v>EPA
United States
Environmental Protection
Agency
Environmental Sciences Research
Laboratory
Research Triangle Park NC 27711
EPA-600/8-78-01 6a
November 1 978
d Development
User's Guide *
For RAM
Volume I.
Algorithm
Description and Use
* »'*
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9, Miscellaneous Reports
This report has been assigned to the SPECIAL REPORTS series. This series is
reserved for reports which are intended to meet the technical information needs
of specifically targeted user groups. Reports in this series include Problem Orient-
ed Reports, Research Application Reports, and Executive Summary Documents.
Typical of these reports include state-of-the-art analyses, technology assess-
ments, reports on the results of major research and development efforts, design
manuals, and user manuals.
EPA REVIEW NOTICE
This report has been reviewed by the U.S. Environmental Protection Agency, and
approved for publication. Mention of trade names or commercial products does
not constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/8-78-016a
November 1978
User's Guide For RAM
Volume I.
Algorithm Description and Use
by
D. Bruce Turner and Joan Hrenko Novak
Meteorology and Assessment Division
Environmental Sciences Research Laboratory
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental
Sciences Research Laboratory, U. S. Environmental Protection
Agency, and approved for publication. Mention of trade names
or commercial products does not constitute endorsement or
recommendation for use.
AUTHORS' AFFILIATION
The authors are on assignment with the U. S. Environmental
Protection Agency from the National Oceanic and Atmospheric
Administration, U. S. Department of Commerce.
-------
FOREWORD
Within the Office of Air, Land, and Water Use, the Environ-
mental Sciences Research Laboratory conducts a research program
in the physical sciences to detect, define, and quantify the
effects of air pollution on urban, regional/ and global atmos-
pheres and the subsequent impact on water quality and land use.
This includes research and development programs designed to
quantitate the relationships between emissions of pollutants
from all types of sources and air quality and atmospheric
effects.
The Meteorology and Assessment Division conducts research
programs in environmental meteorology to describe the roles and
interrelationships of atmospheric processes and airborne pollu-
tants in effective air, water, and land resource management.
Developed air quality simulation models (in Fortran computer
code) are made available to dispersion model users in computer
compatible form by availability of a magnetic tape from NTIS
(See Preface).
RAM is one of the five dispersion algorithms added to
UNAMAP in March 1978. RAM is based upon Gaussian dispersion
concepts of steady-state modeling. Limitations are imposed on
use of the algorithm by the assumptions of non-reactive pollu-
tants and one wind vector and one stability class as representa-
tive of the area being modeled. Also computations made under
light wind conditions should be interpreted skeptically. In
spite of these limitations RAM provides a useful short-term
(hours to a day) algorithm for point and area sources for air
pollution impact assessment.
gy and Assessment Division
m
-------
PREFACE
One of the research activities of the Meteorology and
Assessment Division focuses on the development, evaluation,
validation, and application of air quality simulation, photo-
chemical, and meteorological models capable of describing air
quality and atmospheric processes affecting the disposition of
airborne pollutants, on scales ranging from local to global.
Within the Division, the Environmental Applications Branch
adapts and evaluates new and existing meteorological dispersion
models and statistical technique models, tailors effective
models for recurring user application, and makes these models
available through EPA's computer network system.
RAM, an adaptation of Gaussian techniques previously used
in point source modeling, uses a rapid executing algorithm for
area sources based on the narrow plume hypothesis and has numer-
ous options available to increase user utility. RAM is one of
the atmospheric dispersion models on the User's Network for
Applied Modeling of Air Pollution (UNAMAP) system. The UNAMAP
system may be purchased on magnetic tape from the National
Technical Information Service (NTIS) for use on the user's
computer system.
Although attempts are made to thoroughly check out computer
programs with a wide variety of input data, errors are occasion-
ally found. In case there is a need to correct, revise, or up-
date this model, revisions may be obtained as they are issued by
completing and sending the form on the last page of this guide.
Comments and suggestions regarding this publication should
be directed to:
Chief, Environmental Applications Branch
Meteorology and Assessment Division (MD-80)
Environmental Protection Agency
RESRCH TRI PK, NC 27711
Technical questions regarding use of the model may be asked
by calling (919) 541-4564. Users within the Federal Government
may call FTS: 629-4564. Both volumes of the User's Guide are
available from NTIS, Springfield, Va. 22161.
The magnetic tape containing all Fortran source codes for
RAM, as well as for ten other dispersion models, may be ordered
from Computer Products, NTIS. Ask for UNAMAP (Version 3),
PB 277 193.
IV
-------
ABSTRACT
The information presented in this user's guide is directed
to air pollution scientists having an interest in applying air
quality simulation models. RAM is the three letter designation
for this system of efficient Gaussian-plume multiple-source air
quality algorithms and also the primary algorithm for urban
areas. RAM is a method of estimating short-term dispersion using
the Gaussian steady-state model. These algorithms can be used
for estimating air quality concentrations of relatively non-
reactive pollutants for averaging times from an hour to a day
from point and area sources. The algorithms are applicable for
locations with level or gently rolling terrain where a single
wind vector for each hour is a good approximation to the flow
over the source area considered. Calculations are performed
for each hour. Hourly meteorological data required are wind
direction, wind speed, temperature, stability class, and mixing
height. Emission information required of point sources consists
of source coordinates, emission rate, physical height, stack
diameter, stack gas exit velocity, and stack gas temperature.
Emission information required of area sources consists of south-
west corner coordinates, source side length, total area emission
rate and effective area source-height. Computation time is kept
to a minimum by the manner in which concentrations from area
sources are estimated using a narrow plume hypothesis and using
the area source squares as given rather than breaking down all
sources into an area of uniform elements. Options are available
to the user to allow use of three different types of receptor
locations: 1) those whose coordinates are input by the user,
2) those whose coordinates are determined by the model and are
downwind of significant point and area sources where maxima are
likely 'to occur, and 3) those whose coordinates are determined
by the model to give good area coverage of a specific portion of
the region. Computation time is also decreased by keeping the
number of receptors to a minimum. Volume I considers the use
and capabilities of RAM, its basis, the organization of the
computer program, and data requirements. Volume II presents RAM
example outputs, typical run streams, variable glossaries, and
Fortran source codes.
-------
CONTENTS VOLUME I
Foreward iii
Preface 1v
Abstract v
Figures - . . . ix
Acknowledgements x
1. Introduction 1
2. Recommendations 3
Uses 3
Algorithm Assumptions 3
Proper use of RAM 8
Special problems that may be encountered using
the frequency distribution version (RAMF) of
RAM 9
3. Theoretical Basis for RAM 11
Dilution by the wind 11
Dispersion results in Gaussian distributed cross
sections 11
Steady-state conditions 11
Concentration - sum of individual contributions . 11
Plume rise for point sources 12
Effluent rise for area sources 17
Concentrations from point sources 17
Concentrations from area sources 18
4. Organization of Computer Programs 19
Interrelationship of eight main programs 19
Brief descriptions of programs 22
Brief descriptions of subroutines 28
5. Data Requirements 32
RAMBLK 32
RAMQ 32
Meteorological data and RAMMET 35
RAM and RAMR 37
RAMF and RAMFR 42
CUMF 43
6. Algorithm Capabilities 44
Typical use of RAM 44
Discussion of RAM example 46
Typical RAMR use 50
Typical RAMF use 50
References 52
Appendix 54
vn
-------
CONTENTS VOLUME II
Foreward iii
Preface iv
Abstract v
Figures ix
Acknowledgements x
1. Introduction 1
2. System overview and glossary 2
3. RAMQ 20
Input data formats (RAMQDOC) 20
Run-stream example 21
Output example 22
Fortran source program listings 26
4. RAMMET 36
Input data formats (RAMMETDOC) 36
Run-stream example 38
Output example 40
Fortran source program listings 41
5. RAM 47
Input data formats (RAMDOC) 47
Run-stream example 52
Output examples 53
Fortran source program listings 72
Program to read the partial concentration out-
put file (PARTC) Ill
Format of punch card output (PUNCHCARD) 113
6. RAMF 115
Input data formats 115
Run-stream examples 120
Output example 126
Fortran source program listings 132
Program to read the hourly output file (RAMFHOUR) . 146
Program to read the daily output file (RAMFDAY) . . 148
7. RAMR 150
Fortran source program listings 150
8. RAMFR 176
Fortran source program listings 176
9. CUMF 188
Input data formats 188
Run-stream example 188
Output example 189
Fortran source program listings 201
10. RAMBLK 207
Input data formats 207
Run-stream example 207
Output example 208
Fortran source program listings 212
viii
-------
FIGURES
Number Page
1 Interrelationships of RAM system main programs 20
2 Subroutine structure of RAM and RAMR 25
3 Subroutine structure of RAMF and RAMFR 27
TABLE
Number Page
1 Exponents for Wind Profile 13
-------
ACKNOWLEDGMENTS
The authors appreciate the assitance of Adrian D. Busse for placing RAM
into the UNAMAP system, and the helpful discussions with John S. Irwin and K.
L. Calder.
We appreciate the suggestions for changes received from the Regional
Meteorologists and other contributors in EPA regions. These suggested
changes arose from their experience in running provisional versions of RAM.
RAM bears the initials of Robert A. McCormick who directed the meteor-
ology program of the Federal air pollution control effort from June 1958
until the time of his retirement in January 1973. His dedicated leadership,
encouragement, advice, and counsel throughout this peirod are greatly appre-
ciated.
The assistance of Theresa Burton, Caryl Whaley, Lea Prince, Nancy
Beasley, Bonnie Kirtz, Carolyn Johnston, Sandy Bryant, Tom Pierce, and
especially that of Pamela Hinton and Joan Emory is gratefully acknowledged.
We thank Ralph Seller for the use of his slide showing point and area sources
for the cover.
-------
1, INTRODUCTION
The RAM system includes four dispersion algorithms. The purpose of
formulating RAM is to provide a readily available computer program based on
the assumptions of steady-state Gaussian dispersion. The principal algorithm,
RAM, can be used for short-term (one-hour to one-day) determination of urban
air quality resulting from pollutants released from point and/or area sources.
The algorithms are applicable for locations with level or gently rolling
terrain where a single wind vector for each hour is a reasonable approxima-
tion of the flow over the source area considered. A single mixing height and
a single stability class for each hour are assumed representative of the
area. The use of the RAM is restricted to relatively nonreactive pollutants.
Emission information required of point sources consists of source co-
ordinates, emission rate, physical height, stack diameter, stack gas exit
velocity, and stack gas temperature. Emission information required of area
sources consists of south-west corner coordinates, source side length, total
area emission rate, and effective area source height. Output consists of
calculated air pollutant concentrations at each receptor for hourly averaging
times and a longer averaging time as specified by the user. Contributions to
the concentration in the two categories 1) from point sources, and 2) from
area sources are also given on output. The contributions to the concen-
tration from specific point and area sources can be obtained at the option of
the user.
Computations are performed hour by hour as if the atmosphere had achiev-
ed a steady-state condition. Therefore, errors will occur where there is a
gradual buildup (or decrease) in concentrations from hour to hour, such as
with light wind conditions. Also under light wind conditions the definition
of wind direction is likely to be inaccurate, and variations in the wind flow
from location to location in the area are quite probable.
-------
Briggs1 plume rise equations are used to estimate effective height of
point sources. Concentrations from the point sources are determined using
distance crosswind and distance upwind from the receptor to each source.
Considerable time is saved in calculating concentrations from area
sources by using a narrow plume simplification which considers sources at
various distances on a line directly upwind from the receptor to be repre-
sentative in the crosswind direction of the sources at those distances
affecting the receptor. Area source sizes are used as given in the emission
inventory in lieu of creating an internal inventory of uniform elements.
Options are available to allow the use of three different types of
receptor locations: 1) those whose coordinates are input by the user, 2)
those whose coordinates are determined by RAM and are downwind of significant
point and area sources where maxima are likely to occur, and 3) those whose
coordinates are determined by RAM to give good area coverage of a specific
portion of the region. Options are also available concerning the detail of
output produced.
Urban planners may use RAM to determine the effects of new source loca-
tions and of control strategies upon short-term air quality. If the input
meteorological parameter values can be forecast with sufficient accuracy,
control agency officials may use RAM to predict ambient air quality levels,
primarily over the 24-hour averaging time, to 1) locate mobile air sampling
units, and 2) assist with emission reduction tactics. Especially for con-
trol tactics, diurnal and day-to-day emission variations must be considered
in the source inventory input to the model. For most of these uses, the
optional feature to assist in locating maximum points should be used.
Computations are organized so that execution of the program is rapid, thus
real-time computations are feasible.
-------
2, RECOMMENDATIONS
USES
The principal algorithm in the system, RAM, is a short-term (one-hour to
one-day) urban algorithm for estimating air quality from point and area
sources.
Effects of either control strategies or tactics for specific short-term
periods may be examined by users. The expected effect of a proposed source
or sources can also be determined.
The spatial variation in air quality throughout the urban area or in a
portion of the area for specific periods can be estimated.
In a forecast or predictive mode such as over a 24-hour period, the
algorithm can assist in locating mobile or portable air samplers and to
assist with emission reduction tactics. Successful use of RAM in the forecast
mode is contingent on the validity of the algorithm assumptions and certainly
on the ability to forecast both the input meteorological parameter values and
the input emission parameter values.
ALGORITHM ASSUMPTIONS
Gaussian Plumes
Calculations of concentrations from point sources may be made by diluting
the emissions with the mean wind speed and considering the time-averaged
plumes over 1-hour periods to have Gaussian (normal) distributions perpendic-
ular to the plume centerTine in the horizontal and vertical.
Narrow Plume Simplification
Calculations of concentrations from area sources are made by considering
area sources at various distances on a line directly upwind from the receptor
to be representative of the sources at those distances that affect the
receptor. This assumption is best fulfilled by gradual rather than abrupt
-------
changes in area emission rates from adjacent area sources. The narrow plume
simplification is considered in more detail in the next chapter.
Meteorological Conditions Representative of Region
The meteorological input for each hour consists of a value for each of
the four parameters: wind direction, wind speed, temperature, stability
class, and mixing height, all of which should be representative of the entire
region containing the sources and receptors. Mixing height is required only
if the stability is neutral or unstable.
Steady-state
Calculations are made as if the atmosphere had reached a steady state.
Concentrations for a given hour are calculated independently of conditions
for the previous hour or preceding hours.
Concentration, Sum of Contributions
The total concentration for a given hour for a particular receptor is
the sum of the estimated contributions from each source.
Vertical Stability
Except for stable layers aloft, which inhibit vertical dispersion, the
atmosphere is treated as a single layer in the vertical with the same rate of
vertical dispersion throughout the layer. Complete eddy reflection is assumed
both from the ground and from the stable layer aloft given by the mixing
height.
Mixing Height
If vertical temperature soundings are available from a representative
location, they should be used with hourly surface temperatures to estimate
hourly mixing heights for periods with neutral or unstable stability. If a
series of National Weather Service hourly data are being processed by the
program RAMMET, two values of mixing height per day are required. These are
the maximum and minimum mixing heights as defined by Holzworth (1972). RAMMET
provides a crude interpolation to obtain hourly mixing heights. This inter-
polation does not consider hourly surface temperatures. Substitution of an
improved method for mixing height will be made in a later version of RAM.
-------
Wind Speeds and Directions
Wind speeds and directions should be hourly averages (National Weather
Service hourly observations are really not hourly averages but are averages
of a few minutes at the time of the observation, usually 5 to 10 minutes
prior to the hour). Input winds should be representative of the entire
region being modeled and for a height 10 meters above the ground surface.
The increase of wind speed with height is included based upon a power
law profile. The exponent is dependent upon the stability classification and
surface roughness effects. For any given hour, winds at various altitudes
above ground are likely to deviate considerably from this climatological mean
profile. The exponents are given in Section 3.
There is no inclusion of directional shear with height. This means that
the direction of flow is assumed to be the same at all heights over the
region. The taller the effective height of a source, the larger the expected
error in direction of plume transport. Although the effects of surface
friction are such that wind direction usually veers (turns clockwise) with
height, the thermal effects (in response to the horizontal temperature gradient
in the region) can overcome the effect of friction and cause backing (turning
counterclockwise with height) instead of veering.
In the program RAMMET, which processes National Weather Service hourly
observations, the wind directions, which are reported to the nearest 10°, are
altered by use of a random generated number from 0 to 9 which is used to
add -4° to +5° to the wind vector. The purpose of this is to prevent an
extreme overestimate of concentration at a point downwind of a source during
a period of steady wind when sequential observations are from the same
direction. Rather than allow the plume centerline to remain in exactly the
same position for several hours, the alteration allows for some variation of
the plume centerline within the 10° sector. Although this can in no way
simulate the actual sequence of hourly events (wind direction to 1° accuracy
cannot be obtained from wind direction reported to the nearest 10°), such
alterations can be expected to result in concentrations over a period of
record to be more representative than those obtained using winds to only
the 10° increments reported. (Sensitivity tests of this alteration for single
-------
sources have indicated that, where a few hours of unstable conditions are
critical tc producing high concentrations, the resulting concentrations are
extremely sensitive to the exact sequence of random numbers used, such as two
wind directions 1° apart versus two wind directions 9° apart. Differences of
24-hour concentrations from a single source by 40 to 50 per cent have appeared
in the sensitivity tests due just to the alteration.) It is therefore desir-
able to use wind information as accurate as possible as input into RAM.
Dispersion Parameter Values
The dispersion parameter values representative for urban areas are those
recommended by Briggs and included in Figure 7 and Table 8 of Gifford (1976).
These are used in the two urban algorithms in the RAM system, RAM and RAMF,
which are discussed in Section 4.
The dispersion parameter values representative for open countryside are
the Pasquill-Gifford curves (Pasquill, 1961; Gifford, 1960) which appear in
the Workbook of Atmospheric Dispersion Estimates (Turner, 1970) and also
appear as Figure 2 in Gifford (1976). These are used in the two rural
algorithms in the RAM System, RAMR and RAMFR, which are discussed in Section
4. The subroutines used to determine the open-countryside parameter values
are the same as in the UNAMAP programs PTDIS and PTMTP (U.S. Environmental
Protection Agency, 1974).
Plume Rise
Plume rise from point sources is calculated using the methods of Briggs
(1969, 1971, 1972, 1973, 1975). Although the plume rise from point sources
is usually dominated by buoyancy, plume rise due to momentum is also consid-
ered. Merging of nearby buoyant plumes is not considered. Stack downwash is
considered, but building downwash is not.
The variation of effective height of emission from area sources as a
function of wind speed is thought to be an important factor in properly
simulating dispersion in urban areas. Since this effect is seldom considered
in the compilation of urban area emission inventories, it is difficult to
have the appropriate parameters to estimate this effect. This effect can be
accounted for in RAM. The methodology used is explained in Section 3.
-------
Emission Inventories
Since for similar meteorological conditions the contribution to the
concentration at a receptor from a source is directly proportional to emis-
sion rate from that source, it is imperative to have emissions expressed
accurately. Since many air pollutant sources vary emissions with time, such
as by hour of the day or weekdays vs. weekends, attempts should be made to
include such variations. For cases that have complete emissions information,
it is usually necessary to devise a system that will calculate and store
hourly emissions for the period of record to be simulated. This can be
accomplished independently of RAM. RAM is designed to accept hourly emis-
sions in addition to annual emission rates that establish a correspondence
between emissions and exit velocity.
Removal or Chemical Reactions
Transformations of a pollutant resulting in loss of that pollutant
throughout the entire depth of each plume can be approximated by RAM. This
is accomplished by an exponential decrease with travel time from the source.
The input parameter is the time expected to lose 50% (half-life) of the
emitted pollutant. RAM does not have the capability to change this parameter
value during a given run. If the loss to be simulated takes place through
the whole plume without dependence upon concentration, then the exponential
loss may provide a reasonable simulation if the loss rate is realistic.
However, if the loss mechanism is selective, such as impaction with features
on the ground surface, reactions with materials on the ground, or dependence
on the concentration in a given small parcel of air (requiring consideration
of contributions from all sources to this parcel), the loss mechanism built
into RAM will not be very adequate.
Topographic Influences
RAM is designed for application over level or gently-rolling terrain
where the assumption of a flat plane used in the algorithm is reasonable.
Dispersion parameters for the urban algorithms, RAM and RAMF, in the RAM
systems are representative of surface roughness over urban areas (z ^ 1 m).
Dispersion parameters for the rural algorithms, RAMR and RAMFR, are represent-
ative of surface roughness over rural areas (z ^ 0.03 m). Heights of re-
-------
ceptors (a single height for all receptors is used for a given run) are
heights above ground level, not the height of the local ground level with
respect to ground level at some other point. The algorithms in the RAM
system have no influences of topography incorporated, and some difficulties
might be expected in attempting to do so. Under unstable conditions plumes
may tend to rise over terrain obstructions. Under stable conditions leveled-
off plumes may remain at nearly the same mean-sea-level height but may be
expected to alter the plume path in response to the terrain features, thus
having a different wind direction locally than that specified for the region.
Fumigation
The transitional phenomenon of fumigation, the elimination of an inver-
sion layer containing a stabilized plume from below causing mixing of pollu-
tants downward which results in uniform concentrations with height beneath
the original plume centerline, is not included in calculations made by RAM.
Conditions specified for each hour are calculated as if a steady-state had
been achieved for those specified conditions.
PROPER USE OF RAM
The closer the situation to be simulated agrees with the assumptions
stated above, the greater the expectation of reasonable results. The narrow
plume simplification is most reasonable for situations where there are no
great variations in area emission rates for adjacent area sources.
The higher the physical and resulting effective heights of point sources,
the greater the chance for poorer results since actual directional shear in
the atmosphere, not included in the algorithm, will cause plumes to move in
directions different from the direction input to the model. Also, the higher
the source height, the greater the potential for encountering layers in the
atmosphere having dispersion characteristics different from those being used.
As pointed out above, it is necessary to properly consider variations in
emissions.
Reliable meteorological inputs are also necessary. The light wind
situation is most likely to violate assumptions, since variations in the flow
over the region are likely to occur, and the slower transport may cause
8
-------
buildup of pollutants from hour to hour. Unfortunately, these are the kinds
of conditions that are likely to be associated with maximum 3-hour and
24-hour concentrations in urban areas. These light wind situations do not
conform to the assumptions of Gaussian steady-state models. Research is
underway on models more appropriate for these situations.
RAM is not appropriate for making concentration estimates for topo-
graphic complications. The greater the departure from relatively flat
terrain conditions, the greater the departure from the assumptions of the
algorithm.
RAM is most applicable for pollutants that are quite stable chemically
and physically. A general loss of pollutant with time can be accounted for
by the algorithm. Selective removal or reaction at the plume-ground inter-
face or dependence upon concentration levels is not capable of being well
handled by RAM.
SPECIAL PROBLEMS THAT MAY BE ENCOUNTERED USING THE FREQUENCY DISTRIBUTION
VERSION(RAMF) OF RAM
RAMF, a version of RAM that has been changed so that many of the options
of RAM cannot be used, calculates 24-hour concentrations (by hour-by-hour
simulation) over a relatively long period of record, such as a year. A
principal purpose of utilizing this version of RAM would be to identify
24-hour periods that produce high concentrations so that these periods can
be further examined to compare concentrations with air quality standards.
Current air quality standards are written so that the extremes are very
important. The concentrations for the day out of the year with the highest
concentrations and the day with the second highest concentrations are re-
quired for comparisons with Federal ambient air quality standards. Because
of the difficulties with light winds discussed above, daily concentrations
for such periods may be underestimated.
In order to make computations as economical as possible, it is desir-
able to keep the number of receptor locations to a reasonably small number.
Computer costs of running RAM are approximately directly proportional to the
number of receptor points. Also in order to have the same receptor numbers
refer to the same locations (coordinates) throughout a run of RAMF, the
-------
option to locate receptors downwind of significant sources cannot be used.
Therefore, with a limited number of receptors and difficulty with the light
wind days, it is difficult to determine both the days with the highest and
Second highest concentrations and the location of the maximum concentration
on those days. It appears that several runs with careful examination of
output may be required for optimum analysis.
10
-------
3, THEORETICAL BASIS FOR RAM
The basis for RAM is also discussed in Novak and Turner (1976) which is
included in this users guide as the Appendix.
DILUTION BY THE WIND
Emissions from continuous sources are assumed to be stretched along the
direction of the wind by the speed of the wind. Thus the stronger the wind,
the greater the dilution of the emitted plume. To account for an increase in
wind speeds with height from point of measurement to stack top, a power law
increase with height is used. The exponent used is a function of stability.
DISPERSION RESULTS IN GAUSSIAN-DISTRIBUTED CROSS SECTIONS
The time-averaged concentration distributions through a dispersed plume
resulting from a continuous emission from a point source or an area element
are considered to be Gaussian in both the horizontal and vertical directions.
Modification of the vertical distribution by eddy reflection at the ground or
at a stable layer aloft is allowed. This eddy reflection is accomplished by a
"folding back" of the portion of the distribution that would extend beyond the
barrier if it were absent. This is equivalent to a virtual image source
beneath the ground (or above the stable layer).
STEADY-STATE CONDITIONS
Concentration estimates are made for each simulated hourly period using
the mean meteorological conditions for that hour as if a steady-state condi-
tion had been achieved. Steady-state Gaussian plume equations are used for
point sources, and the integrations of these equations are used for area
sources.
CONCENTRATION SUM OF INDIVIDUAL CONTRIBUTIONS
The total concentration of a pollutant at a receptor is taken as the sum
of the individual concentration estimates from each point and area source
11
-------
affecting that receptor, that is, concentrations are additive. Concentration
estimates for averaging times longer than one hour are determined by arith-
metic averaging the hourly concentrations during the period.
PLUME RISE FOR POINT SOURCES
The methods of Briggs have been used to include effects of downwash in
the lee of the stack, plume rise due to momentum, and plume rise due to
buoyancy.
In the preprocessing of the emission data which ranks the point sources
in order of the significance of their expected impact, only the rise due to
buoyancy is considered since this is expected to be the dominant effect in
plume rise. Also, in determining the distance to maximum concentration that
will allow the algorithm to place receptors downwind of point sources, only
the rise due to buoyancy is considered.
However, in the computation of the effect of each point source upon
receptors for each simulated hour, all three of the above mentioned effects--
stack downwash, momentum plume rise, and buoyant plume rise-- are considered.
These computations will now be discussed in some detail.
Wind Speed
In RAM the input wind speed data are assumed representative for a height
of 10 m above ground. The wind speed at the physical stack height h is
calculated from:
u(h) = u (h/10)p
where u is the input wind speed for this hour, and the exponent p is a func-
tion of stability. If u(h) is determined to be less than 1 m s~ , it is set
equal to 1 m s ~ .
12
-------
TABLE 1. EXPONENTS FOR WIND PROFILE
Stability class
A
B
C
D
E
F
URBAN (RAM)
exponent
0.15
0.15
0.20
0.25
0.40
0.60
RURAL (RAMR)
exponent
0.07
0.07
0.10
0.15
0.35
0.55
Stack Downwash
Modification of the physical stack height to allow for stack downwash is
done following the suggestion on page 4 of Briggs (1973). The h1 is found
from:
h1 = h + 2 f (vs/u(h)) - 1.5 | d for v$ < 1.5 u(h)
h1 = h for v > 1.5 u(h) ,
where v is stack gas velocity, m s~ , and d is inside stack-top diameter, m.
This h1 is used throughout the remainder of the plume height computation.
Unstable or Neutral - Crossover between Momentum and Buoyancy
For most plume rise situations the value of the Briggs buoyancy flux
4 -3
parameter, F, in m s is needed. The following equation is equivalent to
equation (12), page 63 of Briggs (1975):
2
g v d AT
where AT = T - T, T is stack gas temperature, K, and T is ambient air
o j
temperature, K.
For cases with stack gas temperature greater than ambient air temperature,
it is necessary to determine whether the plume rise is dominated by momentum
or buoyancy. The cross-over temperature difference (AT) is determined for F
\f
less than 55, and for F greater than or equal to 55. If the difference
between stack gas temperature and ambient air temperature, AT, exceeds the
13
-------
(AT)c, the plume rise is assumed to be buoyancy dominated; if less than this
amount, the plume rise is assumed to be momentum dominated (see below): 1) For
F less than 55, the crossover temperature difference is found by setting
Equation (5.2), page 59 of Briggs (1969) equal to the combination of Equations
(6) and (7), page 1031 of Briggs (1971) and solving for AT. The result is:
(AT) - 0.0297 T v 1/3/ d2/3
U 5 S
2) For F equal to or greater than 55, the crossover temperature difference is
found by setting Equation (5.2), page 59 of Briggs (1969) equal to the com-
bination of Equations (6) and (7), page 1031 of Briggs (1971) and solving for
AT. The result is:
(AT) = 0.00575 T vc2/3/d1/3
\j j O
Unstable or Neutral Buoyancy Rise
For situations where AT exceeds (AT) as determined above, buoyancy
trf
is assumed to dominate. The distance to final rise xf in kilometers,
is determined from the equivalent of Equation (7), page 1031, of Briggs
(1971), and the assumption that the distance to final rise is 3.5 x*, where
x* is the distance at which atmospheric turbulence begins to dominate
entrainment.
For F less than 55:
xf = 0.049 F5/8
For F equal to or greater than 55:
xf = 0.119 F2/5
The plume height, H, in meters, is determined from the equivalent of the
combination of Equations (6) and (7), page 1031 of Briggs (1971):
For F less than 55:
H = h1 + 21.425 F3/4/u(h)
For F equal to or greater than 55:
H = h1 + 38.71 F3/5/u(h)
Unstable or Neutral Momentum Rise
For situations where the stack gas temperature is less than or equal to
the ambient air temperature, the assumption is made that the plume rise is
14
-------
dominated by momentum. If AT is less than (AT) , the assumption is also made
\f
that the plume rise is dominated by momentum. The plume height is calculated
from Equation (5.2), page 59 of Briggs (1969):
H = h1 + 3d vs/u(h)
Briggs (1969) suggests on page 59 that this equation is most applicable when
v /u is greater than 4. Since momentum rise occurs quite close to the point
of release, the distance to final rise is set equal to zero.
Stable Crossover between Momentum and Buoyancy
For stable situations the stability parameter s is calculated from the
equation on page 1031 of Briggs (1971):
T
As an approximation, for stability class E, or 5, 96/9z is taken as 0.02 K
nf , and for stability class F, or 6, ae/9z is taken as 0.035 K m~ .
For cases with stack gas temperatures greater than ambient air temper-
ature, it is necessary to determine whether the plume rise is dominated by
momentum or buoyancy. The crossover temperature difference (AT) is found by
setting Equation (59), page 96 of Briggs (1975) equal to Equation (4.28), page
59 of Briggs (1969) and solving for AT. The result is:
(AT)c = 0.01958 T vs s1/2
If the difference between stack gas temperature and ambient air temper-
ature, AT, exceeds the (AT) , the plume rise is assumed to be buoyancy domi-
nated; if less than this amount, the plume rise is assumed to be momentum
dominated.
Stable Buoyancy Rise
For situations where AT exceeds (AT) as determined above, buoyancy is
assumed to dominate. The distance to final rise, in kilometers, is deter-
mined by the equivalent of a combination of Equations (48) and (59), page 96,
in Briggs (1975):
xf = 0.00207 u(h) s ~1/2
15
-------
The plume height is determined by the equivalent of Equation (59), page 96
of Briggs (1975):
H = h1 + 2.6
u(h)s
Stable
Momentum 'Rise
Where the stack gas temperature is less than or equal to the ambient air
temperature, the assumption is made that the plume rise is dominated by momen-
tum. If AT is less than (AT)c> the assumption is also made that the plume
rise is dominated by momentum. The plume height is calculated from Equation
(4.28), page 59, of Briggs (1969):
H = h1 + 1.5
vs d T
4Tsu(h)
1/3
.-1/6
The equation for unstable-neutral momentum rise is also evaluated. The
lower result of these two equations is used as the resulting plume height.
All Conditions Distance Less Than Distance to Final Rise
For unstable, neutral, or stable conditions, if the distance upwind from
receptor to source x, in kilometers, is less than the distance to final rise,
the equivalent of Equation (2), page 1030 of Briggs (1972) is used to deter-
mine plume height:
160. F1/3 x 2/3
H = h
u(h)
This will be used only for buoyancy dominated conditions. Should this value
exceed the final rise for the appropriate condition, the final rise is sub-
stituted instead.
General
In working through the receptors to determine concentrations for a given
hour, the first time a source is found to lie upwind of a receptor the fol-
lowing quantities are determined and stored for that source: u(h), h', F, H,.
16
-------
and x,:. These quantities are then used each time this source is encountered
during this hour without recalculation. Only if the upwind receptor-source
distance is less than xf, the effective plume height is determined for each
occurrence by the last mentioned equation.
EFFLUENT RISE FOR AREA SOURCES
RAM can include changes in effective height with wind speed for area
sources. The input area source height HA is assumed to be the average phy-
sical height of the area source plus the effluent rise with a wind speed of 5
m s~ . The user specifies the fraction f of the input height that represents
the physical height, h . This fraction is the same for all area sources in
the inventory.
hp=fHA '
The difference is the effluent rise for a wind speed of 5 m s~ .
AH (u = 5) = HA - hp .
If f = 1, there is no rise and the input height is the effective height for
all wind speeds. For any wind speed u the rise is assumed to be inversely
proportional to wind speed and is determined from:
AH (u) = 5
-------
eddy reflection at the ground is assumed. For unstable or neutral conditions
where vertical dispersion is great enough that uniform mixing is assured
beneath an elevated inversion, Equation (A4) in the Appendix is used. For
unstable or neutral conditions where vertical dispersion is still small com-
pared to the mixing height, Equation (A5),in the Appendix is used, which
incorporates multiple eddy reflections from the ground and the base of the
stable layer aloft. The simplifications to the above mentioned equations,
which occur if the height of the receptor z is assumed at ground-level, are
incorporated into RAM.
CONCENTRATIONS FROM AREA SOURCES
The total concentration at a receptor from the two-dimensional area
source distribution is calculated using the narrow plume simplification
discussed by Gifford and Hanna (1971). This simplification is assumed because
the upwind zone of influence affecting a receptor (an upwind oriented point
source plume) is normally quite narrow in comparison with the characteristic
length scale for appreciable changes in the magnitude of the area-source
emission rate itself. Under these circumstances the two-dimensional integral
that expresses the total area-source contribution to concentration at a
receptor can be replaced approximately by a one-dimensional integral. This
integral involves only two things: 1) knowledge of the distribution of the
area-source emission rates along the line in the direction of the upwind
azimuth from the receptor location, and 2) the meteorologically dependent
function that specifies the crosswind-integrated concentration in the Gaussian
plume from a point source. In using this area source technique, Gifford and
Hanna assumed area source emissions at ground level allowing integration
upwind to be accomplished analytically. In RAM the area sources are con-
sidered to have an effective height, requiring the integration to be accom-
plished numerically. Equations used to perform the calculations are given in
the Appendix. Internal tables of integrations for one to three effective area
source heights are calculated at the beginning of each simulated hour using
the specific meteorological conditions for that hour. The total concentration
from all area sources is determined by performing the integration piecewise
over each source in the upwind direction from the receptor until the farthest
boundary of the source region is reached.
18
-------
4, ORGANIZATION OF COMPUTER PROGRAMS
INTERRELATIONSHIP OF EIGHT MAIN PROGRAMS
The eight main programs in the RAM system consist of the four versions
of RAM: RAM, RAMR, RAMF, and RAMFR; three preprocessor programs: RAMQ,
RAMMET, and RAMBLK; and one postrun processor: CUMF. The relationships are
shown in Figure 1.
The four versions of RAM are similar, but their differences can be
categorized by the following 2 by 2 matrix.
NORMAL FREQUENCY
RAM DISTRIBUTION RAM
URBAN a's
RURAL a's
RAM RAMF
RAMR RAMFR
Two of the versions (RAM and RAMF) have dispersion parameters, a,
representative of urban areas. These a's are based upon the tracer exper-
iments performed in St. Louis and reported by McElroy and Pooler (1968) and
are most representative for z of about 1 m. The equations used to repre-
sent these a values are those of Briggs as reported in Gifford (1976). The
urban a's are functions of distance between source and receptor, and of
atmospheric stability class where the class is specified by open country
conditions.
The other two versions (RAMR and RAMFR) are for rural conditions and
utilize the dispersion parameter values of Pasquill-Gifford (Pasquill, 1961;
Gifford, 1960) representative for ZQ of about 0.03 m as used in the UNAMAP
programs PTMAX, PTDIS, and PTMTP. These values are equivalent to the disper-
19
-------
EMISSION DATA
R
A M Q
DISK FILE OF EMISSION
DATA, AREA SOURCE MAP
ARRAY, ETC.
I
HOURLY
MET
CARDS
4
HOURLY SURFACE DATA
AND
DAILY MIXING HEIGHT DATA
FOR 1 YEAR
FROM NATIONAL CLIMATIC CENTER
RAMBLK
±j
OR
RAM OR RAMR
BLOCK DATA
R A M M
E T
DISK FILE
OF MET. DATA
r
1
\r
i i
~\ i
r
HOU
ME
DA
OR
i
RAMFOR RAMFR
r
HOURLY
AND
DAILY
CONCENTRATIONS
365 OR 366
DISK RECORDS
OF 24-HOUR
CONCENTRATIONS
C U M F
MAGNETIC TAPE
WITH HOURLY
CONCENTRATIONS
FOR 1 YEAR
*NOTE: RAMBLK run by
user only for alternate
dispersion parameters.
PRINT
CUMULATIVE
FREQUENCY
DISTRIBUTIONS
PLOT
CUMULATIVE
FREQUENCY
DISTRIBUTIONS
Figure 1. Interrelationships of RAM system main programs
?n
-------
sion parameter values given in Figures 3-2 and 3-3 of the Workbook of
Atmospheric Dispersion Estimates (Turner, 1970), and may be revised in the
future according to Pasquill and Smith (Pasquill, 1974) in order to include
effects of surface roughness.
Looking at the other dimension of the matrix, the normal RAM is intended
for application to one or several days but not for long periods of record.
These versions have a full range of options available to the user.
The RAMF and RAMFR versions calculate frequency distributions of concen-
trations for a full year's data; therefore, to insure some degree of effi-
ciency, most of the options of RAM are not available. For this version the
receptor coordinates must be specified by the user at the beginning of the
run to insure that frequency distributions of concentrations are for a fixed
set of receptor locations.
The RAM system has four other main programs in addition to the four
versions of RAM. Figure 1 shows the interrelationships of these four programs
to the four versions of RAM.
RAMBLK is executed to produce BLOCK DATA for the RAM programs, which
help determine distance td maximum concentration. Several subroutines are
called including appropriate subroutines to determine dispersion parameter
values as functions of stability class and source-receptor distance. The
data produced are coefficients and exponents for the various ranges of
effective height of emission and are used to determine maximum xu/Q (relative
concentration normalized for wind speed) for point sources and distance to
maximum concentration for point and area sources as functions of stability
class and effective height of emission. This program is executed for each
given method of determining dispersion parameter values a and a . Therefore,
RAMBLK was executed once using the subroutines for urban dispersion parameters
to produce BLOCK DATA for RAM and RAMF. RAMBLK uses the same subroutines as
RAM. RAMBLK was executed a second time using the subroutines for rural
dispersion parameters to produce BLOCK DATA for RAMR and RAMFR. Both of
these outputs from RAMBLK urban parameters and rural parameters -- are
used in RAMQ. However, unless the user needs to modify one of the RAM
models in order to use some other system of dispersion parameters, there is
21
-------
no need to execute RAMBLK since the outputs of RAMBLK are already incorpor-
ated in the RAM versions as a result of the initial executions of RAMBLK.
RAMQ processes emission data for all four versions of RAM. Its prin-
cipal task is to set up the area source map array. The area source map array
provides a correspondence between locations (referred to by coordinates) and
area source number. Other tasks such as ranking sources according to set
criteria are also accomplished.
If meteorological data, including hourly mixing height, are entered into
any of the versions of RAM with data for each hour on a card, RAMMET is not
used. If a period of record of hourly surface data from the National Clima-
tic Center in card image format is being used, RAMMET is executed. Input
consists of both 1) one year's surface data in the form of one observation
per hour, and 2) one maximum and one minimum mixing height per day. The
program primarily determines hourly stability and interpolates to obtain
estimates of hourly mixing height. The output data are organized so that a
single disk record is produced for each day. Output from RAMMET may be used
as input to any of the versions of RAM.
The remaining main program in the RAM system is CUMF. Although RAMF and
RAMFR can be run for a number of sampling times, CUMF is used with disk
output for runs of RAMF or RAMFR for 24-hour sampling times to determine
cumulative frequency distributions for the 24-hour concentrations for each
receptor location. A table for each receptor is produced ranking the
24-hour concentrations for the year. The program also utilizes a Cal-Comp
plotter to produce cumulative frequency plots of estimated concentrations.
The plot for each receptor is concentration on a log scale versus frequency
on a probability scale.
BRIEF DESCRIPTIONS OF PROGRAMS
The interrelationships of the eight main programs are shown in Figure 1.
Brief descriptions of these main programs follow:
RAMBLK - This main program determines xu/Q maxima and distance to the point
of maximum for point sources as functions of stability class and
effective height of emissions. Coefficients and exponents relating
22
-------
these two parameters to effective height of emission are determined
for various stability and effective height range combinations.
These coefficients and exponents as well as ones for determining
the distance of the maximum concentration downwind from the edge
of an area source are output from this program in the form of punch
cards to be used as block data in RAMQ and the four versions of
RAM. The subroutines DBTRCB and JMHCZB, which are similar to
subroutines DBTRCU and JMHCZU in RAM, are used for estimation of
maximum concentration. In order to produce the block data, BLOCK,
for urban areas, RAMBLK calls the subroutines BRSYSZ and BRSZ for
calculation of urban dispersion parameters. And similarly, in
order to produce the block data, BLOCKR, for open countryside,
RAMBLK calls the subroutines PGSYSZ and PGSZ for calculation of
rural dispersion parameters.
RAMQ - This program primarily processes emission inventory information so
that it can be used later in any of the four RAM versions. An
important aspect of this is the construction of the area source map
array which allows a correspondence between any location in the
area source region and the number of the area source at that
location. All source coordinates in units convenient to the user
(user units) are converted to internal units. An internal unit is
a length such that any area source side length used in a given run
can be expressed as an integer multiple of an internal unit. The
internal unit is generally equal to the length of the side of the
smallest area in the emission inventory. The user must determine
the internal unit length and specify it in user units. Both point
and area sources are ranked according to expected impact at ground
level. The 25 point sources and the 10 area sources with the
greatest expected ground-level impact are listed. Also, the total
emissions from various physical heights for both point and area
sources are listed. This aids the user in determining area source
heights and the number to be used. Generated information is placed
on disk files to be used as input to the RAM programs. RAMQ
requires BLOCK, the data generated by RAMBLK for urban areas,
23
-------
and BLOCKR, the data generated by RAMBLK for open countryside.
RAMQ output is required as input to all four versions of RAM.
RAMMET - This program processes meteorological data for one year. The data
input consists of hourly meteorological records in the standard
card format 144 of the National Climatic Center and twice-a-day
estimates of mixing height (minimum and maximum). Hourly stability
class is determined using the objective method of Turner (1964)
based on Pasquill's technique (Pasquill, 1961). Shifts by only one
stability class are allowed for adjacent hours. Hourly mixing
height is interpolated from the twice-a-day estimates. Hourly
meteorological data of wind direction, wind speed, temperature,
stability class, and mixing height are written into a file with one
record per day for the entire year. The subroutine RANDU which
generates random numbers is called by RAMMET. RANDU is a library
subroutine of UNIVAC 1110's MATH-PACK. (For use on other computers
this call must be replaced by a call to a suitable random number
generator, or, to be consistent with outputs generated by EPA test
data, use the set of random numbers furnished in the CRSTER file of
UNAMAP.) Meteorological data may be input into the RAM versions
either as the output from RAMMET or as card input, one card per
hour. (Data files produced by RAMMET are also compatible as input
to two other dispersion programs MX24SP and CRSTER. These are not
directly related to RAM.)
RAM - This short-term Gaussian steady-state model estimates concentra-
tions of stable pollutants from urban point and area sources. The
general structure of RAM and the subroutines called by RAM (and the
rural version RAMR) are given in Figure 2. Hourly meteorological
data are required. Hourly concentrations and averages over an
averaging time less than or equal to 24 hours can be estimated.
RAM is normally not executed for a time period exceeding several
days. Briggs plume rise is used. Pasquill-Gifford dispersion
equations with dispersion parameters considered valid for urban
areas are used in the model. Concentrations from area sources
determined using the narrow plume hypothesis; that is, sourc^
24
-------
RAM (RAMR)
BLOCK (BLOCKR)
(PREPARED BY RAMBLK)
-READ DATA FROM DISK FILES
(PREPARED BY RAMQ)
-READ DATA FROM CARDS
-LOOP ON DAYS
-READ MET DATA FROM CARDS
(OR FROM DISK PREPARED BY RAMMET)
-ANGARC
-JMHREC
-JMHHON
-LOOP ON HOURS
-READ HOURLY EMISSIONS
-JMHPTU (JMHPTR)
' DBTRCU (DBTRCR)
BRSYSZ (PGSYSZ)
JMH54U (JMH54R)
I JMHCZU (JMHCZR)
I BRSZ (PGSZ)
-JMHARE
-JMHPOL
-JMHOUR
-JMHFIN
EXIT
Figure 2. Subroutine structure of RAM and RAMR
25
-------
directly upwind are considered representative of area source
emissions affecting the receptor. Special features of the model
include determination of receptor locations downwind of significant
sources and determination of locations of uniformly spaced recep-
tors to ensure good area coverage with a minimum number of recep-
tors. RAM allows use of 250 point sources, 100 area sources, and
150 receptors.
RAMR - This program differs from RAM in that it is applicable for loca-
tions in basically rural surroundings. Because of this the Pas-
quill-Gifford dispersion parameter values are used. The subrou-
tines PGSYSZ and PGSZ will duplicate values of a and a from the
curves in the Workbook of Atmospheric Dispersion Estimates (Turner,
1970). Also the mixing heights generated for rural areas by RAMMET
are used. In general no area sources would be expected for most
applications of RAMR.
RAMF - This program is designed to allow computations for a full year of
record. The general structure of RAMF and RAMFR along with the
subroutines called by these two programs are given in Figure 3.
Many of the options of RAM are not available in this version
because the receptor locations must remain the same from period to
period. All receptor locations must be read in as input; none can
be generated. This is to prevent changing receptor locations from
day to day as in RAM when using the option to generate receptors
downwind of significant sources. Although computations are
performed hour-by-hour, emphasis is upon averaging times longer
than one hour. Only output for the selected averaging time is
printed (note that only output for a 24-hour averaging time is
compatible with the program CUMF, see below.) However, hourly
concentrations and averages for the averaging time selected for
each receptor are transferred to a magnetic tape and can be pro-
cessed by the user to yield averages for other time periods.
RAMFR - This program is similar to RAMF but differs in that dispersion
parameter values and mixing heights representative for rural areas
26
-------
RAMF (RAMFR)
-BLOCK (BLOCKR)
(PREPARED BY RAMBLK)
-READ DATA FROM DISK FILES
(PREPARED BY RAMQ)
-READ DATA FROM CARDS
-LOOP ON DAYS
-READ MET DATA FROM DISK
(PREPARED BY RAMMET)
-ANGARC
-LOOP ON HOURS
-READ HOURLY EMISSIONS
JMHPTU (JMHPTR)
' DBTRCU (DBTRCR)
BRSYSZ (PGSYSZ)
-JMH54U (JMH54R)
JMHCZU (JMHCZR)
I BRSZ (PGSZ)
-JMHARE
-JMHPOL
-JMHFD
EXIT
Figure 3. Subroutine structure of RAMF and RAMFR
27
-------
are used. This program should be applied only to sources in rural
surroundings.
CUMF - This program uses disk output from RAMF or RAMFR to print and plot
cumulative frequency distributions of 24-hour concentrations over
the one-year period for each receptor. Print-out consists of
ranking concentrations from lowest to highest with Julian day
number associated with each concentration also printed. The annual
arithmetic average concentration is also given. The plot is log of
concentration against cumulative frequency on a probability scale.
Subroutines GRAPH and PROB are called by CUMF. Also the following
subroutines that are part of the CALCOMP Basic Software package are
called by CUMF: PLOTS, SYMBOL, NUMBER, PLOT, and LINE.
BRIEF DESCRIPTIONS OF SUBROUTINES
The subroutine and function descriptions that follow are in the order in
which they are called by RAM.
ANGARC - This function determines the appropriate arctan of the east resul-
tant wind component over the north resultant wind component with
the resulting angle between 0° and 360°.
JMHREC - This subroutine called by RAM and RAMR determines receptor locations
downwind of significant sources based upon the resultant meteoro-
logical conditions for the averaging period, usually 3 or 24 hours.
Plume rise and effective height of emission are calculated. The
distance of the maximum concentration is determined as a function
of the stability and the effective height of emission in order to
locate the position of a receptor. Two receptors are generated for
each significant point source, one at the expected distance of
maximum concentration and one at twice this distance. One receptor
is generated for each significant area source at the expected
distance of maximum concentration.
JMHHON - This subroutine called by RAM and RAMR generates additional recep-
tors within a specified area in order to give adequate coverage of
that area with the minimum number of receptors. Receptors are
placed equidistant from nearby receptors resulting in a honeycomb
28
-------
array. The distance between receptors is an input to the main
program. Proposed receptors located closer than half this distance
to any other receptor are not included.
JMH54U - This subroutine called by RAM and RAMF generates tables of xu/qn
(relative concentration normalized for wind speed) from area
sources that extend from a receptor to various upwind distances.
A table is produced for each area source height. One to three
heights can be used. This subroutine calls subroutine JMHCZU which
performs the urban dispersion calculation obtaining a from BRSZ.
JMH54R - This subroutine called by RAMR and RAMFR is similar to JMH54U
except that it calls JMHCZR which calls PGSZ to determine rural
dispersion parameter values.
JMHCZU - This subroutine called by JMH54U calculates concentrations from
infinite crosswind line sources at a distance x upwind from a
receptor. To obtain the vertical dispersion parameter value a ,
subroutine BRSZ is called.
JMHCZR - This subroutine called by JMH54R is similar to JMHCZU except that
rural dispersion parameter values are determined from PGSZ.
JMHCZB - This subroutine called by RAMBLK uses subroutines for both urban
and rural dispersion parameters, BRSZ and PGSZ.
BRSZ - This subroutine called by JMHCZU in RAM and RAMF determines the
value of the vertical dispersion parameter a for a given upwind
distance of receptor to source. The parameter values are for urban
areas from the experiments of McElroy and Pooler (1968).
PGSZ - This subroutine called by JMHCZR in RAMR and RAMFR determines the
value of the vertical dispersion parameter a for a given upwind
distance of receptor to source. The parameter values are for rural
areas and are the same as those given in the Workbook of Atmospheric
Dispersion Estimates (Turner, 1970).
JMHARE - This subroutine called by all four versions of RAM integrates along
the line upwind from the receptor in order to obtain the effect of
all area sources along this line. This is accomplished by finding
29
-------
the closest and farthest distance of each source along this path
and calling the subroutine JMHPOL for each distance. JHMPOL inter-
polates between values in the tables generated by JMH54U or JMH54R
in order to obtain the effect of area sources to the specific
distance.
JMHPOL - This subroutine called by JMHARE interpolates for a given distance
from the values in the tables generated by subroutine JMH54U or
JMH54R. This yields the effect of an area source at the given
height extending upwind to this distance.
JMHPTU - This subroutine is called by RAM and RAMF to determine concen-
trations at receptors from point sources in urban areas. Subrou-
tine DBTRCU and the dispersion parameter routine BRSYSZ are called
to complete the computations.
JMHPTR - This subroutine called by RAMR and RAMFR determines concentrations
at receptors from point sources in rural areas. Subroutine DBTRCR
and the dispersion parameter routine PGSYSZ are called to complete
the computations.
DBTRCU - This subroutine is called by JMHPTU to determine the relative
concentration at a receptor from a point source in an urban area at
a given upwind and crosswind distance. This subroutine calls
BRSYSZ.
DBTRCR - This subroutine is called by JMHPTR to determine the relative
concentration at a receptor from a point source in a rural area at
a given upwind and crosswind distance. This subroutine calls
PGSYSZ.
DBTRCB - This subroutine is called by RAMBLK to determine the relative
concentration at a receptor from a point source in either rural or
urban surroundings. This subroutine calls BRSYSZ and PGSYSZ.
BRSYSZ - This subroutine called by DBTRCU of RAM and RAMF, determines the
values of a . and a for a given upwind distance of source to
receptor. The parameter values are for urban areas.
30
-------
PGSYSZ - This subroutine called by DBTRCR of RAMR and RAMFR determines the
values of a and a for a given upwind distance of source to
receptor. The parameter values are for rural areas.
JMHOUR - This subroutine called by RAM and RAMR writes hourly output.
JMHFIN - This subroutine called by RAM and RAMR produces output for the
basic averaging time, usually 24 hours.
JMHFD - This subroutine called by RAMF and RAMFR is very similar to JMHFIN
but also writes concentrations for the selected averaging time to
disk for further processing by CUMF.
GRAPH - This subroutine called by CUMF uses the plotter to draw the base
chart for the plot of log of concentration against probability.
The following subroutines that are part of the CALCOMP Basic
Software package are called by GRAPH: PLOT, SYMBOL, and NUMBER.
Also; subroutine LGAXS, which is part of the Scientific Applica-
tions Category of CALCOMP1s Functional Software Library, is also
called by GRAPH.
PROB - This function called by CUMF determines plotting position on the
probability scale.
31
-------
5, DATA REQUIREMENTS
This chapter is intended to show the user the general data requirements
for the RAM programs.
RAMBLK
The program RAMBLK requires input of a single digit (1 if data for
rural areas are to be generated, or 2 if data for urban areas are to be
generated). Four subroutines are called, two (one for urban, one for
rural) that determine both dispersion parameter values a and a as functions
of stability class and source-receptor distance, and the other two (one for
urban, one for rural) that determine a only.
If a user finds it necessary to run RAMBLK (because of a need to use
dispersion coefficients not routine in RAM), it will be necessary to closely
follow the instructions on modification of the output in order for it to be
compatible with RAMQ. Specific instructions are given in section 10 of
Vol. II.
RAMQ
Program RAMQ processes the emissions. Either point sources, or area
sources, or both types may be included. When both types of sources are
included in the output from RAMQ, the user still has the option in RAM to
select one or both source types. Most any rectangular coordinate system can
be used provided that the positive quadrant is used, that is, that all
coordinate values are positive and a single coordinate system is used for
both point and area sources. The scale of the coordinate system is com-
pletely arbitrary. The input is as follows (variable names and card formats
are given in explanatory material with the RAMQ runstream example).
RAMQ-CARD 1 Up to 80 alphanumeric characters for a title.
32
-------
RAMQ-CARD 2-4 variables
a) Estimated number of point sources (this must be set to zero if the
run is for area sources only.)
b) Estimated number of area sources (this must be set to zero if the
run is for point sources only.)
c) Pollutant-indicator (3 is used for SCL, and 4 for suspended
particulate).
d) Dispersion parameter indicator (1 is used for rural, and 2 for
urban).
RAMQ-CARD 3-2 variables
a) Number of user units per internal unit. (An internal unit is a
length such that any area source side length in this run is a
multiple of the internal unit.)
b) Multiplier constant to convert length in user units to kilometers
(for example, if the user units are in miles, this constant would
be 1.609344).
RAMQ-CARD TYPE 4-9 variables (one of these cards for each point source).
a) Point Source Identification (12 characters)
b) East Coordinate (user units)
c) North Coordinate (user units)
d) SCL Emission Rate (grams/second)*
e) Particulate Emission Rate (grams/second)*
f) Physical Stack Height (meters)
g) Stack Gas Exit Temperature (K)
h) Stack Inside Diameter (meters)
j) Stack Gas Exit Velocity (meters/second).
To indicate the end of point source cards, the words "END POINTS" are
punched in card columns 1 to 10.
*Emission rates for other pollutants may be substituted for sulfur diox-
ide and particulates. If substitutions are made, changes in data statements
are necessary in order to have the proper pollutant names on printed output.
33
-------
RAMQ-CARD TYPE 5-6 variables (one of these cards for each area source).
a) East coordinate of Southwest Corner (user units)
b) North Coordinate of Southwest Corner (user units)
c) Area Source Side Length (user units)
d) S02 Emission Rate (grams/second)**
e) Particulate Emission Rate (grams/second)**
f) Height of Emission (meters)
Although only one pollutant can be considered for a given run of either
RAMQ or any of the versions of RAM, both of the entered emission rates are
listed on RAMQ output. One of the emission rates may be left off and will
appear as zeros on the output.
Area sources can vary in size, but certain requirements must be met.
There must be a definable internal unit such that the side length of all
other area sources is an integer multiple of the side length of this internal
unit. For example, if an emission inventory consists of area source squares
having side lengths of 1, 2.5, 5, and 10 km, the internal unit must be
chosen to equal 0.5 km. It is better to conduct emission inventories so that
area source squares have side lengths that are multiples of the side lengths
of the smallest area source squares. Also if a grid is constructed of UNIT
SQUARES, squares having side length of one internal unit, the boundaries of
all area sources must coincide with lines in that grid; there can be no
overlap of one area source over another. Although these statements may seem
restrictive, the area source entries to RAMQ are quite versatile. Concen-
trations from area sources are calculated by performing computations for
each area source encountered in proceeding from a receptor in the upwind
direction until the upwind boundary of the area sources is encountered. If
there are large areas (larger than the UNIT SQUARE) of zero emissions within
the rectangle that includes all area sources (area source region), it is
desirable to define these as area sources with zero emissions in squares as
**The emission rate is a total rate for the entire area. It is later
-1 -2
transformed into g s m . As with point sources, emission rates for other
pollutants may be substituted for sulfur dioxide, and particulates with
appropriate name changes made in data statements to affect titles on printout,
34
-------
large as is possible. This will save considerably in computer running time.
For further clarification on area sources, see p. A-2 in the Appendix.
If the height of emission is the effective height of the area source at
a wind speed of 5m s , and if the physical height of the source is a set
fraction of this value, which is the same for all area sources, it will be
possible to consider the variation of effective height of area sources with
wind speed in RAM. Otherwise the fraction will be 1.0, and it will be
assumed that the input height of emission is the effective height for all
wind speeds.
METEOROLOGICAL DATA AND RAMMET
Meteorological data for all four versions of RAM can be furnished in
either of two ways: 1) punched cards containing the meteorological data
for each simulated hour (one punch card per hour), or 2) magnetic disk or
tape output from the program RAMMET.
Meteorological data output from the program RAMMET may be used as input
to all four RAM versions. RAMMET requires one year of hourly surface obser-
vation data and one year plus two days of daily maximum and minimum mixing
height data. The hourly surface data normally on magnetic tape in card
image format, CARD DECK 144, can be obtained from the National Climatic
Center in Asheville, N.C.
All required surface data for each hour must be included on the tape;
therefore, all data flagged as missing by RAMMET must be determined and
included in the data set before proceeding. The data used from the surface
observation tape for each hour are: Year, Month, Day, Hour, Cloud Ceiling
Code, Wind Direction, Wind Speed, Temperature, and Opaque Cloud Cover.
The mixing height data is expected in card format, one card per day
containing the minimum and maximum mixing height for that day.
A more detailed description of required data follows, but variable
names and card formats are given with the RAMMET example run stream.
35
-------
RAMMET-CARD 1 - 7 variables
a) A 5-digit identification of the meteorological tape to be used. For
tapes generated by the National Climatic Center, this will normally
be the surface station number.
b) A two-digit year of the meteorological data.
c) The latitude (degrees) of the site to be modeled.
d) The longitude (degrees) of the site to be modeled (positive for
west longitude, negative for east longitude).
e) The time zone of the site equivalent to the Greenwich Meridan Time
minus the local standard time.
f) The number of days to be processed (the same as the number of days
in the year).
g) An initial number to be used as the beginning point for the random
number generator. If the same initial number is used, the same set
of random numbers will be generated each time the program is run.
Any odd numbers greater than three digits are suggested as appro-
priate seeds.
RAMMET-CARD 2 - 2 variables
a) Yesterday's minimum mixing height (meters).
b) Yesterday's maximum mixing height (meters).
This card will contain data for December 31 of the previous year.
The mixing heights will normally be determined using the methods
of Holzworth (1972). The maximum mixing height is used for both
urban and rural applications. The minimum mixing height is used
only for urban applications.
RAMMET-CARD 3-4 variables
a) Identification for the radiosonde station used to determine the
mixing height.
b) A 2 digit year for the mixing height data.
c) Minimum mixing height (meters).
d) Maximum mixing height (meters).
This card will contain data for January 1.
36
-------
RAMMET-CARD TYPE 4-2 variables
a) Minimum mixing height (meters).
b) Maximum mixing height (meters).
There will be as many type 4 cards as the number of days in the
year, and data will be for January 2 through January 1 of the next
year.
RAM and RAMR
In addition to requiring emission information from RAMQ and accepting
meteorological information from either punch cards or a file generated by
RAMMET, RAM and RAMR require some punch card input. This is described and
commented upon.
RAM-CARDS 1 - 3. Each card has up to 80 alphanumeric characters.
Information is written on all output and can suit the user. Normal use
has been to identify the user and the run date on card 1, the location
and date of the emissions data on card 2, and the Tocations and dates
of both surface and upper air meteorological data on card 3.
RAM-CARD 4-9 variables, 22 values
a) Values for 13 different options, 1 is used to employ the option, 0
is for nonuse.
Option 1 - used if computations are to be performed for point
sources.
Option 2 - used if computations are to be performed for area
sources.
Option 3 - used to indicate that permanent receptor coordinates
will be entered.
Option 4 - used if receptors downwind of significant point sources
are to be generated.
Option 5 - used if receptors downwind of significant area sources
are to be generated.
Option 6 - used if receptors (referred to as honeycomb receptors)
are to be generated by the program in order to insure good
area coverage. Receptors generated under this option are
placed equidistant in staggered rows over a specified area.
37
-------
Candidate receptor positions are checked against other recep-
tors (either input or generated by other options of the
program) and if the distance between the proposed receptor and
any other receptor is less than one-half the normal distance
between honeycomb receptors, then that candidate receptor is
not added to the list. The boundaries of the area to be
covered by these receptors are specified by the user using
card type 11.
Option 7 - used if hourly output is desired by the user. If not
employed, output will occur for the specified averaging time,
normally from 3 to 24 hours.
Option 8 - used if partial concentrations are to be written on
disk. Partial concentrations are the individual concentration
contributions due to each source at each receptor. This
option should be used with a recognition for the tremendous
quantities of data which will be generated.
Option 9 - used if hourly summaries are to be printed rather than
the entire hourly output. This results in one page of printed
output instead of three or more pages for each hour. Used
only if option 7 is used.
Option 10 - used if cards containing concentrations for each
receptor location for the averaging time selected are to be
punched. These cards can be used externally such as with the
CALCOMP contouring routines.
Option 11 - used if meteorological data is input on cards with one
card for each simulated hour (see card type 12.)
Option 12 - used if the user will specify source numbers (from the
input emission list) that he wants to consider as significant.
This will allow the examination of the individual contributions
to each receptor from each of the specified sources. Both
point and area sources may be specified (see card types 6
and 9.)
Option 13 - used if emissions will be read hourly. This is the
preferred method of operation of RAM since it allows consider-
ation of all known variability in emission rates. The exit
38
-------
velocity is scaled according to the varying emission rates
during program execution in order to allow for an appropriate
plume rise calculation.
b) The number of periods to be run. For example, if the run will use
data for 2 days (48 hours) and the averaging time is to be 3 hours,
the number of periods will be 16.
c) The number of hours in the averaging time (<_ 24). If the averaging
time is 24 hours, the number of periods is equivalent to the number
of days to be run. For averaging times of 1, 2, 3, 4, 6, 8, 12, or
24 hours, any number of periods may be run. For averaging times
other than those specified above, the total number of hours run
cannot exceed 24 hours.
d) Receptor height above ground (meters). All receptors must be at
the same height for a given run.
e) Pollutant half-life (seconds).
f) The number of significant point sources (maximum of 25).
g) The number of significant area sources (maximum of 10).
h) A 2-digit year and a 3-digit starting Julian day for this run.
i) The start hour for this run.
When using meteorological data from RAMMET, there are greater restric-
tions on certain input parameters than there are when meteorological data
from cards are used. Using RAMMET data, one averaging time must be used, and
it must be evenly divisible into 24. The start hour must be 1. Periods must
be sequential in the time series. The starting day may be any in the file.
The file will be positioned to the correct start day based upon the Julian
day entered on card 4.
When using meteorological data from cards, one averaging time must be
used, but it can be any integer value from 1 to 24. The start hour can be
any hour from 1 to 24. Day and hour values must be punched correctly on
input cards and must be in sequence within each period. Data from period to
period need not be in sequence. For example, calculations for two two-hour
periods could be made for first: day 181, hour 24, followed by day 182, hour
1, then secondly: day 23, hour 13, followed by day 23, hour 14.
39
-------
RAM-CARD TYPE 5-4 variables. This card is required only if option 11 is
not used, meaning that meteorological data from RAMMET will be used.
a) Surface meteorological station identification - (normally 5 digits)
b) Year of surface data - 2 digits.
c) Identification of radiosonde station used to determine mixing
height.
d) Year of mixing height data - 2 digits.
RAM-CARD TYPE 6 - 2 variables - from 2 to 26 values
This card is required only if both options 1 and 12 are used.
a) The number of user specified significant point sources (maximum of
25).
b) The point source numbers that the user wants to consider signifi-
cant. There will be as many sources in this list as indicated in
a).
RAM-CARD TYPE 7-4 variables - 4 to 6 values.
This card is required only if option 2 is employed (area sources).
a) Fraction of area source height that is physical height (1.00 or
less).
b) Distance limit for integration of the area source contribution
(user units). The distance should be equal to or exceed the
greatest possible distance from a receptor (including receptors
generated by RAM) to the farthest corner of the area source region
for this run.
c) Number of heights to be used for area sources (from 1 to 3).
d) Height(s) in meters for the area source integrations. There must
be as many heights as specified by the previous variable. Look at
RAMQ output to help decide on these heights.
RAM-CARD TYPE 8 - 1 variable - 1 or 2 values
Heights between the area source height classes. One value is read
if the number of heights on the previous card is one or two. If
only one area source height is to be used (1 entered in c, and one
value in d of card Type 7), the height read here must have a value
higher than any area source height in the data set for this run.
40
-------
Two values are read if the number of heights on the previous card
is three. These heights are to distinguish between area height
classes and can be chosen by examining the output from RAMQ to
determine the quantities of area source emissions that are released
from various height ranges.
RAM-CARD TYPE 9-2 variables - 1 to 11 values
This card is required only if options 2 and 12 are used.
a) The number of user specified significant area sources (maximum of
10).
b) The area source numbers that user wants to consider significant.
There will be as many sources in this list as indicated in a).
RAM-CARD TYPE 10-3 variables
If option 3 is used there will be one card for each receptor that the
user specifies.
a) Receptor identifier in eight alphanumeric characters.
b) East coordinate of receptor in user units.
c) North coordinate of receptor in user units.
Both coordinates of receptors should be positive. Receptors may be
either inside or outside the area source region. A blank card
signals that the receptor list has been completed. Therefore, a
receptor at the origin cannot be used since it would cause the same
program flow as a blank card.
RAM-CARD TYPE 11-5 variables - 5 values
This card is needed only if option 6 is used to generate additional
receptors for area coverage.
a) Distance between honeycomb receptors in user units.
b) Minimum east coordinate of boundary for area to be covered by
receptors in user units.
c) Maximum east coordinate of boundary.
d) Minimum north coordinate of boundary.
e) Maximum north coordinate of boundary.
If b through e are entered as zero, the boundaries considered for
these area sources will be the same as those of the area source
region.
41
-------
RAM-CARD TYPE 12-8 variables - 8 values
Cards of this type are needed only if option 11 is used. There should
be a card for each hour to be modeled.
a) Year of meteorological data (2 digits).
b) Julian day of meteorological data (3 digits).
c) Hour of meteorological data (2 digits).
d) Stability class for this hour.
e) Wind speed (meters/second) for this hour.
f) Ambient air temperature (Kelvin) for this hour.
g) Wind direction (degrees) from which the wind is blowing, for
this hour.
h) Mixing height (meters) for this hour.
To account for variability in emission rates with time in order to
simulate emissions most accurately, it is possible to enter new emission
rates for each of the sources for each simulated hour using option 13. In
order to employ this option, emissions for each source must have been deter-
mined and written on two tape or disk files (one for point sources, one for
area sources) with one record for each hour that is to be simulated. The
emission information from RAMQ is still required and must be a "normal"
emission rate in order that the exit velocity of the source can be scaled up
or down in proportion to the hourly emission rate. Also, all permanent
information about sources such as coordinates, physical stack height, and
diameter are furnished by output from RAMQ.
RAMF and RAMFR
With few exceptions, which are easily noted in the example run streams,
data inputs to RAMF and RAMFR are similar to those required by RAM and RAMR.
In order to keep input lists the same for RAM and RAMF, even the list of
option variables are the same, although values for options not available in
RAMF are not even examined. (Values must be entered just to keep the
variable lists the same length.) Options 4, 5, 6, 7, 8, 9, and 10 are not
used in RAMF and RAMFR.
42
-------
When using meteorological data from RAMMET, the start hour must be 1,
and averaging times must be evenly divisible into 24. The averaging time
must be 24 if the output data are to be used with the program CUMF. Days
must be run in order but can be broken up into separate runs if desired.
When using meteorological data from cards, the start hour can be any
integer from 1 to 24. The averaging time must be evenly divisible into 24 to
run for multiple days. Days and hours on the meteorological data must be
consecutive. The input parameter for the last day of the run must be input
as the Julian day on the last hour's card to be processed. Care must be
exercised in specifying this value when the start hour is not equal to 1.
CUMF
In addition to the disk file of 24-hour concentrations from RAMF or
RAMFR, the data requirements are quite simple for CUMF. Two cards are re-
quired.
CUMF - CARD 1 - Up to 80 alphanumeric characters for the title to be used on
the printed output and the plots.
CUMF - CARD 2-4 variables
a) The number of log cycles on the concentration scale.
b) The horizontal size of the plot (inches).
c) The vertical size of the plot (inches).
d) The minimum value of concentration on the concentration scale
3
(micrograms/meter ).
43
-------
6, ALGORITHM CAPABILITIES
The capabilities of RAM are discussed by considering typical uses of
various RAM programs and discussing an example problem for RAM.
TYPICAL USE OF RAM
The normal operation of RAM is to simulate the dispersion of pollutants
released from point and area sources in an urban area over a period of one or
two days. The meteorological data would be entered on cards, with a card for
each simulated hour, using option 11. General emission information would be
disk files generated by RAMQ, and hourly emission data would be entered from
tape using option 13.
Most applications would consider both point and area sources using
options 1 and 2. The locations of any existing air quality sampling stations
would be used as specified receptor locations using option 3.
The use of options 4 and 5 to locate additional receptors downwind of
significant point and area sources would assist in determining locations of
maximum concentration. Since the resultant wind vector for the averaging
period selected by the user is used to determine the direction of these
receptors from the sources, averaging times as long as 12 to 24 hours that
contain significant wind shifts may result in misleading calculated concen-
trations.
For example, if the wind is generally from 200° for a period of 10 to 12
hours (pollutant flow toward 20°), suddenly shifts to 280° (pollutant flow
toward 100°), and remains in that general direction for the remainder of the
24-hour period, the resultant direction calculated by the model will be
around 60°, although there were almost no hours of wind blowing pollutants in
this direction. Therefore, concentrations will be very low in this direc-
tion from a given source but will be higher in the directions of 20° and 100°
from the source. If attempting to locate high concentrations, it would be
44
-------
better to run RAM for two periods, before and after the wind shift, using
options 4 and 5. The locations of the higher concentrations could then be
entered as receptor locations using option 3 for an additional run covering
the entire period.
The user should note that when using options 4 and 5 to locate receptors
downwind from significant sources, the locations of these receptors will
shift for each averaging period dependent upon the resultant meteorological
conditions for each period. Therefore, receptors with the same numbers will
be at different locations for different averaging times.
If the user wants to obtain sufficient density of concentration esti-
mates for a specific area so that pollution patterns are discerned, option 6
can be used to add additional receptors. The pattern used is such that
adjacent receptors are equidistant. We refer to this as a honeycomb pattern.
The distance between receptors is selected by the user. The boundaries of
the area covered are also selected by the user. If the four boundaries are
entered as zeros, the boundaries will be set to coincide with the boundaries
of the area source map array.
Most modelers are aware, but it should be pointed out, that concentra-
tion gradients may be very steep, especially in the vicinity of plumes from
point sources. Therefore, the addition of more receptors generally will
reveal a more complex concentration pattern and usually some areas of higher
concentrations. Therefore, in searching for maximum concentrations the
individual user must decide on the point of diminishing returns as to recep-
tor spacing commensurate with resources, analysis time, and the purpose of
the project.
For the typical run, hourly output would be desired, so option 7 would
be used. If option 7 is not employed, output is printed only for the speci-
fied averaging time. The use of option 8 to write partial concentrations
onto a disk file will be used only if additional computer analysis is in-
tended using the individual contributions of sources upon particular recep-
tors. Computer programs to do this analysis must be written by the individ-
ual user to suit his purpose.
45
-------
Option 9 is checked only if option 7 is used to get hourly summaries.
The use of option 9 will print only a summary page for each hour. In this
summary, in addition to giving the total concentration for each receptor,
four other concentrations are given for each receptor: 1) the contribution
to the concentration from all point sources, 2) the contribution to the
concentration from all area sources, 3) the contribution to the concentration
from all the significant point sources together, and 4) the contribution to
the concentration from all the significant area sources together. The prin-
cipal information that will be obtained by using option 7 but not option 9 is
the contributions to the concentrations at each of the receptors from each of
the significant sources. The maximum of 10 significant area sources results
in an additional page of output per simulated hour. The maximum of 25 sig-
nificant point sources results in three additional pages of output per sim-
ulated hour (one page for every 10 significant point sources or fraction
thereof). Unless the concentration contributions are specifically needed for
analysis of contributions from particular sources, option 9 should be used to
reduce the quantity of output.
Option 10 would not be employed for this run unless the punched cards
with concentrations for each receptor are desired for further analysis or
they are to be used with graphics software to produce maps with concentration
isopleths. As discussed previously, option 11 for meteorological card input
and option 13 for entry of hourly emission data would be used. If the con-
tributions to the concentrations at receptors from particular sources are of
interest, and if these particular sources are not included high enough in the
significant source lists from RAMQ to be included in the number of signifi-
cant sources used in the run, option 12 may be used, and the sources of
interest specified. In this case in addition to getting concentration
contributions for the averaging time, it is probably desirable to use option
7 to obtain hourly output but not use option 9 so that the full hourly output
is available.
DISCUSSION OF RAM EXAMPLE
An example of a test run for RAM is given in Volume II, Section 5 of this
publication. This example is for a run simulating 2 hours. All options are
46
-------
employed in this run with the exception of option 9. Option 9 would delete
part of the hourly output.
For this run five disk files are assigned for the following five func-
tions: 1) RAM program file, 2) output file from RAMQ containing primarily
emission data, 3) an output file to receive the partial concentrations, 4) an
input file containing the 1-hour emissions for each point source, and 5) an
input file containing the 1-hour emissions for each area source.
The emission inventory for this example consists of 12 point sources and
15 area sources. The smallest rectangle that will include all area sources
defines the area source region. The size of the smallest area source square
is two user units on a side. The user unit is one mile since it is stated on
the first page of the example output (see Volume II, Section 5) that there
are 1.6093440 kilometers per user unit. The area source emission inventory
has one source defined with a zero emission rate. This is area source 6.
Time is saved in executing RAM by specifying areas of zero emission within
the area source region that are larger than the smallest area source squares,
with squares as large as possible. Note that point sources and receptor
locations can be placed anywhere without regard as to whether they are inside
or outside the area source region.
Note that, on the output under general input data, the height of the
receptors above ground level must be the same for all receptors. The assumed
pollutant half-life is used for all hours for the period simulated by the
run.
The example run is executed by reading 14 data cards. All possible data
cards are represented with the exception of card type 5 which is not needed
to identify sources of meteorological data to compare with the disk file
since option 11 is used reading meteorological data from cards. Cards 1, 2,
and 3 are used to provide headings. Card 4 specifying choices for options
and other required information is needed. Note that in this example the
number of significant sources is specified as 5 for point sources and 10 for
area sources.
47
-------
A card of type 6 is required since both options 1 and 12 are used. It
indicates that one point source is user specified and that it is source
number 7.
The information on card type 7 on area sources is specified from a
knowledge of the emission inventory and examination of the table from RAMQ
output that shows the distribution of emissions with area source height.
Three area source heights were selected with the 11-m height chosen to
represent both the 10-m and 12-m sources.
The input information on area heights of emission may be confusing to
the user. Area source heights may be expected to vary with wind speed, but
little information pertaining to this is included in most emission inven-
tories.
If the user wants the area source emission heights to remain constant
throughout his run with no variation with wind speed, the first variable on
card type 7, the fraction of the area source height that is physical height,
should be entered as 1.
If the user wants to vary the area source height with wind speed, the
area source heights input to RAMQ should represent the effective emission
height from each area at a wind speed of 5 m s . The fraction entered as the
first variable on card type 7 should approximate as closely as possible the
average physical height of each area source when the fraction is multiplied
by the input area source height. To most effectively use this feature, the
fact that both physical and effective heights are of interest should be known
when conducting the emission inventory.
In the example the two heights used to determine which area source
integration table to use appear in card type 8. It should be noted by the
user that if only one area source height is used, one height should be read
from card type 8, and its value should be higher than any area source height
on the data used.
A card type 9 is required because both option 2 (area sources) and
option 12 to specify specific numbered sources as significant are used. In
this example option 12 is being used to specify only one point source which
48
-------
was done with card type 6. The card of type 9 indicates that no area sources
are being specified. The card of type 9 is still required.
Three cards of type 10 are used in this example to specify coordinates
of two receptors input by the user. Note that the third card of this type is
blank to signify the end of card type 10 input.
Since option 6 is used to generate additional receptors, one card type
11 is required. The first value on the card is the distance between recep-
tors. The remaining variables are the boundary coordinates. If these are
set to zeros on the input card, the boundaries of the area source region will
be used. In this example an area smaller than the area source region is
used.
The remaining two cards of input to the example are two hours of meteor-
ological data on cards of type 12.
The output from the test run is reasonably easy to understand. The
information under the headings: General Information from RAMQ, General Input
Data, Point Source Information, Area Source Information, Area Source Map
Array (IA), and Area Source Information are primarily from the RAMQ output
transferred by disk.
The receptor locations input by the user are listed next. All location
coordinates on RAM output are in the user's units that have been used for
input.
The meteorological data and resultant conditions for the period are
listed next. These resultant conditions are used in locating receptors
downwind of significant sources.
The receptors located by the algorithm downwind of significant point and
area sources are listed next. Note that RAM generates two receptors for
each significant point source, one at the expected point of maximum for the
resultant meteorological conditions, the other at twice this distance. The
second receptor's placement is to allow for the possible interaction of
pollutant plumes from several sources.
49
-------
For the purpose of insuring good area coverage, the receptors are listed
next under the heading "Generated Honeycomb Receptors." Note that in this
example these cover only a portion of the area source region.
The concentration outputs follow next with three pages for each one-hour
simulation: a page of concentrations from point sources, a page of concen-
trations from area sources, and one page with a summary table of concentra-
tions.
Following the output for each hour are three pages of output for the
averaging time of two hours: one page of concentrations from point sources,
one page of concentrations from area sources, and one page with a summary
table.
In this example the highest concentrations for bcth hours and for the
two-hour period were at a receptor downwind of point source number 5.
TYPICAL RAMR USE
The only difference between RAM and RAMR is in the dispersion parameters
used. RAM uses dispersion parameter values representative for urban areas;
RAMR uses dispersion parameter values representative for open countryside.
The full range of 13 options is available in both programs. A typical run
using RAMR will not contain area sources. The presence of area sources will
often, though not always, signify a sufficiently built-up area to require the
use of urban dispersion parameters. An exception to this could be situations
of pollutant releases from open areas such as particulate matter being raised
by the wind from plowed fields or desert areas or other similar situations
generally referred to as "fugitive" emissions.
TYPICAL RAMF USE
Many of the options available to RAM and RAMR are not available in RAMF
and RAMFR. However, in order to keep input lists the same, dummy entries
even for those options not used are made on input. Options not used are 4,
5, 6, 7, 8, 9, and 10. Options 1 and 2 are still used to indicate use of
point and/or area sources. All receptors must be read as input to RAMF and
RAMFR, so option 3 must be employed. Although it is generally expected that
meteorological input will be from disk files prepared by RAMMET, option 11
50
-------
may be used to enter data by punch card if more representative data are
available. Because of the large number of card entries required for a
lengthy period of record, input simulating the form of the data from RAMMET
or a modification of the program to accept the meteorological data on tape or
disk records may be considered as opposed to input on cards. In this case,
the user must modify the program. Option 12 may be used giving the concen-
tration contributions for the sources specified for the averaging period
selected. Option 13 can also be used to input hourly emissions.
RAMF is primarily used to generate concentrations for an averaging time
greater than an hour (generally 24 hours) for a period of record of one year
so that this data can be input to the program CUMF to produce cumulative
frequency distributions.
The RAM example, typical run streams, variable glossaries, and FORTRAN
source codes are given in Volume II of this document.
51
-------
REFERENCES
Briggs, Gary A., 1969: Plume Rise, USAEC Critical Review Series, TID-25075,
National Technical Information Service, Springfield, Va. 81 pp.
Briggs, Gary A., 1971: Some recent analyses of plume rise observation, in
Proceedings of the Second International Clean Air Congress, edited by
H.M. Englund and W.T. Beery. Academic Press, New York. pp. 1029-1032.
Briggs, Gary A., 1972: Discussion on chimney plumes in neutral and stable
surroundings. Atmos. Environ. 6: 507-510.
Briggs, Gary A. , 1973: Diffusion Estimation for Small Emissions. Atmos.
Turb. and Diff. Lab.Contribution File No. (Draft) 79.Oak Ridge, Tenn.
59 pp.
Briggs, Gary A., 1975: Plume rise predictions, Chapter 3 (pp. 59-111) in
Lectures on Air Pollution and Environmental Impact Analysis. Duane A.
Haugen, editor, Amer. Meteorol. Soc. Boston, Mass. 296 pp.
Gifford, Franklin A., Jr., 1960: Atmospheric dispersion calculations using
the generalized Gaussian plume model, Nucl. Saf. 2 (2): 56-59.
Gifford, Franklin A., and Hanna, Steven R., 1971: Urban air pollution
modeling, in Proceedings of the Second International Clean Air Congress,
edited by H.M. Englund and W.T. Beery. Academic Press, New York.
pp 1146-1151.
Gifford, Franklin A., 1976: Turbulent diffusion-typing schemes: a review,
Nucl. Saf., 17_ (1): 68-86.
Holzworth, George C., 1972: Mixing Heights, Wind Speeds, and Potential for
Urban Air Pollution through the Contiguous United States, Office of Air
Programs Publication No. AP-101. U.S. Environmental Protection Agency,
Raleigh, N.C. 118 pp.
McElroy, James L., and Pooler, Francis, 1968: St. Louis Dispersion Study.
Volume II-Analysis, National Air Pollution Control Administration. Pub-
lication No. AP-53. U.S. Dept. of Health Education and Welfare, Arlington,
Va. 51 pp.
Novak, Joan Hrenko, and Turner, D. Bruce, 1976: An efficient Gaussian-plume
multiple-source air quality algorithm, J. Air Poll. Control Assoc., 2_6_
(6): 570-575.
52
-------
Pasquill, Frank, 1961: The estimation of the dispersion of windborne
material, Meteorol. Mag., 90 (1063): 33-49.
Pasquill, Frank, 1974: Atmospheric Diffusion. 2d ed., Halstead Press, New
York. 429 pp.
Turner, D. Bruce, 1964: A diffusion model for an urban area. J. Appl.
Meteorol., 3., 83-91.
Turner, D. Bruce, 1970: Workbook of Atmospheric Dispersion Estimates,
Office of Air Programs Publication No. AP-26. US Environmental Protection
Agency, Research Triangle Park, NC. 84 pp.
U.S. Environmental Protection Agency, 1974: User's Network for Applied
Modeling of Air Pollution (UNAMAP). (Computer Programs on Tape for Point
Source Models, HIWAY, Climatological Dispersion Model and APRAC-1A),
NTIS PB 229-771, National Technical Information Service, Springfield, Va.
53
-------
APPENDIX
An Efficient Gaussian-Plume Multiple-Source
Air Quality Algorithm
Joan Hrenko Novak and D. Bruce Turner
U S Environmental Protection Agency
The information presented in this paper is directed to air pollution
scientists with an interest in applying air quality simulation models.
RAM is the three letter designation for this efficient Gaussian-plume
multiple-source air quality algorithm. RAM is a method of estimating
short-term dispersion using the Gaussian steady-state model. This
algorithm can be used for estimating air quality concentrations of
relatively stable pollutants for averaging times from an hour to a day
in urban areas from point and area sources. The algorithm is appli-
cable for locations with level or gently rolling terrain where a single
wind vector for each hour is a good approximation to the flow over
the source area considered. Calculations are performed for each hour.
Hourly meteorological data required are wind direction, wind speed,
stability class, and mixing height. Emission information required of
point sources consists of source coordinates, emission rate, physical
height, stack gas volume flow and stack gas temperature. Emission
information required of area sources consists of south-west corner
coordinates, source area, total area emission rate and effective area
source height. Computation time is kept to a minimum by the manner
in which concentrations from area sources are estimated using a
narrow plume hypothesis and using the area source squares as given
rather than breaking down all sources to an area of uniform elements.
Options are available to the user to allow use of three different types
of receptor locations: 1) those whose coordinates are input by the
user, 2) those whose coordinates are determined by the model and
are downwind of significant point and area sources where maxima
are likely to occur, and 3) those whose coordinates are determined
by the model to give good area coverage of a specific portion of the
region. Computation time is also decreased by keeping the number
of receptors to a minimum
Reprinted from APCA JOURNAL, Vol. 26, No 6, June 1976
The purpose of formulating RAM is to provide a readily
available computer program based on the assumptions of
steady-state Gaussian dispersion. RAM can be used for any
short-term (one-hour to one-day) determination of urban air
quality resulting from pollutants released from point and/or
area sources. Urban planners can use RAM to determine the
effects of new source locations and of control strategies upon
short term air quality. If the input meteorological parameter
values can be forecast with sufficient accuracy, control agency
officials can use RAM to predict ambient air quality levels,
primarily over the 24-hour averaging time, to 1) locate mobile
air sampling units, and 2) assist with emission reduction tac-
tics. Especially lor control tactics, diurnal and day-to-day
emission variations must be considered in the source inventory
input to the model. For most of these uses, the optional feature
to assist in locating maximum points should be utilized.
Computations are organized so that execution of the program
is rapid, thus real-time computations are feasible.
Hriggs' plume rise equations are used to estimate effective
height of point sources. Concentrations from the point sources
are determined using distance crosswind and distance upwind
from the receptor Considerable time is saved in calculating
concentrations from area sources by using a narrow plume
Mrs Novak is systems analyst. Model Development and
Assessment Branch, and Mr Turner is Chief, Environmental
Applications Branch, Meteorology and Assessment Division,
I'S Environmental Protection Agencv, Research Triangle
Park. N(" L'7711 Both authors are on assignment from the
National Oceanic and Atmospheric Administration. I' S
Department of Commerce Phis paper was presented as
Paper No 75-04 .'! at the B9th Annual Meeting of APCA at
Boston in .lune I97">
54
-------
simplification which considers sources upwind from a receptor
to be representative of those affecting the receptor. Area
source sizes are used as given in the inventory rather than
creating an internal inventory of uniform elements.
The algorithm is applicable for locations with level or gently
rolling terrain where a single wind vector for each hour is a
reasonable approximation of the flow over the source area
considered. A single mixing height and a single stability class
for each hour are assumed representative of the area. The use
of the RAM is restricted to relatively stable pollutants.
Options are available to allow the use of three different-
types of receptor locations: 1) those whose coordinates are
input by the user, 2) those whose coordinates are determined
by RAM and are downwind of significant point and area
sources where maxima are likely to occur, and 3) those whose
coordinates are determined by RAM to give good area cov-
erage of a specific portion of the region. Options are also
available concerning the detail of output produced.
The Algorithm
Inputs Required
The algorithm always requires emission and meteorological
data, and depending on receptor options used, it may also
require receptor data. Any convenient east-north rectangular
coordinate system may be used since all conversion from user
units to meters is performed internally by use of an input
conversion factor.
A. Point source information consists of the following for each
source:
1. East coordinate of source location, user units
2. North coordinate of source location, user units
,'!. Stack height (above ground), meters
4. Stack inside top diameter, meters
5. Stack gas temperature, °K
6. Stack gas velocity, m sec"1
7. Pollutant emission rate, g sec"1
B. Area source information consists of the following for each
source:
1. East coordinate of the southwest corner of the area
source, user units
2. North coordinate of the southwest corner of the area
source, user units
3. Effective emission height, meters
4. Side length of area source, user units
f>. Total pollutant emission rate for the area, g sec"1
Area sources must be squares. They can be of various
sizes, but their side length must be an integer multiple of a
common side length. The term UNIT SQUARE refers to a
source with this minimum common side length. The effective
emission height of the area sources is assumed to be the ef-
fective height that occurs with a 5 m sec"1 wind. The effective
height of the area sources can be varied with wind speed. Area
emission rates are converted internally to g sec"1 m
-2
(' Meteorological data, representative of the region being
considered, consists oi hourly values ol the following:
1 Wind direction, cleg clockwise from North
2. Wind speed, m sec"1
li Stability class, dimensionless
4. Mixing height, meters
The stability class is that of Pasquill.
IX Receptor information, if required by user specification,
consists ol the following for each receptor:
i Kast coordinate of the receptor location, user units
2. North coordinate of the receptor location, user units
Only one receptor neight, z. above ground is allowed tor a
given execution of the model. This height can be zero or pos-
itive.
Basic Principles
The following assumptions are made: 1) Dispersion from
points and area elements result in Gaussian distributions in
both the horizontal and vertical directions through the dis-
persing plume, and therefore steady-state'Gaussian plume
equations can be used for point sources and the integration
of these equations for area sources. 2) Concentration estimates
may be made for each hourly period using the mean meteo-
rological conditions appropriate for each hour. 3) The total
concentration at a receptor is the sum of the concentrations
estimated from all point and area sources, that is, concen-
trations are additive.
For point sources, the plume rise is calculated from the
stack gas temperature, stack diameter, and stack gas velocity
using the equations of Briggs.1"3 The effective emission height
is the physical stack height plus the plume rise.
In order to calculate contributions from point sources the
upwind distance, x, and the crosswind distance, y, of each
source from each receptor are calculated using Eq. Al and A2
PLAN VIEW OF AREA SOURCES
SOURCE NUMBER IN LOWER LEFT CORNER
EMISSION RATE (G/s.c) IN PARENTHESES
(446)
;
(131 31
1
(1261
8
10.0)
6
(4311
2
1263)
3
1231)
9
i
(32)
5
(389)
4
\
\
\
AREA SOURCE MAP ARRAV
V
7 8
6 i 5
= 1 2 3 4 5
Figure 1. Plan view of area sources and area source map array
55
-------
in the Appendix. The dispersion parameter values, n^ and nz
are determined as a function of the upwind distance, x, and
stability class (See p 374 of Pasquill4). The three equations
used to estimate concentrations under various conditions of
stability and mixing height (Equations A3, A4, and A5) are
discussed in the Appendix. These equations are for a receptor
height, z, above ground and simplify considerably when the
receptor height is assumed to be at ground level, z = 0. (Those
simplifications are incorporated into RAM.)
The total concentration at a receptor arising from the
two-dimensional area-source distribution is calculated using
the narrow plume simplification of Gifford and Hanna/1 This
simplification is assumed because, on an urban scale, the
plume from a point source release is normally quite narrow
in comparison with the characteristic length scale for appre-
ciable changes of the magnitude of the area-source emission
rate itself. Under these circumstances the two-dimensional
integral that expresses the total area-source contribution to
concentration can be replaced approximately by a one-di-
mensional integral that only involves knowledge of the dis-
tribution of the area-source emissions along the line in the
direction of the upwind azimuth from the receptor location,
and the meteorologically-dependent function that specifies
the crosswind-integrated concentration in the Gaussian plume
from a point source. Further evidence for the validity of this
approximation for treating area-source concentrations has
been provided by some numerical tests of Thayer and Koch.K
In the use of this area source technique by Gifford and
Hanna,'' area source emissions were assumed at ground level
allowing integration upwind to be accomplished analytically.
However, in our application of this technique within RAM,
the area sources are considered to have an effective height,
thus requiring the integration to be accomplished numerically.
The equations used to estimate concentrations from area
sources (Eq. A10 through A13) are given in the Appendix. The
total concentration from all area sources is determined by
performing the integration in the upwind direction until the
farthest boundary of the source region is reached.
Concentrations at a receptor for periods longer than 1 hr
are determined by averaging the hourly concentrations over
the period of interest.
How Computations Are Made
Initially, a preprocessor program is used to store the emis-
sion inventory in a convenient form and perform any neces-
sary conversions. A most important function of the prepro-
cessor is to arrange the area sources in such a way as to mini-
mi/e computation time for area source concentrations. Each
area source number (area sources are numbered sequentially
us the sources are input) is stored in a two dimensional array
which essentially forms a map of the relative locations of all
the area sources. Kach element in the array corresponds loan
area the si/.e of a unit square (previously defined). Therefore
a unit source will have its source number stored into one ele-
ment of t he array, whereas an area source that is 4 units by 4
units will have its source number stored into 16 elements of
the array (4 X 4) Obviously area sources must be mutually
exclusive; they must not overlap. Array elements corre-
sponding to areas of the source region not covered by area
sources will have a xcro stored in the array. As will he ex-
plained later, it is to the advantage of the user to define areas
'2 X '2 units or larger with no emissions as specific source areas
will) xcro emissions (source (>, Figure 1). An example of a
simplified source region and the result ing array are shown in
Figure 1.
Concentration estimates are made hour-by-hour for up to
'2\ hr. This algorithm is not designed to determine average
-AREA SOURCE REGION
12 34567
X, UPWIND DISTANCE (km)
Figure 2. Features of area source estimates
concentrations over periods longer than 24 hr. First, concen-
trations resulting from area sources are calculated. In an effort
to reduce the total amount of computer time, tables (arrays)
which contain relative concentrations, V, normalized for
emission rate and wind speed, are calculated only once for each
simulated hour using the appropriate stability and mixing
height, and thereby eliminating all repetitive calculations. The
function V is calculated from:
V(d)
fd
Jo
(1)
and is the non-dimensional concentration resulting from an
area source of given effective height extending upwind from
a receptor to the distance, d. The function /, whose form de-
pends on stability, and mixing height, is defined in the Ap-
pendix (Eq. All, A12, and A13). The stored tables contain
values of this integral obtained by numerical integration for
a number of values of d. Both / and V for 3 area source heights
are shown in Figure 2. Because V(d) changes rapidly for small
values of d, the numerical integration using the trapezoidal
rule is done using varying size intervals, as small as 1 meter
56
-------
for or less than 100 meters, and as large as 1 km for x greater
than 15 km. The values of V are also stored for varying in-
tervals of d (ranging from 10 m to 1 km), so that linear inter-
polation between stored values will result in an accurate es-
timate of V For each effective area source height, up to a
maximum of three, a V table is generated and stored at the
beginning of each simulated hour.
The concentrations from the area sources are computed
receptor by receptor. If the receptor is outside the source re-
gion (the rectangular region containing all the area sources),
it is first determined if the upwind ray (the line pointed in the
wind direction) intersects the source region. If it does not in-
tersect the source region, no contribution from area sources
at this receptor is calculated. If the upwind ray does intersect
the source region, the distance, d\, (See Figure 2) along the
ray to the source region is determined using Eq. A6 and A8 in
the Appendix. The coordinates of this intersection point and
consideration of wind direction provide direct access, through
the area source map array, to the source number of the par-
ticular area source at this intersection point. Since all other
source information is stored in arrays indexed on source
number, the1 area source location (coordinates of SW corner),
size, effective height, and emission rate are readily
available.
Knowledge of the location and size of the area source per-
mits the calculation of the intersection point of the upwind
ray from the receptor with the area source boundary on the
other side of the source (See Equations A6 through A9 in
Appendix) and subsequently the calculation of distance (d->)
from the receptor to this point (Figure 2) These two distances,
d i and d >, are then used to obtain linearly interpolated values
of V from the tables, \'(d>> and V(d\) The concentration
from this source (assume this is source number i ) is then given
by:
X \, = \q\,/u)\V(d,) - V{dt)\ (2)
where x.ii 's the concentration at the receptor from the ith
area source, q \, is the area source emission rate from the ith
area source, and u is the mean wind speed. V(d{) is subtracted
since it represents the area source contribution not present.
II, however, the emission rate is zero or the source number
stored in the area source map array is zero, the source does not
contribute to the concentration, but the intersection with the
boundary and the distance to this intersection is determined
as before.
After estimating the contribution of this area source to the
receptor, the coordinates at the boundary furthest from the
receptor are used to determine the next adjacent source en-
tered by the upwind ray The procedures are then repeated
tor this source and all other sources until the boundary of the
area source region is reached by working upwind along the
upwind ray
In the case where the receptor is initially within the area
^ource region, the coordinates of the receptor are used to de-
termine within which area source the receptor lies. If the
source number is zero, indicating no source area, the inter-
section point ot the upwind ray and the upwind boundary of
a unit square is determined and computation proceeds as
above. If the receptor is within a numbered source area, the
intersection [joint ol the upwind ray and the upwind area
source boundary, see Figure 2, as well as the distance, d (, to
this point are determined Then by interpolation in the V
table corresponding lo the appropriate area source height, the
contribution to the concentration is computed as follows:
The next area source upwind is determined and computations
proceed lor the other upwind sources as above. The advantage
of specifying large areas of no emission, rather than leave them
numbered as zero in the area source map array, is that the
intersection of the upwind ray and the far boundary can be
determined directly rather than stepping across a number ot
unit squares.
After the influence of area sources upon all receptors is
calculated for a simulated hour the contribution from point
sources is determined. Concentrations from point sources are
also calculated receptor by receptor; and tor each receptor,
calculations are made source by source. The upwind distance,
x, of the point source trom the receptor is determined for this
hour from the coordinates of the point source, the coordinates
of the receptor, and the wind direction (See Eq. Al in the
Appendix). If this distance is negative, the source does not
contribute to the receptor and the next source is examined
However, if the upwind distance is positive, the crosswmd
distance, v, and the ratio y/'o\ are determined next. It'y/n\ is
greater than 10, the factory (See Appendix) is always so small
that the contribution from this point source to the receptor
is negligible. But with v/o\ less than 10 an additional test must
be made to see if the concentration is significant. If the factor
#1 multiplied by the point source emission rate is less than
some specified threshold concentration, no further calcula-
tions are made for this source.
In most cases the concentration is above the threshold, and
plume rise must be calculated for the source being considered
provided that it was not calculated previously for estimates
at another receptor for this simulated hour. A table of final
plume heights and distance to the final rise is filled in as plume
rise calculations are required, thus final plume rise is calcu-
lated only once for each source for each hour's simulation. If
the upwind distance of the source from the receptor, x, is less
than the distance to final plume rise, the gradual rise of the
plume from stack top to final rise is considered, and the plume
height at this nearer distance is used for estimates for this
receptor. After the appropriate plume rise is obtained, the
concentration at the receptor from this point source is calcu-
lated using the equation appropriate for stability class and
mixing height as discussed in the Appendix. Concentrations
from other point sources are similarly determined. Similar
procedures are repeated then for each of the other receptors.
The total concentration at a receptor is the sum of the
concentrations from area sources and from point sources. If
any background concentration exists that is caused by sources
outside the source region, it must be added to the concentra-
tion estimates from RAM.
Options
Three options are available regarding use of receptor loca-
tions in RAM. The first option allows coordinates of specific
receptors to he entered as input.
Use of the second receptor option allows the user to specify
how many significant point and how many significant area
sources he wants to consider. The model then calculates the
location of the maximum concentration from each significant
point source using a plume rise calculation, the resultant wind
direction, and the most frequently occurring (modal) stability
class during the period modeled (24 hr or less). (It is not de-
sirable to use this option if there are significant shifts of the
wind during the period modeled, because the resultant di-
rection will not represent the mean transport.) A receptor is
located at the estimated point of maximum from each sig-
nificant source, and another in the same direction nut twice
as lar away. A receptor at this second distance may also have
high concentrations for cases of overlapping plumes from
several sources, using this second receptor option there are
I wo receptors established for eacn significant point source.
57
-------
The second receptor option also determines the location of
a single receptor downwind of each significant area source.
Since the effective height of area sources are generally lower
than point sources, the maximum concentration from the area
source is calculated quite near the boundary of the source.
The location of the maximum concentrations from specific
point and area sources will, of course, not necessarily be a
location where the contribution from all sources will result in
a maximum. Since the location of the maxima are highly de-
pendent upon the dispersion parameter values, o\ and az, any
modification of the algorithm that changes the way in which
these dispersion parameters are calculated will also require
extensive modifications to the subroutines, which determine
the maximum distances from point and area sources, if the
second receptor option is to be used.
The third receptor option allows for good area coverage of
a specified portion of the region. The boundaries of the region
to be covered and the spacing between receptors, w, are
specified by the user. In order to cover the maximum area with
the fewest number of stations, a hexagonal or 'honeycomb'
grid is used. Receptor locations are at equal distances from
nearby receptors so that if lines are drawn to all nearby re-
ceptors, six equilateral triangles will result. Also in order to
keep the total number of receptors to a minimum, any po-
tential receptor locations generated by the third option are
deleted if they are within one - half w of any other existing re-
ceptor.
Several other options available are mainly used to delete
special output when not required. These options are not as
significant as the receptor options and will not be discussed
here.
Summary
RAM is a steadv state Gaussian algorithm applicable to
urban areas for pollutants emitted from point and area
sources Calculations are made for one-hour time periods.
Average concentrations mav be obtained for time periods up
to 21 hr
Kstimation ot concentrations from point sources is
straightforward. Hnggs' plume rise equations are used. Up-
wind and crosswmd distances ol each source from each re-
ceptor are determined and concentration is estimated from
various Gaussian equations.
Innovative techniques are used in keeping the number of
receptors to a minimum and in the treatment of the area
cnussion inventorv F.xcept lor the area source map array used
lor coordinating area source number with location, area source
in I or mat ion is stored and used directlv lor a number of pos-
sible source si/es A narrow plume simplification with con-
sideration of source height of each area is used. The emission
rates ol the area soun es in t he -.ource region along the upwind
a/imulh are considered representative of the area emission
ates affecting the receptor Irotn various distances upwind
1 narrow plume hvpothesisi Determination, at the beginning
M( each simulated hour, ot t he effect ol area sources extending
'o dillerent distances upwind are stored in tabular form with
.1 fill I < rent tahU- lor cac h el I eel ive area source height (up to
', heights allowed) 1.1 near interpolation ol these tabular values
'or >;!( n ^ource, mil re< eplor i>\ receptor, to obtain concen-
' r, 11 ions i rom , i rea sources saves considerable computer tune
! lie various re< eptor opt ions in I he model allow lor versa -
! ilil v in ! he use ol HAM Coordinates corresponding to fixed
'<>< .-dions. sued .is ;ur qualilv sampling locations mav be used
'n .it tern pi ing '< est limit e maximum ( o nee nt rat ions lor par-
iimlar snort term i>eric> = exp |-0.5U - H)-/T,-!] + exp [-0.5(2 + H)-/a^]
= V jexp |-0.5U -H + '2NLV'/n,'\
\ = - .
exp|-0.5(r +
+ IN D- /<,,*}
(This infinite series converges rapidly and evaluation with N
varying from 4 to -M is usually sufficient.)
where
H = effective height of emission, meters
L = mixing height, the top ol the unstable layer, meters
v = crosswmd distance, meters
/ = receptor height above ground, meters
n^ = standard deviation of plume concentration distri-
bution in the horizontal, meters
n, = standard deviation ot plume concentration distri-
bution m the vertical, meters
58
-------
Point Source Computations
The upwind distance, x, and the crosswind distance, y, of
a point source from a receptor are given by:
x = (Sp-Sr) cos 6 +(Rp-Rr) sin0 (Al)
y = (Sp -Sr) sin (9- (Rp - Rr) cos 8 (A2)
where Rp, Sp are the coordinates of the point source; Rr, Sr
are the coordinates of the receptor, and 8 is the wind direction
(the direction from which the wind blows). The units of A: and
y will be the same as those of the coordinate system R, S.
Frequently a conversion is required in order to express x, and
y in meters or kilometers.
The contribution to the concentration, Xp, from a single
point source to a receptor is given by one of the three following
equations where \p is in g m,~:!' Q is point source emission rate
in g sec"1, u is wind speed in m sec"1, and trv ar>d az are eval-
uated for the upwind distance x, and the stability class.
For stable conditions or unlimited mixing:
tersection is:
XP =
(A3)
In unstable or neutral conditions and if o> is greater than
1.6 times the mixing height, L, the distribution below the
mixing height is uniform with height provided that both the
effective height, H, and the receptor height, z, are below the
mixing height:
(If H or 2 is above the mixing height, XP = 0.)
In all other unstable or neutral conditions, that is, if az is
less than 1.6 times the mixing height:
k',/?t/(27T,T,0>U) (A5)
Area Source Computations
Some analytic geometry relationships are used in estimating
concentrations from area sources
The distance, d\, along an upwind ray in the direction 0
from a receptor Rr, Sr to a north-south boundary given by R
d , =
- Rr)/sm(l
(A6)
The east coordinate of the locus of the boundary and the
upwind ray is. of course, Ri, The north coordinate of this in-
SL = Sr + dicos6
(A7)
The distance, d-2, along an upwind ray in the direction 0
from a receptor Rr, Sr to an east-west boundary given by S =
Sh is:
d2= (Sb -Sr)/cos6 (A8)
The north coordinate of the locus of the boundary and the
upwind ray is, Sb. The east coordinate of this intersection
is:
RI. = Rr + d^ sin 6 (A9)
(Depending upon the units of the coordinate system R, S, the
results of these equations may have to be multiplied by a
factor to convert to meters).
The contribution of the concentration, XA, from a uniform
area source directly upwind of a receptor is:
XA =
fx>
\ fdx
*/n
(A10)
where XA is in g m
is area source emission rate in g sec
m L>, u is wind speed in m sec ', x } is the distance in meters
from the receptor to the locus of the upwind ray (extending
from the receptor) and the closest boundary of the area source,
x 2 is the distance in meters from the receptor to the locus of
the upwind ray (extending from the receptor) and the distant
boundary of the area source, and / is given by one of the three
equations below. The integral in the preceeding equation is
evaluated numerically.
For stable conditions or unlimited mixing:
f = g.1/\a!(W-} (All)
In unstable or neutral conditions and if a, is greater than
1.6 times the mixing height, L, the distribution below the
mixing height is uniform with height provided that both the
effective height, H, and the receptor height, z, are below the
mixing height:
/= l/L (A12)
(If H or z is above the mixing height, / = 0.)
In all other unstable or neutral conditions, that is, if n, is
less than 1.6 times the mixing height:
F = ^.,/[
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/8-78-016a
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
USER'S GUIDE FOR RAM
Vol. I. Algorithm Description and Use
5. REPORT DATE
November 1978
6. PERFORMING ORGANIZATION CODE
7 AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
D. Bruce Turner and Joan Hrenko Novak
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Trianglp Park, NC 27711
10. PROGRAM ELEMENT NO.
1AA603 AB-25 (FY-78)
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory - RTF, NC
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Tn-hnn^p
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT The information presented 1n this user1 guide is directed to air pollu-
tion scientists having an interest in applying air quality simulation models. RAM is
the three letter designation for this system of efficient Gaussian-plume multiple-sourc
air quality algorithms and also the primary algorithm for urban areas. These algo-
rithms can be used for estimating air quality concentrations of relatively nonreactive
pollutants for averaging times from an hour to a day from point and area sources. The
algorithms are applicable for locations with level or gently rolling terrain where a
single wind vector for each hour is a good approximation to the flow over the source
area considered. Calculations are performed for each hour. Computation time is kept
to a minimum by the manner in which concentrations from area sources are estimated
using a narrow plume hypothesis and using the area source squares as given rather than
breaking down all sources into an area of uniform elements. Options are available to
the user to allow use of three different types of receptor locations: (1) those whose
coordinates are input by the user, (2) those whose coordinates are determined by the
model and are downwind of significant point and area sources where maxima are likely
to occur, and (3) those whose coordinates are determined by the model to give good area
coverage of a specific portion of the region. Computation time is also decreased by
keeping the number of receptors to a minimum. Volume I considers the use and capa-
bilities of RAM, its basis, the organization of the computer program and data require-
ments.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COS AT I Field/Group
*Air pollution
*Atmospheric models
Algorithms
*Dispersion
13
14
12
13. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
UNCLASSIFIED
21. NO. OF PAGES
70
20 SECURITY CLASS (This page)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
60
-------
Date
Chief, Environmental Applications Branch
Meteorology and Assessment Division (MD-80)
U.S. Environmental Protection Agency
RESRCH TRI PK, NC 27711
I would like to receive future revisions to the
User's Guide For RAM, Vol. I.
Name
Organization
Address
City
State Zip
Phone (Optional) ( )
J. .OVERVILNT POINTING OFFICE: 19 '4 - -74O- 2 -1 /41n L Regain No. 4
------- |