coming across the area L of the top of
the inversion that is converted to sensible heat. The
solar irradiance may be approximated by a sine func-
tion having a certain amplitude A\ and period T, so
that the total rate of energy input becomes
  dO,
  -^ = AoLF = A0(l
   at
                                                                         hC)A, sin - (/ - t,).   (14)
                                             160

-------
292
                   JOURNAL OF APPLIED  METEOROLOGY
                                                                                  VOLUME 21
An energy balance for the valley inversion is obtained
by equating (14) to the sum of (12) and (13). Al-
ternatively, a fraction of the energy input is available
to dnve the growth of the CBL while the rest of the
incoming energy is  used  to remove mass from the
valley. The fraction of energy input used to drive the
CBL growth is assumed to be of the form
where k is a number between 0 and 1. This form is
chosen in order to simplify later equations. Equating
this fraction of the energy input to (12) and the re-
mainder to (13) and solving for dH/dt and dh/dt
results in the final model equations
 dt
dt
1JL
Tpc.
                  hC~
                    1 +
                             HCY1
                      —• sinT- (r - r()] .   (16)
                       T«    LT       J
These equations specify the dependence of the rate
of ascent of the CBL and the rate of descent of the
inversion top on inversion characteristics,  incoming
energy and valley topography. An integration of the
coupled equations allows the simulation of the time-
dependent behavior of the  heights of the CBL and
inversion top. If the potential temperature #/, at the
top of the inversion is known and is independent of
time, and 7 is constant, knowledge of the variation
of h and H with time  is sufficient  to  specify how
vertical profiles of potential  temperature change with
time. When k  = 0, the equations  provide an ap-
proximate simulation of Pattern 2 inversion destruc-
tion in  which  destruction occurs solely due to the
removal of mass from a valley  in the slope flows,
resulting in a descent of the inversion top. When k
=  1, the equations provide a simulation of Pattern
1 inversion destruction, in which destruction occurs
mainly  due to the growth of a CBL over the valley
floor. When k = 1 and  the valley floor becomes very
wide, the simulation approaches that  of  inversion
destruction over the plains. When k is between 0 and
1,  the equations provide a  simulation of Pattern 3
inversion destruction in which the inversion is de-
stroyed  by the combined effect of a growing CBL
and a descending inversion top. A  more  complete
description of the characteristics  of the  model equa-
tions for Pattern 1, 2 and  3 temperature  structure
evolution follows.
                                           b. Pattern 2 inversion destruction

                                             The physical hypothesis of Pattern 2 inversion de-
                                           struction requires a shallow  CBL to form over the
                                           sidewalls so  that additional  energy can  be used to
                                           cause mass to flow up them. Pattern 2 destruction
                                           may be approximated by assuming that all the energy
                                           available to destroy the inversion goes solely to move
                                           mass up the sidewalls, causing the top of the inversion
                                           to descend. This can be accomplished by setting k
                                           equal to zero in (15) and (16) resulting in the two
                                           equations

                                                               f-a
                                            dh
                                            dt
                                                             'I + hC
                                                                        pc.yh
                                                                              sin -(t~t,)\.
                                                                                LT        J
                                            (18)
Following these equations,  the CBL does not grow
as a function of time, and the inversion is destroyed
as the top of the inversion sinks. The rate of descent
of the top of the inversion increases as the inversion
descends. The descent rate  is faster when more en-
ergy is available and when the potential temperature
gradient  of  the inversion is weaker. The factor in
large parentheses in ( 1 8 ) is a topographic factor that
varies from  1  to 2 depending on the shape of the
valley cross  section.  This accounts for the reduced
volume of air within  the mountain valley relative to
that over the plains  for the same energy flux on a
horizontal surface. Since the valley  has less volume
to be heated by the same incoming energy, it warms
more  rapidly.  By separating variables h and / and
integrating from the  initial  conditions h = h, and H
= 0 at / = r, to h - h and H = H at  r = t, analytical
expressions are obtained which describe how H and
h vary with  time, i.e.,
                                                                      H=Q,
                                                        pc,y ir
                                                                                      (19)
                                                                                      (20)
                                             Fig. 2 illustrates the shapes of the curves of h vs
                                           t for a reference simulation and for two cases where
                                           one parameter in the reference simulation is changed.
                                           The reference simulation  uses representative values
                                           of valley parameters observed in western Colorado
                                           including / = 1000 m, a,  = a2 =  15°, h, = 500 m,
                                           7 = 0.025 K m-',  r = 12  h = 43200 s, T/B = 1 and
                                           p = 1  kg m~!. The reference inversion takes nearly
                                           4'/> h to be destroyed when A^A\/pc,, = 0.25 K m s"'.
                                           Fig. 2 was obtained by a numerical integration of
                                           (18) using a forward finite difference scheme and a
                                           time step of  10 min.
                                             An analytical expression for the time required to
                                                 161

-------
MARCH 1982
 C.  DAVID WHITEMAN  AND THOMAS  B. McKEE


 500 r
                                                                                                    293
                       400
                       300
                       200
                        100
                                                            REFERENCE  SIMULATION
                           012345
                                                 t-tj (hours)

                      FIG. 2. Descent of inversion top as a function of time for the reference inversion
                    simulation and for two simulations for which the single parameters indicated were
                    changed. Pattern 2 destruction.
destroy an inversion can be obtained by integrating
(18) from the initial conditions to the final conditions
of h = 0 at i = to. This expression,
tD — t, = - cos"
         IT
8
                                              «.»
enumerates the factors  affecting inversion  breakup
time. The sensitivity of inversion breakup time (mea-
sured from sunrise) to the various parameters is il-
lustrated  in  Fig.  3 using the reference  simulation
above. The reference inversion takes 4.4  h  to break
(vertical line in Fig. 3).  The effect  on the  time re-
quired to  destroy the reference inversion  by varying
the individual  parameters is obtained by following
the labeled curves. Thus, varying the initial height
of the inversion from 400 to 600 m, other parameters
being equal, changes the time required to destroy the
inversion  from 3.5  to 5.4 h.  For the  reference inver-
sion, the most sensitive parameters affecting the time
required to break an inversion are the available en-
ergy and  the initial potential temperature  gradient
                                                                              X

                                                                               04O
                                                                               030
                                                                               020
                                                                               010
                                                                  h;
                                                                  -800
                                                                  SOO


                                                                   (m)


                                                                  4OO




                                                                  200
                                  3        4       ~S

                                        tD-tj (hours)

            FIG. 3. Sensitivity of inversion destruction time to various model parameters for Pattern 2 destruction.
                                                 162

-------
294
      JOURNAL OF APPLIED METEOROLOGY
                                                                      VOLUME 21
   500
                     t-t|( hours)

  FIG. 4. Ascent of CBL and descent of inversion top as a function
of time  for Pattern 1  inversion destruction in a valley for the
reference simulation and for two simulations in which single pa-
rameters were changed to the values indicated.
and inversion height. In the normally dry Colorado
valleys the most important factors affecting the avail-
able energy are  albedo (snow versus no snow) and
latent heat flux. The  effect of valley shape on the
breakup time is relatively small for normal  ranges
of a and / encountered in valleys of western Colorado.
Nevertheless, it is apparent that the valley width and
sidewall angles may affect the mode of inversion de-
struction since they control, to a certain extent, the
divergence of mass in the CBL's and thus determine
whether inversion destruction  more nearly follows
Pattern 1  or Pattern  2. Overall,  predictions  of the
time required to destroy inversions, given typical val-
ues of the parameters observed in field experiments,
are consistent with the observed range of 3.5-5 h.

c.  Pattern I inversion destruction—Valley case

  A useful approximation to Pattern 1 inversion de-
struction can be obtained from the model equations
by setting k = \  in (15) and (16). The general equa-
tions then reduce to the two equations
 dH=9_ f I + HC
 dt  ~ T\l
dh = _9_ \~(h-H)C~\
dt~    Tll
AoA,  .  fir .
— —• sin -(r-
pCpyh    ]_r
,   (22)
                                             ,   .
                                          •   (23)
  Eq. (22) is formulated so that the entire fraction
AQ of the energy coming across the area (/ -t- HC~)
of the top of the CBL is used to cause the CBL to
grow. The energy used to  cause the top of the in-
                                                   version to descend is the fraction A<> of the difference
                                                   between  the energy coming across the top of the in-
                                                   version and the energy coming across the top of the
                                                   CBL. The time-dependent behavior of the height of
                                                   the CBL can be obtained by an integration of (22)
                                                   from the initial condition of H =  0 at t = t, to the
                                                   final condition of H = H at /  » t, such that
                                                                                                (24)
                              The integration of (22) and (23) can be accomplished
                              numerically to determine how h and H change with
                              time in a Pattern 1 inversion destruction in a moun-
                              tain valley. This is done for the reference simulation
                              and for two simulations in which a single parameter
                              of the reference simulation is changed. The resulting
                              plots are shown in Fig. 4.  The characteristics of the
                              plots include  a  near-linear growth of the CBL with
                              time, a slow descent of the inversion top, and a more
                              rapid breakup than for Pattern 2 destruction. Pattern
                              1 destruction takes  3.7  h  versus the 4.4  h required
                              for Pattern 2 breakup.  The same total  amount of
                              energy is required to destroy the reference inversion,
                              whether it is destroyed following Pattern 1 or Pattern
                              2. However, since the energy available to destroy the
                              inversion comes across  the  area  of  the  top of  the
                              inversion, and this area  is larger when the inversion
                              top sinks more slowly, the total amount of energy
                              required  to destroy  the  inversion  is attained earlier
                              in the day, resulting  in an earlier inversion breakup.

                              d. Pattern I inversion destruction—Flat plains case

                                Application of (22) and (23) to a valley that is
                              very  wide or  approaches a plain (/ — oo) results in
                              the equations
                                       dH   8
                                        dt
                                                      (25)
                                        dt
                                                                          (26)
                              Thus, over flat terrain where no topographically in-
                              duced mass  divergence  occurs from the CBL,  the
                              inversion is destroyed solely by the growth of a CBL.
                              Performing integrations on (25) as described in  the
                              previous section results in the analytical expression
                  J2£l4>dl  i-cos-U-r,)     , (27)
                     TT  —
                              and an expression for the breakup time,
                                                                      ,   (28)
                                                  163

-------
MARCH 1982             C. DAVID  WHITEMAN  AND THOMAS B.  McKEE


                   500 r


                                                   REFERENCE SIMULATION x
                   500



                   400
                   300
                I


                   200
                    100
                                                                                                    295
                      01        234567

                                                 t-t, (hours)

                  FIG. 5. Growth of CBL over Sat terrain as a function of time for the modified reference
                inversion (a " 0, / —• vo) and for two simulations in which single parameters were changed
                to the values indicated. Pattern ! destruction.
Eq. (27) is nearly identical to an equation for CBL
growth over  homogeneous  terrain  as derived  by
Leahey and  Friend (1971). Following (27),  Fig. 5
presents plots of H vs time  for the reference simu-
lation and for two cases where one parameter in the
reference simulation has been changed. The refer-
ence inversion is destroyed in 5.65 h by a near-linear
increase in the depth of the CBL, If the  incoming
energy is doubled, the inversion takes 3.8 h to break,
and if the potential temperature gradient is increased
to 0.035 K. m"',  the inversion is broken in ~7.1  h.
Fig.  6  indicates the sensitivity of the  time required
to break an inversion on the different parameters of
(28).  The  effect on the breakup time  of changing
individual  parameters in the reference simulation is
obtained by following the individual curves.

e.  Pattern 3 inversion destruction

   A simulation of Pattern  3  inversion destruction
uses the general model [ Eqs. (15) and (16) ], in which
a  partitioning  of  energy is  required to allow both
CBL  growth and  inversion top descent. In order to
use the general model equations, the fraction of sen-
                       •340
                                  456
                                       t0-ti (hours)
                                                                            30


                                                                            (•Km/sec.)


                                                                           1 20
                                                                           -lo
                      FIG. 6. Sensitivity of inversion destruction time to various model parameters
                                for Pattern 1 inversion destruction over flat terrain.
                                                     164

-------
296
                           JOURNAL OF  APPLIED METEOROLOGY
                                                                                           VOLUME 21
      500
     400
     300
   r
   o
     aoo •
      100 •
        012345
                      t-f, (hours)

  Fie. 7 Ascent of CBL and descent of inversion top as a function
of tune for Pattern 3 destruction of the reference inversion for
different values of k.
sible heat flux K = k[(l + HC)/(l + he)] that drives
the growth of the CBL  must be determined.  It is
apparent that the fraction K is a function of time,
since the initial energy input must be used primarily
to develop the CBL's  before appreciable mass can
be carried up them. Factor K also depends  on the
topographic characteristics of the  valley, since K
must approach 1  as the valley width approaches in-
finity. It seems probable that K may also be a func-
tion  of  sensible heat flux. Since the functional de-
pendencies of K are not yet known, it is assumed that
k is a constant and, by  comparing model simulations
to actual  data, the constant value of k that  results
in the best fit  to data is determined. This approach
allows an  investigation of the effect of  k on the sim-
ulation. Further research is necessary  to determine
the actual functional form of K, so that a better un-
derstanding of the energy partitioning phenomenon
can be  obtained, resulting in more accurate simu-
lations.
  The equations used to simulate Pattern 3 destruc-
tion  are thus
dH
 dt  ''
      8  k f l + HC
                         X rinjj (t - t,) \ ,   (29)
^L = _ 1 _L f/ + hC-k(l +
                                  J  ~fh

                         x sinTj (f - r,)l,   (30)


             0 < (k = constant) « 1.          (31)

These  equations can  be integrated  numerically to
determine how H and h vary with time from a given
where
                                                   initial state. Fig. 7 shows several numerical integra-
                                                   tions using this method,  time steps of 10 min, the
                                                   reference simulation, and several values of k.  Re-
                                                   plotted  on the same figure are some of the limiting
                                                   cases of the general equations for destruction of the
                                                   reference inversion discussed earlier.  The time re-
                                                   quired  to break the  reference  inversion decreases
                                                   from 4.4 to 3.7 h as the value of k is increased from
                                                   0 to 1. For all values of k the growth of the CBL  is
                                                   nearly linear.
                                                     Fig. 8 shows  the effect of varying individual pa-
                                                   rameters  in the reference simulation  with  k fixed.
                                                   The inversion is destroyed when the ascending CBL
                                                   and the descending inversion top meet at a height
                                                   of HD = h0 = 205 m. The fact that  the fraction  k
                                                   uniquely determines the height at which the ascend-
                                                   ing CBL  meets  the descending inversion top  at the
                                                   time of inversion  destruction, suggests a means of
                                                   fitting the model results to  actual data that will be
                                                   used in  a later section.

                                                   / Model modification to account for warming of the
                                                        neutral layer

                                                     Field observations  show  that the potential tem-
                                                   perature 8),  at  the top of the inversion usually in-
                                                   creases  slowly with time. The warming rate  varies
                                                   from valley to valley  and from day to day with the
                                                   average warming rate being about 0.4 K h"1. From
                                                   the physical hypothesis, this warming requires that
                                                   more energy be spent to move mass up the sidewalls,
                                                   since the  parcels must be warmed to a higher tem-
                                                   perature 9h(t) = 8k(t,) -I- (d8h/dt)6t to be removed
                                                   from the valley.  The energy requirement can be  cal-
                                                      500
                                                     400
                                                      300
                                                      200
                                                      100
                                                                                              035

                                                                                   REFERENCE SIMULA!,ON
                                                         01         2345
                                                                         f-t, (hours)

                                                     FIG. 8. Ascent of CBL and descent of inversion lop for Pattern
                                                   3 destruction (k = 0.2) of the reference inversion, and for two
                                                   simulations in which single parameters were changed to the values
                                                   indicated.
                                                165

-------
MARCH 1982
                      C. DAVID  WHITEMAN AND THOMAS B. McKEE
                                                                                              297
culated by considering that the mass of air removed
from the valley to allow the  top of the inversion to
sink from h, to h is given by
                                                 dQ±
                                                 —r-
                                                            T Q\~
                                                         c!,--\(
 mass removed
               pV

               p(h, - h)(l
                                                                     ~(t- O    (/ +
                                                                                          .  (35)
                             Vi(A, + h)C],  (32)
and that the energy required to move this mass across
a potential temperature jump at the top of the in-
version, 0»(r) — #*('() can be approximated by
                                            (33)
   If the wanning in the neutral layer occurs linearly
in time, the potential temperature jump is given by
j3(/ — r,) where ft = d8hjdt is  the rate of wanning.
Then
                       X [/ + H(h, + h)C],   (34)
                                                  The model equations are modified to account for this
                                                  extra energy requirement by specifying that the frac-
                                                  tion k((l + HC)/(l +  hC)] of the inversion energy
                                                  input dQt/dt drives the CBL growth, or
                                                               *(
                                                                 I + HC
                                                                \l + hCj  dt
                                                                               dQ2
                                                                               dt  '
                                                                                             (36)
                                                 while the remainder of the energy  input drives the
                                                 descent of the inversion and carries parcels across
                                                 the potential temperature jump, such that
                                                                   hCj   dt
                                                                                dt
                                                                                       dt
                                                                                             (37)
                                                  Substituting Eqs. ( 1 2 ), ( 1 3 ), ( 1 4) and ( 3 5 ), the mod-
                                                  ified model equations are
                                                 dt
                                                              /
dh^ _^J_
dt '   Tpcf
                                           sin -
                                               r
                                                  -O-pc,-^(Af-A){/
                                                                           1 sinj - (t - t,]
                                                                                   h)C] ]
                 I
                                                         - f,)(/ + AC)
                                                                                             (38)
                                                                                              (39)
These  equations can be used to simulate inversion
breakup when significant wanning occurs in the neu-
tral layer above. Eq. (39) reduces to (30) when /3 is
zero. It is important to note that (38) has the same
form as before but, since more energy is required to
move mass up the sidewalls, the partitioning of en-
ergy may be affected. If this actually occurs in  na-
ture, then the functional dependency of K  is more
complicated than previously discussed, since it will
depend not  only  on time, energy input and valley
width, but also on the rate of warming of the neutral
layer above  the inversion. As before, Eqs. (38) and
(39) may be integrated numerically to simulate val-
ley inversion breakup if Bh(t) is known. If 9* does not
vary with time, Eqs. (29) and (30) can be used for
the simulation.

3. Comparison of model results with data

   In this section a finite difference form of the model
equations (38) and (39) will be applied to simulate
actual  data  collected  in   the  valleys of  western
Colorado. The input parameters needed to solve the
equations are given in Table 1 along with a summary
of how they may be obtained. The output of  the
model  is h(t) and H(t). From these outputs  and the
assumptions that 7 is constant for altitudes between
H and h, that potential  temperature  is independent
                                                 of height in the CBL and neutral layer, and  that
                                                 B*(t,) and ddhjdt are known,  8(t,  z)  for the CBL.
                                                 stable core  and  neutral layer can be determined.
                                                 Unfortunately, since two of the input  parameters to
                                                 the thermodynamic model (k  and  A0) were not  ob-
                                                 served in the field programs,  the  equations cannot
                                                 be applied directly.  Instead,  arbitrary values for k
                                                 and A0 are  chosen  until the best simulation of the
                                                 data is obtained with the model. It is then determined
                                                 whether the values of k and A0 are reasonable for
                                                 the situation at hand. Both k and A0 are bounded,
                                                 since they are fractions between 0  and 1. The value
                                                 of k specifies the constant fraction of sensible heat
                                                 flux that is  used to cause the CBL to deepen.  The
                                                 model results are presented below  for a winter Pat-
                                                 tern 2 inversion destruction in  the wide Yampa Val-
                                                 ley and for  a fall Pattern 3 inversion  destruction in
                                                 the Eagie Valley.

                                                 a.  Pattern 2 simulation—Yampa  Valley, 23  Feb-
                                                    ruary 1978

                                                   The model input parameters required to simulate
                                                 the Pattern  2 inversion destruction observed in  the
                                                 snow covered Yampa Valley on 23 February  1978
                                                 were  obtained as follows.  First, 9T~{ =  1.07  and
                                                 pc? = 1040 J m"J K"1 were calculated using the ap-
                                                166

-------
298
                           JOURNAL OF  APPLIED METEOROLOGY


                                      TABLE 1. Model input parameters.
                                                                                             VOLUME 21
                         Model input
External conditions (neutral layer warming)



Energy partition

Surface energy balance

Numencal
(viii)
(ix)

(x)
(xi)

(xii)

(xiii)
(xiv)
                                                                                   Source
Constants

Valley topography

Initial inversion characteristics

Solar irradiance
(i)
(ii)
(iii)
(iv)
(v)
(vi)
(vii)
8 IT
PC,
I
C
1
h,
^
from average P and T of

from topographic maps

from sunrise sounding

from extraterrestrial solar
valley atmosphere





i [radiance model
  or field notes for t,


from sequential soundings taken during
  observation of inversion breakup or from
  climatic data

comparisons of theory and data

measurements, if available

arbitrary
proximate mean pressure  (780 mb)  of the early
morning inversion and an average temperature of the
valley atmosphere (—10°C) from the defining equa-
tion for potential temperature and from the equation
of state, respectively. Second, the valley topographic
parameters (/  = 2580 m,  a, = 9°,  a2 =  16°, C
=  9.80)  were  obtained from  topographic maps.
Third, the initial inversion parameters  (h, =  530  m,
1  = 0.0345 K  m"1) were estimated from a straight
line fit to  the top of the 0714 MST sounding on this
date.  The  hyperbolic  lower region of  the  sounding
could not be adequately fit with a straight line, so
it  is ignored in the analysis. Fourth, a  sinusoidal fit
to the extraterrestrial solar  flux curve obtained from
a  standard solar irradiance model (e.g., Sellers,
          1965) provides  solar  irradiance  parameters  (At
          = 878 W  m'2,  t, = 0655 MST, r =  10.9  h). The
          output of  the solar irradiance  model  for the hori-
          zontal surface of interest is presented in Fig. 9.  For
          reference, the solar fluxes on extraterrestrial surfaces
          with the same aspect  and inclination angles as the
          valley sidewalls are indicated on the figure. The input
          required to run  the solar irradiance  model includes
          the date and the latitude and longitude of the Yampa
          Valley site. The remaining parameters necessary to
          run the inversion destruction model  are fractions k
          and A& and the neutral layer  wanning rate 0. To
          simulate a  Pattern 2 destruction, k must be zero. The
          neutral layer warmed only 1.1 K during the inversion
          destruction, for  a warming rate $  of 2.8 X 10~5 K
                           ' SOM8HCRO RANCH
                            YAMPA VALLEY
                            23 FEBRUARY 1978
                                                 II      13
                                              TIME (MST)
                                                                       17
                                                                               19
                    FIG. 9. Extraterrestrial solar flux calculated for valley floor and sidewall surfaces
                                   of the Yampa Valley, 23 February 1978.
                                                 167

-------
MARCH 1982
                      C  DAVID WHITEMAN AND THOMAS  B.  McKEE
                                                                                                 299
          SCO

          500

        J 400
        t-
        | 300

        I 200

          100

            0
                  YAMPA VALLEY NEAR STEAMBOAT SPRINGS, COLORADO
                                      FEBRUARY Z5, 1978
THEORETICAL SUNRISE   0655
SUNRISE ON  VALLEY FLOOR   0802

0  DATA  POINTS
m  MODEL
                0700
                         0800
                                 0900
                                                            1200
                                                                     1300
                                                                              1400
                                                                                       1500
                                           1000      1100

                                            TIME (MST)

         FIG. 10. Comparison of model simulation -of h(t) with actual data for the Yarapa Valley, 23 February 1978.
s~'. The best fit of the model output h(t) to the in-
version top data  of  Fig. 10 was obtained with A9
= 0.19 by  a  trial-and-error procedure.  Using this
value, the simulation of the height of the top of the
inversion agrees to within  25 m of the data over a
7 h period. In mid-afternoon, however, the simulation
continues to call for the descent of the inversion top,
when the  actual data indicate that  the descent
stopped. Thts is probably due to afternoon shading
of the valley  by a mountain peak southwest of the
site.  The model, using a simple sine function to sim-
ulate solar flux, does not account for  this shading.
In Fig. 11 the potential temperature profile simula-
tions are  compared  to actual sounding data. The
sounding data consist of eight consecutive potential
temperature  soundings  taken   at  •—1 h  intervals
throughout the  day. The characteristics of the initial
inversion were obtained from the 0714 MST sound-
ing and are indicated by the straight line fit to the
sounding in the figure. The  slow neutral layer warm-
ing is also apparent  in  the figure. The excellent fit
of the model simulations to the  data  in the stable
core above the  50 m  deep surface layer is shown for
soundings 4 and 6. The value of A0 required to obtain
these results  seems reasonable for the snow-covered
valley with evergreen forests covering  much of the
valley sidewall  above the observation site. It is par-
ticularly interesting  that the model assumption of
constant potential temperature gradient was satisfied
well during the long period of inversion destruction
in  the complicated topography of the Yampa Valley.
The case study is an excellent example of the  effect
of enhanced  albedo due to snow cover in retarding
the normal breakup of an inversion. Despite the fact
that the temperature inversion was not destroyed on
this  clear day,  the diurnal  range of temperature at
the ground was quite large.

b.  Pattern 3  simulation—Eagle Valley,  16 October
     1977

   Pattern 3  inversion  breakup  in the  Eagle Valley
is  remarkablv  consistent in all seasons when snow
                                  cover is not present in the valley. To test the math-
                                  ematical model, the Pattern 3 breakup on the clear
                                  day of 16 October 1977 is chosen for simulation. The
                                  input parameters to the model are  obtained as for
                                  the Pattern 2 simulation above. Thus, values of the
                                  constants 07*"' (1.08) and pcf (990 J m'3 K~') are
                                  determined from  the approximate mean pressure
                                  (768 mb) and  temperature (0°C) of the valley in-
                                  version. The valley topographic parameters (/ = 1450
                                  m, a,  = 21°, «2 = 10°, C = 8.28) are determined
                                  from topographic maps. The initial values of the in-
                                  version parameters (h, =  650 m and 7 = 0.0269 K
                                  m~') were taken from the 0650-0719 MST sounding
                                  since  the pre-sunrise  sounding was of insufficient
                                  height to determine h,. A solar irradiance  model was
                                  used to determine the parameters A\ = 906 W irT2,
                                  t, = 0621 MST and r =  11 h  (Fig.  12).  The value
                                  of /3 (8.3 X 10"5 K s"1) was taken from valley tem-
                                  perature sounding data.
                                    The fitting of the model output to the data on Fig.
                                  13 was accomplished first by choosing the value of
                                          |-   YAMPA VALLEY
                                              FEBRUARY 23, 1978
                                        300 -


                                        200 -


                                        iooL
                                                	SIMULATION
                                         ze:
                                              270   275   23O   285   290   295
                                               POTENTIAL TEMPERATURE  (  K!
                                    Fio. 11. Comparison of model simulation of potential temper-
                                  ature structure with data for the Yampa Valley, 23 February 1978
                                  Upsoundings 1-3 were initiated at 0714, 0905. 0959, 1100, 1202.
                                  1259, 1359 and  1508 MST, respectively.
                                                168

-------
300
                           JOURNAL OF APPLIED METEOROLOGY
                                                                                             VOLUME 21
                    u.
                    IT
                    Id
                    CC
                    o:
                    uj
   I4OO


   1200


   IOOO


Z~ aoo
 6
J 500


   400


   200
                              STEVE MILLER RESIDENCE
                              EASLE VALLEY
                            . 16 OCTOBER 1977
                                                    jO'SLOPE. 194-ASPECT

                                                          S-FACIN6 SLOPE
                          0600   0800
                                         IOOO     1200   .1400
                                              TIME (MST)
                       Fic.  1 2. Extraterrestrial solar flux calculated for valley floor and sidewall
                                  surfaces of Eagle Valley, 16 October 1977.
k so that the ascending CBL and descending inver-
sion top met at  the  proper observed height at the
time of inversion destruction. The value of A<> was
then varied until the model outputs h(t) and ff(t)
fit the data well. The fit in Fig. 13 was obtained with
k  = 0.14 and  A0  » 0.45. The  value of A0 seems
realistic considering the dry nature of the valley sur-
face on this date. The equations were integrated from
the initial conditions of h = 650 m and H  =  10 m
at i = 0705 MST using a time step of 10 min. From
Fig. 13 it is clear that a good fit to the data is obtained
with the chosen values of the two parameters Aa and
k. The  simulation of CBL height and inversion top
height agrees with the data within 50 m over most
of the  period  of inversion destruction. The  CBL
height was overpredicted  early in the inversion pe-
riod, due to the bulk nature of the model. Following
sunrise, the CBL develops first over the illuminated
sidewall or  sidewalls, and somewhat  later over the
valley floor when it is sunlit. The bulk model, how-
ever, does not differentiate between CBL growth over
the three different  valley surfaces. All energy going
into CBL growth  is attributed, in  the  model, to
growth of the valley floor CBL. Thus, the initial
      700i

      600^
      400
      200

      100
    STEVE MILLER RESIDENCE
    EAGLE VALLEY
    OCTOBER 16,1977
              0700
                      0800    09OO
                     TIME (MST)
                                    IOOO
                               overprediction of CBL growth over the  valley floor
                               is a characteristic feature of the model equations.
                               The behavior of the simulation at mid-levels of the
                               valley atmosphere near the time of inversion destruc-
                               tion should also  be  mentioned. The data typically
                               show a more sudden inversion breakup  is to be ex-
                               pected in nature since the final remnants of the stable
                               core will break up  in  convective overturning,  once
                               the stable core becomes thin enough and the con-
                               vective plumes rising from the valley floor become
                               vigorous enough. Due to  the chaotic nature of the
                               breakup, actual soundings taken during this time will
                               often  show deformations in  the  vertical potential
                               temperature profiles. Estimates of heights h and H
                               from soundings are difficult to make during this time.
                               A continued  flux of energy into  this region  is nec-
                                     EAGLE VALLEY
   FIG. 13 Comparison of model simulation of H(t) and hd)
    with actual data for the Eagle Valley, 16 October 1977
                                    285     290     293    300    305     310
                                            POTENTIAL  TEMPERATURE («K)

                                 FIG. 14. Comparison of model simulation of potential temper-
                               ature structure with  data  from the Eagle Valley, 16 October
                               1977
                                                 169

-------
MARCH 1982
                      C.  DAVID  WHITEMAN AND THOMAS B. McKEE
                                                                                               301
essary  before the deformations in the profiles  are
destroyed and well-organized convection through the
entire depth of the valley results in smooth neutral
profiles. When this occurs, the breakup can be con-
sidered as finished.
  The  potential temperature profiles corresponding
to the data and simulation of Fig.  13 are presented
in Fig. 14. The  potential temperature data are from
the tethersonde up-soundings taken at 0650-0719,
0829-0848, 0924-0948, 1013-1033 and 1106-1121
MST. The corresponding potential temperature sim-
ulations are plotted for the approximate midtimes
of the tethersonde up-soundings at 0840, 0935, 1025
and 1115. Again, the model provides a  good simu-
lation of the actual data, reproducing the wanning
and growth of the CBL, the warming of the stable
core, and the descent of the inversion top.

4. Summary and conclusions
   A thermodynamic  model of valley  temperature
structure evolution has been developed. By differ-
entiating the first law of thermodynamics, the rate
of energy input into the valley atmosphere is equated
to the rate of descent of the inversion top and the
rate of ascent of the  CBL under the following as-
sumptions:
   I) The potential temperature gradient in the sta-
ble core is constant during the period of inversion
breakup.
   2) The temperature structure is horizontally ho-
mogeneous in the cross-valley direction.
   3) The valley topography can be adequately rep-
resented by a  horizontal  valley floor of  arbitrary
width and two linear sidewalls of arbitrary slope.
   4) The initial inversion  at  sunrise can be  ade-
quately represented by a constant potential temper-
ature gradient layer of arbitrary depth.
   5) Convective boundary layers can be adequately
represented as constant potential temperature layers
of arbitrary height.
   6) The rate  of energy input into the valley at-
mosphere is a constant fraction of the solar energy
flux (assumed to be a sinusoidal function of time
from sunrise to sunset) coming across the horizontal
upper surface of the inversion.
   To complete the model, the partitioning  of the
energy input into CBL growth and mass  transport
must  be estimated. As  a first approximation, it is
assumed that a constant fraction of the energy input
is used to cause the CBLs  to grow. The remaining
energy is used for mass transport in the upslope flows.
Energy going into the growth of the CBL causes the
CBL depth to increase with time. Energy causing air
parcels to flow  up the sidewall CBLs  results in the
descent of the top of the inversion. When most of the
available energy drives  the growth  of the CBLs, a
temperature structure evolves in which the inversion
is destroyed predominantly by the  upward growth
of a CBL from the ground. When most of the avail-
able energy drives mass up the sidewalls, a temper-
ature  structure evolves in which the inversion is de-
stroyed by  the descent of the  inversion top.
  The inputs to the model are 1) the initial inversion
characteristics (depth and average potential temper-
ature  gradient); 2) valley topography (floor  width
and inclination angles of the  two sidewalls); 3) the
rate of energy input;  and 4)  the fraction of energy
available to increase the depth of the CBL. The effect
on inversion evolution of warm air advection  above
the inversion layer  can be investigated by using a
modified version of the model in which the wanning
rate is input.
  The model simulates the changes with time of the
height of the inversion top and the depth of the CBL
during the inversion breakup period. From these sim-
ulations potential temperature profiles of the  valley
atmosphere can be constructed for any time during
the period.
  Sensitivity analyses were conducted for the lim-
iting cases of the model. The results indicate that the
time required  to  destroy an inversion depends pri-
marily on the  initial height of the inversion,  on  its
potential temperature gradient, and on the amount
of energy available  to destroy it. Using a reference
simulation  in  which model parameters were given
values typical of valley inversions, inversion destruc-
tion took approximately  3'/i to 4Vi  h  after sunrise.
These  times correspond  well  with  actual observa-
tions.  Less time is required to destroy a valley  in-
version than an inversion of like dimensions over the
plains, because the available energy  is  used to warm
a smaller volume of air. For the dry valleys of western
Colorado, the  amount of energy available depends
to a large extent on the presence or absence of snow
cover  or surface  moisture in  the  valley.  Valley  in-
versions were  destroyed sooner by the growth of a
CBL than by the descent of the inversion top. Valley
width and inclination angles of the sidewalls had only
a limited effect on the time required  to destroy  an
inversion. Increased  valley width and steeper side-
walls  both  increased slightly the time  required.
  The thermodynamic model  was used to simulate
two specific sets of inversion breakup data for Pattern
2 and  3  temperature structure evolution in the  to-
pographically  diverse  Eagle  and Yampa Valleys.
Simulations were obtained by fitting two constants
in the  model (relating to the  surface energy budget
and energy partitioning) to the data. The model out-
put fit the Pattern 2 inversion breakup in the  snow-
covered Yampa Valley very  well using  an energy
input  equal to 19% of the extraterrestrial solar flux
on a horizontal surface, and assuming that all of this
energy was used to drive the slope flows.  A good fit
to the  Eagle Valley data was  obtained using an en-
ergy input equal to 45% of the extraterrestrial solar
                                                  170

-------
302
                           JOURNAL OF  APPLIED METEOROLOGY
                                                                                           VOLUME 21
flux and assuming that 14% of this energy was used
to cause the valley floor CBL to grow. The remaining
energy was used to remove mass from the valley in
the slope flows.
  Results of the present study provide new  insights
into the evolution of valley temperature structure and
quantify the influence of the various parameters af-
fecting temperature inversion breakup.  The model
explains the importance of the initial sunrise inver-
sion characteristics; the observed timing of the be-
ginning of inversion destruction; the mean time re-
quired to destroy  typical inversions  in the  deep
valleys of western Colorado; the weak seasonal de-
pendence of the time period required to destroy the
inversions; the effects of snow cover and ground mois-
ture and of valley topography; the patterns of warm-
ing observed in the various layers of the temperature
structure; the typical observed inversion top descent
rates of 40-150 m h~'; and the  retarded growth of
the valley CBL's relative to the flat plains case. The
thermodynamic model,  while  implicitly  incorporat-
ing up-slope mass transport, is able to simulate tem-
perature structure evolution in a wide range of valley
topography  without  taking account of  along-valley
wind systems.

  Acknowledgments.  The  research   was  accom-
plished at the Department of Atmospheric Science,
Colorado  State  University,  under funding from
Grant  ATM76-84405,  Atmospheric Sciences  Sec-
tion, National Science Foundation. The manuscript
was prepared under funding from the Environmental
Protection Agency through Interagency Agreement
AD-89-F-0-097-0, with the U.S. Department of En-
ergy.

                  REFERENCES

Leahey,  D. M., and J. P  Fnend, 1971: A model for predicting
   the depth of the mixing layer over an urban heat island with
   applications to New York City. / Appl. Meteor., 10, 1162-
   1173.
Sellers, W. D.,  1965: Physical Climatology. University of Chicago
   Press, 272 pp.
Whiteman, C. D., 1982: Breakup of temperature inversions in deep
   mountain valleys: Part I.  Observations. J. Appl.  Meteor., 21,
   270-289.
                                                171

-------
            APPENDIX D



SUMMARY OF MODIFICATIONS TO VALMET
               173

-------
                                   APPENDIX D

                       SUMMARY OF MODIFICATIONS TO VALMET

VERSION 1.0, DATED APRIL 30,  1983
     Original version of VALMET.

VERSION 1.1, DATED DECEMBER 1. 1984
     Major code changes:
      1.  Main Program
            -  The call to subroutine PRISE was  fully  incorporated .into
               the code (it had  been "commented  out" in the original
               version).
            -  The code which corrects concentrations  to  standard
               conditions was moved from subroutine GAUSS to the main
               program.
            -  The call to subroutine GAUSS for  the calculation of  the
               plume center!ine  concentration was corrected to allow the
               plume to be offset from the valley centerline.  Correction
               to standard conditions is now accomplished following the
               call to GAUSS.
      2.  Subroutine INPUT
               New code was developed to check the model  parameter  values
               that are input by the user.  An error message is written
               when the user  inputs values outside normal atmospheric
               ranges.
            -  Calculations of NBOX, NTOTI and AC were moved from the main
               program  into the  INPUT module to  facilitate the new  error
               checking code.
            -  A new error message scheme was developed for the INPUT
               module.
                                      175

-------
Subroutine PRISE
  -  A new plume rise module was developed from  the MPTER  code
     to replace the old module derived from the  CRSTER code.
  -  A new code was developed in order to bypass the  plume  rise
     calculations when SRAD and/or SVEL are set  to zero.
Subroutine GAUSS
  -  The section of code that corrects concentrations to
     standard atmospheric conditions was removed and  placed  in
     the main program.
Subroutine INGRAT
  -  A statement was added to this subroutine.
Subroutine BRKUP
  -  A new exponential decay algorithm was added to handle
     variable time step lengths.
                           176

-------
                                   TECHNICAL REPORT DATA
                            ffletue read Imtructions OH tfie rtvtrte be fort comptellngj
  REPORT NO.
                              2.
                                                            3. RECIPIENTS ACCESSION NO.
4. TITLE AND SUBTITLE
  GREEN RIVER AIR QUALITY MODEL DEVELOPMENT
  VALMET - A Valley Air  Pollution Model
                                                            6. REPORT DATE
                                                            6. PERFORMING ORGANIZATION CODE
7. AUTMOR(S)
                                                             . PERFORMING ORGANIZATION REPORT NO.
  C.  D.  Whiteman and K. J.  All wine
                                                               PNL-4728
 . PERFORMING ORGANIZATION NAME AND ADDRESS
 Battelle, Pacific Northwest Laboratory
 Richland, Washington 99532
                                                            10. PROGRAM ELEMENT NO.

                                                               CDTA1D/09-0726 (FY-85)
                                                            11. CONTRACT/GRANT NO.
                                                               IAG AD89F20970
12. SPONSORING AGENCY NAME AND ADDRESS
  Atmospheric Sciences  Research Laboratory, RTF,  NC
  Office of Research  and  Development
  Environmental Protection  Agency
  Research Triangle Park, North Carolina 27711
                                                            13. TYPE OF REPORT AND PERIOD COVERED
                                                              Final  FY-85	
                                                            14. SPONSORING AGENCY CODE
                                                               EPA/600/09
18. SUPPLEMENTARY NOTES
16. ABSTRACT

       Following a thorough  analysis of meteorological  data obtained from  deep valleys
  of western Colorado,  a  modular air pollution  model  has been developed  to simulate
  the transport and diffusion  of pollutants released  from an elevated point source in
  a  well-defined mountain valley during the nighttime and morning transition periods.
  This initial version  of the  model, named VALMET, operates on a valley  cross section
  at an arbitrary distance down-valley from a continuous point source.   The model  has
  been constructed to include  parameterizations  of the  major physical processes that
  act to disperse pollution  during these time periods.   The model has not  been fully
  evaluated.  Further testing,  evaluations, and  development of the model are needed.
  Priorities for further  development and testing are  provided.
17.
                                KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
                                               b.IDENTIFIERS/OPEN ENDED TERMS  C.  COSATI Field/Croup
18. DISTRIBUTION STATEMENT
                                               19. SECURITY CLASS (Tilts Report)
                                               UNCLASSIFIED
                                                                           21. NO. OF PAGES
                       RELEASE TO PUBLIC
                                               20. SECURITY CLASS (Tills page I
                                                                           22. PRICE
                                               UNCLASSIFIED
EPA F*rm 2220.1
               . 4-77)   ^««VIOU1 BDITION i« OBIOLCTC

-------