OCR error (C:\Conversion\JobRoot\00000AX2\tiff\2000XF7O.tif): Unspecified error
-------
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 ENVIRONMENTAL MONITORING series
This series describes research conducted to develop new or improved methods
and instrumentation for the identification and quantification of environmental
pollutants at the lowest conceivably significant concentrations It also includes
studies to determine the ambient concentrations of pollutants m the environment
and/or the variance of pollutants as a function of time or meteorological factors.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161
-------
EPA-600/4-80-013C
February 1980
EVALUATION OF THE REAL-TIME
AIR-QUALITY MODEL
USING THE RAPS DATA BASE
Volume 3. Program User's Guide
by
Ronald E. Ruff and Hisao Shigeishi
Atmospheric Science Center
SRI International
Menlo Park, California 94025
and
Rodney H. Allen
Comp-Aid, Inc.
Research Triangle Park, North Carolina 27711
Contract No. 68-02-2770
Project Officer
John S. Irwin
Meteorology and Assessment Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research
Laboratory, U.S. Environmental Protection Agency, and approved for publica-
tion. Approval does not signify that the contents necessarily reflect the
views and policies of the U.S. Environmental Protection Agency, nor does men-
tion of trade names or commercial products constitute endorsement or recom-
mendation for use.
ii
-------
ABSTRACT
The theory and programming of statistical tests for evaluating the
Real-Time Air-Quality Model (RAM) using the Regional Air Pollution Study
(RAPS) data base are fully documented in four volumes. Moreover, the
tests are generally applicable to other model evaluation problems. Vol-
ume 3 presents the software used in the statistical tests for evaluating
the RAM. Six statistical tests are described, with attention to the pro-
gramming philosophy behind them. Also presented is a review of the
auxiliary software that sort, retrieve, format, and display the data.
This report was submitted in fulfillment of Contract No. 68-02-2770 by
SRI International under the sponsorship of the U.S. Environmental Pro-
tection Agency. This report covers a period from 1 October 1977 to 1
April 1979, and work was completed as of 1 April 1979.
111
-------
CONTENTS
Abstract iii
Figures vi
Tables viii
1. Overview 1
Background 1
Discussion of statistical programs 2
Programming philosophy 3
2. Data-Base Considerations 4
Content 4
Structure and formats 4
Display and retrieval software 10
Auxiliary software 13
3. Statistical Package 16
Introduction 16
Accuracy score 16
Residual time series 30
Chi-square goodness-of-fit 34
Bivariate regression and correlation 38
Interstation error correlation 43
Multiple regression of error residuals 45
References 50
Appendices
A. Listing of test data base 51
B. Modified EPA frequency-distribution software 54
C. Accuracy score listings 66
D. Residual time series listings 93
E. Chi-square goodness-of-fit listings 105
F. Bivariate regression and correlation listing 114
G. Interstation error correlation listings 125
-------
FIGURES
Number Page
1 SPSSDAT program 8
2 SPSSSORT program 11
3 Examples of SPSS histograms 12
4 Frequency distribution of observed and predicted
concentrations on log-probability axes 14
5 Frequency distributions of residuals (percent)
on log-probability axes 15
6 Accuracy score equations 17
7 Flow diagram for the accuracy-score program 19
8 Results of Test 1 displayed on a map of St. Louis 21
9 First part of the accuracy score session 22
10 Accuracy score session (MODE CODE =1) 23
11 Accuracy score session (MODE CODE = 2) 24
12 Software structure for the accuracy score program 25
13 Display of the autocorrelation function
for Station 1 of the test data base 31
14 Display of the normalized cumulation periodogram
for Station 1 of the test data base 32
15 Program structure for the residual time series ....".. 33
16 Histogram of concentrations for the test data base .... 35
17 Histogram of concentration residuals
for the test data base 36
18 Structure of the chi-square goodness-of-fit program .... 37
vi
-------
Number Page
19 Scattergram with least-squares, confidence,
and probability lines 39
20 Structure of the regression and correlation program .... 40
21 COREL2 output for test data base 43
22 Structure of the interstation error
correlation program 44
23 Example of a regression procedure 46
24 Summary output for regression of RESID
' on ASC, WDC, and WSC 47
25 Summary output for regression of RESID
on TIME and SITE 47
26 Example of SPSS SCATTERGRAM output 49
vii
-------
TABLES
Number Page
1 Definition of User-Entered Values for the
Accuracy Score Test . 23
2 Description and Format of Test Data Base 27
viii
-------
SECTION 1
OVERVIEW
BACKGROUND
The programs described in this volume are meant to be general-purpose
tools useful to anyone with a computer supporting the FORTRAN IV program-
ming language and a graphics display. However, from a practical stand-
point, we had to tune our programming efforts to EPA's National Computer
Center (NCC) in Research Triangle Park, North Carolina. The mainframe is
a Univac 1110 computer. (The executive system, Exec 8, 'is described in
i
Volume VII of the NCC User Reference Manual. ) Standard peripherals were
used whenever possible. For graphics, the Tektronix 4014 terminal is
the only device on which the plotting programs can operate. (Graphic
output in this text are from the 4014 using a Tektronix 4631 hardcopy
unit.)
Most of the programs were written in Univac"s FORTRAN V language.
(See Volume III of the NCC User Reference Manual.) Certainly the pro-
grams as they now exist would require only minimal effort for conversion
to FORTRAN IV. For graphics, FORTRAN-callable Tektronix subroutines in
the Advanced Graphing II and the Terminal Control System3 manuals were
used. A few programs were written in the special Statistical Packages
4
for the Social Sciences (SPSS) language. Many major computer facilities
have this software package or one that is quite similar.
From the above description, it should be clear that the software
presented will operate only on the NCC system. However, with minimal
modification it should be readily adaptable to most other major computer
systems. For the remainder of this text, we assume that the reader is
familiar with the hardware and basic software used. We do attempt to
point out code that is usually specific to the NCC computer system.
-------
DISCUSSION OF STATISTICAL PROGRAMS
The six statistical tests recommended in Volume 2 were programmed,
debugged, and tested. The tests and program names are given below:
*
Test Name Program Name
Accuracy score M21ADO*STAT01.ACSCOR
Residual time series M21ADO*STAT02.TIMSER
Chi-square goodness-of-fit M21ADO*STAT03.CHIFIT
Bivariate regression
and correlation'1' M21ADO*STAT04.REGANA
Interstation error
Correlation M21ADO*STAT05.COREL2
Multiple regression
of error residuals M21ADO*STAT06.SPSSRUN
The Pearson's correlation coefficient and linear regression tests were
combined because (1) it was computationally efficient to do so and (2)
results from both tests have historically been presented together when
evaluating air-quality models. The six programs are fully described in
Section 3 of this volume. (Note that the O's are zeros in M21ADO and
STAT01 through STAT06 and are the letter 0 elsewhere.)
*
The program names are given in typical Exec 8 file designation format--
FILENAME.ELEMENT. On the NCC computer, one must often qualify the file
with an account code--ACCOUNT*FILENAME.ELEMENT. The account qualifier
for these statistical programs is M21ADO. All the source, relocatable,
and absolute elements found in files M21ADO*STATO1 through M21ADO*STAT06
are backed up in a single file labeled SRI*SRI. In this file an absolute
such as M21ADO*STAT01.ACSCOR is also known as SRI*SRI.ACSCOR. The test
data base in data element M21ADO*STAT01.COQS2N is also known as
SR1*SRI.COQS2N. All other source, data, and relocatable elements follow
the example for M21ADO*STAT02.CASE2 which is also contained in
SRI*SRI.CASE2/STAT02.
Combines the features of two tests: (1) linear regression with confi-
dence and prediction bands and (2) Pearson's correlation coefficient.
See Volume 2 for a complete description.
-------
*
Each of the program files (STAT01 through STAT06) contains all the
subroutines and run streams (control language statements) required to run
the statistical test in question. In addition, a test data base is con-
tained in M21ADO*STAT01.COQS2N. All programs except those contained in
STAT06 are written in FORTRAN V. STAT06 programs are written in the SPSS
language. In addition to the multiple regression program, SPSS is also
used to plot frequency distributions of any variable, to plot scattergrams,
and to retrieve subsets of the data base. These auxiliary data-retrieval
and display programs are discussed in Section 2.
PROGRAMMING PHILOSOPHY
Before discussing the details of the software, we wish to impart
some of the philosophy used during development. Our approach is a compro-
mise between practical considerations in computing efficiency and the
ideal software from the users' perspective.
The ideal programs would never need to be modified regardless of the
data base used and the application in question. Furthermore, they would
work on a large number of computers, given that these computers were
equipped with certain software.
Having defined our concept of programs ideal to a user community, we
now alert the reader that our programs do not satisfy that definition.
However, we have standardized where practical, and through the documenta-
tion provided in this volume, the user can readily customize the programs
to his application and system, assuming that the system supports FORTRAN
IV, Tektronix graphics, and SPSS. (As stated earlier, we attempt to
point out sections of the program that are specific to the NCC Univac
system.)
*
For brevity, the account qualifier--M21ADO--is often omitted.
-------
SECTION 2
DATA-BASE CONSIDERATIONS
CONTENT
In their simplest form, the data for most statistical tests consist
of n pairs of observed and predicted concentrations. These data are
denoted (OC, PC) ,(OC2,PC2),...,(OCn,PCn). (This data set may include all
the data gathered or only a selected subset thereof.) The concentrations
might refer to hourly averages, three-hour averages, daily averages, etc.,
at a particular monitoring station or from any number of stations. Except
for the residual time series test, the data pairs need not be sequential
nor time ordered in any particular fashion.
Since it is desirable to assess model performance under a variety of
meteorological and emission conditions, the data base should be arranged
so that it can be subdivided into many nonexclusive subsets, according
to the following parameters:
• Date, D
• Time period, T
• Location, L
• Pasquill-Gifford stability class, G-)
• Wind-speed class, 62
• Wind-direction class, 63
• Mixing-height class, G^
• Emission class, Q
• Observed concentration, OC
• Predicted concentration, PC.
STRUCTURE AND FORMA1S
General Structure
The exact structure (or hierarchy) of the data depends on the fea-
tures of the data base to be tested. Consequently, the format chosen for
-------
our statistical programs was determined by features of our test data base.
The data file (in card image form) is ordered as follows:
Card/Line Number Data
1 (D,T,L,GltG2,G3,G4,Q,OC,PC)
2 (D,T,L,G1,G2,G3,G4,Q>OC>PC)2
n (D,T,L,G1,G2,G3,G4)Q,OC,PC)n
(We will refer to each record by line number rather than card number,
since normally the records will be stored on disk.)
The only requirement for this structure is that all the observed and
predicted concentrations--(OC,PC) through (OCn,PCn)--be distinct. Now,
we must recognize that the data file can include concentrations for a
number of monitoring locations, denoted by L. Also, for each time period,
T, we may choose to concern ourselves with only one set of meteorological
and emission parameters. (This would typically be the case when evalu-
ating the RAM.) Therefore, if we expand our subscript notation somewhat,
our data file would typically be ordered as follows:
Card/Line Number Data
1 D1,T1,1,(G1,G2,G3,G4,Q)1,OC1,PC1
2 D1,T1,2,(G1,G2,G3,G4,Q)1,OC2,PC2
ns D1,T1,nSj(G1,G2,G3,G4,Q)1,OCns,PCns
ns + 1 D2,T2,l,(G1,G2,G3,G4,Q)2,OCns+1,PCns+1
2*ns D2 ,T2 ,ns ? (G1 ,G2 ,G3 ,G4 ,Q)2 ,
,Tp ,ns, (GL ,G2 ,63 ,G4 ,Q)p ,OCp,Vng ,PCp*ns
-------
where ns = number of monitoring stations (locations), p = number of time
periods, and ns*p = n = number of observed and predicted concentration
pairs. Generally speaking, our test data base is structured as above.
While the structure described above is functional, it is probably
not always computationally efficient. This is particularly true if we
must apply the programs to a large data base (such as that for the RAPS).
The inefficiency stems from having to store the time, date, meteorolog-
ical, and emissions data once for each monitoring site. Thus repetition
consumes a moderate amount of storage space. (In a one-year period, we
are storing over 50,000 words of redundant data.) Consequently, for
larger data bases, we would recommend storing the data in a more effi-
cient manner.
For the RAPS data base, we note that meteorological data are col-
lected at each of the monitoring stations. As discussed in Volume 4,
it may be desirable to store these station-specific meteorological data
for use in our analysis even though these data are not used directly by
the model. Therefore, depending on the application, we may wish to retain
additional data that will require format changes. In Volume 4, we dis-
cuss specific recommendations for evaluating the RAM using the RAPS data
base.
As suggested in the above discussion, the exact structure (and even
content) of the data base may vary according to application, user conve-
nience, and computational efficiency. It generally requires modification
of only one or two lines in the program code to accommodate a slightly
different data structure. When we discuss the individual programs (in
Section 3), we point out the sections of code that control the data-
base interface and the exact data requirements. We stress that the user
should have a flexible attitude concerning the exact format, allowing for
changes that befit the application at hand.
6
-------
Test Data Base
Standard (FORTRAN) Format--
The data base is stored in free-form format in STAT01.COQS2N
(file.element). Appendix A contains a listing of the data. In all, there
are 101 line (or card) images. Each line image contains the following
information, in the order presented:
• Date — six digits
• Time — four digits
• Station—one digit whose value is 1, 2, or 3
• Atmosphere stability—one digit whose value varies from 1 to 6
• Wind-speed class—one digit whose value varies from 1 to 6
• Wind-direction class — two digits whose value varies from 1 to 16
• Observed concentration—a number between 0.0 and 1000.0 (one
significant figure to the right of the decimal point)
• Predicted concentration--same format as above.
The mixing-height class and emissions class were not required when the
test data base was acquired. (Refer to Volume 1 for a description of
the field study during which the test data base was acquired.)
SPSS Format—
The SPSS system of computer programs is used to perform the multiple
linear regression test, which relates meteorological parameters to
observed and predicted concentration residuals. The SPSS is also useful
in retrieving and displaying subsets of the data base. (This is described
on pages 10 and 11.) Prior to using any of the SPSS procedures, the input
data must be converted to a special SPSS format. This can be done within
each program or, alternatively, done once and then stored in the special
SPSS format. This latter approach is more efficient when an assortment
or SPSS programs is continually applied. In the next paragraph, we dis-
cuss the program that converts a standard FORTRAN file into an SPSS file.
-------
The program to reformat the data is stored in STAT06.SPSSDAT. (In
discussing the SPSS procedures and codes, the reader should refer to the
4.
SPSS manual for a complete description. Our descriptions will be brief
and will not ~ver all the features of the statements involved.) The pro-
gram stored in SPSSDAT appears in Figure 1 below.
^DELETE HSPRNT.
$ASG,PU HSPRNT.
3ASG,A SPSSINPUT.
3USE 8.,SPSSINPUT.
$ASG,PU SPSSFILE.
3USE 4.,SPSSFILE.
06RKPT PRINIS/HSPRNI
espss
RUN NAME TEST MODEL
FILE NAME SPSSDATA
VARIABLE LIST DATE , T IMF., SITE , ASC , IvSC , WOC , HC , PC
INPUT MEOIIJVI TAPF
N OF CASES UNK'JOht'i
INPUT FORMAT FIXED (12X,2F6.0 , M .0,2F3 . 0 , F4 . 0 , 7X , 2F8 .1 )
MISSING VALUES ALL (BLANKS)
COMPUTE RES ID = O'C-PC
VAP LABF.'LS PLS1D , RES1DOAI
ASC, SFABILIIY CLASS
WSC, 'AlKD SPEED CLASS
WOC, ft:It,D DIFFCTIU^ CLASS
OC, OBSERVED C0r> CK'-T^ A'l IGM
PC, PREDICTED CUNCKNTPAT10N
LIST CASES CASES = JOOO/
VARIABLES = AF.L
READ INPUT DATA
SAVE ULE
FINISH
P3RKPI PRIMTS
?FRfcE HSPRNT.
T., ,tD04PH
Figure 1. SPSSDAT program.
Note that two files were assigned and related to peripheral devices (4
and 8). The first file SPSSINPUT is simply a copy of the test data base
contained in STAT01.COQS2N. (SPSS cannot work directly with data con-
tained in elements. Therefore, a system utility was used to copy data
-------
from STAT01.COQS2N to SPSSINPUT.) The second file, SPSSFILE, is the output
file, which will become the specially formatted SPSS file used in subse-
quent SPSS programs. The @SPSS control card is the manner in which the
NCC system calls the complete SPSS package. Subsequent cards will then
be. recognized as SPSS procedure cards. The remainder of the Exec 8 state-
ment* control the routing of the print file, HSPRNT, to the desired termi-
nal (FD04PR in this case).
In Figure 1, the statements following @SPSS are the SPSS control
statements. (These begin in column 1. The specification of the desired
parameters begins in column 16.) In our example, the RUN NAME is arbi-
trarily titled TEST MODEL, which thus becomes an identifier/header on the
output listing. The FILE NAME has been called SPSSDATA, which becomes
the output SPSS file referred to by future SPSS programs. Through the
VARIABLE LIST statement we choose the desired parameter names to corres-
pond with the data in our input file. The INPUT MEDIUM, for Exec 8,
recognizes only a CARD or TAPE specification. TAPE simply identifies
that the input will be read from a mass storage file, always named "8".
(Thus, through the @USE statement, SPSSINPUT is identified to SPSS as the
input file.) The N OF CASE statement that follows can be set to UNKNOWN
if cards are not used for the input. By specifying FIXED on the INPUT
FORMAT statement, we are able to use a standard FORTRAN format for the
parameters in the VARIABLE LIST statement.
The preceding paragraph fully specifies the input file. Next, we
describe how SPSS is used to manipulate the data and store them on the
output file. First, by the MISSING VALUES statement, we could assign
values to any parameters when they are missing in the input data. In our
case, this is not a consideration, so we leave such data blank. The
COMPUTE statement enables us to compute any desired parameters. Since we
use concentration residuals in another SPSS program, we compute the RESID
variable as shown. •
*
Alternatively, for the NCC only, routing of the output file can also be
controlled by using the @AB*US.SUSPEND and @AB*US•RESUME statements as
shown in Figure 2, page 11.
-------
The VAR LABELS statement is used to assign meaningful names to the
parameters and subsequently cause these parameters to be better documented
in the ouput of the program. The LIST CASES enables us to produce an
output listing of a user-specified number of cases (e.g., 1000) and the
desired variables (e.g., ALL).
The READ INPUT DATA statement instructs the program to begin reading
the data. The SAVE FILE instructs the system to save the processed file.
Using Exec 8, SPSS expects that the file available for writing will be
named "4". (Thus, the @USE statement causes the file to be called
SPSSFILE.)
The above SPSS program is by no means the only way to generate an
SPSS file. Some of the statements are specified in a somewhat arbitrary
fashion. We leave it to the reader to further familiarize himself with
the SPSS statements involved.
DISPLAY AND RETRIEVAL SOFTWARE
Prior to selecting, applying, or interpreting a particular test, we
may wish to examine the data base in some detail. For instance, we may
wish to stratify the data for a given meteorological class and then run
the test on that subset; or we may wish to closely examine the meteoro-
logical conditions associated with concentration comparisons that fall
outside the 90 percent probability bounds in a regression analysis.
A number of computer languages could be used to satisfy the above
types of requirements. Standard FORTRAN programs could be written or a
5
data-base management language (e.g., System 2000 ) could be applied.
For our immediate needs, the most attractive choices of procedures are
those available within SPSS. Two general SPSS data-retrieval and display
programs are discussed in the following paragraphs.
A program stored in STAT06.SPSSSORT is an example of how SPSS can
be used to selectively sort, retrieve, and display data in an SPSS file.
The program is given in Figure 2 below.
,10
-------
PASG,A HGSSOPT.
@bSt 15.,HGSSriHT.
«AB*US.SUSPFMD
PASG,A SPSSFILK.
3. ,SPSSF K.L.
P U Fi H A vi t S U K 1 p i L fr".
GU' FILE SPSSDAfA
PA* OUTPUT LIMIT Ib
tSfc.Lt.Cr IF (ASC F.I.: 4)
•vPlTL CASFS (2F6.0,M.f>,2M.O,M.O,7X,2F8.] ) RATE , II vt , S 1TF. , ASC
*SC,wDC,OC,PC
*ShLECT IF (ASC EC 4 A'JN SJIt t^ 1)
FRtQUE'MCIKS ItMTtGfc-P = /DC ( 1 , 1 fc ) «SC(l,b)
OfliONS 3,8
FINISH
P FD04PH
Figure 2. SPSSSORT program.
The input file, SPSSFILE, is named "3". (Note the @USE control state-
ment.) Within the SPSS program, the GET FILE statement automatically
looks to Unit 3 for the input. SPSSDATA is the file name previously
assigned by the SPSSDAT program (see Figure 1). The RAW OUTPUT UNIT 15
statement, coupled with the @ASG and @USE Exec 8 statements, identifies
HGSSORT as the output file. (Note that the SPSS output will omit decimal
points. Hence, the output as written must be divided by 10 in the
FORTRAN program. Alternatively, we could drop the least significant
figure and use integers.) The first "SELECT IF statement, followed by
WRITE CASES, causes an output file to be written on HGSSORT in the format
and order specified in the WRITE CASES field. The resulting file will
contain those cases where the atmospheric stability classes (ASC) were
class 4. This file can then be read by a FORTRAN program.
The second -SELECT IF statement, followed by the FREQUENCIES state-
ment, causes the SPSS to generate a histogram on the line printer for the
identified variables—wind direction and speed — for station 1 and an
atmospheric stability class of 4. The resulting output is shown in
Figure 3. OPTION 3 causes the output to be left-justified on an 8-1/2
by 11 format; OPTION 8 causes a histogram to be printed for each variable
specified on the FREQUENCIES statement.
11
-------
Oc CCT 73
.-' ILL - SPS5DM* - CREATED 0<» OC T 78
P*GE 2
WDC
CC3E
1 **•**»*»»*»«* ( 5)
I
I
I
^
3
]
i
****» ( 2 )
r
[...... ... I.. ...... .1.
1
-------
There are a number of variations that can be used with the above
program. Some of these are considered in Volume IV, in which we describe
techniques for applying the statistical tests.
AUXILIARY SOFTWARE
As will be discussed thoroughly in Volume 4, simple graphic compari~
sons can provide very useful information during the model-evaluation pro-
cess. Examples, plotted on a Tektronix 4014 terminal, are shown in Fig-
ures 4 and 5. The first graph (Figure 4) displays on logarithm-probability
axes the frequency distribution of both observed and predicted concentra-
tions; the second graph (Figure 5) displays the frequency distribution
of the percentage difference (residual) in observed and predicted
concentrations.
The programs to draw the axes and plot the data were developed by
EPA. SRI's subcontractor, Comp-Aid Incorporated, adapted the program to
our test data base and added some annotations. Options for three differ-
ent graphs can be exercised. (An example of the third type of graph,
frequency distribution of the absolute value of the concentration residual,
is discussed in Volume 4.) The listings of the programs are given in
Appendix B. The mapped (or linked) program is stored in SRI*SRI.FREQ
(account*file.element). The run stream, stored in SRI*SRI.DATAFREQ,
which shows its execution with our test data base (contained in
M21ADO--STAT01.COQS2N) is given in Section 1 of Appendix B.
The figure for SRI*SRI.MAPFREQ, Section 2 of Appendix B, shows how
the program is "mapped" and serves to identify the subroutines. The main
program, SRI*SRI.FREQ, is given the name of the executable. Its listing
is given in Section 3 of Appendix B. The subroutine listings are given in
Section 4 of Appendix B. The principal subroutines are contained in
SRI*SRI. ORDER, SRI* SRI. GRAPH, and SRI-'-SRI.GRALIN . The other subroutines,
SRI*SRI.BOX, SRI*SRI.HEADER, and RAPS*UTILITY.COMPOZ, are used to draw the
frame around the plot and annotate the data executed (upper-right-hand
corner in Figures 4 and 5). These last three subroutines are discussed
further in Section 3.
13
-------
Figure 4. Frequency distribution of observed and predicted concentrations
on log-probability axes.
14
-------
or-in
1*5
z
a
UJU
o z
cc o
•- u
z
uj a
CJ UJ
a: >
UJ K
a. lu
ir o
o a
u.
o
ZK
O
00
a;
W
O
u
z
LLI
3
a
ui
a;
a
a
•^
in
UJ
u.
o
99'L
00'
WO
l
m
i
i
i
•?
s
S
•
> •
•
c
a
UJ
et
u.
OI'O- If 1-
00'C-
ft] 1«nOlS38
Figure 5. Frequency distributions of residuals (percent) on log-probability axes.
15
-------
SECTION 3
STATISTICAL PACKAGE
INTRODUCTION
In this section we discuss the structure and logic of the programs
and their data requirements. Each program has been tested using our test
data base stored in M21ADO*STAT01.COQS2N. The programs can operate on
any other data base in the same way, provided that the data base is for-
matted in the same way.
We assume chat the data base has been arranged to include the de-
sired conditions. As discussed in Section 2, the data-sorting procedures
within SPSS can be used to stratify the data according to any parameter
(e.g., meteorological class, emissic •<- ciass). Because it is often
desirable to conduct the statistical test for each station, we have
included provisions within some of the programs to sort the data by sta-
tion. Such provisions are discussed for the specific test in question.
Also, it is important to realize that in some cases the software may
be somewhat specific to the test data base and the features of the NCC
system. We feel that these cases are minimal; however, we do point out
these areas in our discussion or by comments in the code itself.
ACCURACY SCORE (STATOl.ACSCOR)
The accuracy score is the recommended "final evaluation statistic."
It consists of eight separate statistical tests, which are computationally
quite simple. The tests are fully described in Volume 2. Their equa-
tions are briefly summarized in Figure 6, which is a reproduction of a
portion of a terminal session.
16
-------
ft SELECT
1.
2.
3.
5.
b.
7.
LIME TEST JIEAM SOU APE EHROP ...
E s (1 ,0/'l)*(Slif/'( (DC( ! )/PC( 1 ))**2) ) ivHhPL 1=1
ABSOLUTE EHHUP THPESH>Ll< ...
E = ( ) ,0/N)*f SOM(C( I) ) ) wHEkE 1=1
/.HKPfc C([J = f, IF AbS(OC(J)-PC( I ) Kl'E.L
xWPt. C(l) = 1 IF
PERCENTAGE F.RHT^ IHHrShCHO
•-MERE C(I) =
.vHtPt C(J) =
11- ABS( (OC(I )-PC( I ))/OCCn
IF APS((OC( J )-PC(i)
E = (1 .0/fO*(6'J^(C(l ) ) )
f. C(I) = 0
AHFRE C(i) = 1
ASYM^EIF 1C HIG'-i-rcu LOSS
E = (1 .0/M)*( Sl".'(C( 1 ) ) )
E C(I3 = LI
CU) = I?
IF PC (I ) .GT.
Af«b flC( I ) .LT
IF l-C(I ) .LT.NURM
A^o ucu ) .Gr
USEP SDPt-LIF? IDSS "ATPI> ...
E = (1 ,0/\i) *(J?UM(C( 1 J ) )
C( I ) = LU,K )
E PC(1) = J Af D f'C(U = K
CONC*- *" rpAl lOu Li'CATiLr1 ...
E = (l.lV'O*(Sjr-'('Hl ) ) )
/HI: Ft 1 =
1J- PC(J) AND bC ( I ) . Lt . u
IF PC(1) AM? HC( U .GT.i-.
IF OTHhhwlSL.
ivhtHfc 1 = 1
E4
v-i'EHF 1 = 1, N
wriFPE 1 = 1
i-.HEPF D(I
.E.2
E3
, N
)=DISTANCE
Figure 6. Accuracy score equations.
17
-------
The program is sufficiently simple that it lends itself to interac-
tive operation, permitting the user to exercise several options, as
follows:
• Modify constants within six of the eight tests. (The other two
tests involve no such constants.)
• Perform the test for the combined data from all monitoring sta-
tions or from each of the stations, or from any selected station.
• Analyze any file in the correct format and order.
• Choose tutorial or brief-prompting interactive modes. (Figure 6
is an example of the summary printed in the tutorial mode.)
• Select any or all the statistical tests.
• Display the accuracy score on a rudimentary map. (For illustra-
tive purposes and subsequent application with the RAPS data base,
a St. Louis map was selected.)
Program Logic
The logic of the program structure is illustrated in the flow chart
of Figure 7. The FORTRAN V code is given in Section 1 of Appendix C. To
simplify relating the flow chart to the source listing, we give the line
of the code next to the logic blocks in Figure 4.
Summarizing the logic of Figure 7, we note that the first operator
entry is the number of data points and monitoring stations. Knowledge of
the number of data points enables us to eliminate computer-specific end-
of-file checks or the need for inserting special identifiers indicating
the end of the data. The number of stations is obviously well known.
The number of data points will be known after prescreening the data.
(The SPSS data-display software will list the size of the data base.)
The next interaction requires the user to select whether the tests will
be conducted on the entire data base, or for all or any one of the moni-
toring stations. If tests are to be done for each station, the user has
the option of displaying the results on a geographical map, assuming a
Tektronix 4014 terminal is available.
18
-------
(000163)
ENTER NUMBER OF DATA
LINES AND STATIONS
SELECT
INTERACTIVE
MODE
SELECT DESIRED TEST(S)
TEST NUMBER 1 2
i !
ENTER TEST-SPECIFIC
THRESHOLDS AND CONSTANTS
.....
I—J—i L__ 4—i i—
(000746)
YES
NO
(000752)
NEW
MODE
7
YES
NO
(000756)
YES
NO
END
Figure 7. Flow diagram for the accuracy-score program.
19
-------
Prior to opting for a graphical display, the user must enter the
data base either directly or by attaching a file. Using FORTRAN V and
Exec 8, the user can attach the test data base during the interactive
session by simply entering
@ADD M21ADO*STAT01.COQS2N
Once the data base has been entered, the user must then choose between
the tutorial (detailed) or abbreviated modes. This is followed by a
choice of the desired test(s). (As discussed later, there are options
for changing constants within most tests.)
The program now proceeds to perform the simple calculations specified
by the user. After the results have been printed, the program checks to
see if the map option was selected and computations were made. If these
conditions are met, accuracy scores are then plotted on a geographical
background map of St. Louis, on a Textronix 4014 terminal (see Figure 8).
After all of the output is presented, the user is given the option
of continuing the session by selecting another test, mode, or data sort.
The user may choose to conduct the same test with different constants.
In this case he would respond "yes" when asked "Would you like to try
another mode?" He would then opt for the same test and have the chance
to choose different constants. Definitions of constants for each test,
along with typical values for the test data base, are given in Table 1.
Sample Session
The first part of the terminal session is illustrated in Figure 9.
As shown, the user must answer questions on the number of data points and
stations, the sort code, and the mode code. Prior to entering the mode
code, the user must enter the data base.
The mode code determines the degree of detail in future interaction.
If mode-code 1 is entered, the most detailed messages are printed. The
next message, after mode selection, will be a listing of the tests and
20
-------
Figure 8. Results of Test 1 displayed on a map of St. Louis.
21
-------
***** STAIJS1ICAL 'IFChMyibLS FOR EVALUATING MODELS *****
« ENTER THE N(j. OF DATA (ND) AMJ THE NO. UF "MjNITOPlMG STATIUM5 (MS)
>101 3
it SELECT A DATA SORTING CODE |VO. F PON1 1HE FOLLOWING ...
-1, EXECUTE TESKS) ON DAIA FOH LACH STATION.
0, EXECUTE TESI(S) ON DATA FOP ALL STATIONS.
N, EXECD1E TESI(S) ON DATA FOF THF NIH STATION (WHERE f. = l,fv&)
i sot0
# F.NTEP ND LINES OF THE FOLLG,MfjU DATA ...
1. DATE (Yf-MQDA)
2. Tli'E (HP«*N)
3. MONITOP1NG SITE NO.
4. OPSFRVED VALUE
S. PREDICTED VALUE
>£ADD SrAT01.COOS?fJ
It SELECT ONE ^ODt. NO. FROM THF. FOLLOWING ...
1. DETAILED-PUN (FOP UN-INITIATED MSFPS)
2. yUlCF-RUN (FOR IMTIATF.D USERS)
» MODE CODF?
1
* SELECT ONE TEST NO. FROM THE FOLLOWING ...
Figure 9. First part of the accuracy score session.
equations (See Figure 6). In Figure 10, a sample of the remaining part
of the session is given. (Test 7 was arbitrarily selected for illustra-
tive purposes.
A sample session for mode-code 2 is given in Figure 11. Note that
we have the ability to specify any group of tests (by a series of 1's and
O's). The program then proceeds to conduct the tests, interacting and
outputting results as shown.
22
-------
TABLE 1. DEFINITION OF USER-ENTERED VALUES
FOR THE ACCURACY SCORE TEST
Test
3
4
5
6
7
8
Constant
El
E2
E3
(NORM)
E4
(NORM)
LI
L2
CMAX
Description
Error threshold
Percentage error threshold
Concentration cutoff
Concentration cutoff
Overprediction loss
Underprediction loss
Loss matrix
Anticipated maximum concentration
Distance matrix
*
Test Value
50
0.2
400
400
1
2
t
STAT01.ERRLOS
1000
STAT01.DISTAN
^Suggested test values for the test data base, STAT01.COQS2N.
Test elements in file STAT01.
NOTE
. TESTS IS VALID ONLY FOR SORT CODE -1.
# TEST CODE?
SITE ALL, TEST1: MEAN ABSOLUTE ERROR IS .458+02
95% CONFIDENCE INTERVAL: .384+02 TO .532+02
/' WOULD YOU LIKE TO TRY ANOTHER TEST ... TYPE YES OR NO.
>NO
WOULD YOU LIKE TO TRY ANOTHER MODE ... TYPE YES OR NO.
>YES
f MODE CODE?
Figure 10. Accuracy score session (MODE CODE = 1).
See Figure 9 and then Figure 6 for the first part of the session.
23
-------
/MODE CODE?
>2
^SELECT T¥.S~ NOS. WITH A SERIES OF 1 (YES) AND 0(NO).
1. MEAN ABSOLUTE ERROR.
2. MEAN SQUARE ERROR.
3. ABSOLUTE ERROR THRESHOLD.
4. PERCENTAGE ERROR THRESHOLD.
5. SYMMETRIC HIGH-LOW LOSS FUNCTION.
6. ASYMMETRIC HIGH-LOW LOSS FUNCTION.
7. USER SUPPLIED LOSS MATRIX.
8. MAXIMUM CONCENTRATION LOCATION.
>1 0101010
SITE 01, TEST1: MEAN ABSOLUTE ERROR IS .858+02
85% CONFIDENCE INTERVAL: .633+02 TO .109+03
SITE 02, TEST1: MEAN ABSOLUTE ERROR IS .280+02
95% CONFIDENCE INTERVAL: .215+02 TO .345+02
SITE 03, TEST1: MEAN ABSOLUTE ERROR IS .232+02
95% CONFIDENCE INTERVAL: .191+02 TO .272+02
#TEST3, ENTER El:
>25
SITE 01, TEST3: \BSOLUTE ERROR THRESHOLD IS .765+00
95% CONFIDENCE INTERVAL: .695+00 TO .835+00
SITE 02, TEST3: ABSOLUTE ERROR THRESHOLD IS .364+00
95% CONFIDENCE INTERVAL: .230+00 TO .497+00
SITE 03, TEST3: ABSOLUTE ERROR THRESHOLD IS .294+00
95% CONFIDENCE INTERVAL: .204+00 TO .385+00
/'TEST5, ENTER E3:
Figure 11. Accuracy score session (MODE CODE = 2).
24
-------
Program Structure
The program (software) structure is given in Figure 12. (The Exec 8
control statements that "map" or link the subroutines are contained in
file STAT01.MAPIT. A listing is given in Section 2 of Appendix C.) As
shown in Figure 12, the main program, STAT01.ACSCOR, calls for statistical
subroutines, STATOl.TESTS and STAT01.CONLIM, and, as a user option,
plotting-routine SRI*SRI.RAMMAP. This plotting routine calls two FORTRAN
subroutines, SRI*SRI.BOX and SRI*SRI.HEADER, a special RAPS utility,
RAPS*UTILITY.COMPOZ, and the standard Tektronix software, GRAPH*TEKTRONIX.
MAIN PROGRAM
M21 ADO*STAT01 .ACSCOR
1
DATA ELEMENTS
M21 ADO*STAT01 .COQS2N
M21ADO*STAT01.ERRLOS
M21ADO*STAT01.DISTAN
MAir
MAIN PLOTTING ROUTINE
SRI'SRI.RAMMAP
I
TINE
STATISTICAL SUBROUTINES
M21ADO*STAT01. TESTS
M21ADO*STAT01.CONLIM
SRI*SRI.HEADER I-*
SRfSRI.BOX
RAPS'UTILITY.COMPOZ
Figure 12. Software structure for the accuracy score program.
The data elements in STATOl are available as a user option for
testing the software. The first, COQS2N, is the test data base previ-
ously described. The second element, ERRLOS, can be used in Test 7 as
the user-supplied loss matrix, while the third, DISTAN, can be used in
Test 8 as the distance-loss matrix. These data elements are further
discussed in the next subsection, Data Entry.
25
-------
The programs are written in the FORTRAN V language. The main pro-
gram and statistical subroutines are written for general application on
any computer. The plotting programs, however, do use some specialized
software (e.g., RAPS*UTILITY.COMPOZ). Also, the map is specific to the
St. Louis area and the RAPS monitoring locations. Therefore, when apply-
ing the accuracy score programs to other geographical areas, one must, of
necessity, write new plotting software for the specific area in question.
The primary function of the main program, STAT01.ACSCOR, is to con-
trol the interaction between the user and the computer. This interaction
includes the data entry, which in turn controls which of the eight tests
will be conducted, in addition to all the other user-specified options.
The statistical calculations are performed within the subroutine contained
in STAT01.TESTS. (The subroutines are listed in Section 3 of Appendix C.)
This subroutine has a separate entry point for each of the eight tests.
Shortly after calling the TESTS routine, the main program calls
STAT01.CONLIM, which performs the key confidence-interval calculations.
The mapping subroutine, RAMMAP, is called at the user's option.
(Refer back to Figure 8 for an example of the output display.) Most of
the plotting is controlled in the RAMMAP routine, listed in Section 3 of
Appendix C. The program contained in SRI*SRI.HEADER annotates the plot
by producing the small box and test shown in the upper^right-hand corner
of the plot. The RAPS*UTILITY.COMPOZ routine retrieves the date and time
information displayed by the HEADER program. The SRI*SRI.BOX routine,
called by both RAMMAP and HEADER, is used to draw frames around the map
and header. All the above routines are listed in Section 3 of Appendix C.
Data Entry
Some of the information on data entry has just been described.
As stated, the user, upon interrogation, must enter the number of data
points and monitoring stations. The other interrogations also involve
entry of one or two variables, as described in Table 1.
,26
-------
The data base to be tested is requested during the session by the
p romp t:
# ENTER ND LINES OF THE FOLLOWING DATA...
1. DATE (YRMODA)
2. TIME (HRMN)
3. MONITORING SITE NO.
4. OBSERVED VALUE
5. PREDICTED VALUE
The contents and format of the test data base is given in Table 2. (See
Appendix A for the listing of the test data base.) While the ISTC, IWSC,
and IWDC parameters are not used within the program, we must allow for
them when entering our data. Using the first line of data in Appendix A
as an example, we could format the first line of our data entry at the
terminal as follows (where the three consecutive 2's are arbitrary but
necessary for formatting purposes):
741012,0900,1,2,2,2,309.7,322.8
TABLE 2. DESCRIPTION AND FORMAT OF TEST DATA BASE
Item
1
2
3
4
5
6
7
8
Name
IDATE
ITIME
ISITE
ISTC
IWSC
IWDC
OCS
PCS
Format
16
14
11
11
11
12
F5.1
F5.1
Description
Calendar date--YRMODA
Time, 0000-2300
Monitoring station number, 1-3
Stability class, 1-6
Wind speed class, 1-6
Wind direction class, 1-16
Observed concentration
Predicted concentration
27
-------
The above format would be repeated for each set of comparisons ("ND" lines
total). For data entry, the user has the option of attaching a data base
in an existing file or entering the data base during the session. In the
former case, as previously stated, a data base can be attached by entering
a statement such as:
@ADD M21ADO*STAT01.COQS2N ,
which will enter our test data base. Alternatively, we can enter data
as specified by the "READ" statement (see lines 000185, 000215, and so
forth, of the ACSCOR code):
READ(INP,3)IDATE(I),ITIME(I),ISITE(I),ISTC,IWSC,IWD,OCS ,PCS !
where the first three parameters (IDATE,ITIME,ISITE) and last two para-
meters (OCS,PCS) are specified, and dummy variables are entered for
ISTC, IWSC, and IWDC.
After data are entered, each record is examined to ensure that the
observed and predicted concentrations are valid (or missing). Invalid
data are designated as such by negative numbers. The ACSCOR program
checks to see if either of the concentrations is negative (e.g., see
line 000190). It negative, the data are not used in the calculation.
The remaining input data are entered as shown in the sample inter-
active sessions (see Figure 9). Usually, each query involves the input
of one or two numbers, but there are two exceptions, Test 7 and Test 8.
When Test 7 is specified, the program will request a user-supplied
loss matrix. As the program now stands, this is a 5 by 5 loss matrix
that can be entered directly during the session or by attaching a file.
For instance, our test file (see the listing in Section 4 of Appendix C)
can be entered by:
@ADD M21ADO*STAT01.ERRLOS
28
-------
Alternatively, we could directly enter five data lines, or rows
(J = 1 to 5), each with five loss indexes (K = 1 to 5). Each index is
a loss associated with predicting concentrations in class J while observ-
ing them in class K. (The loss is usually set to zero when J = K).
As shown in line 000627 of the ACSCOR program, the loss matrix is
read into the AS array. This corresponds to five evenly divided cate-
gories of predicted and observed concentrations. The categories J and K
are calculated in lines 000059 and 000061 of the TESTS program. Note
that 5 and 0.2 (1/5) determine how many categories are entered. CMAX is
the user-entered expected maximum concentration (see Table 1). The num-
ber of categories can be respecified by changing lines 000059 through
000062 and the size of the AS array in the ACSCOR and TESTS programs.
For Test 8, the distance-loss matrix is currently specified as a
3 by 3 array because there are three stations. When asked for the
distance-loss matrix, the user can attach the existing file (see Section
4 of Appendix C) by entering:
@ADD M21ADO*STAT01.DISTAN
Alternatively, the user can enter three rows (J = 1 to 3), each containing
three loss indexes (K = 1 to 3). The resulting matrix should be symmetric.
Each index in the matrix is a loss associated with predicting the maximum
concentration at one location, J, while observing it at another, K.
(The loss is usually set to zero when J = K.) This information is read
into the DD array (see line 000703 of the ACSCOR program). The number
of monitoring stations previously entered determines the required size
of the entered matrix. The DD array needs to be dimensioned so that the
number of rows and columns equals the maximum number of monitoring sta-
tions. (Even though a 3 by 3 matrix is sufficient, the DD array is cur-
rently dimensioned by a 5 by 5 matrix in the COMMON statements labeled
DAT.) The above changes are all that are required to adapt the program
for inclusion of additional stations.
29
-------
RESIDUAL TIME SERIES (STAT02.TIMSER)
The residual time series program is an "intermediate evaluation
statistic" used to detect cyclical components in the data. An unbroken
sequence of observed-predicted concentration pairs is required. Missing
concentration values are accepted by the program if flagged in sequence
with a negative number. The output is entirely graphic, so it is assumed
that a Tektronix 4014 terminal will be used. There are two graphs dis-
played for each station. The first, shown in Figure 13, is a plot of the
autocorrelation function against time lag (in user units which equal the
concentration averaging time). The second, shown in Figure 14, is a plot
of the cumulative periodogram. These plots are generated for each moni-
toring station of interest.
From a programming standpoint, the logic is quite simple because it
contains few options. The program structure for the Residual Time Series
Program is shown in Figure 15. Execution of the program and data entry
are controlled by the Control Element, M21ADO*STAT02.CASEl. (Parameters
within the Control Element may be varied to specify different options or
data. Hence, we refer to CASEn in Figure 7 because there are many varia-
tions. However, the structure remains the same.) As shown in the listing
for CASE1, Section 1 of Appendix D, the first line specifies the execution
of the program contained in M21ADO*STAT02.TIMSER. Execution for CASE1
is begun by entering on the Tektronix terminal:
@ADD,L M21ADO*STAT02.CASE1
Execution of the TIMSER program will then begin, with the data beginning
in line 000002 being read in under program control.
The mapping of the subroutines required by the TIMSER program is
shown in Section 2 of Appendix D. The source listing for the main pro-
gram, TIMSER, is given in Section 3 of Appendix D. The main program per-
forms all the calculations, most of the plotting, and the data entry.
30
-------
o
Ul
o
LD
LT)
o
I
Figure 13. Display of the autocorrelation function for Station 1 of the test data base.
31
-------
e»F-<
ov-
-------
CONTROL ELEMENT
M21ADO*STAT02.CASE
MAIN PROGRAM
M21ADO*STAT02.TIMSER
DATA ELEMENT
M21 ADO'STATCH .COO.S2N
PLOTTING SUBROUTINES
M21 ADO*STAT02.SETUP
AUXILIARY PROGRAMS
M21ADO*STAT02.USESET
SRI'SRI.BOX
SRI*SRI.HEADER
RAPS*UTILITY.COMPOZ
Figure 15. Program structure for the residual time series.
Referring to the TIMSER listing, we see that comment statements
(lines 000041 through 000081) are used to define the input data require-
ments and program options. The corresponding READ statements are imme-
diately before and after the comment statements. The input data will be
read line by line, starting with line 000002 of the Control Element,
Section 1 of Appendix D. The data base for observed and predicted con-
centration is specified in line 000011, our COQS2N test data base. Alter-
natively, other elements could be specified or, as discussed earlier on
page 27, the data could be specified directly, starting in line 000011
and continuing for ND lines.
The format requirements of the TIMSER program for data input are
identical to those of the other programs. However, the time sequence of
the data is important. In general, the data should be read in equal,
33
-------
ascending time intervals. When a data base contains more than one site,
then for each time period one data record per site must be presented.
After the record for the last site, the records for the next time period
should follow, starting with the first site.
Referring to the TIMBER listing, we see that the data are read on
line 000092. The next two lines of code check to see if the data fall
within the specified timeframe (see line 000082). After more time and
data checks, the program computes the parameters of the residual time
series, starting in line 000119. The calculations for the autocorrela-
tion function begin at line 000121; its values, ACF(K) are computed on
line 000159. The calculations for the normalized cumulative periodogram
begin on line 000209; its values, CNP(J), are computed on line 000231.
The principal subroutine, SETUP, controls plotting of the axis and
some of the annotation. In turn, it calls the HEADER, BOX, and COMPOZ
subroutines (described earlier). The listing for SETUP is given in Sec-
tion 4 of Appendix D.
CHI-SQUARE GOODNESS-OF-FIT (STAT03.CHIFIT)
The chi-square goodness-of-fit program can be useful as either an
intermediate or a final evaluation statistic. As was the case for the
residual time series program, the output is completely graphic. So a
Tektronix 4014 terminal must be used. The output consists of two dis-
plays, each containing two plots. The first output display, Figure 16,
consists of histograms for both the observed and predicted concentration
data sets; the second output display, Figure 17, consists of histograms
of the difference between observed and predicted concentrations, absolute
and percentage.
2
The chi-square statistic, X , and the number of samples, N, are
printed in the lower-right-hand corner of the second plot. (The statis-
2
tic, X , is simply the summation of the square of the observed and pre-
dicted concentration differences divided by the observed concentration.)
34
-------
r^i»*i
UIN
ox ~
LU
I
U.
O
I
V)
UJ
z:
o
o
o
UJ
a:
ot
I-
U")
I-H
o
a:
ui
CD
ca
o
ru
z
o
UJ
o
z
o
o
- o
Z)
CO
V)
o
UJ
I-
o
t-t
o
UJ
QC
Q-
D
-------
OJ
UJ
«
cu
Ul
o
o
o
«
cu
ru
i
Figure 17. Histogram of concentration residuals for the test data base.
36
-------
The program structure for the chi-square program is shown in
Figure 18. The Control Element is used to execute the program and supply
the requisite input data. To initiate the program from the Tektronix
terminal, the user gives the Exec 8 command (see listings for the Control
Element in Section 1 of Appendix E):
@ADD,L M21ADO*STAT03.CASE1
Note that the first line specifies execution of the program
M21ADO*STAT03.CHIFIT. Its elements are shown in the mapping routine,
listed in Section 2 of Appendix E,
CONTROL ELEMENT
M21 ADO*STAT03.CASE
MAIN PROGRAM
M21ADO*STAT03.CHIFIT
DATA ELEMENT
M21 ADO'STATOI .COQS2N
PLOTTING SUBROUTINE
M21ADO*STAT03.HISTO
AUXILIARY PROGRAMS
M21 ADO*STAT03.SETUP
SRI*SRI*BOX
Figure 18. Structure of the chi-square goodness-of-fit program.
37
-------
2
The main program, CHIFIT, controls the data flow, computes X ,
divides the data into categories, and computes the frequency of occurrence
in each category. The listing for CHIFIT is given in Section 3 of Appen-
dix E. The input data for CASE1 starts in line 2 (Section 1, Appendix E).
Each line corresponds to a READ statement, the first of which appears as
line 000043 of CHIFIT. The last READ statement (line 000073) specifies
the entry of the concentration data base. (For CASE1, our test data base
STAT01.COQS2N is specified.) Each read-in parameter is defined in the
listing for CHIFIT (lines 000023 through 000042 and lines 000055 through
000066).
The four frequency distributions are calculated within CHIFIT,
beginning in lines 000087, 000113, 000132, and 000142. The X2 sta-
tistic is calculated along with the last distribution (CHI2 in line
000148). After the second and fourth distributions are calculated, the
dual histograms are plotted by calling subroutine HISTO, whose listing
appears in Section 4 of Appendix E.
Within the HISTO subroutine, subroutines SETUP and BOX are called
(see Section 4 of Appendix E). Subroutine SETUP performs some of the
standard functions used in all histograms, such as framing the plots and
partially annotating the graphs.
BIVARIATE REGRESSION AND CORRELATION (STAT04.REGANA)
The bivariate (linear) regression and correlation program is most
often used as an intermediate evaluation statistic. The graphic output
consists of a scatterplot with options to display the least-squares fit,
probability and confidence limits, and "sensitivity" bounds. (The util-
ity of the options is described in Volume 4 of this report.) As for the
two previous programs, this program must be executed on a Tektronix 4014
terminal. An example of the program output is given in Figure 19. Note
that the correlation coefficient, regression coefficients, and number of
samples are displayed in the lower-left-hand corner of the plot.
38
-------
OJ
I
uJ o
o ~
* «-H
V) ^
^_l
il
i i
ii:
ri
OJ
U3
O
U)
n
OJ
•or-w*
Figure 19. Scattergram with least-squares, confidence, and probability lines.
39
-------
The program structure for the bivariate regression and correlation
program (REGANA) is shown in Figure 20. Execution of the program and
data entry are controlled by the Control Element M21ADO*STAT04.CASE6.
(This particular CASE results in the plotting of confidence and proba-
bility limits about the regression line, as shown in Figure 19.) The
listing of the CASE6 Control Element is given in Section 1 of Appendix F.
CONTROL ELEMENT
M21 ADO*STAT04.CASEn
MAIN PROGRAM
M21ADO*STAT04.REGANA
DATA ELEMENT
M21 ADO*STAT01 .COQS2N
STATISTICAL SUBROUTINES
M21 ADO*STAT04.LINLSQ
M21ADO*STAT04.STAB
AUXILIARY
PLOTTING SUBROUTINES
SRI'SRI. HEADER
SRI*SRI.BOX
Figure 20. Structure of the regression and correlation program.
Execution is begun by the command:
@ADD,L M21ADO*STAT04.CASE6
This will cause the REGANA program to be executed, with the data begin-
ning in line 000002 being read in under program control.
40
-------
The mapping of the subroutines required by the REGANA is shown in
Section 2 of Appendix F. The source listing for the main program is
given in Section 3 of Appendix F. The main program controls the entering
of data, calling of statistical subroutines, and plotting of scattergrams,
regression lines, and confidence limits.
Referring to the REGANA listing, we see that comment statements
(lines 000023 through 000045 and lines 000060 and 000071) are used to
define the input data requirements and program options. The corresponding
READ statements start in line 000046 and culminate in line 000076. (These
two lines correspond to the data given in lines 000002 and 000017 in the
CASE6 element.) In line 000103, REGANA calls the LINLSQ subroutine,
which computes all the statistical parameters of interest. The program
then proceeds to plot the scattergram (line 000138) and limit lines,
depending on the options (the ICODE parameters).
In plotting the scattergram, the program distinguishes among the
site codes (ISITE) for each of the three stations in the test data base.
They are stored in three separate arrays--(Xl, Yl), (X2, Y2), and
(X3, Y3)—and plotted with different symbols. The logic starting in line
000081 and ending in line 000096 determines that the array suffix cor-
responds to the monitoring station number. The actual plotting of the
arrays is programmed in lines 000216, 000224, and 000228.
The parameter ICODE determines which, if any, of the regression or
limit lines will be plotted on the scattergram. The READ statement for
ICODE is in line 000048. (For our example, the data are contained in
line 000003 of the CASE6 element. The first parameter, "6", is ICODE
for this case. The second parameter specifies that the X and Y axes will
be linear.) The logic starting in line 000232 shows how the ICODE value
controls the six different options available to the user.
The ICODE options are defined in the comment statements (lines
000027 through 000033). Specification of ICODEs of 3 through 6 causes
statistical subroutines to be called. Subroutine LINLSQ, which is called
for all ICODEs, is used somewhat differently when an ICODE of 3 is
41
-------
specified. Note that after sensitivity bounds are calculated (lines
000116 to 000124), the LINLSQ program is called (line 000134) for only
those points falling outside the sensitivity bounds. (Note the sensi-
tivity bound check in line 000129.) The significance of this calculation
is discussed in Volume 4.
For ICODEs of 4 through 6, the FORTRAN function routine, STAB, is
called in line 000256. (See Section 4 of Appendix F for a listing of
STAB.) STAB assists in computing the 90 percent confidence intervals
using Student's t-distribution read into the TAB array.
The subroutine LINLSQ, called in line 000103, computes most of the
statistical parameters of interest. The listing for LINLSQ is given in
Section 4 of Appendix F. The label COMMON LSQ contains most of the
parameters of interest, as follows:
RECNOP - reciprocal of the number of samples
A - constant in the regression-line equation
B - X coefficient in the regression-line equation
XSD - standard deviation of observed concentrations
YSD - standard deviation of predicted concentrations
CIL - lower 95 percent confidence limit for the correlation
coefficient
CIU - upper 95 percent confidence limit for the correlation
coefficient
SXMXB - square of the accumulated differences between the actual
and mean observed concentrations
RC - estimate of Pearson's Correlation Coefficient
RMSE - root-mean-square error
BSD - estimated standard deviation of predicted concentrations
about the regression limit
XBAR - average observed concentration.
Not all the above parameters are used in subsequent calculations by the
main program REGANA. However, they are useful statistical parameters
that can be displayed as an option.
-------
INTERSTATION ERROR CORRELATION (STAT05.COREL2)
The interstation error correlation (COREL2) test computes the corre-
lation coefficient between the residual concentration (observed minus
predicted concentrations) at different monitoring stations. The output
consists of two matricies. The sample output for our test data base,
COQS2N, is shown in Figure 21. The first matrix is the correlation coef-
ficient between each of the true stations in our data base. The second
matrix shows the upper and lower 95 percent confidence limits for the
correlation coefficients.
CORRELATION MATRIX ...
Silt 1 2 3
1 1.00 .78 .75
2 .78 1.00 .74
3 .75 .74 1.00
r.MCK Ll'-UTS W
SITE 1 2 3
1
2
3
L
U
L
U
L
U
1 .00
1.00
.bO
.89
.54
.87
.60
.89
J .00
1.00
.5;
. ^ "'
.•34
.47
.5;
.*7
1 .;)">
1 . 0('
Figure 21. COREL2 output for test data base.
Figure 22 displays the structure of the COREL2 program, whose sub-
routines are shown mapped in the listing contained in Section 1 of Appen-
dix G. The main COREL2 program (listed in Section 2 of Appendix G) con-
trols data input and output and calls the statistical subroutine XYCORR,
which controls the calculation of the correlation coefficient.
43
-------
MAIN PROGRAM
M21ADO*STAT05.COREL2
SUBROUTINE
M21ADO*STAT05.XYCORR
DATA ELEMENT
M21 ADO* ST AT01 .COQS2N
SUBROUTINE
M21ADO*STAT05.LINLSQ
Figure 22. Structure of the interstation error correlation program.
The input parameters (parameters are defined in lines 000016 through
000030) are entered as specified by the READ statements in lines 000033
and 000046 of the main program. The first input requires the number of
data points, the number of stations, the option code, and the first and
second monitoring station numbers. For our test data base, if we want
the comparisons for all three stations, the following input would be
entered:
101,3,2,1,2 .
The next read statement requires 101 lines of input data in the pre-
scribed order. Using our test data base, we would enter:
M21ADO*STAT01.COQS2N
44
-------
Entering our input data in this order will produce the output shown in
Figure 21.
The subroutine stored in M21ADO*STAT05.XYCORR computes the residual
concentrations for each monitoring station. As shown in the listing
(Section 3 of Appendix G), XYCORR computes the DC1 and DC2 arrays (lines
000013 and 000014) and then calls the LINLSQ subroutine (line 000019).
The LINLSQ subroutine, listed in Section 4 of Appendix F, is the same
one used in the STAT04.REGANA program.
MULTIPLE REGRESSION OF ERROR RESIDUALS (STAT06.SPSSRUN)
The multiple regression of error residuals, using the meteorological
and emission classes as the independent variables, is an intermediate
evaluation statistic. Its purpose is to detect the model input parameter
that contributes most to the difference between observed and predicted
concentrations. These residuals were computed in the STAT06.SPSSDAT pro-
gram described in Section 2.
The procedure for conducting a regression analysis on the residuals
in SPSSFILE is given in Figure 23.
The general format of an SPSS program was described earlier. The
program simply finds the correct file, SPSSDATA, which is contained in
SPSSFILE. (Using Exec 8, the program automatically looks to Unit 3 for
the desired file.) Three specifications are found in the field of the
SPSS REGRESSION statement. The first, VARIABLES, identifies the parame-
ters required. If any are missing, a diagnostic will be given. The
second and third specifications are REGRESSIONS. Each of the REGRESSION
statements identifies the type of regression to be performed. The first,
RESID with ASC, WSC, WDC (1), specifies the regression of the concentra-
tion residuals on the three meteorological parameters; the second
45
-------
•aDELETE PPINT2.
OASG,PU PKINT2.
«ASG,A SPSSFJLF.
(aUSE J.,SPSSFTLF.
068KPT P*1UT$/PH I'M F2
BSPSS
KUiM NA^E TEST »
-------
UJ O O
• • •
I
u? o
U? M
1^ O
• •
CI3
o
CO
f*. a *
(N ^
o » ^
» n ^ a
5 ^gg
CJ
CO
c
o
CO
UJ
cc
Z -> »
•* -i a
5 S§
H
c
o
Q
CO
c
o
c
o
en
0)
01
0)
O
>
CD
E
E
CO
O
E
E
CO
CN
aj
D
O)
2
LJ
a
X
UJ
CL
LU
o
UJ UJ Z
r >- o
•-« i-l LI
*~ IA —
47
-------
The output for the SCATTERGRAM procedure is given in Figure 26. Note,
as specified in the SCATTERGRAM field, that the RESID scale adjusts to
the values between the lowest and highest in the file while the ASC scale
is from 3 to 6. Again, refer to the SPSS manual for a complete descrip-
tion of the SCATTERGRAM procedure.
48
-------
a
•
•
a
0
• Q
" f
if) I
I
;
:> T
i
i
; T
- I
r
I o
I*
I e S
3
a
o
oc
a
oc
in
i-
<
CJ
Q-
CO
H—
O
_5
a
E
TO
X
UJ
CC
CM
! 3
49
-------
REFERENCES
1. "NCC User Reference Manual," EPA Contract No. 68-2378 (9 Volumes)
Systems Research and Development Corporation, Research Triangle
Park, North Carolina (1977).
2. "Plot 10 - Advanced Graphing II, User's Manual," Tektronix, Inc.,
P.O. Box 500, Beaverton, Oregon (1975).
3. "Plot 10 - Terminal Control System, User Manual," Tektronix, Inc.,
P.O. Box 500, Beaverton, Oregon (1978).
4. N. H. Nie et al., Statistical Packages for the Social Sciences,
(2nd Ed.) McGraw-Hill Book Company, New York, New York, 1975).
5. "System 2000 Reference Manual," MRI Systems Corporation, Austin,
Texas .(1973).
50
-------
APPENDIX A
LISTING OF TEST DATA BASE M21ADO*STATOl.COQS2N
51
-------
iELT.L S1
.COQS2N
EL12G7 SL73R1 11/01/78 09
OOC001
COCC02
COCC03
CO COO*
COLOC5
COOC06
CQGC07
COGCG8
COCCG9
COGC10
C00011
COG012
COC013
IOC014
COC015.
COCC16
COCD17
COCC18
COG019
COC020
C0u021
COC022
COC023
COGC24
COCC25
C00026
COCC27
COCC28
COCC29
LQCC30
COCJC31
COQG32
COGC33
COC.034
COGC35
COC036
CQOC37
C00038
COOC39
COOOAO
COCC41
COCQ42
COC043
COC044
COCC45
COC046
COG047
COCOAS
COGC49
COGC50
COOC51
COC052
COCC53
COC054
COL055
COO
000
DOU
000
000
000
000
000
000
000
000
ooc
JOO
ooc
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
ooc
000
000
000
000
OOG
000
000
000
000
000
000
000
000
OOCJ
000
000
000
UOG
000
741012
741012
741012
741012
741012
741012
741012
741012
741012
741012
741012
7*1012
741014
741014
741014
741014
741014
741014
741015
741015
741015
741015
741015
741015
741015
741025
741025
741025
741025
7*1025
741025
741024
741024
74 1024
741025
741025
741025
741025
741025
741025
741025
741025
741025
741025
741025
741025
741025
741025
741025
741031
741031
741031
741031
741031
741031
:3S:12
0900
0900
0900
1000
1000
1000
11 JO
1100
1100
1200
1200
1200
10uO
1000
1000
1100
1130
1100
0900
0900
0900
1000
1000
11CO
1100
0800
0800
0800
0900
0900
0900
1000
1000
1000
1100
1100
1100
1200
1200
1200
1300
1300
1300
1400
1400
1400
1500
1500
1500
1700
1700
1800
1800
1800
1900
tc.
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
T
1
3
1
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
1
2
i
1
)
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
4
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
1
1
1
1
1
1
?
2
2
2
2
2
2
2
2
3
3
3
1
1
1
1
1
1
3
3
7
3
T
3
7
2
2
2
2
2
2
2
2
2
2
2
2
2
2
?
2
2
2
2
2
2
2
2
2
^
£
?
T
T
7
7
309.7
130.4
40. £
230.3
1C7.9
66. fc
216.1
70.2
24.9
221.4
97. C
44.4
106.6
39.8
11.1 '
129.3
38.8
8.4
229.6
4C.3
51.6
262.2
30.0
269. C
62. C
477.3
168.9
69.4
464.6
141.2
36.1
411.3
164.2
31. 6
467.2
147.4
43.7
359. C
156.6
73.3
498.8
186.4
71.6
442. C
195.5
40.8
313.1
140.8
48.9
194.7
40.0
206.8
11fc.2
25.9
119.0
322. fc
12V.1
56.6
24£.5
106.2
50.4
239.1
103.2
49.5
283.7
117.1
53.3
198. B
86.6
45.5
192.2
90.4
53.1
584.4
264.7
117.3
457.3
97.6
553.6
112.6
350. 8
137.5
56. 8
393.8
15U.4
62.2
393.8
150.4
62.2
417.7
157.4
64.1
417.7
157.4
64.1
417.7
157.4
64.1
417.7
157.4
64.1
417.7
157.4
64.1
277.3
130.6
491 .1
224 .5
103.0
145.5
52
-------
C00056
COOC57
coucse
COOQ59
COCC60
COCC61
COCC62
COOC63
COCQ64
COCC65
COGG66
COOC67
COC068
COC069
COGC70
COCQ71
COCC72
COOG73
COC074
COCC75
COCC76
COC077
CO CO 78
COCC79
iocceo
C00061
COCC82
COC083
COOG84
C00085
COCC86
COC087
LOC088
COC089
COQC90
COCC91
COC092
C00093
COGC94
COC095
COCC96
C00097
COCQ98
COOQ99
COC100
COC101
000
000
000
000
000
000
OOu
000
000
000
000
000
000
000
000
000
OOu
000
000
ooc
000
000
000
ooc
000
000
000
000
000
000
000
000
000
000
000
000
ooc
000
000
000
000
000
000
000
000
000
741031
7*1031
741031
741031
741031
741031
741031
741031
7411C4
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741114
741120
741120
741120
741120
741120
741120
741120
741120
19oO
1900
20GO
2000
2000
2100
2100
2100
C700
0300
0300
0300
0400
0400
0500
0500
0500
0600
0600
0600
0700
0700
0700
0800
0800
0800
0900
0900
0900
1000
1000
1000
1100
1100
1100
1200
1200
1200
0200
02CO
0200
0300
0300
0300
C400
0400
2
T
1
2
3
1
2
3
3
1
2
3
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
i
2
3
1
2
3
1
2
3
1
2
4
4
5
5
5
5
5
5
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
1
1
1
1
1
1
1
1
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
1
1
1
1
1
1
1
1
•t
i
i
3
3
•»
3
7
3
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
1
1
1
1
1
1
1
1
73.1
22.6
201.1
77.4
36.6
137. £
137.1
34.3
16.6
331.3
153.4
58.5
140.3
61.4
597.6
170.4
52.1
434.2
157.3
54.6
467. C
146.4
53.3
470.0
163.6
40.8
519. i
134.2
55.7
462.7
170. C
54. 8
454.0
161.6
57.7
538.6
160. S
54.3
72.8
74.6
28.6
65. C
49.5
12.4
37. C
23.0
73.4
41.3
234.1
111.7
61.3
292.5
137.5
71.0
53.7
291.0
119.4
53.9
141. A
59.9
417.7
157.4
64.1
417.7
157.4
64.1
417.7
157.7
64.1
417.7
157.7
64.1
417.7
157.7
64.1
417.7
157.7
64.1
417.7
157.7
64.1
417.7
157.7
64 .1
73.0
40.8
31.9
69 .5
38.7
30.9
6JS.3
37.9
END ELT.
53
-------
APPENDIX B
MODIFIED EPA FREQUENCY DISTRIBUTION SOFTWARE
54
-------
B.I Run Stream Example, SRI--SRI.DATAFREQ
" r. 1^ i , 1 1 .->.-/i
0 U 0 u 0 1
OOnoO />
OoOOO 3
UOO-Ji; t
0 0 0 0 <• S
fioOOOb
0 (' 0 0 () 7
G0.'jt<0:i
0 U 0 * ' 0 ^
GO" 0 1 0
GOOUl t
00001 2
00001 3
OOdOl i
0 0 0 0 1 "5
Ot) ">•"> 1 b
OOOOl 7
OOOijl 6
•0000 I 4
0 0 0 0 2 0
0 u 0 0 2 1
00002/>
* K 1 11
;\ M /
• <• 7
no 7
•-/') 7
• i ! ' /
0 ;/ 7
')') /
Ol> i
" ,) 7
•">{; 4
• ;u ^
0 0 i
f<0 /
uO 3
•M' ^
•JO 7
0 0 <
'H'3
1)0 ?
007
0 0 3
00 3
/"I//- ,1V
'•)!,', 1 - -
1 -it / 1
rAl:.) ci'
-1 , -1
1 ] . S , ) ', •
C
'<•<::!' .^^
1 '1 / 1
?Al>0 l^t-
-1 ,-1
1 1 . ^ , " 1
r
' X r i S h
IM / 1
?ii;l> Sr-
\
-t ,-1
1 1 . *> , 1 ( •
C
c
hLI.
: v.i: v - i 7, )
I *.-. t i . r r i ,
^ ! f - -3 }
i *.1.^ i .r<.' r-/
1 • •
* 5 (•• T . (•' ^ !• '
4 1 • • « .1
*S(- I .C'",'^
55
-------
B.2 Map of Program, SRI*SRI.MAPFREQ
•ELT.L
ELT007
COOOC1
COGC02
COGG03
COQG04
COOC05
C00006
COC007
C00008
C00009
C00010
COOQ11
END ELT.
S.MAPFREO
SL73R1 11/01/78 09:36:48 (4,)
DOO 8FOR SRI*SRI.FRE6
000 SHAP ,SRI*SR1.FRE6
000 IN SRI«SRI.FREQ
000 IN SRI*SRI.ORDER
000 IN SRI*SRI.6RALIN
000 IN SRI*SRI.GRAPH
000 IN SRI*SRI.PROB
OOA IN SRI*SRI.HEADER
004 IN SRI*SRI.BOX
004 IN RAPS'imtlTY.CCMPOZ
000 SADD 6RAPH*TEKTRON1X.PREVIEU/MAPECL
56
-------
B.3 Listing of the Main Program, SRI*SRI.FREQ
•ELT.L
ELT307
COC001
COCG02
C00003
COC004
COOOOS
C00006
COG007
C00008
COCQ09
COC01D
CQ0011
COCQ12
C00013
C00014
C00015
C00016
000017
C00018
COC019
COL320
COC021
COC022
COG023
000024
C00025
C00026
COCG27
C00028
COC029
C00030
C00031
COC032
C00033
COG034
COQQ35
COC036
COC037
COC038
OOG039
C00040
GOC041
COG042
COG043
C00044
COCOAS
COC046
COOOA7
COCOAS
COCOA9
COC050
COC051
OOG052
C00053
C0005A
C00055
S. FREQ
SL73R1 11/1
OA9
OA9
OA9
OA9
OA9
OA9
049
OA9
OA9
OA9
OA9
OA9
049
049
OA9
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
049
04V
049
01/7
C P
C M
C P
C M
C T
C T
C t>
C 1
800
500
C Ni
C X
C Y
C Yl
C N
C N
C N
C I
C I
C N
C Ci
801
100
C CI
1
15
25
35
01/78 09:35:38 (53,)
PROGRAM SRI *SR1 .FPECi
MODIFIED FROM M 21 A D 0*GMHta Y . FR EQ
PROGRAM FROM BILL PETERSON
MODIFIED BY ROD ALLEN, COHP-AID INC., OCTOBER 1978
TO COMPILE AND PAP SADD SR I*SR I .MAP F RECi
TO EXECUTE AND DATA SADD.L SR I* SR I . D ATA FRE Q
DECLARATIONS
DIMENSION OB(999),PR(999),P!«0<999>,PMODO<999),OPMODO(999>,
10PMO<999),OOB(999),FREC(999),X<999),Y(999),YY(999),OPR(999)
INITIALIZATION
URITE(6,8CG)
FORMATC ENTER LINE OF CONTROL PARAMETERS')
READ(5,500) N,IOP,IGRAPH,NLOG,XDIM,YDIM,YBASE,NPLT
FORMAT( )
NLOG=NUMBER OF LOG CYCLES.
XOIPI = X DIMENSION IN INCHES OF PLOT.
YDIM = Y DIMENSION IN INCHES OF PLOT. FOR NPLT=3, YDIM = 10.
YBASE = LOWEST Y VALUE. FOR NPLT = 3 , YB ASE =-3 ONLY.
NPLT, NPLT = 1 PLOT OB S E R VED ,PR ED ICTE D VS. CUM. FREQ.
NPLT = 2 PLOT ABS(PR-OB) VS. CUM. FREQ.
NPLT=3 PLOT (PR-OB)/OB VS. CUM. FREQ.
IOP =1, PRINTOUT OF TABLE. IOP = 2,NO PRINTOUT.
IGRAPH=1, GRAPH DISPLAY. IGRAPH=2,NO GRAPH DISPLAY.
N= NUMBER OF DATA POINTS.
COMPUTATION
URITE(6T801 )
FORMAT!* ENTER OBSERVED AND PREDICTED DATA VALUES')
DO 1 1=1, N
READ(5,10C) OBCI),PR(I)
FORHATC33X.2F8.C)
CHECK OBSERVED AND PREDICTED VALUES BECAUSE OF LOGS
IF(OB OB(I) = 0.001
1F(PR(I).LT.O.C01) PR(I)=C.C01
PMO(I)=PR(I )-OB(I )
PMODO(I)=PMO(I)/OB(I)
J=MOD(N,2>
IF(J.Efi.l) GO TO 25
NN=N/2
J=N*1
DO 15 1=1, NN
FREO(I)=100.*((I-0.4)/N)
J = J-1
FRE8(J)=100.-FREa(I)
CONTINUE
60 TO 45
CONTINUE
NN=(N-1)/2
DO 35 I=1,NN
FREO(I)=100.*((I-0.4)/N)
J = J-1
FREQ(J)=1CC.-FREQ(I)
CONTINUE
J=NN+1
= 100.*«J-0.4)/N)
57
-------
GOC056 049 AS CONTINUE
000057 0*5 00 47 1 = 1,N
COC058 349 FREtt
COOG68 049 DO 55 1 = 1,N
C00069 049 WRITE(6,3CO) I,OB ,OPR (I> ,
COOQ70 049 10PMOCI),OPBODO(I> ,FREQ (I)
000071 049 300 FOR«AT(1X ,13,T5,9(F10.3.3X))
C00072 049 55 CONTINUE
OOOC73 049 56 CONTINUE
C00074 049 IFU6RAPH.E0.2) GO TO 4
COG075 049 CALL PLOTS(BUF,1,14)
C00076 049 CALL HEADER
000077 OS1 CALL SYBBOL(2.,-.8,.14,'CUHULATIVE FREQUENCY <(Z)'VQ.,24>
COCC78 049 CALL BOX(0,C,1023,780)
COG079 049 IF(NPLT.EQ.3) GO TO 95
C00080 049 CALL GRAPH (XD1H,YDI* ,NL06,YBASE.1)
COC081 049 IFCNPLT.EQ.2) GO TO 75
COOQ82 049 C NPLT=1
COC083 051 CALL SYMBOL(-.5,1.5,.14.'CONCENTRATION* ,90. , 13 )
000084 051 YZ=YDIH+.5
COC085 051 CALL SYMBOL<-.5,Y7 ,.14,'FREOUENCY DISTRIBUTION FOR OBSERVED AND P
C00086 049 1REDICTED CONCENTRATION',0.,63)
C00087 049 DO 65 1 = 1,N
COOQ88 049 A = FREhU)
000089 049 B=OOB(I)
C00090 049 C=OPR(1)
C00091 049 X(I>=PROB(A)
C00092 049 Y(I)=AL061C(B)
000093 049 YV(1)=AL0610(C)
COC094 049 65 CONTINUE
C00095 049 X(N*1)=PROB(C!.001>
C00096 049 X(N+2)=(PROB(0.999)-PROB(0.001))/XDlM
COC097 049 Y(N»1)=ALOG1C(YBASE)
C00098 049 YY(N+1)=Y(N+1)
COCQ99 049 Y
-------
COG113
COC114
C00115
C00116
COC117
COC118
000119
COC120
COC121
C00122
C00123
C00124
000125
C90126
COQ127
COC128
C00129
C00130
COC131
C00132
C00133
C00134
C00135
C00136
C00137
C00138
COC139
COCUO
COOU1
C00142
COOU3
COC144
COOU5
C00146
COG147
CO I) 148
C00149
C00150
C00151
COU152
C00153
C00154
C00155
C00156
100157
COQ158
L00159
COCUO
C00161
COC162
000163
C00164
052
053
053
049
049
049
049
049
049
049
049
049
049
049
049
049
049
052
049
052
052
052
052
052
052
049
049
049
052
052
053
049
053
053
049
049
049
049
04V
049
049
049
049
049
049
049
049
049
049
049
049
049
85
95
C N
105
C T
999
4
CALL SYMBOL*-.6,Y7 , .14 , 'FREQUENCY DISTRIBUTION FOR RESIDUAL (OBSER
1VED-PREDICTED) CONCENTRATION',J.,70)
IN=1
DO 85 1=1,N
A=FREO(0.001)
X(N + 2) = (PRCB(C.999)-PROB(O.Ci01))/XDlM
Y(N-»1)=ALOG1CUYBASE)
Y(N+2)=NLO€/YDIM
CALL PLOMC., Q.,-3)
CALL LINECX,Y,N,1 ,-1,4)
NIN=N-IK+1
CALL LINE(X(IN),Y(IN),NIN,1,-1,C)
CALL SY«BOL(XD1M-2.5,1.,.2,4,0.,-1)
CALL SYHBOL(XDIM-2.2,1.,.14,'UNDERESTIMATED',0.,14)
CALL SY«BOL
-------
B.4 Subroutine SRI*SRI.GRALIN
iELT.L
ELT007
GOG001
COCC02
C00003
COGCOA
COL'COS
COGC06
COGC07
COG008
C00009
COOQ10
GOG011
C00012
COGQ13
C0001A
C00015
C00016
C00017
C00018
C00019
COCG20
COCQ21
C00022
C00023
COOC2A
COOC25
000026
COCC27
COC028
COOQ29
GOC030
COOC31
COCC32
COG033
COCC3A
COOC35
C00036
COC037
C00038
COC039
COGOAO
GOGCA1
GOOOA2
COCOA3
COGGAA
GOOCA5
COCCA6
COGQA7
COGOA8
COGOA9
GOGC50
C00051
COC052
COGQ53
COG05A
COGC55
S.GRALIN
SL73R1 1
008
012
008
011
008
008
008
008
008
008
008
012
012
012
008
008
008
008
008
012
012
012
012
008
008
008
008
008
008
OOfc
008
008
038
006
008
008
008
008
OOfa
008
008
008
008
008
008
008
008
008
OCb
OOfc
G08
008
OOfe
008
008
1/01/
C I
c
c
c
c
CR
CR
17
C
C
CR
CR
CR
18
C
C
C
c
c
c
c
c
11
12
13
SUBROUTINE CRALIN(XDIM,YDII"l,NLOG,Y6ASE,NPLOT)
MODIFIED BY ROD ALLEN, COMP-AID INC., OCTOBER 1978
GRAPH DRAWS THE BASE CHART WITH PROBABILITY SCALE ON THE
ABSCISSA AND LINEAR SCALE ON THE ORDINATE. THIS ROUTINE WAS
DEVELOPED BY DALE COVENTRY.
DIMENSION XfcAL(21)
DATA XVAL/C.1.C.2.O.5,1.0,2.0,5.0', 1C.0,20.0,30.0,AO.0,50.0,60.0,
U 70.0,80.C,90.C,95.f,98.0,99.0,99.5,99.8,99.9/
DATA DUMT/* '/
IF iNPLOT.GT .1)60 TO 17
SET TO BASE PLOTTER POSITION AT BOTTOM OF PAGE.
CALL PLOTO2.0,-36.0,3)
CALL PLOT((xDin/2.),-33.0,-3)
CALL PLOTt(XDIM/2.),0.,-3)
FCTR = ABS«XDIM/2.)/PROB(O.G01))
60 TO 18
ICHECK=HOD(NPLOT,3)
GO TO CORRECT POSITION r-R MODULUS
(THREE PLOTS ARE MADE ACROSS WIDTH
IFUCHECK.EQ.OCALL PLOT (5.,11.,-3)
IFUCHECK.EC.1KALL PLOT(17.,-22.,-3)
IF(ICHECK.E0.2)CALL PLOT (5.,11.,-3)
INDICATE ON PRINTER THAT PLOT IS BEING ATTEMPTED.
POSITION AT CENTER OF PLOT
CALL SYMBOL(( -X DIM/2.), 0.0, 0.1,13, 0.0,-1)
DRAW CENTER LINE
CALL PLOTU-XDIM/2.), 0.0,3)
CALL PLOT((-XDIM/2.),YDIM,2)
CALL SYMBOL((-XDIM/2.)tYDIM,0.1,13,0.0,-1)
LOOP TO DRAW LINES AND PLOT TOP AND BOTTOM NUMBERS
PROBABILITY PART OF GRAPH.
2 1=2,21
3 OF PLOT NUMBER
OF PAPER.)
FOR
DO
GRAPH01G
GRAPH020
6RAPHC3G
GRAPHOAC
GRAPHC5C
GRAPH060
GRAPH070
GRAPH110
GRAPH120
GRAPH130
GRAPHUC
GRAPH150
GRAPH160
6RAPH170
6RAPH180
GRAPH190
6RAPH200
GRAPH210
GRAPH220
6RAPH230
6RAPH2AO
6RAPH260
(•RAPH270
GRAPH280
GRAPH290
GRAPH300
6RAPH310
GRAPH32C
GRAPH330
&RAPH3AO
M IS
GRAPH3SO
A 'FLIP-FLOP' SWITCH TO 60 FROM TOP TO BOTTOM, THEN BOTTOMGRAPH360
TO TOP, ETC.
ANUM=XVAL(I)
BNUM=100.0-XVAL(I)
CNUM=XVAL(I )*0.01
DETERMINE LINE POSITION
XP=PROB(CNUB)*FCTR
IF(M.LT.Q)GO TO 1
IF(ANUM.LT.10.)60 TO 11
CALL NUMBER((XP-0.17),-0.15,0.07,ANU«,G.0,1)
GO TO M
CALL NUMBER ((XP-Q.07),-0.15,0.37,ANUH.C.u.D
CALL SYMBOL (XP.O.C.C.1,13,O.Ct-D
CALL PLOT (XF,?.0,3)
CALL PLOT(XF,YDIM,2)
CALL SYMBOL(XP,YDIN,0.1,13,3.0,-1)
IF(BNUM.LT.10.)60 TO 13
CALL NH-BER((XP-C.17),0 TO c
CALL NUKt3ER((XP-O.C7),(YDII'i*G.05),G.07tPNUM,j.C,1)
GO TO c
6RAPH370
6RAPH380
GRAPH390
GRAPHAOO
GRAPH410
6RAPHA20
GRAPHA3C
6RAPH4AC
GRAPHA5C
GRAPHA60
GRAPHA70
6RAPHA80
GRAPH490
GRAPH500
GPAPH510
GRAPH52C
GRAPH53C
bRAPHSAO
&RAPH550
GR*PH560
60
-------
OOUC56
CCOQ57
C00058
COCC59
COCC60
COOC61
C00062
COC063
COOC64
COCC65
COGC66
CCOC67
COC068
COG069
COOC70
COCC71
C00072
COCC73
C0007A
C00075
COOQ76
C00077
COCC78
COOC79
OOb
006
006
008
OOfi
008
008
008
008
OOo
008
00£
006
OOfc
00£
OOb
008
008
010
010
006
008
008
008
1
1A
15
16
2
C
50
31
IF (BNUM .LT. 1Q.)GO TC 1
CALL S»*BOL
1F(ANUn.LT. 1C. )60 TO 16
CALL NUMBER ((XP-0.17),-0.15,0.07,ANUH,0.0,1)
60 TO 2
CALL NUHBER((XP-C. 07) ,-0.15,0.07,AMUM, 0.0,1)
CONTINUE
WRITE ORDINATE LEGEND.(TEXT AND NUMBERS)
XCOR=XDl«/<-2.)
CALL AXIS(XCOR,0.0,DUHTt1,VDin,90.,YBASE,(1Q./YDin))
CALL PLOT(XCOR,0.0,3)
00 50 1=1,11
CALL PLOT( (XCOR + XDIH) ,(1-1 .) ,2)
CALL PLOT
-------
B.4 Subroutine SRI*SRI.GRAPH
iELT.L S.
GRAPH
ELTC07 SL73R1 11/01/78 09:36:11 (6.)
COQ001
C00002
COC003
C00004
coooos
COCC06
COOC07
C00008
COC009
COC010
COC011
COCQ1Z
COC013
COCQU
C00015
COC016
COOC17
COC018
C00019
COC020
IOCQ21
COC022
COC0.23
COC024
COC025
OOCC26
C00027
C00028
000029
COC030
C00031
C00032
C00033
COC034
COOC35
COOQ36
C00037
COC038
COOQ39
COCC40
COOOO
000042
GOC043
C00044
COC045
00 C 046
EOOG47
COCC48
COOQ49
COG050
000051
COCC52
C00053
COG054
000055
002
006
002
002
OC2
003
002
002
005
002
002
006
006
006
002
002
002
002
002
006
006
006
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
c
c
c
c
c
CR
CR
17
C
C
CR
CR
CR
18
C
C
C
C
C
c
c
c
11
12
13
SUBROUTINE 6RAPH(XDII",YD1«,NLOG,YBASE,NPLOT)
MODIFIED BY ROD ALLEN, COUP-AID INC., OCTOBER 1978
GRAPH DRAWS THE BASE CHART WITH PROBABILITY SCALE ON
ABSCISSA AND LOG SCALE ON THE ORDINATE. THIS ROUTINE
DEVELOPED BY DALE COVENTRY.
DIMENSION XVAL(21)
DATA XVAL/C. 1,0. 2, 0.5, 1.0, 2. 0,5.0, 1C. 0,20. 0,30. 0,40. 0,5
U 70. 0, 80. G, 90.0, 95. C, 98. 0,99. 0,99. 5, 99. 8, 99. 9 /
DATA DUHT/- '/
IMNPLOT.6T.1J60 TO 17
SET TO BASE PLOTTER POSITION AT BOTTOM OF PA6E.
CALL PLOT(12.0,-36.C,3)
CALL PLOTUXDIM/2.),-33.G,-3)
CALL PLOT ((XDIM/2.) ,0. ,-3)
f CTR=ABS«XOIB/2.)/PROe
BNUM=10C.O-XVAL(I )
CNUM=XVAL (I)*0.01
DETERMINE LINE POSITION
XP=PROB(CNUM)*FCTR
IF(H.LT.O)60 TO 1
IF(ANUM.LT.10.)GO TO 11
CALL NUMBER((XP-0.17>,-0.15,0.07,AHU«,G.G,1>
60 TO 12
CALL NUKB ER « XP-0. 07), -0.15,0. 07, ASUM, 0.0,1)
CALL SYMBOL (XP.0.0 ,0.1 ,13,0. 0,-1)
CALL PLOT (XP, 0.0, 3)
CALL F>LOTCXP,YDIM,2>
CALL SYMBOL
IF(BNUM.LT.10.)60 TO 13
CALL NUMBER <(XP-0.17),< YDIM-tQ. 05), O.C7,BNUM, 0.0,1)
60 TO 2
CALL NUMBER «XP-u.Q7>,(YDIM + 0.05)*O.G7,BNUM,G.C,1)
GO TO 2
GRAPH010
THE bRAPHC2C
WAS GRAPH03G
GRAPMC40
GRAPH050
0.0,60.0, GRAPH060
GRAPH070
6RAPH110
GRAPH120
GRAPH13C
GRAPHUO
GRAPH150
6RAPH160
GRAPH170
6RAPH180
GRAPH190
6RAPH200
6RAPH210
6RAPH220
GRAPH230
GRAPH24C
6RAPH260
6RAPH270
GRAPH28Q
GRAPH290
GRAPH3QO
6RAPH310
OR 6RAPH320
GRAPH33C
6RAPH340
6RAPH350
THEN BOTTOMGRAPH360
6RAPH370
GRAPH380
GRAPH390
GRAPH400
GRAPH41C
6RAPH42C
GRAPH43C
&RAPH440
6RAPH450
GRAPH46C
6RAPH47C
GRAPH480
GRAPH490
GRAPH50Q
GRAPH510
6RAPH52C
GRAPH53C
GRAPH540
GRAPHS'50
6RAPH56C
62
-------
C00056
C00057
C00058
C00059
C01060
COC061
C00062
COOC63
COCC64
COCC65
C00066
C00067
C00068
C00069
C00070
C00071
COOC72
C00073
COC074
COCC75
COCC76
COCQ77
COC078
COC079
cocoeo
C00081
C00082
COC083
COOOE4
C00085
C30086
COCC87
COCC88
C00089
C00090
C00091
COCC92
C00093
COGQ94
COCQ95
COOQ96
COC097
002
002
002
002
002
002
002
002
002
002
002
002
002
002
005
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
1
14
15
16
2
C
C
C
3
4
C
30
31
1 F (BNUh .LT. 10.)GO TO 14 t,R/>PH57
CALL NUMBER UXP-0.17) , (YDIM + 3.05),:.i.7,PNUh,'j.C,1) CRAPH58
GO TO 15 C.RAPH59
CALL NUMBER ((XP-O.J7) , (YDIM+0.05),O.C7.RNUM,C.C,1) GRAPH60
CALL SYMBOL&C TO 30 GRA.PH9
CALL PLOT(XPLOT,YDIM,3) 6RAPH9
CALL PLOKZPLOT,YDIM,2) 6RAPH9
60 TO 31 GRAPH9
CALL PLCKZPLOT,YDIP,3> GRAPHS
CALL PLOT(XPLOT,YDIM,2) GRAPH?
CALL PLOKZFLOT,O.C,-3 ) GRAPH^
RETURN GRAPH'
END GRAPH'
END ELT.
63
-------
B.4 Subroutine SRI*SRI.ORDER
•ELT.L S.ORDER
ELTC07 SL73R1 11/01/78 09:37:01 <2,)
COG001
COC002
COGC03
C00004
COOC05
COC006
COOC07
C00008
COCC09
CC0010
C00011
C00012
C00013
COOOU
000015
COC016
C00017
END ELT.
000
001
001
002
000
000
000
000
300
000
000
000
000
000
000
000
000
SUBROUTINE ORDER(A,E,N)
C SUBROUTINE ORDER FROM BILL PETERSON
C USED BY PROGRAM fREG
DIMENSION A(1),B(1>
DO 10 1=1,N
1Q B(I ) = A(I)
NN=N-1
DO 20 K=1,NN
DO 30 J=1,NN
IF(B(J>-B(J«1)> 30,25,25
25 X=B(J)
B(J)=B(J+1)
B(J+1)=X
30 CONTINUE
20 CONTINUE
RETURN
END
64
-------
B.4 Subroutine SRI*SRI.PROB
iELT.L
EL T 007
COGOC1
COGG02
COOC03
COCQ04
COOC05
COQC06
COOC07
COC008
COOC09
COC010
CO C0 11
COC012
COG013
COQOH
C00015
C00016
COGG17
C00018
C00019
COOC20
C00021
COOC22
COCC23
COG02A
COC025
COC026
C00027
COC028
COCC29
S.PROB
SL73R1 1
oco
000
000
000
000
000
000
000
000
000
000
003
002
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
1/01/7
c*«
c**
c**
c«*
c**
c**
c**
c
9CO
1
2
3
A
5
6
7
8
9
10
FUNCTION PRCB(Z)
C*«*TH£ FUNCTION PROB(Z) TAKES THE INPUT Z IN FREQUENCY
DM 0 TO 1.0 AND DETERMINES THE PLUS OR MINUS
C***STANDARD DEVIATIONS (PROB) ON THE NORMAL CURVE OF
C***FREQUENCY THUS GIVING A PLOTTING POSITION FOR GRAPHING
C***ON THE PROBABILITY SCALE.
C***THIS SUBROUTINE WAS DEVELOPED BY JOE SANTNER AND OBTAINED
C***FROM RALPH LARSEN.
y(A)=S6RT(AL06<1./(A*A)>)
X(A>=W(A)-(t2.515517+0.802853*W.E8.Q) 60 TO 15C
IF (ICODE.LT.O) 60 TO 1*4
COMPUTE FOR ONE/ALL SITES.
CALL TEST1 (N,OC,PC,E>
WRITE (OUT, 11) NOS(ISN),E
IF (LN.LT.NCRIT) 60 TO 142
DETERMINE THE ERROR AND CONFIDENCE LIMITS (C.L.).
SUM2=0.0
DO 141 1=1, f
CALL TEST1
IF (NN.6T.O 60 TO 145
ERR(IS)=-1.C
60 TO 148
CALL TEST1 (NN , SOCC (1 , IS ) , S PCC ( 1 , IS ) , E )
ERR(1S)=E
WRITE (OUT, 11) NOS(IS),E
IF (LM(IS).LT.NCRIT) 60 TO 147
DETERMINE THE ERROR AND C.L.
SUM2=0.0
DO 146 1=1, f
CALL TEST1 (NODSd.IS) ,SOC A (1 , I , IS ) ,SPC A ( 1 , I,IS),A(I»
t: £=A<1)-E
CONTINUE
CALL CONLIM (E , FM ,ROOTM ,SUM2)
IF (MOD(1S,5) .E8.C.AND .IMAP.EQ.3HYES) CALL HCOPV
60 TO 148
PRINT 32
IF = -1 .C,
CALL RAMHAP ( 1 , E R R , END RM>
IF (MODE.E6.2) 60 TO 150
GO TO 290
2. COMPUTE THE MEAN SQUARED ERROR.
IF (MODE ,EO .2 .AND .NOT( Z) .FC .0) GO TO 161
IF (ICODE.LT.C) 60 TO 154
COMPUTE FOR ONE/ALL SITES.
CALL TEST2
72
-------
COC341
C00342
C00343
COC344
COC345
COU346
OOC347
C00348
C00349
C00350
C00351
COC352
COC353
COC354
C00355
000356
COL357
000358
C00359
C00360
COC361
C00362
C00363
COC364
C00365
000366
COC367
C00368
COC369
COC370
COC371
C00372
C00373
COC374
00&375
C00376
C00377
C00378
COG379
COG380
000361
C00382
000383
COC384
COC365
C00386
C00387
C00388
COC389
COG390
L00391
COC392
COU393
COC394
C00395
C00396
C00397
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
025
024
024
024
025
024
025
024
024
024
024
024
024
024
024
024
024
025
024
024
U24
C
151
152
C
154
155
C
156
157
158
159
C
C
C
160
161
F=SQRT(E)
WRITE (OUT, 13) NOS(ISN),E
IF (LN.LT.NCRIT) 60 TO 152
DETERMINE THE ERROR AND C.L.
SUf"2 = O.G
DO 151 1 = 1, P
CALL TtST2 (NODS ( I , IS ) ,CC A ( 1 , I ) ,PC A ( 1 , 1 ) , A ( I ) )
AME=A(I)-E
SUM2=SUM2+AKE*AME
CONTINUE
CALL CONLIM (E , FM ,kOOTM , SUM2)
PRINT 37, F
60 TO 159
PRINT 32
PRINT 37, F
60 TO 159
COMPUTE FOR EACH SITE.
DO 158 IS=1 ,NS
NN=KK(IS)
IF (NN.6T.C) 60 TO 155
ERR
-------
1 fl 0 * V H
cooivv
100400
(00401
L0u402
COG403
1.01404
100405
(.00406
100407
COL.40R
000409
L00410
COC411
CDG412
C00413
000414
C00415
L0041A
COG417
IOC418
L00419
COG420
C00421
COG422
COG423
LOC424
COC425
100426
C00427
100428
COC429
000430
COG431
C00432
COCI433
COC434
COG435
000436
C00437
COC43B
COC439
COC440
C00441
100442
COG443
C00444
COL445
C00446
C00447
C0044H
COG449
CC0450
C00451
COG452
G00453
G00454
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
025
025
024
025
024
025
024
024
024
024
024
024
024
024
024
024
024
024
025
024
024
024
J24
024
(.
t
16?
163
C
164
165
C
166
167
168
169
C
C
C
170
171
C
COUP lit I fO(< ONI/All ', I n r. .
1*1 I If ST 1 (b ,G( ,('( .t >
WRIlt (OUT, 16) NOS(I!>N),(
II (IN.lI.MfKII) f,C 10 1C. '
UFllHMINt IHI tkROK AND I. I.
sn«?- n.O
00 162 1=1, f
C«IL TIST3 =-1.C
fcO TO 168
CALL TEST3 (UN , SOC C ( 1 , I S ) , 5 PC C ( 1 , IS ) , f )
ERR(IS)*E
WRITE (OUT, 16) NOS(IS),f
IF (LM(IS).LT .NCR IT ) GO TO 167
DETERMINE THF FRROR AND C.L.
SU»2*O.C
DO 166 1=1,*
CALL TEST3 (NODS ( I , I S ) , SOC A ( 1 , I , I S ) , S PC A ( 1 , I , I S ) , A ( I ) )
AHE=A( I )-E
SUM2 = SUCI2«AI'E *A«fc
CONTINUE
CALL CONLIH (E ,F B , WOOTM, SUM ?)
IF (HOD (IS , r ) .ttl . C. AND . IflAP .10. ',HY£ S ) CALL HCOPY
GO TO 168
PRINT 32
IF (HOD (I S, 5) .LG .O.AND . IHAP .EO. 3HYE S > CALL HCOPY
CONTINUE
IF (HOD(NS,5) .NE .O.AND.IHAP.E0.3HY t S) CALL HCOPY
IF (IHAP.EO .3HNO ) GO TO 169
CALL RAHMAP ( 3 , E H R , FNO MM )
IF (HODF.EQ.2) GO TO 171
WRITE (OUT, 17)
READ (INP.18) IYON
IF (1 YON.E8 .3HY(t S) GO TO 16C
GO TO 290
4. COMPUTE THF PTRCFNTAGt tRKOR THKLiHOLD.
WRITE (OUT,19)
IF (MODE .EC .
-------
COL455
COG456
COG457
CQC458
COG459
COGA60
COOA61
COC462
COC463
CQG46A
COG465
COCA66
COCA67
COCA68
OOGA69
COCA70
COGA71
COQA72
COC473
COC47A
C00475
COOA76
COCA77
COGA78
COCA79
COGA80
COGA81
COCA82
COGA83
COCA84
000485
C00486
COC487
COCA88
COOA89
C00490
000491
COGA92
COG493
COGA94
COC495
COOA96
COOA97
COGA98
COC499
C00500
COG501
C00502
COC503
C00504
COG505
C00506
COG507
COG508
COC509
COC510
£00511
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
025
024
024
025
024
025
024
024
024
024
024
024
024
024
024
024
024
024
025
024
024
024
024
024
024
024
C
172
173
C
174
175
C
176
177
178
179
C
C
C
180
181
C
WRITE (OUT, 20) NOS(1SN),E
IF (LN.LT.NCRIT) 60 TO 173
DETERMINE THE ERROR AND C.L.
SUM2=O.C
DO 172 1=1, r
CALL TESTA (NODS ( I , IS ) ,OC A { 1 , I ) , PC A ( 1 , 1 ) , A < I ) )
AME=A
-------
.it, M/
IMMA1 *.
(OHM4
I (HIM*.
l.dl'MC,
UK. Mr
( 0 1 M K
OOdMV
cm'. /n
KM. './I
COk.V,'
cnu'KM
ooi, ">/4
in ()'./'>
LOU'./'.
COO'..V
COU'-/H
(0(I'./V
(Oil'. M)
UHl'.M
((!(]•. V
( n n '. ^
LOO', U
UHr, V>
COd'. V,
r.nu'i w
i n4H
IOU'.4V
uur-Mi
(Ol.'.M
IIMI1,',/
( fid'.', ^
( 0(,V>4
IOC'.1.1.
i,lli:',V,
iioc1,', ;
I III.'-1..1!
cm v-v
I 01 '.Ml
IDliV 1
( Oil'././
( (III1, C, ^
I0il'.< 4
KH.'.f '•
I Oi '.<• C.
1 (M.'-f, '
I III '.f.M
1 1 .' '.
il,''.
0 / '.
(1/4
0/4
11/4
'1/4
0/4
0/4
0/4
0/4
0/4
0/4
0/4
0/4
0/4
0/4
0/4
0/4
0/4
0/4
0/4
0 / 4
0/4
0/4
0/4
0/4
0/4
0/4
0 / '•
0/4
0/4
0,">
0/4
0 / ',
0/4
0/4
0 / 4
0/4
0/4
11/4
0/4
(1/4
0/4
0/4
0 / 4
0/4
0 / ',
0/4
0/4
0/4
0/4
0/4
0 / '.
,).' ,
1 1 / '.
0 .' •.
1 III 1 1 KM 1 t 1 Mil 1 1. HDL AM. ( . 1 .
•,»M/-(I -I'
|) 0 I H / 1 - 1 , ~
(AM 1 I ', 1 ". «NOI>'. ( I , I1, ) ,0( A ( 1 , I ) ,IT A ( 1 , 1 ) , A ( 1 ) )
AMI -A ( 1 > -1
MIM/-MIH/ • AMI * AMI
I)'-/ ( ON M Mill
(AII CONIIM o ,i n ,w no i M ,MIM ' )
(.0 10 1HV
1 HI ('MINI ', /
1,0 10 1HV
( roni'ii ii ( OH i ATH ',111.
1 H 4 IX) 1 1> H 1 S - 1 , N •>
N N - r K < I '. )
If
i R H < i r, ) - - i . f
(.0 10 UH
1H'> (AM 1IMS ( ( ( 1 , IS) ,M't f < 1, 1 ', ) ,1 )
1 H H ( 1 r. ) - 1
WHIM (0111 , ,'/ ) NO'. < 1!. ) ,1
II 0 1H(S 1^1, *
(All II MS (N 0 1) !, 1 | , 1 '., ) , S 0 ( A ( 1 , 1 , 1 1 ) . I, I1 ( A ( 1 , 1 , I ', ) , A ( 1 ) )
AMI -A< 1 ) -1
MIM/-MIN/ « AMI 'AMI
1fU. ( ONI INIII
(All fOKIlIM ( 1 .1 M ,HOO 1M ,!. DM /)
II ( M 0 II ( 1 S , ' ) . 1 U . T . A N D . 1 M A )' . | U . Ul Y 1 ', ) (AM HCOI'Y
«, O 1O 1 I' fi
1H7 I'M INI V
11 (MOI)( 1 ',,'.). 1 L .(!. ANn . 1MAT .1 U . 'HY 1 '.) (All IHOI'Y
U'.fl ( ONT INIII
II (MOU (N '. , ' ) .Nl . 1 . AND . 1 MAI' .1 U . MlY 1 ', ) 1 Al I 1KOIY
I . « 1 MAI' .1 b . \ UNO ) (.0 10 1 H '1
(All R A MM AC ( '• .1 U M , 1 Ndl.M >
1 H V || ( M 0 II 1 . 1 U . / ) (, O 10 1 V 1
WN 1 11 ((Ml 1 , 1 M
K 1 A I) (INI'.IH) IYON
II ( 1 VON.I U . MIVI S ) (,O 10 I.1'1'
1,0 10 /V()
1
( C. IIIMIMIII IHI A'.YMMI IKM III(.I!-|(1U 1 ()',', lUNCTlON
I
1 Vf) UIJ 1 1 1 ( Oil I , ," )
1'J' II (MOI> 1 . 1 U ., ./\N|i .NO 1 ( M . 1 U .(I ) (.0 10 />M
WR 1 1 1 t (ill 1,44)
N 1 A 1) ( 1 NT , \ ) 1 4 , III 1 ,111 '
1 NOHM ( 1 )- I 4
1 NOHM ( .' ) - III 1
1 NO UM ( ' > - HI ,'
ii ( i ( iini .1 i . n ) (,o 10 lv.
( ( uf I'll 1 1 1 HI' ONI / A 1 1 ', 1 1 1 '. .
1 AM 1 1 '. 1 f: Nil '. ( 1 ', N ) , 1
II < 1 N . 1 1 .N I l< 1 1 ) (.() 10 1V'
-------
COC569
COC570
COG571
COC572
C00573
COC574
C00575
COC/576
C00577
COC578
COC579
COC580
000561
COC582
COC5S3
COC584
COC565
COG586
C00587
COL588
COC569
C00590
C00591
COC592
C00593
C00594
COOS95
COC596
000597
C00598
COC599
CQC600
C00601
C00602
000603
100604
COC605
C00606
C00607
COC608
C00609
COOdIO
C00611
COC612
C00613
COU614
000615
COC616
C00617
COC618
COC619
COC620
LOC621
COC«22
COC623
C00624
COC625
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
02*
024
024
024
024
024
025
024
024
025
024
025
024
024
024
02*
024
024
024
024
024
024
024
324
024
024
024
025
024
024
024
024
025
025
C
C
C
C
C
C
C
C
DETERMINE
SUK2=O.C
DO 192 1 = 1,
THE EfcROK AND C .L
,
CALL TEST6 (NODS ( I , IS > ,OC A ( 1 , I ) ,PC A ( 1 , I ) , A ( 1 ) )
AME=A(I)-E
192 CONTINUE
CALL CONLIM (E , FM , R COTM , SUK 2)
GO TO 199
193 PRINT 32
60 TO 199
COMPUTE FOR EACH SITE.
194 00 198 IS=1,NS
NN=KK (IS)
IF (NN.6T.U) 60 TO 195
ERRCIS)=-1.C
60 TO 198
195 CALL TEST6 (NN , SO CC (1 , I S } , S PCC ( 1 , I S ) , E )
ERR(IS)=E
yRITE (OUT, 24) NOS(IS) ,E
IF (Ifl(lS).LT.KCRIT) 60 TO 197
DETERMINE THE ERROR AND C.L.
SUM2=0.0
DO 196 1 = 1, B
CALL TEST6 (NODS ( 1 , IS ) , SOC A ( 1 , 1 , 1 S ) , S PC A ( 1 , I , I S ) , A ( I ) )
AME=A(I)-E
SUM2=SUM2+APE*AWE
196 CONTINUE
CALL CONLIM ( E , FM , R OOTM , SUM ?)
IF
-------
COC626 025 GO TO 216
COG627 024 210 READ (1KP.3) ( (AS (I, J ) , J = 1 t 5) .1 =1 , 5 )
C00628 02* WRITE (OUT,33) ((AS(I,J),J = 1,5) ,1 = 1 , 5>
COC629 024 WRITE (OUT,34)
C00630 024 READ (INP,1£> IVON
C00631 024 IF (ITON.EO.3HNO > 60 TO 20C
C00632 025 215 WRITE (OUT,25)
C00633 024 216 READ (INP,3> CKAX
C00634 024 ENORM(1)=CNAX
COC635 024 IF (ICODE.LT.O) 60 TO 219
C00636 024 C COMPUTE FOR ONE/ALL SITES.
C00637 024 CALL TEST? (N,OC,PC,E)
C00638 024 WRITE (OUT,26) NOS(ISN),E
C00639 024 IF (LN .LT.NCRIT) 60 TO 218
COC640 024 C DETERMINE THE ERROR AND C.L.
COC641 024 SUH2=O.C
COC642 024 DO 217 1=1,f
COC647 024 CALL TEST? (NODS (I , IS) ,OCA(1,I),PCA(1,I),A(I))
COG644 024 AME=A(I)-E
COC645 024 SUM2 = SUM2*A!>E*AME
COC646 024 217 CONTINUE
OOC647 024 CALL CONLIM (E ,FM ,RCOTM,SUM2)
C00648 024 60 TO 224
C00649 024 218 PRINT 32
C00650 024 60 TO 224
C006S1 024 C COMPUTE FOR EACH SITE.
000652 024 219 DO 223 IS=1,NS
COG653 024 NN=KK(1S)
COL654 024 IF (NN.6T.G) 60 TO ?20
COC655 024 ERR(IS)=-1.C
COC656 024 GO TO 223
C00657 024 220 CALL TEST? (NN ,SOCC<1,IS),SPCC(1 , IS) , E)
COC658 024 ERR(IS)=E
C00659 024 WRITE (OUT,26) NOS(IS),E
COG660 024 IF (LM(IS).LT.NCRIT) 60 TO 222
COC661 024 C DETERMINE THE ERROR AND C.L.
C00662 024 SUK2=0.0
COG663 024 DO 221 1=1,f
C00664 024 CALL TEST? (NODS(I , IS) ,SOCA(1,1,IS),SPCA(1,I,IS),A(I))
C00665 024 AME=A(I)-E
CQ0666 024 SUM2=SUM2*AP£*AME
COC667 024 221 CONTINUE
000668 024 CALL CONLIM (E ,FM ,ROOTM,SUM2)
C00669 025 IF (MOD(IS, 5).EC .C. AND .IMAP.EQ.3HYES) CALL HCOPY
C00670 024 60 TO 223
C00671 024 222 PRINT 32
C00672 025 IF (MOD(IS,5).EG .0.AND .IMAP.EQ.3HYES) CALL HCOPY
C00673 024 223 CONTINUE
C00674 025 IF (MOO(NS,5).NE .C.AKD .IMAP.EQ.3HYES) CALL HCOPY
C00675 024 IF (IMAP.EG .3HNO ) 60 TO 224
CQC676 024 CALL RAPIHAP ( 7 ,ERR , ENORH)
COC677 025 224 IF (MODE.EC.2) 60 TO 225
COC678 024 WRITE (OUT, 29)
COC679 024 READ (INP,1P) IYON
C00660 024 IF (1 YON.Eft .3«YES ) GO TO 215
C00681 024 WRITE (OUT,31)
COC682 024 READ (INP,1P) IYON
78
-------
COC663
COC684
C00685
C00686
COC687
LOC688
COC689
COC690
C00691
C00692
COC693
COC694
C00695
C00696
C00697
COC698
COC699
000700
COC701
C00702
COC703
C00704
COC705
COC706
COC707
C00708
COC709
COC710
COC711
C00712
C00713
COC714
COG715
COC716
C00717
COC718
£00719
C00720
C00721
C00722
C00723
COC72«
COC725
C00726
C00727
100728
C00729
COC730
COC731
C00732
COC733
C00734
C00735
COC736
C00737
C00738
C00739
C2A
024
024
024
025
025
025
025
025
025
024
025
025
025
025
024
025
325
025
024
025
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
024
C24
024
024
024
024
024
024
024
U25
C
C
C
225
226
C
227
228
229
230
C
C
235
240
C
250
C
255
IF (IYON.EU .3HYtS > GO TO 200
8. DETERMINE THE LOCATION «HtFt THE MAX CONCENTRATION OCCURS
If (ICOOE .£C.-1.AND.NOT(6).EQ.1 ) GO TO 226
IF (NOT(8). EQ .C) GO TO 2£ 0
WRITE (GUT, 46)
GO TO 280
IF (HODE.EC.2) GO TO 227
WRITE (OUT, 27)
READ (ISP.U) IYON
IF (irON. EG .3HYES) 60 TO 229
QUICK-RUN USES PROGRAM DISTANCE MATRIX.
DO 228 1 = 1, NS
DO 228 J=1,NS
DD , J = 1 ,N S)
WRITE (OUT, 34)
READ (INP,1f> IYON
IF (IYON.EC.3HNO ) GO TO 225
IF (ICODE.LT.O) GO TO 250
COMPUTE FOR ONE/ALL SITES.
CALL TESTS (N,OC,PC,E)
WRITE (OUT, 28) NOS(ISN),E
IF (LN.LT.NCRIT) GO TO 2*0
DETERMINE THE ERROR AND C.L.
SUM2=0.0
DO 235 1 = 1, M
CALL TESTS (NODSd.IS) ,OCA< 1,I>,p{A(1,I),A( I»
AME=A(I)-E
CONTINUE
CALL CONLIM ( E , FM ,ROOTM ,SUM 2)
60 TO 280
PRINT 32
GO TO 280
COMPUTF FOR EACH SITE.
00 265 IS=1,NS
NN=KK(IS>
IF (NN.EQ.O) 60 TO 265
CALL TESTS ,E)
WRITE (OUT,2b) NOS(IS),E
IF (LM(IS) .LT .NCRIT) 60 TO 260
DETERMINE THE ERROR AND C.L.
SUM2=0.0
DO 255 I=1,P
CALL TESTS (NOD S ( I , IS ) ,SCC A (1 , I , I S ) , S PC A ( 1 , I , I S ) , A ( I ) )
AME=A(1 )-E
SUM2=SUM2*AfE*AME
CONTINUE
CALL CONLIM ( E , F M , R OOTM , SUX, 2 )
IF (MOD (IS,5) .tG.C.AND .IMAP.EQ.3HYES) CALL HCOf-Y
79
-------
C00740
COC741
C3C742
COC743
00074/,
C00745
COC746
COC747
COC7A8
COC7A9
COC750
C00751
COC752
C00753
C0ii75*
COC755
I.00756
C00757
C00758
C00759
COC760
000761
024
024
025
024
325
024
024
024
024
024
024
024
024
025
024
024
024
025
024
025
024
024
60 70 265
260 PRINT li
IF (MOD( IS, 5) .EG .C.AND.IHAP .EG .3HYES) CALL HCOPY
265 CONTINUE *
IF (nOD(NS,5) .ME .O.AND .IKAP.EQ .3HYES) CALL HCOPY
260 IF (MODE.E0.2) 60 TO 3CO
; TRY ANOTHER TEST?
290 WRITE
READ (INP.18) IYON
IF (IYON.EO .3HYES) 60 TO
: TRY ANOTHER NODE?
300 WRITE (OUT, 36)
READ (INP,18) IYON
IF (IYON.EU.3HYES) GO TO
: TRY ANOTHER SORT?
WRITE (OUT, 38)
(0,2C)
13C
99
READ (INP,1fr) IYON
IF (IYON.EQ.3HYES) 60 TO 54
WRITE (OUT, 86)
IF (INAP.Ea.3HVES) CALL FINITT
STOP
END
END ELT.
80
-------
C.2 Program Map, M21ADO*STAT01.MAPIT
SELT,L
ELTC07
COCC01
COGC02
COC003
cocoo*
cocoes
COOC06
CO&G07
coocoe
:OCCD9
C00010
COOC11
S1 .MAPIT
SL73R1 1
003
001
001
001
001
001
001
001
301
001
003
1/01/7P
SWAP
IN M
IN «
IN M
IN S
IN S
IN S
IN R
LIB
END
iXST
09:38:21 («,)
,M21ADO*STATC1.ACSCOR
M21ADO*STAT01.ACSCOR
«2lADCi*STATQ1.CONLIH
M21ADO«STAT01.TESTS
SRI*SRI.RAMMAP
SRI*SRI.BOX
SRI*SRI.HEADER
RAPS*UTILITY.CCMPOZ
eRAPH*TEKTRONIX.
«21ADC*STAT01.ACSCOR
END ELT.
81
-------
C.3 Subroutine M21ADO*STAT01.TESTS
5ELT,L St. TESTS
ELTC07 SL73R1 11/01/7?
LOGC01 000
COGC02 000
COCC03 000
C00004 000
COOC05 000
000006 300
COOC07 000
COCC08 000
C00009 000
COC010 000 100
COOC11 000
C00012 000
COGC1J 000
COG01A 000
COC015 000
COOC16 000
COOC17 000
COOG18 000 110
C00019 000
COOC20 000
C00021 000
COCC22 000
COOG23 000
COCC24 000
COGG25 000 120
CCG026 000
C00027 000
COOG28 000
C00029 000
COCC30 000
C00031 000
C00032 000
C00033 000
COOC3A 000
COC035 000 125
COOC36 000 130
000037 000
COCC38 000
COQ039 000
000040 000
COCG41 000
£00042 000
COGOA3 000
COCOA4 000
C00045 000 UO
COOOA6 000
COOOA7 000
00 03 AS 000
COOOA9 000
COC050 000
CQG351 000
100052 000
COOG53 000 150
COGC5A 000
COCG55 000
09:38:22 (1,)
SUBROUTINE TESTS
DIMENSION OC(N),PC(N)
COMMON /DAT/El,E2,E2tE4,PChIN,hLl,HL2,Cf»AX,DDtAS(5,5)
COMMON /VAR/IDATE(123>,ITIME(123),1SITE<123)
RETURN
ENTRY TEST1 (N,OC,PC,E)
sun=o.o
00 100 1=1, N
SUn=SUM*ABS(OC (I)-PC(I ))
CONTINUE
E=SUM/N
RETURN
ENTRY TEST2 (N,OC,PC,E)
DO 110 1=1, N
OCHPC=OCCI)-PC(I)
SU«=SU«*OC«PC*OCMPC
CONTINUE
E-SUM/N
RETURN
ENTRY TEST3 (N,OC,PC,E)
SUM=0.0
DO 120 1=1, N
IF (ABS(OC( I)-PC(I)).6T.E1)
CONTINUE
E=SUH/N
RETURN
ENTRY TESTA (N,OC,PC,E)
SUM=0.0
DO 130 1=1, N
IF
SUH=0.0
DO 1AO 1=1, N
IF (OC(I).LE.E3.AND.PC(I) .LE.E3) GO TO 1AO
IF (OC( I) .6T.E3.ANO.PC (D.GT.E3) GO TO 140
SUM=SUM+1 .u
CONTINUE
E=SUH/N
RETURN
ENTRY TEST6 (N.OC.PC.E)
SUM=0.0
DO 150 1=1, N
IF (OC(I) .LT.EA . AND. PC (I) .GT.E4) SUM = SUM-»HL1
IF (OC(I).GT.EA.AND.PC (I) .LT.EA) SUM = SUW+HL2
CONTINUE
E=SUh/N
RETURN
82
-------
C00056
COOG57
COGD58
COCC59
LOUC60
COCC61
COCC62
COCC63
COOCtA
COGC65
COLC66
0 COO 67
COC068
CCGC69
COuC70
COCC71
COCC72
COOC73
LOOC74
C00075
COC076
COQC77
COCG78
COC079
COG080
coooei
COC062
C00083
COC084
COOC85
COG066
C00087
000068
COC089
COC090
C00091
COOC92
COOC93
COGC94
COC095
COC096
COCC97
000098
OCG
ooc
000
000
000
000
ooc
000
000
000
000
300
000
000
000
UOO
000
000
000
000
000
000
ooc
000
000
000
000
000
000
000
000
000
000
000
000
ooc
000
000
000
000
000
000
000
cND ELT.
iHDG.N
170
175
160
ENTRY TEST? (N,OC ,PC,E )
S U M = j . 0
DO un 1 = 1,N
J = 5.0*( +t.2*CI'iAX)/£MAX>
IF U.fcT.5) K=5
SUM = SUI"l + AS ( J, K)
1tO CONTINUE
1&5
RETURN
ENTRY TESTfc (N,OC,PC,E)
sur,=o.u
NT = 0
X = -1 .CE1 C
PMAX=-1 .QE1 C
IDATEO=IOATE(1)
ITIMED=ITIME<1)
00 185 1=1, N
IF (1DATE (I ).NE.IDATED )
IF (ITIME(I).NE.lTirED)
60
GO
TO
TO
18C
IF (OC(I).LT.OHAX) 60 TO 175
OHAX=OC(I)
«OC=ISITE (I)
IF (PC11) .LT.PMAX) GO TO 185
PHAX=PCtI)
«PC=IS1TE (I )
60 TO 185
NT=NT+1
= SUM+OD(MOC ,MPC)
MPC = 1
OMAX=-1 .CE1 C
PI«AX = -1 .OE1 C
IOATEO=IDATEtI)
60 TO 170
CONTINUE
E=SUM/NT
RETURN
END
83
-------
C.3 Subroutine M21ADO*STATO1.CONLIM
«ELT,L S1.CONLIf
EL1GG7 SL73R1 11/31/78
UOCC01 DOO
CO&OC2 OCO
COGC03 000
COGOO« 000
COG005 000
COC006 000
COCQ07 000
COCC08 000
COC009 000 40
COCC10 000
COQ011 OOU
END ELT.
09:33:08 (2,)
SUBROUTINE CONLIM (E,Ff,ROOTM,SUM2 )
S=SQRT(SU«2/(FM-1.3))
CLU=1.96«S/ROOTM
CLL=-CLU
CL1=E+CLU
CL2=E+CLL
IF (CL2.LT.C) CL2=0.0
PRINT *C,CL2,CL1
FORMAT (/12X24HV5X CONFIDENCE INTER VAL :E10 .3 , AH TO E10.3)
RETURN
END
84
-------
C.3 Subroutine M21ADO*STATO1.RAMMAP
•ELT.L
ELTG07
COOC01
COC002
C00003
COG004
000005
C00006
COOC07
COGC08
COCC09
COOC10
COGC11
COC012
000013
COGOU
C00015
COCC16
COCC17
COGG18
COOG19
COGG20
COCC21
COGQ22
COG323
COC024
COCC25
COCQ26
C00027
C00028
COG029
COG030
C00031
C00032
COOC33
COCQ3A
COG035
C00036
COQQ37
COQC38
COOC39
COCC40
COG041
C00042
COOG43
COCC44
COG045
COCG46
COC047
COCOAS
C00049
COG050
n0051
COGG52
COGG53
C00054
COGC55
S. RAMMAP
SL73R1 1
026
026
02c
026
026
026
326
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
028
028
028
028
026
026
026
026
026
026
026
026
026
026
026
1/01/
C
C
C
C
C
C
C
C
C
75
76
C
C
1/01/78 G9:37:1j (30.)
SUBROUTINE RAMMAP(NTEST,VALUES,PARM)
ROD ALLEN, COMP-A1D INC., OCTOBER 1978
PLOT VALUES ON RAMS STATION HAP OF ST. LOUIS FOR RAM MODEL TESTS
FOR RON RUFF OF SRI
ASSUMES TEKTRONIX 4G14
3 E1 THRESHOLD, 4 E2 THRESHOLD, 5 E3 NORM, 6 E4 NORM, L1 LOSS 1, L2 LOSS 2,
7 LOSS MATRIX SUPPLIED BT USER, NOT LISTED HERE
DECLARATIONS
DIMENSION TEST<6,7),ZERO(4),VALUES<25),LOC<25>,PARM<3),AOUT(5>
DIMENSION RAMSX(25),RAMSY(25),XLAB<4),YLAB<4),NUMBER<25)
DIMENSION MOX(5),MOY(5),MSX<14),HSY(14),MSLX(6>,MSLY<6)
INITIALIZATION OF DATA
DATA MMS/14/.MMO/5/,MSL/6/
DATA MOX/696,710,720,732,750/,MOY/4276,4284,4297,4306,4:»On/
DATA MSLX/ 7 39, 735,733. 734, 7 38,7 46 / .MSLYM263,4270, 4275, 4282,4288,
1 4294/
DAT A MSX/73C,732,739,744,746,744,744,746,75C,750,735,723,714,708/
DATA MSY/4235,4251,4268,4273,4230,4286,4290,4294,4300,4306,4310,
1 4314,4314,433C/
DATA RAMSX/744.2,742.5,747.6,747.3,743.7,
1 738.7,740.2,748.4,755.8,747.2, 738.8,733.9,737.7,744.3,757.1,
2 762.8,760.6,743.1,729.8,723.1, 732 .4 ,741 .6 ,777.3,749.3 ,697,4/
DATA RAHSY/426C.,4286.,4282.,4277.,4276.,
1 4278.,4283.,4291.,428C.,4273., 4272 . ,4281. ,4290.,4297.,4298.,
2 4290.,4273 .,4263.,4271.,4286., 4302. ,4329. ,4286.,4237 . ,4282./
DATA XLAB/'700",'720','740','760'/
DATA NUMeER/'01',"02','03','04','05",*06','37','08*,'09','10',
1 '11',*12','13",'14',"15','16','17','18','19','20','21','22',
2 '23','24','25'/
DATA LOCM, 1,1,1,3,3,3,1,1,1,4,3,3, 3,1,1,1,1,3,3,1,4,3, 2, 1/
DATA ZERO/*C.',' 0.',' 0.',' Q.'/
DATA YLAB/'425G','4275','43CO','4325'/
DATA TEST/'MEAN ABSOLUTE ERROR
'MEAN SQUARE ERROR
'ABSOLUTE ERROR THRESHOLD
'PERCENTAGE ERROR THRESHOLD
"SYMMETRIC HI6H-LOU LOSS FUNCTION '
'ASYMMETRIC HIGH-LOW LOSS FUNCTION'
'USER SUPPLIED LOSS MATRIX
INITIALIZATION
WRITE(6,75)
FOR«AT*64
CALL ftOUTST(1,N)
CALL MOVABS(20,680>
CALL AOUTST(33,TEST(1 ,NTEST))
PRINT PARAMETERS
85
-------
NOT LISTED HERE*)
COC056 026 IF(NTEST.LE .2) 60 TO 8C
COOCS7 026 CALL MOVABS(20,660)
C00058 026 If(NTEST.LE .6) 60 TO 81
C00059 026 CALL AOUTST(45,'LOSS MATRIX SUPPLIED BY USER,
COC060 029 CALL MOVABS(20,640)
COQC61 026 81 ENCODE(30,82,AOUT) PARK
COCC62 026 82 FORMATdPGS .2 , 4X , 1 P268 .2 , 2X )
COOC63 026 IF(NTEST.EC.3> CALL AOUTST(12,'E1 THRESHOLD')
C00064 026 IFtNTEST.EO.4) CALL AOUTST(12,'E2 THRESHOLD')
COC065 026 IF (NTEST.Efl .5) CALL AOUTST(7,'E3 NORM')
COC066 026 IF(NTEST.EG.6> CALL AOUTSTC25,'E4 NORM, L1 LOSS, L2 LOSS')
COOC67 025 IF(NTEST.E8.7> CALL AOUTST(26,'CMAX CONCENTRATION MAXIMUM')
COCC68 026 CALL AOUTST(6,AOUT)
COGC69 026 IF(NTEST.EQ .6) CALL AOUTST<16,AOUT(3))
COCO70 026 80 CALL CHRSIZ (3)
COOC71 026 C USES THE NEXT TC SMALLEST OF 4 CHARACTER SIZES
C00072 026 C WRITE SUBTITLES AND LE6END
C00073 026 CALL MOVABS (20,2CO
COOC74 026 CALL AOUTST(29,'VALUES 6IVEN NEXT TO STATIONS')
COG075 026 CALL HOVABS(22,184)
COCC76 026 CALL AOUTST(15,'SN RAMS STATION')
COC077 026 CALL BOX(20,182,38,196)
C00078 026 CALL NOVABS(20,168)
COCC79 026 CALL AOUTST(34,'WHERE SN+10C IS THE STATION NUMBER')
COCC80 026 CALL MOVABS(20,152)
C00081 026 CALL AOUTST(37,'VALUE EXAMPLES 12-3 = .012 AND 12 + 3 = 120')
COOC82 026 CALL MOVABS(2C,136)
C00083 026 CALL AOUTST(26,'AXES UNITS UTM COORDINATES')
C00084 026 C DRAW FRAME
C00085 026 CALL BOX(0,C,1023,78C)
000086 026 C DRAW TIC MARKS
000087 026 DO 1 1=1,4
COC088 026 IX = 1024*(68C.+ 20*1-696.3)7(778.4-696.3)
COC089 026 IY=781*(4225.+25*1-4235.7)7(4330.3-4235.7)
C00090 026 CALL MOVABS(G,IY>
COCC91 026 CALL ORNABSU.IV)
C00092 026 CALL MOVABS (1023, IV)
C00093 026 CALL DRWABS(1019,I V)
COC094 026 CALL MOVABSUX.O)
000095 026 CALL DRHABS(IX,4)
COCC96 026 CALL MOVABS (IX,776)
C00097 026 CALL DRUABS(IX,780)
COCO 98 026 CALL MOVABSUX-6,5)
COC099 026 CALL AOUTST<3,XLAB(I))
C00100 026 CALL MOVABS<5,IY-6)
C00101 026 1 CALL AOUTST (4,YLAB(I))
C00102 026 C PRINT TIME AND DATE HEADER
C00103 026 CALL HEADER
CO0104 026 C DRAW STATION BOXES AND WRITE STATION NUMBERS
C00105 026 DO 2 1=1,25
C00106 026 IX = 1024*(RAMSX(I)-696.3)7(778.4-696.3)
C00107 026 IY=781*(RAMSY(I)-4235.7)7(4330.3-4235.7)
C00108 026 CALL BOX(IX-9,IY-7,IX+9,IY+7)
C00109 026 CALL HOVABS(IX-7.IY-5)
C00110 026 CALL AOUTST (2,NUKBERd ))
C00111 026 C WRITE VALUES NEAR STATIONS
C00112 026 IF
-------
C00113
COG114
COC115
CC0116
C00117
COG118
C00119
COC120
COC121
C00122
COG123
COC124
COC125
COC126
C00127
C00128
COG129
COC130
C00131
000132
C00133
COC134
000135
COC136
COG137
COC138
C00139
COOHO
000141
000142
COOU3
C00144
C00145
C00146
COG147
COCU8
C00149
C00150
COC151
C00152
C00153
COC154
CC0155
C00156
C00157
C00158
COQ159
C00160
COC161
C00162
C00163
COC164
COG165
COC166
COC167
C00168
COOT69
026
026
026
026
027
027
027
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
026
030
030
030
026
C
C
6
C
C
3
IF(LOCd) .EG.2) CALL N CV A P S < I X - 1 5 , I Y-» 1 1 )
IF(LOCd) .EG.3) CALL M CVA E- S ( I X -43 , 1 Y-5 )
1 F(LOC (I) .E0.4) CALL MOV AF S ( I X- 1 5 , I Y -2 0 )
ENCODE OUTPUT OF VALUES(I)
IF VALUES(I) < f LEAVE BLANK
IF(VALUESU)) 2,6,3
L=LOC(I)
CALL AOUTSTU ,?ERO(L))
60 TO 2
IF VALUES(I) > S9 + C9 OR <= 99-K USE X + XX FORCAT
IF VALUESd) <= 99 + 09 AND > 99-10 USE X X •• X FORKAT
IF(VALUES
-------
C00170
COG171
COCJ172
000173
COC174
COC175
COG176
C00177
000178
END ELT,
027
027
027
027
027
027
030
026
026
CALL flOVABS (10,50)
CALL AOUTSK24,'PRESS RETURN TO CONTINUE')
CALL BELL
CALL TSEND
CALL TINPUT(I)
CALL ERASE
CALL F1MTT(0,7£G)
RETURN
END
88
-------
C.3 Subroutine M21ADO*STAT01.HEADER
oELT.L
ELTC07
COCC01
C00002
C00003
C00004
C00005
COCC06
COC007
COCC08
C00009
C00010
C00011
COC012
COC313
000014
COC015
COC016
COC017
D00018
Q00019
S. HEADER
SL73R1 11/
016
014
016
016
017
018
014
014
014
OU
014
014
014
014
014
014
014
016
016
01/78 C9:36:20 (18.)
SUBROUTINE HEADER
C ROD ALLEN, COUP-AID INC., OCTOBER 1978
C PRINT TIME AND DATE HEADER IN UPPER RIGHT CORNER AND BOX IT
C ASSUMES TEKTRONIX 4CU
DIMENSION DATE(2>, TIME (2)
CALL CHKSIZC3)
BOX(953,736,1023,780)
COMP02(0»IL .DATE >
COMPOZ(18)
COMP02 <0,1L,T1J"E>
COBPOZ(17)
MOVABS(956,767)
AOUTST(7,'PLOTTED')
MOVABS(956,754)
AOUTST(8,6ATE)
MOVABS(956,741)
AOUTST(8,TI«E)
CALL
CALL
CALL
CALL
CALL
CALL
CALL
CALL
CALL
CALL
CALL
RETURN
END
END ELT.
89
-------
C.3 Subroutine M21ADO*STAT01.BOX
iELT.L
ELTCG7
OOCC01
COCC02
COCC03
COC004
COC005
COCC06
COOC07
C00008
C00009
COCC10
COC011
S.BOX
SL73R1 11/01/78
000
000
000
000
000
000
000
000
000
000
000
C RO
C DR
C CO
09:35:19
-------
C.4 Data Element M21ADO*STATOl.DISTAN
iELT,L S1.01STAN
ELTGC7 SL73R1 11/01/78 09:38:16 (C.)
COCC01 000 C.O 0.45 1.15
COC002 000 0.45 C.O 0.70
COCC03 000 1.15 0.70 0.0
tND ELT.
91
-------
C.4 Data Element M21ADO*STAT01.ERRLOS
SELT,L S1.ERRLOS
dLTCOr SL73R1 11/01/78 u9:38:18 (C,)
COC001 000 O.C 1.0 2.0 3.C 4.0
C00002 000 1.0 0.0 1.0 2.0 3.D
COLC03 OOC 2.0 1.0 0.0 1.C 2.0
COGQOA 300 3.0 2.0 1.0 C.O 1.0
COC30S 000 4.0 3.0 2.0 1.C 0.0
END ELT.
92
-------
APPENDIX D
RESIDUAL TIME SERIES LISTINGS
93
-------
D.I Control Element Example, M21ADO*STAT02.CASE1
•ELT.L S2.CASE1
ELTCC7 SL73R1 11/01/78 09:39:0fc U.)
COCCC1 001
COCCG2 001
CO&C03 003
COC004 002
CCCC05 002
000006 002
C00007 00*
LOC008 002
COC009 002
C00010 000
COCQ11 001
END ELT.
SXCT M21ADO*STATJ2.T1MSER
(CHIKU/G)
1 0011
01
OP TIME LAG
15 AUTOCORRELATION
01
13 FREQUENCY (F)
42 VALUE OF NORMALIZED CUMULATIVE PERIOOOGRAM
101 3 741C12 0900 74112C J400
&ADD M21ADO*STAT01.COGS2N
94
-------
D.2 Program Map, M21ADO*STAT02.MAPIT
«ELT,L S2.MAPIT
ELTC07 SL73R1 11/
'J02
302
COG003 001
COCQOA 001
COC005 003
C00006 003
C00007 003
COCC08 000
C00009 000
END EL7.
01/78 09:29:33 (3,)
IK H21ADOSTATQ2.11PSER
IN M21 ADU*STATQ2. SETUP
IN n21ADO*STATQ2.tSESET
IN SRI*SRI.BOX
IN SRI*SRI. HEADER
IN RAPS*UTIL1TY .CCMPOZ
LIB 6RAPH*TEKTRONIX
END
95
-------
D.3 Listing of the Main Program, M21ADO*STAT02.TIMBER
•ELT.L
ELTC07
OOOC01
COCOOZ
COGG03
COQ004
000005
CO D 006
C00007
COC008
COC009
000010
COC011
COCC12
C00013
n riftft i L
U UUU 1 *f
C00015
C00016
C00017
COOC18
COCQ19
C00020
COC021
C00022
COC023
C0002A
C00025
C00026
C00027
LOCC28
C00029
C00030
C00031
COCC32
COC033
OOOC34
COOC35
C00036
COCC37
C00038
COCC39
COCC40
COCC41
COCOA2
C00043
COOCAA
COC045
C00046
COGG47
COCOAS
OOC049
COC050
COC051
COC052
COC053
COG054
COC055
S2.TIMSER
SL73R1 1
011
011
011
011
011
011
011
011
011
311
011
011
011
AI 1
U 1 1
011
008
011
006
008
008
009
OOti
008
006
008
008
008
008
008
008
008
ooe
008
008
008
008
008
008
008
008
008
008
008
008
008
008
006
008
008
006
008
OOb
008
OOt
006
1/01
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
/78 09:39:35 (12,)
*******************************************************************
THIS PROGRAM COMPUTES THE AUTOCORRELATION AND NORMALIZED CUM-
ULATIVE PERIOD06RAM FOR THE RESIDUAL TIME StRlfS.
REFERENCES:
1. BOX, 6.E. AND 6.M. JENKINS, 1976: TIKE SERIES ANALYSIS
FORECASTING AND CONTROL, HOLDEN-DAY, SAN FRANCISCO, CALIF.
2. JENKINS, 6.M. AND D.G. WATTS, 12966: SPECTRAL ANALYSIS AND ITS
APPLICATIONS, HOLDEN-DAY, SAN FRANCISCO, CALIF.
MODIFIED BY ROD ALLEN, COUP-AID INC., OCTOBER 1978
D I PENSION OC(123,3>,PC(123,3),IDATE(123),ITI«£(123),RTS(123),ACF
1(15),CNP(123),FI<123),XX(2),YY(2),IDEN(5),TIT1(10),TIT2(10),XARR1
2(10),XARR2(10),YARR1(1C),YARR2(10)
COMMON /DAT/ XMIN , XMAX , YMIN , YM AX , I T YPE , 1C OD E , I S
DATA PI.ITYPE /3 . U16 , 001 1 /
1 FORMAT
5 FORMAT
6 FORMAT (/6H S ITE : , 13,5 X .5HRBAR : ,F1 0.2 ,5X,6HSUMR2 : , F10.2 /)
7 FORMAT
-------
C00056
COC057
COC058
OOC059
cocoto
COOC61
COOC62
C00063
COOC64
C00065
COG066
COC067
COC068
COC069
COOC70
COOC71
COOC72
LOC073
C0007*
COOC75
COOC76
COC077
COOC78
COOC79
COOC80
COCC81
000082
C00083
COC08A
COOC85
COC'066
COC087
COC088
COOC89
COCC90
C00091
COCC92
COC093
COOC94
COOC95
COOC96
C00097
C00098
C00099
C00100
COC101
000102
100103
C0010*
C00105
COC106
COC107
C00108
C00109
COC110
COC111
C00112
008
008
006
008
008
008
008
008
008
008
008
008
008
008
008
008
008
008
008
008
008
008
006
008
008
008
008
008
008
008
008
008
008
OOfc
008
008
008
008
008
008
008
008
008
008
008
008
008
OOfc
008
008
006
003
OOfc
008
008
006
006
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
CR
100
110
115
120
125
130
135
CR
INPUT FROM PARAMETER FILE ...
IDEN = DATA IDENTIFIER
ICODE = 1, AUTOCORRELATION ANALYSIS.
= 2, NORMALIZED CUMULATIVE PERIOD06RAM
= 3, DO CODES 1 AND 2.
ITYPE = CC01 , X-AXIS IS LINEAR.
= 0010, Y-AXIS IS LINEAR.
= 0100, X-AXIS IS LOGARITHMIC.
= 1COO, Y-AXIS IS LOGARITHMIC.
NTC1 = NO. OF ASCII CHARACTERS IN TIT1.
TIT1 = UPPER TITLE FOR GRAPH 1.
NXC1 = NO. OF ASCII CHARACTERS IN XARR1.
XARR1 = X-AXIS LABEL FOR GRAPH 1.
NVC1 = NO. OF ASCII CHARACTERS IN YARR1.
YARR1 = Y-AXIS LABEL FOR GRAPH 1.
NTC2 = NO. OF ASCII CHARACTERS IN TIT2.
TIT2 = UPPER TITLE FOR GRAPH 2.
NXC2 = NO. OF ASCII CHARACTERS IN XARR2.
XARR2 = X-AXIS LABEL FOR GRAPH 2.
NYC2 = NO. OF ASCII CHARACTERS IN YARR2.
YARR2 = Y-AXIS LABEL FOR GRAPH 2.
READ 3,ND,NS,IDATEB,ITIMEB,IDATEE,ITIMEE
PRINT4,ND,NS,IDATEB,ITIHEB,IDATEE,1TIMEE
DO 110 1 = 1, NS
DO 100 J=1,ND
OC=-99.9
CONTINUE
CONTINUE
NT = 1
DO 135 1=1, NC
READ 3,10ATES,1TIMES,I$ITE,ISTC,IWSC,IWDC,OCS,PCS
IF (10ATES.EO.IDATEB.AND.ITIMES.LT.ITIMEB) GO TO 135
IF (IDATES.EC .IDATEE.AND.ITIHES .GT.ITIMEE ) GO TO 135
IF (I.6T.1) GO TO 115
IDATED=IDATES
ITIMED=IT1MES
IF (IDATES.E8 .IDATED) GO TO 120
IDATED=1DATES
NT=NT+1
IDATE (NT)=IDATED
GO TO 125
IF (IT1MES.EQ .ITIMED) GO TO 130
NT=NT+1
ITIMED=ITIMES
ITIME (NT)=ITIK£D
OC(NT,ISITE ) = OCS
PC (NT.ISITE ) = PCS
CONTINUE
PRINT 5, NT
PI2=2.0*PI
RNT=NT
97
-------
COC/113
C001U
COG115
C00116
C00117
C00118
COG119
C00120
COC121
C00122
COt 123
C00124
C00125
C00126
C00127
C00128
C00129
COC130
C00131
CCC132
C00133
COG134
C00135
C00136
C00137
000138
COC139
000140
COOU1
COC142
COC143
COG144
COOU5
C00146
000147
C00148
COC149
C00150
C00151
C00152
COC153
C00154
OOC155
C00156
C00157
C00158
C00159
C00160
C00161
COG162
000163
C00164
COC/165
C00166
C00167
C00168
C00169
006
008
008
012
006
012
008
008
008
ooe
OOfe
008
008
008
008
008
012
012
008
008
011
011
011
011
011
011
011
008
008
008
008
008
006
012
008
008
008
006
008
008
008
008
012
008
008
008
008
008
008
008
012
012
012
012
012
006
006
CR
C10
c
c
139
140
C
C
145
CR
150
155
CR
160
CR
C
C
C
C
C
C
ROOTN=S6RT(RNT>
JN=NT/2
RJN=JN
PRINT 10,RNT,ROOTN,THON,PI2
FORMAT (/1H .10F12.4)
DO 300 IS = 1fNS
DETERMINE THE RESIDUAL TIME SERIES.
SUMR=0.0
DO 140 1=1, NT
OCS=OC(1, IS)
PCS=PC
IF (OCS.LT.C.O.OR.PCS.LT.G.C) €0 TO 139
RTSCI>=OCS-PCS
SUMR=SUMR+RTS(I>
60 TO 140
WHENEVER OC OR PC IS INVALID SET RTS = -99999.9
RTSU>=-99999.9
CONTINUE
RBAR=SUMR/NT
DETERMINE THE AUTOCORRELATION FUNCTION (ACF) OF THE R.T.S
RTS = RESIDUALS
RBAR = MEAN OF THE RESIDUALS
SUNR2 = VARIANCE OF THE RESIDUALS
MTMK = TIME LAG
NT = NO. OF POINTS
SUMR2-0.0
DO 145 1=1, NT
RMR=RTSCI)-RBAR
CONTINUE
PRINT 6,IS,RBAR,SUMR2
IF (ICODE-2) 15C,19C,150
XMIN=O.O
X«AX=16.0
YMIN=-1.0
TMAX= 1.0
DO 160 K=1,JN
SUM=0.0
NTMK=NT-K
DO 155 L=1,NTMK
IF CRTStL >.LE.-99999.9.0R.RTS(L*K).LE.-99999.9) 60 TO 155
SU«=SUM+ < RT S (L)-R BAR)* (RTS (L*K) -RBAR)
CONTINUE
ACFCK)=SUH/SUMR2
ACFCK)=(SUM/SUHR2)/(NTHK/RNT)
CONTINUE
PRINT 7,(ACF(K),K=1 ,JN)
PLOT THE AUTOCORRELATION FUNCTION AND CONFIDENCE BANDS.
IN THE FOLLOWING DO-LOOP THE UPPER LIMIT OF K DETERMINES
THE MAXIMUM TIME LAG. AS A DEFAULT IT CAN BE SET TO HALF THE
NUMBER OF SAMPLES (JN) IN THE TIME SERIES.
AUL=1.96/ROCTN
ALL=-AUL
98
-------
COC170
C00171
£00172
COC173
£00174
£00175
C00176
COC177
£00178
C00179
C00180
000181
£00182
COG183
£0om
COC185
£00186
C00187
C00188
C00189
C00190
COC191
C00192
COC193
£00194
COC195
£00196
£00197
C00198
COC199
OOQ200
C00201
COC202
C00203
C00204
C00205
C00206
C00207
COC208
000209
£00210
COC211
C00212
C00213
C00214
C00215
C00216
C00217
COQ218
C00219
C00220
C00221
C00222
C00223
C00224
COC225
£00226
OOfc
008
OOfc
006
006
008
006
009
009
008
008
008
008
006
008
006
008
008
008
008
008
008
008
008
006
008
008
008
008
008
008
008
008
006
006
011
011
011
011
012
008
008
008
008
008
008
008
006
008
012
008
008
008
008
008
008
008
c
c
c
C RHA
165
C
C
C
C
C
190
200
210
220
CALL SETUP (NTC1 ,TIT1 ,NXC1,XARR1,NYC1,YARR1)
DRAU THE PLUS AND KINUS 95 PERCENT CONFIDENCE LIMIT?
XX<1)=0.0
YY(1)=AUL
XX (2)=XMAX
YV(2)=AUL
CALL LIKE (12)
CALL XNEAT(C)
CALL XTICS(?)
CALL NPTS (2)
CALL CHECK (XX,YY)
CALL DSPLAY
CALL ORAWA (XC.YC)
CONTINUE
CALL FRAME
CALL HDCOPY
IF (1CODE.E0.1) 60 TO 3CQ
DETERMINE THE NORMALIZED CUMRliLATIVE PERIODOGRAM.
CNP = PARTIAL SUH OF KFD/SUN OF KFI)
FI = FREQUENCY
TWON = 2.0/NT
J=0
J=J + 1
FI(J)=J/RNT
IF (J.6T.JN) GO TO 230
SUM=0.0
DO 220 1=1,J
FIPI2=FI(I)*PI2
SUHC-0.0
SOMS=0.0
CO 210 L=1,NT
IF (RTS(L).LE .-99999.9) 60 TO 210
FIPL=FIPI2*L
SUMC = SUHC-»RTS(L)*COS(FIPL)
SUMS=SUMS+RTS(L)*SIN(FIPL)
CONTINUE
FIF=TWON*(SUMC*SUWC«SUMS*SU"S)
SUM=SUM+FIF
CONTINUE
99
-------
C0t227
C00228
C00229
C00230
COG231
C00232
000233
COG234
C00235
C00236
COC237
C00238
C00239
C00240
C00241
C00242
C00243
000244
COC245
COC246
COC247
COC248
C.00249
C00250
C00251
C00252
C00253
C00254
C00255
C00256
C00257
CQC258
000259
C00260
COC261
COC262
C00263
COC264
000265
L00266
COC267
(.00268
COC269
C00270
COC271
C00272
C00273
C00274
COQ275
COC276
C00277
COC278
C00i79
COQ2&0
C00261
COC282
C002fc3
006
008
008
008
006
008
008
008
DOS
008
008
008
006
008
008
008
008
008
008
009
010
006
008
008
OOfe
008
009
008
009
009
009
009
009
008
008
008
008
008
008
008
008
009
008
006
004)
006
006
038
00&
OCb
00V
ooe
008
009
009
009
009
c
230
240
CR
C
CR
C
C
c
c
c
c
c
c
CR
CR
CR
CR
CNP(.
GO T'
DO 2'
CNPf
CONT
XIKIN
YHIN
XHAX
THAX
DLINI
SLOPI
ANGLI
COSA
B = OL
PRIN
CALL
CALL
CALL
CALL
CALL
CALL
CALL
CALL
CALL
CALL
CALL
XX(1
YY(1
XX(2
YY(2
CALL
BED
XX(1
YY(1
XX(2
YY(2
CALL
CALL
CALL
XX(1
YY<1
XX (2
YY(2
CALL
CALL
CALL
CALL
CALL
2CO
DETERMINE THE STRAIGHT LINE PARAMETERS.
240 J=1,JN
CNP(J)=CNP(J)/CNP(JN)
UE
.0
.0
.5
.0
1.96/ROOTN
YMAX/XHAX
ANGLE=ATAN(SLOPE)
COS(ANGLE)
OLINE/COSA
PRINT 10,DLINE,SLOPE,ANGLE,COSA,B
PLOT THE NORMALIZED CUMULATIVE PERIODOGRAM VS. F8E8UENCY.
CALL SETUP (NTC2,T1T2,NXC2,XARR2,NYC2,YARR2)
EPSd)
ZES (0.5)
-------
COG264 00? CR CALL CHECK (XX,TV)
000285 009 CR CALL OSPLAt (XX,YY)
C00286 008 CALL FRAME
COC287 006 CALL HDCOPY
C00288 008 300 CONTINUE
COC289 008 CALL FINITT (0,700)
COC290 008 END
END ELT.
101
-------
D.4 Subroutine M21ADO*STAT02.SETUP
•ELT.L
ELTC07
COCC01
COCQ02
COCOC3
COOCOA
C00005
QOC006
COC307
COC008
COCG09
COOC10
COCQ11
COC012
C00013
COGC14
COCQ15
COCC16
C00017
COOQ18
COCQ19
COCC20
COCQ21
C00022
C00023
COC024
COOC2S
COOC26
COC027
COOC28
C00029
COOC30
C00031
COOQ32
C00033
COC034
C00035
OOOC36
COC037
C.OCC38
COCQ39
C00040
COC041
COC042
COCC43
COGG44
000045
C00046
CO€C47
COCC48
C00049
C00050
COOC51
COCC52
COC053
COOC54
COUu55
S2. SETUP
SL73R1 11/
004
004
004
004
004
004
006
004
004
004
004
004
004
004
006
006
006
007
006
007
006
006
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
008
008
004
004
01/78
C HOI
C
C
C
C
C AS!
98
C
C
C
CR
CR
C
100
j9:39:31 (8,)
SUbROUTINE SETUP (NTC i TIT,NXC,XARR,NYC,YARR)
[FIED BY ROD ALLEN, COMP-AID INC., OCTOBER 1978
THIS ROUTINE SETS UP THE TEKTRONIX tRAPHICS PACKAGE, SETS
INITIAL LIMITS FOR THE GRAPH AND DRAWS THE FKAME AND LABELS.
COMMON /DAT/ XK1N ,XPAX ,YMIN ,YMA X,ITYPE,1C CDE,IS
DIMENSION XARR(1>,1ARH<1>,TIT(1)
CALL INITT (12O
CALL TERM(2,1024)
ASSUMES TEKTRONIX 4014
CALL HEADER
CALL CHRSIZ<2>
CALL BOX(C,C,1C23,7?0>
CALL MOVABS(10,760)
CALL AOUTSH17,'TIME SERIES, CASE')
ENCODED,98 .CASE) ICODE
CALL AOUTST(2.CASE)
CALL AOUTSTdO,', STATION ')
ENCODEC2.98.CASE) IS
FORMAT(I2)
CALL AOtTST (2.CASE)
CALL BIMTT
SET THE SCREEN LIMITS FOR X AND Y.
CALL PLACE (3HSTD)
CALL XFRM (3)
CALL YFRM (3)
CALL XMFRM (3)
CALL YMFRM (3)
PLOT THE HAJOR-HIHOR TIC MARKS ON THE AXES.
CALL DLIHX (XMIN.XMAX)
CALL DLIMV (YMIN.VHAX)
CALL XNEAT (1)
CALL YNEAT (1)
DETERMINE THE CALL YTYPE(2)
ITYPE=1TYPE-1000
IF (ITrPE.GT.1C) CALL XTYPEC2)
ITYPE=ITYPE-10C
IF (ITYPE.6T.1) CALL YTYPEd)
ITYPE=ITYPE-10
IF (ITYPE.GT.O) CALL XTYPEd)
CALL MOVABS(300,50)
CALL AOUTST(NXC.XARR)
CALL QUOTE(SO. ,1SO.,.1S,YARR,90.,NYC)
CALL MOVABS(300,30)
CALL AOUTST(NTC.TIT)
IF (ICODE.LE.1) GO TO 100
CALL HOVABS(30C.630)
CALL AOUTST(34,'DASHED LINES 951 CONFIDENCE LIMITS')
IF (XMAX.6T.1.C) GO TO 10C
CALL MOVABS (30C.760)
CALL AOUTSTO2, 'PERIOD (1/F)')
DRAW A STANDARD FRAME.
CALL FRAME
THE
102
-------
LOGC56 GC<. RETURN
COCC57 OOA END
END ELT.
103
-------
D.4 Subroutine M21ADO*STAT02.USESET
•ELT,L S2.USESET
ELTC.G7 SL73R1 11/01/78 09:39:55 11.)
UOOC01 OCO SUBROUTINE USESET (FNUK,IbIDTH,NBASE , LAGEL 1 )
C.OCC'02 001 C MODIFIED BY ROD ALLEN, COHP-AID INC., OCTOBER 1978
COCC03 001 DIMENSION LA6ELH1)
COOC04 001 FNUnB=1.0/FNU«
COGC05 001 CALL FfORM (FNUMB , 1W1DTH,1 ,LABELI ,32 )
COCG06 000 RETURN
COCGC7 000 END
END ELT.
104
-------
APPENDIX E
CHI-SQUARE GOODNESS-OF-FIT LISTINGS
105
-------
E.I Control Element Example, M21ADO*STAT03.CASE1
SELT.L
ILTC07
COCC01
IOCC02
COCC03
COC004
000005
COC006
COCC07
COC008
000009
C00010
C00011
COC012
COCC13
S3.CASE1
SL73R1 1
001
000
004
003
003
003
003
003
003
003
003
000
001
1/01/78 Q9:4Q:46 (4.)
SXOT *21AP.O*STAT03.CHIFIT
(CHI)
-------
E.2 Program Map, M21ADO-STAT03.MAPIT
•ELT.L
ELT007
C00001
COGC02
COCD03
COOC04
COOC05
COC006
C00007
cocooe
C00009
COG010
£ND ELT.
S3 .MAPIT
SL73R1 11/01/78 09:41:36 (2.)
001 i*AP ,M21ADO*STATC3.CHIFIT
001 IN M21ADC/*STATOI.CHIFIT
001 IN «21ADO*STATQ3.HISTO
001 IN M21 AD J*STATQ3. SETUP
001 IN M21ADG*STAT03.ILINE
OOi IN SRI*SRI.BOX
002 IN SRI*SRI. HEADER
002 IK RAPS*UTUITY.CCHPOZ
000 LIB 6RAPH*TEKTROMX
000 END
107
-------
E.3 Listing of the Main Program, M21ADO*STAT03.CHIFIT
iELT.L
ELT007
COC001
COC002
C00003
COCOOA
C00005
COCC06
000007
CQCOOB
COOC09
COC010
C00011
C00012
COGQ13
coocu
COOC15
C00016
COG017
C00018
C00019
COC020
COC021
COOC22
C00023
C0002A
COOC25
C00026
C00027
C00028
C00029
COCQ30
C00031
COCQ32
COOQ33
GOG03A
C00035
C00036
COC037
COGG3?
COC039
COCOAO
COOOA1
COCOA?
COCCA3
COOCAA
LOGCA5
COGCA6
COGOA7
COOCA8
COCCA9
COCC50
COC051
COC052
C00053
COCC5A
COCC55
S3.CHIFIT
SL73R1 1
OOA
OOA
OOA
ooe
006
008
006
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
OOA
GOA
OOA
OCA
OOA
UOA
1/01
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
/78 u9:A1:17 (8,)
CHI-S8UARE GOODNESS-OF-FIT TEST.
THIS PROGRAM COMPUTES THE CHI-SQUARE STATISTIC AND PLOTS
FREQUENCY DISTRIBUTION FOR PREDICTED AND OBSERVED CONCENTRATIONS!
THE DIFFERENCES AND PERCENTAGES.
MODIFIED BY ROD ALLEN, COUP-AID INC., OCTOBER 1978
COMMON /DAT/ XMIN.XMAX ,YMIN ,YHAX,NCLAS,XC LA S(21),ICLAS(21>, ITYPE
COMMON /HNM/ XMIN1 , XMI N2 , XM AX1 , XMAX2 , YMIN 1 , YMI N2 , YMAX 1 , YM AX?
DIMENSION IDATE(123),ITIME(123>,ISITE<123>,OC(123>,PC(123),IDEN(5>
1,OBSD<21>,PRED<21),PMOD<21>,TITinC),TIT2<10>,TIT3(10>,TITA<10>,
2YARRK10>,YARR2(10>,YARR3<10>,YARRA<10>,PMOP(21>
DATA NCLAS.KCLAS1 /10.11/
1 FORMAT (//37H CHI SQUARE 600DNESS-0 F-FIT TEST FOR ,5A6/>
2 FORMAT C10A6)
3 FORMAT ()
A FORMAT
-------
100056
IOGC57
COCC58
COGQ59
(.00060
COCOM
C00062
COGC63
C00064
COG065
C00066
COC067
COOC68
COCC69
COG070
COOC71
COOQ72
COC073
COG074
C00075
COOC76
COOC77
COGC78
C00079
cocoeo
COGC61
COC082
COOC83
COOC84
COCC85
COCJC86
C00087
COQQ88
C00089
COCCVO
COUC91
COCC92
COC093
COCQ94
COCC95
COG096
COLC97
COCC98
COCC99
COC100
C00101
C00102
C0010?
C00104
C00105
100106
COC107
cocioe
COC109
C00110
COC111
C00112
^064
00*
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
006
005
004
004
004
004
004
004
004
004
006
004
004
004
004
004
004
004
004
004
004
004
004
004
OG4
004
004
004
004
004
004
004
004
004
004
004
004
004
004
C »!
c
C
c
c
c
c
c
c
c
c
CR
CR
99
50
CR
60
C
c
c
100
110
c
115
C
FROK'CATA EASE FILE ...() II
ND = ->NC. OF DATA.
NS = NC. OF STATIONS.
1DATE = DATE OF OBSERVATION
ISITE = STATION NO.
ITIHE = TIME OF OBSERVATION
ISTC,H.SC,IWDC ARE NOT USED.
OC = OBSERVED CONCENTRATION
PC = PREDICTED CONCENTRATION
READ 3,ND,NS
PRINT 4,ND,NS
XMIN1=*1.CE10
X«AX1=-1.0E10
DO 60 1 = 1,ND
READ 3,lDATE(I),ITIfE
XMAX1=AHAX1(XKAX1,PC(I),OC(D)
XMIN1=AMIN1
-------
COG113
OOG114
£00115
C00116
C00117
COG118
C00119
C00120
COC121
C00122
C00123
000124
C00125
C00126
COC127
C00128
C00129
CDC130
COC131
COC132
COG133
COC134
COC135
COC136
COG137
COC138
C00139
COOUO
C00141
COC142
COGU3
CQ014*
COC145
COOU6
COOU7
C00148
COOH9
C00150
COC151
C00152
C00153
C00154
C00155
C00156
000157
004
004
304
004
004
004
004
004
004
004
004
004
OG4
007
004
004
007
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
004
007
007
004
004
004
004
004
007
004
004
004
c
c
c
c
c
c
c
c
120
125
130
135
140
145
150
DtTERHINE THE NURc-ER OF PC IN EACH OF 10 CLASSES.
DO 120 1 = 1, N
INO=PC(I)/OELTX+1.C
If UNO .6T.NCLAS) IND=NCLAS
PRED(IND)=PREDUND)+1.C
CONTINUE
PRED(NCLAS1>=PRED(NCLAS>
00 125 1 = 1 ,NCLAS
YMAX1=AMAX1 (YHAXl.PPEDU))
CONTINUE
CHI2=0.
IF CItODE-2) 130,135,130
CALL HISTO (MCLAS 1 , OBS D ,PRE O.NTC1 ,NTC2 , TITl ,TI T2 , N YC1 ,N YC ? , YARR1 ,
1YARR2,ICODE,CHI2«N)
IF (ICODE.E0.1) 60 TO 150
DETERMINE THE NUMBER OF PC-OC IN EACH OF 10 CLASSES.
YBAX1=-1.CE10
Y«IN1=+1.0E10
DO 140 I=1,NCLAS1
PflOD=PREt(I>-OBSD(I>
YMAX1=AHAX1 (YPAX 1 .PHOD (I ) )
YHIN1=A«INUYHIN1,PMOO(I))
CONTINUE
DETERMINE THE PERCENTAGES OF (PC-OO/OC IN EACH OF 10 CLASSES,
T*AX2=-1.CE10
DO 145 I=1,NCLAS1
IF(OBSDtI).LE.1.) 06SD(I)=1.
CHI2-CHI2«PHOD(I)*PMOD(I)/OBSD(I)
PMOP(I )=100.0*PHOD( J)/OBSO(I)
YMAX2=AMAX1 (YHAX2,PfOP(I))
YMIN2-AHIN1 (VMIN2 tP^OPd))
CONTINUE
CALL HISTO (NCLAS1,PMOD,PNOP,NTC3,NTC4,TIT3,TIT4INYC3,NVC4,VARR3,
1YARR4.ICODE ,CHI2,N)
CALL FINITT (0,700)
STOP
END
END ELI.
110
-------
E.4 Subroutine M21ADO*STAT03.HISTO
SELT,L
ELTG07
COCC01
C3C-C02
COC003
COC004
C00005
COLC06
COQC07
COLC08
C00009
COCC10
COC011
COG012
COOG13
cocou
i.00015
COC016
OOC017
100018
COLC19
COCG20
C00021
COC022
COCC23
C0002*
COC025
C00026
C00027
COG028
C00029
COC030
COC031
C00032
COC033
COCC34
COCC35
COCC36
CQOC37
COC038
COUC39
COG040
C00041
COOC42
COG043
COC044
COC045
CCOC46
tOuC47
COOOA8
IOCC49
COOC50
C00051
C00052
COCC53
COIC54
COC055
S3.HISTO
SL73R1 11/
009
012
009
009
009
013
U09
009
009
009
009
009
009
009
009
009
009
009
009
009
009
009
009
009
009
009
009
009
009
009
009
009
009
010
010
009
009
009
009
009
009
009
009
009
009
009
009
010
009
009
009
009
009
00V
009
01/78
C MOD
CR 1
CR 2
CR
CR
CR
CR
1CO
CR
CR
CR
CR
110
C RHA
C RHA
C
CR OR
C
C
C
39:41:30 (13.)
SUb ROUTINE HISTO (NCl,YlfY2,NTCl»NTC2tTITl,TITC,NYClfNrC2,¥ARR1,
1YARR2.ICODE ,CH12 ,N)
IFIED BY ROD ALLEN, COMP-AID INC., OCTOBER 1978
COMMON /OAT/ XMIN,X"AX,YMIN,YMAX,NCLAS,XCLAS(?1),ICLAS(21),ITYPE
DIMENSION Y1(NC1),TIT1(1),YARR1(1),Y2(NC1),TIT2(1)f
1YARR2(1),YY(21),AOUT(6)
COMMON /MNM/XMINl,XMIN2,XI«AXl,XMAX2,YMlM,YriN2.YMAXltYMAX2
DATA YY /21*C.C/
FORMAT
-------
COC056
C00057
CD0058
COCC59
COOG60
C00061
COC062
COGC63
COGC64
000065
COOC66
COCC67
COGG68
C00069
C00070
C00071
COC072
COC073
C00074
C00075
C00076
COC077
END ELT.
011 CALL MOVABS (10,760)
009 CALL AOUTST(33,'CHI-SQUARE GOODNESS-OF-FJT, CASE
009 ENCODEd ,101, CASE) ICOOE
009 1C1 FORHATU1)
009 CALL AOUTSTd .CASE)
009 C RHA BOX AROUND PLOT
009 CALL BOXiO.C,1C23,7?0)
009 CALL BOX(0,C,1023,390)
012 C PRINT CH12 AND N IN BOX
012 IF(CHI2.LE.C.) GO TO 80
012 CALL BOX<849,0,1023,33)
013 ENCODE (36,81, ACJUT) CHI2.N
012 81 FORHATCCHI-SOUARE =' , 1P68 . 2 , AX ,'N =',I6,3X)
012 CALL NOVABS(853.A)
012 CALL CHRSIZC3)
013 CALL AOUTST(9,AOUT(5))
012 CALL MOVABSC&53.18)
012 CALL AOUTSK2C.AOUT)
012 CALL CHRSIZ(2)
012 £0 CALL HDCOPT
009 RETURN
012 END
112
-------
E.4 Subroutine M21ADO*STAT03.SETUP
•ELT.L
ELTC07
COCQ01
100002
COCC03
00000*
LOk.C05
OOC006
COCC07
COCG08
C00009
COCC10
COCC11
100012
COOC13
coccn
COCC15
C00016
C00017
COC018
C00019
C00020
COOC21
COCC22
00 CO 23
COCC24
COCC25
COC026
COG027
COC028
COCQ29
C00030
IOCC31
COU032
UOU033
COOC34
C.OOQ35
000036
COCC37
000038
COC039
S3. SETUP
SL73R1 11/
302
002
OC2
302
002
002
002
002
002
006
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
002
C02
002
003
002
002
004
004
005
002
002
002
002
01/78 C9:41:45 (6,)
SUBROUTINE SETUP (NTC ,TIT ,N YC , Y ARR , I PLOT , LAc-X Y ,L AP Y Y ,LAl-T Y )
C MODIFIED BY ROD ALLEN, COUP-AID INC., OCTOBER 157?
C
C THIS ROUTINE SETS UP THE TEKTRONIX GRAPHICS PACKAGE, SETS THE
C INITIAL LIMITS FOR THE GRAPH AMD DPA»S THE FtfAME AND LAPELS.
C
DIMENSION T 1T(1>, YARR ( 1)
CCCIMON /DAT/ XMN,XKAX,YMIN,YMAX,NCLAS,XCLAS<21),ICLAS<21),ITYPF
CALL B1NITT
CALL CHftSI2(2>
C SET THE SCREEN LIMITS FOR X AND Y.
CALL PLACE (IPLOT)
CALL XFRM (3)
CALL YFRM (T)
CALL XKFRM (3)
CALL YMFRM (3)
C PLOT THE KAJOW-fllNOR TIC MARKS ON THE AXES.
CALL DLIMX (XPUN , XM AX )
CALL DLIHY (YMIN.YMAX)
CALL XNEAT (1)
CALL YNEAT (1)
C DETERMINE THE (X,Y) AXES TYPE.
IF (ITYPE .67.100) CALL YTYPE<2)
ITYPE=ITYPE-1000
IF (ITYPE .GT. 10) CALL XTYPE<2>
ITYPE=ITYPE-10C
IF CITYPE.6T.1) CALL YTYPE(1)
ITYPE=ITYPE-1D
IF (ITYPE. 6T.O) CALL XTYPE<1)
CALL HOVABS <300,LA8XY)
CALL AOOTST(19, 'CONCENTRATION CLASS')
C RHA 6UOTE IS EPA NCC CALCOMP SYfBOL ROUTINE FOR TEKTRONIX
YPAGE=LABYY
CALL QUOTE(50.,YPA6E,.15,YARR,VC.,NYt)
CALL TSEND
CALL MOVABS <3TG,LABTY)
CALL AOUTST (NTC.TIT)
RETURN
END
fcND ELT.
113
-------
APPENDIX F
BIVARIATE REGRESSION AND CORRELATION LISTING
114
-------
F.I Control Element Example, M21ADO*STAT04.CASE6
«ELT,L SA .CASCC
ELTC07 SL73R1 11/01/78 C9:<.3:4<:
LOGOG1 o01
COCC02 000
C00003 OCA
COOC04 000
C00005 000
COC006 000
COCC07 003
COCCOf U03
COG309 003
C0001C 003
COC011 003
COC012 003
LOOC13 303
C0001A 003
COC015 003
COC016 002
COOQ17 001
O.X8T
(CHI ) (U)/ ((-)
6 0011
6.31 2.92 2.35 2.13 2.02 1.9*. 1
1.72 1.72 1.7J 1,
V l.f-6 1.53 1.M 1.3C 1.7» 1.77 1.76 1.?5 1.75
71 1.71 1.71 1.71 1.7J 1.70 1.70 1.70 1.6? 1.67
1.7* 1.73 1.73
1.66 1.645
36 (CHI)(U/0) -- I'MTS (ir**-6)«<**»-?)
22 ObSERVEO CONCtNTRAT ION
23 PREDICTED CONCFNTHAI ION
11 SCATTER&RAM
23 REGRESSION LINE (SOLID LINE)
3C SENSITIVITY PAKDS (SMALL DASH)
33 90* CONHOENCE BANDS (SMALL DASH)
3C PROBABILITY BANDS (LARfE DASH)
58 90X CONFIDENCE (SMALL DASH) PROBAttlLITY (LARGE BASH) oANDS
101 3
SADD M21ADO*STATCJ1.COQS2N
END ELT.
115
-------
F.2 Program Map, M21ADO*STAT04.MAPIT
. S4.MAPIT
ELTC07 SL73R1 11/01/78 C9:44:00 (2,)
OOOC01 001 iMAP ,M21ADO*STATU.REGANA
COC002 001 IN W21ADC*STATQ4.REGANA
COCC03 001 IN «21AOO*STAT04.STAB
COOC04 001 IN K21ADG*STAT04.LINLS6
COCQOS 002 IN SDI*SRI.BOX
COCC06 002 IN SRI*SRI.HEADER
COCC07 002 IN RAPS*UTILITY.COHPOZ
COC008 000 LIB 6RAPH*TEKTRONIX
100009 000 END
ENC ELT.
116
-------
F.3 Listing of the Main Program, M21ADO*STAT04.REGANA
»ELT,L
ELTCC7
OOGG01
COCC02
COCG03
COG004
COL005
C00006
COOC07
C00008
C00009
COC010
C00011
COUG12
COG013
L00014
C.00015
C00016
£00017
COC018
COC019
COC020
COOQ21
COC022
COGG23
OOC02A
COOC25
COOQ26
COCC27
COOC28
C00029
COCC30
GOCQ31
COGC32
COCD33
COC034
C00035
COOC36
COCC37
LOCC38
COCC39
COOOAO
CCGC41
COCC42
COCC43
C00044
COCOAS
COG046
CO CCA 7
£00048
COQC49
COC050
COOC51
C00052
COCQ53
COOC54
COOC55
SA .REGANA
SL73R1 11/01/7? 09:44:08 C15,)
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
012
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
c
c
REGRESSION ANALYSIS PROGRAM
MODIFIED BY ROD ALLEN, COUP-AID INC., OCTOBER 1978
COMMON /LSC/ RECNOP ,A ,B,XSD ,YSO ,C I L ,C IU ,S XKX6 , RC , RMSE ,ESD
DIMENSION X1(39),X2<39),X3(39),Y1(39),Y2(3V),Y3<39),YRL(1
,XPAR
23),IDATE
U123), niME<123),ISlTE<123),OCU23),PC(123), ID EN(5), TAB (34), XARR
2<10),YARR<1C),TIT<1C),XX(123),YY<123),ZARR(10>,7AR<10>,AOUT<8)
1
2
3
U
5
6
7
8
9
C
C
C
C
C
C
C
C
C
C
C
C
c
c
c
c
c
c
c
c
c
c
c
CR
CR
DIMENSION Y CLP (123) ,YC LK ( 1 23) ,Y PLP ( 1 23 ) , Y PLM ( 1 23) ,XS(123)
DATA XM1N,YHN,XMAX,YMAX / 1 0. ,1 Q. , -1 .OE1 0 ,- 1 .OE1 O/
DATA N1 ,N2,N3,NE6,YOIG /0,0 ,0, 0 ,-1 . CE1 O/
,YS(123)
FORMAT
FORMAT
INPUT FROM PARAMETER FILE ...
IDEN = DATA IDENTIFIER.
ICODE = 0, NO GRAPHICAL OUTPUT.
= 1, PLOT SCATTER GRAPH.
= 2, ADD THE REGRESSION LINE TO CODc 1.
= 3, ADD THE SENSITIVITY BOUNDS TO CODES 1-2.
= A, ADD THE CONFIDENCE BOUNDS TO CODES 1-2.
= 5, ADD THE PROBABILITY BOUNDS TO CODES 1-2.
- *, ADD THE CONFIDENCE AND PROBABILITY BOUNDS
ITYPE = 0001, X-AXIS IS LINEAR.
= 001C, Y-AXIS IS LINEAR
= 0100, X-AXIS IS LOGARITHMIC.
= 1CCO, Y-AXIS IS LOGARITHMIC.
NTC = NO. OF ASCII CHARACTERS IN THE TITLE.
NXC = NO. OF ASCII CHARACTERS IN THE X-AXIS LABEL.
NYC = NO. OF ASCII CHARACTERS IN THE Y-AXIS LABEL.
ZARR = GRAPH LOWER TITLE.
XARR = X-AXIS LABEL.
YARR = Y-AXIS LABEL.
TAB = STUDENT-T TABLE FOR 90 PERCENT C.L. AND N-2
READ 2, IDEN
PRINT 1.1DEN
READ 3, ICODE, ITYPE
READ 3, (TAB (L),L = 1 ,34)
READ 9, NTC, TIT
RLAD 9, NXC, XARR
READ 9, NYC, YARR
PRINTS, ICODE, ITYPE, NTC, NXC, NYC
DO 60 1 = 1 ,6
IF (l.Nt.ICCDF) GO TO 50
3X4HNYC:
4X4H R =
E10.3//
.3/>
TO CODES
D.O.F .
1-2.
117
-------
COOC56
000057
COOC58
COC059
C00060
C00061
COOC62
C00063
CQGC64
COC065
COC066
C00067
C00068
C00069
C00070
C00071
C00072
COOC73
COOG74
C00075
C00076
C00077
COCC78
C00079
COCG60
COOC81
C00082
COC083
COOC8A
COOQ85
C00086
COOC87
COG088
COC089
COGC90
100091
COOC92
COC093
COGC94
COC095
COOC96
C00097
COOC98
COCC99
COC100
COC101
C00102
COC103
C00104
C00105
COC106
COC107
C00108
C00109
COC110
COC111
C00112
011
011
011
011
011
011
011 •
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
READ 9.NZC, 7ARR
GO TO 6«,
50 READ 9.NZ.2AK
60 CONTINUE
C
C INPUT FROM DATA BASE FILE ...
C
C ND * NO. OF DATA.
C NS = NC. Of STATIONS.
C IDATE = DATE OF OBSERVATION
C ISITE = STATION NO.
C ITIHE = TIME OF OBSERVATION
C ISTC ,IbSC ,IbDC ARE NOT USED.
C OC = OBSERVED CONCENTRATION
C PC = PREDICTED CONCENTRATION
C
1CO READ 3,ND,NS
CR PRINT 4,ND,NS
N = 0
DO 120 1=1, ND
READ 3,lDATE(I),ITIWE.LT.O.O.OR.PCU).LT.O.O> GO TO 115
N = N*1
OC(V )=OC( I)
PC(N)=PC(I)
YRAX=AHAX1(YKAX,OC(N> , PC(N) )
IF (ISITE(I).GT.I) GO TO 105
N1=N1+1
X1 (N1)=OC GO TO 11C
N2=N2*1
X2(N2)=OCU )
Y2(N2)=PCCI)
JF (Y2(N2).GT.YBIG) YBIG=Y2(N2)
60 TO 120
110 N3=N3+1
X3(N3)= OCU )
Y3
-------
(.30113
COC1U
COC115
000116
COC117
C00118
C00119
COC120
C00121
C00122
C00123
£00124
C00125
C00126
COL127
COC128
C00129
COC130
C00131
COC132
C00133
C00134
C00135
COC136
C00137
000138
C00139
C00140
coom
C00142
000143
000144
COOU5
C00146
COG147
C00148
COG149
COG150
COG151
030152
C00153
000154
C00155
C00156
C00157
COC158
COC159
C00160
COC161
C00162
C00163
COU164
C00165
COC166
C00167
COC168
COC169
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
Q11
011
011
011
011
012
011
011
011
011
011
011
011
011
011
011
011
011
011
IF CICODE.EU.'") £0 TO ',3d
IF (ICODE.NE.3) GO TO 145
c
C PLOT THE SENSITIVITY BOUNDS ( '_ . 1 * Y"A X >
C
13C CONTINUE
CR PRINT 6
YFACT=J.1*YB1G
00 135 1=1, NN
YPLP(I)=YY
CALL AOUTST (33, 'LINEAR REGRESSION ANALYSIS,
ENCODE(1,99,CASE) ICODt
99 FOR«AT(I1)
CALL AOUTSTd .CASE)
C PRINT LEGENDS
CALL CHRSIZO)
AN = N
IFdCODE.EQ .3) AN=n
ENCODE (48 ,95, AOUT) B,A,RC,AN
95 FORHAT(4( 1PG8.2.4X))
CALL nOVABS(4,4)
CALL AOUTST(3,*N =*)
CALL AOUTST(8,AOUT(7))
CALL MOVABS(4,18>
CALL AODTST(3,'R =')
CALL AOUTST (8 ,AOUT(5) )
CALL MO*ABS<4 ,32)
CALL AOUTST(3,'B =')
CALL AOUTST
-------
C0u170
C00171
100172
COC173
COC174
C00175
COC176
£00177
C00178
COC179
C00180
COC181
COC182
COC183
C00184
C00185
C00186
COC187
COC.188
000189
C00190
000191
COC192
COC193
C00194
COC195
C00196
COC197
000198
C00199
COC200
COC201
COC202
C00203
C00204
COC205
COC206
COC207
CO 0208
000209
COC210
COC211
COC212
C00213
C00214
IOC215
C00216
COC217
COC218
COC219
C00220
COI221
C00222
C00223
COG224
COC225
COC226
011
011
011
011
011
011
C11
ou
014
011
011
011
011
011
014
015
011
011
013
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
E5
8C
c
c
c
c
c
c
9CX CONFIDENCE BOUNDS'
IDEAL LINE")
SENSITIVITY POUNDS')
IF(ICODE.Eft .1.OR.ICODE .E8.5) 60 TO 8C
CAi_L HOVA6S (72C,4)
CALL AOUTST (31t'SOLID LINE - LEAST SQUARES LINt')
CALL MOVABS (720,18)
IF(ICODE.E6 .4) CALL AOUTST ( 33,'DASH LINE
IF(ICODE.EO .2) CALL AOUTST(22 ,'DASH LINE
IFUCODE.E6 .3) CALL AOuTST(30,'DASH LINE
IFdCODE.NE.6) CALL B OX (716 ,0 ,1 023 , 3 '(. )
IF
-------
CGC227
CCU22P
(.00229
COC230
COG231
CCG232
CDC233
OOJ234
COG235
COL236
COD237
COG238
COG239
C00240
COG241
£00242
CQC243
000244
COG245
COC246
COC247
COG248
COG249
COC250
C00251
COC252
000253
C3C254
GOC255
C00256
C00257
COC258
C00259
COC260
C0u261
CCl'262
COC263
COC264
COC265
COC266
C00267
£06268
OOC269
COG270
OOG271
COC272
COQ273
C00274
OOG275
C00276
COC277
C.OG278
CQ1279
CCG2SO
COG2S1
COG202
COC2S3
011
011
011
011
01 1
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
Oil
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
011
C11
011
011
011
011
011
011
011
011
011
011
011
C11
011
c
c
c
c
CK 0
c
c
c
150
c
t
c
160
c
t
c
165
CALL CPLOT (X2,Y<:>
3KD STATION (SI.UARCS).
CALL SYMBL (4)
CALL NPTS (S3)
CALL CPLOT (X7.Y3)
IF (ICODE.Et.D GO TO 18C
PLOT THE REGRESSION LINE (SOLID)
CALL STtPL (1)
ORIGINAL AGII FANUAL
CALL LINE (C)
CALL NPTS (NN)
CALL SYMBL (0)
CALL CPLOT (XX,YY)
IF (ICODE.GT.2) GO
USED C FOR SOLID LINE
TO 150
PLOT THE PERFECT LINE (DASH).
XX (1)=XMIN
YY(1)=XMN
XX (2)=XMAX
YY(2)=XMAX
CALL LINE (12)
CALL NPTS (2)
CALL CPLOT (XX,YY)
GO TO 180
IF (ICODE.EQ.3) 60 TO 175
RECNP1=RECNOP+1
FACT1=STAB(N,TAB)*ESD
IF (ICODE.EC.5) GO TO 165
PLOT THE 90 PERCENT CONFIDENCE BOUNDS.
DO 160 1=1,NN
XBXB = XX (I)-XBAR
DIV I = XMXB*XWXB/SXMXP
FACT2=SCRT(fiECNOF+DIVI)
FACT3=FACT1*fACT2
YCLPd ) = YY( D + FACT3
YCLM I )=YY( I) -FACT3
CONTINUE
CALL LINE(12)
CALL NPTS(NN)
CALL CPLOT (XX,YCLP)
CALL CPLOT (XX.YCLPl)
IF (ICODE.EQ.4) GO TO 160
PLOT THE PROBAPILITY BOUNDS.
DO 170 1=1,KN
XMX9 = XX (I )-XBAR
D I V I = X M X b * X f" X B / S X f X P
FACT6 = SGRT(RECNP1+DIVI )
FACT7=FACT1*FACT6
YPLPd )=YY( D + FACT7
YPLWd )=YY(I)-fACTi>
121
-------
011 170 CONTINUE
jOD2£5 011 175 CALL LINE <3O
100286 011 CALL NPTS (NM)
000287 011 CALL CPLOT (XX.YPLP)
COC288 011 CALL CPLOT (XX.TPLM)
COQ289 011 180 CONTINUE
COC290 011 CALL HOCOPV
C00291 011 CALL F1N1TT (0,700)
C00292 011 C END OF PLOT.
C00293 011 END
END ELT.
122
-------
F.4 Subroutine M21ADO*STAT04.LINLSQ
iELT.L
ELTCC7
COG001
COGCG2
COCG03
C3CC04
COLCC5
iOCGG6
COGCG7
CObOOS
CQ0309
COCC10
LOGC11
COGJ12
COCC13
COCCU
>.OLC15
LOC016
CQC017
COCCU
COUC19
COCC20
LOIC21
COCC22
COLJ23
COLG24
000325
COC026
L3CC27
COG02?
COOC29
COGC30
COC031
COG032
COCC33
COIC34
COGC35
COCC36
COG037
COGQ38
LOGC39
COGC40
COGC41
COCCA2
COCGA3
COGQ44
COGOA5
COOC46
OOGOA7
COCC48
COCOA9
COCC50
COOG51
COOC52
COOC53
LOCD5A
SA .LINLSG
SL73R1 11/
30C
000
000
OOG
000
oco
ceo
OOG
000
000
000
oco
000
GOu
000
300
OOG
000
000
000
000
000
000
000
000
000
000
000
000
000
000
coo
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
DOC
ooc
OGC
000
000
OOU
01 17".
t
C
C
50
100
105
J9 :43;5t (C, )
SUBROUTINE LINLSu ( X , Y ,NO D , YR L )
YEAR = A •> fc'XTAR
/LSu/ PtCNOP,A,bfXSD,YSC,CILiCIUrSXi^Xr,nC,RMSt,tSD.
DIMENSION X (NOD),Y(NOD),YKL(NOD)
SUMX=O.C
SUMY=O.C
SUMXt=D.O
xr-AR
SUMY2=0.0
SWX«Y2=C.C
NBAO=0
DO 100 1=1, NOD
IF
-------
F.4 Subroutine M21ADO*STAT04.STAB
sELT.L S4.STAB
ELT007 SL73R1 11/01/78
COCC01 000
COC002 000 C
COOG03 OOC C
COOOOA 000 C
COOG05 000 C
COOQ06 000
COC007 000
C00008 000
C00009 000
C00010 000 1
C00011 000
C00012 000 100
COOC13 000
000014 000
COOQ15 000 110
C00016 000
C00017 000
000018 000 120
C00019 000
COOC20 000
COOC21 COO 130
C00022 000
C00023 000
COC024 000 140
C00025 000
000026 000
END ELT.
09:44:14 (C,)
FUNCTION STAB UNO,TAB)
ESTIMATION Of POPULATION MEAN USING THE STUDENT-T DISTRIBUTION
yiTH 90 PERCENT CONFIDENCE INTERVAL AND 2 DEGREES OF FREEDOM.
DIMENSION TAB(34)
NU=IND-2
IF (NU.6T.O) GO TO 100
PRINT 1,NU
FORMAT 60 TO 13C
STAB=TAB(32>
RETURN
IF (NU.GT.120) GO TO 140
STAB=TAB<33>
RETURN
STAB=TAB(34)
RETURN
END
124
-------
APPENDIX G
INTERSTATION ERROR CORRELATION LISTINGS
125
-------
G.I Program Map, M21ADO*STAT05.MAPIT
•FLT.L S5.MAP1T
EL-U07 SL73R1 11/01/78 09:46:01 (0.)
COOC01 000 IN STATG5.COKEL2
COC002 OOG IN STATG5 .LINLSC
COLCO? 000 IN STAT05.XYCORR
COLG04 000 END
END ELT.
126
-------
G.2 Listing of the Main Program, M21ADO*STAT05.CORHL2
iELT.L
ELTC.07
LOCL01
COuC02
COC303
COCCCK
COLC05
COL306
COCC07
C00008
COCC09
IOC010
COCC11
CQCC12
LOCG13
COCCU
COC'C15
C00016
COC017
COCC18
COCQ19
C0t020
coorzi
COOC22
COC023
COGC24
COCC25
COLC26
COGC27
COCG28
COOC29
C00030
LOG031
COC032
COG033
COG034
IOCG35
COGC36
COCG37
LOGu38
COGC39
COCC40
CPG041
COGC42
COCC43
COuC<,«
IOCC45
COCC46
LOOCA7
C.OUC48
COCC49
COCQ50
COGC51
COC052
IOG053
000054
IOUC55
S5 .COREL r
SL73R1 11/
000
001
C01
001
001
001
001
001
000
001
001
001
001
001
001
000
000
000
001
001
001
001
001
001
OC1
001
001
000
001
001
300
OOG
000
coo
001
001
001
u01
oco
OOu
000
oco
ooc
001
001
000
001
001
001
001
001
001
001
000
001
01 /7,fi
1
?
T
t,
5
6
7
8
9
c
t
c
c
t
c
c
c
c
c
c
c
c
c
c
c
1CO
110
c
c
c
115
uo
:<9:45:5u (1.)
COMMON /LSC,/ Rt Cr,OP ,A ,t ,X SD ,Y SD ,C I L ,t I U , i, Xf At , KC , f * ? t ,L '. D
DIMENSION PC(123,3),OC(12:',3),IOATE(123),niKCl12''),COI
-------
CXC56
IOCC57
COGC58
COCC59
CQG060
COC061
CQCC62
COG063
COGS64
000365
COCC66
G00067
LOCC68
COC069
COCC70
C.OCC71
C00072
COC073
COCC74
COOC75
IDOC76
COCC77
CCGC78
C00079
COOObO
C00081
COCC62
COOG83
COOCbA
C00085
COG086
COL.387
CCCCbB
OOU089
COGC90
COC091
COGC92
COOC93
CD0094
COG095
COGC96
LOCC97
C0tu98
C00099
C00100
COC101
COU102
J01
U01
001
001
001
001
UOL
OC1
000
000
GOO
001
000
000
000
000
000
001
001
001
001
001
001
001
000
001
001
001
000
oco
000
000
000
001
001
001
000
000
OOG
000
001
001
001
001
000
000
OOG
125
130
135
c
c
c
c
c
c
2CO
210
220
230
C
c
c
240
C
C
C
250
ITIMtD=lTIMcS
ITIKE (NT)=ITIKED
IF (OCS.LT.C.C.OR.PCS.LT.r.C) GO TO 135
PCCNT, ISITt ) = PCS
OC(NT,ISITE ) = OCS
CONTINUE
IF (IOPT.GT .1) 60 TO 210
OPTION 1: DETERMINE THE CORRELATION 6ETUEEN AIvY TkO SITES.
CALL XYCORR (NT,OC<1,IS1),PC<1,IS1) ,OC (1,IS2)tPC(1, I S2 ) )
PRINT A ,NT,RC ,CIL,CIU
STOP1
OPTION 2: DETERMINE THE CORRELATION AND CONFIDENCE LIMITS MATRIX,
DO 230 1 = 1, NS
DO 220 4 = 1,NS
IF (I.Ett.J) GO TO 210
CALL XYCORR (NT.OC(1,1), PC(1,1),OC(1,J), PC (1 ,J>>
COR»(I,J)=RC
CONL(I,J)=CIL
CONUd, J)=CIU
60 TO 220
coRnd ,J) = 1.0
CONL(I,J) = 1 .0
CONUd, J)=1 .0
CONTINUE
CONTINUE
PRINT CUT THE CORRELATION MATRIX.
PRINT 5,d,I = 1,NS)
DO 240 1=1,NS
PRINT 6 ,1 , SCORP (I ,J),J=1,NS)
CONTINUE
PRINT OUT THE LOWER AND UPPER CONFIDENCE LIMITS.
PRINT 7,(I,I=1,NS)
DO 250 1=1,NS
PRINT 9,1 ,(CONL(I ,J),J = 1,NS)
PRINT 8,(CONU(I,J),J = 1 ,NS)
CONTINUE
STOP;;
END
ENt. ELT.
128
-------
G.3 Subroutine M21ADO*STAT05.XYCORR
«ELT,L
ELTC07
OOOC01
COCQ02
COC003
COC004
COCG05
COC006
C00007
COC008
LOC009
COC310
COG011
£00012
COCC13
COCQU
COU015
COC016
COCC17
COCC18
COGG19
C00020
COOC21
S5 .XYCORR
SL7iR1 11/
001
000
001
001
000
001
OOu
001
001
001
001
001
001
001
jOO
000
000
000
001
ooc
000
01/76
C
C
C
100
C
C
C
09:^6 -.03 C1,)
SUEROUT1NF XKCORR (NT , C,C1 ,P C 1 , OC2 , FC2 )
COMMON /LSU/ RECNOP.A, B.XSO , Y S D , C I L , C I U , S XM Xb , R C .RflSE ,t SD
DIMENSION OC1,PCl(NT>tOC2(NT),PC^,DC2(12?>,YRl(12
13)
COhPUTf (OC-PC) FOR FIRST SITE.
00 100 J = 1|M
IF (OC1(J).LT.C.C.OR.PCl(J).LT.O.O.OK.OC2(J).LT.:.u.OR.PC?(J).LT
10.0) 60 TO 100
NN=NN+1
DC1(NN)=OC1 (J)-PCK J)
DC2(NN)=OC2 (J)-PC2( J)
CONTINUE
COMPUTE THE LINEArt CORRELATION BETWEEN TUO SITES.
CALL LINLSQ ( DC 1 , D C 2 ,N N , Y RL )
RETURN
END
END ELT.
129
-------
TECHNICAL REPORT DATA
(Please read instructions on the reverse before completing)
1 REPORT NO
EPA-600/4-80-013C
3. RECIPIENT'S ACCESSIOf»NO.
4. TITLE AND SUBTITLE
EVALUATION OF THE REAL-TIME AIR-QUALITY MODEL USING
THE RAPS DATA BASE
Volume 3. Program User's Guide
5. REPORT DATE
February 1980
6. PERFORMING ORGANIZATION CODE
7 AUTHORIS)
8. PERFORMING ORGANIZATION REPORT NO.
R.E. Ruff, H. Shigeishi, and R.H. Allen (Comp-Aid, Inc.)
Final Report
SRI Project 6868
9 PERFORMING ORGANIZATION NAME AND ADDRESS
SRI International
333 Ravenswood Avenue
Menlo Park, California 94025
10. PROGRAM ELEMENT NO.
I AA603 AA-26 (FY-77)
11. CONTRACT/GRANT NO.
68-02-2770
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory -
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park. North Carolina 27711
13. TYPE OF REPORT AND PERIOD COVERED
RTP, NC
FINAL 8/77-4/79
14. SPONSORING AGENCY CODE
EPA/600/09
15 SUPPLEMENTARY NOTES
16. ABSTRACT
The theory and programming of statistical tests for evaluating the Real-Time Air-Quality Model (RAM) using the
Regional Air Pollution Study (RAPS) data base are fully documented in four volumes. Moreover, the tests are
generally applicable to other model evaluation problems. Volume 3 presents the software used in the statistical
tests for evaluating the RAM. Six statistical tests are described, with attention to the programming philosophy
behind them. Also presented is a review of the auxiliary software that sort, retrieve, format, and display the data.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
* Air pollution
* Mathematical models
* Evaluation
* Tests
* Computer systems programs
* Statistical tests
b.IDENTIFIERS/OPEN ENDED TERMS c. COSATI Field/Group
Real-Time Air-Quality Model
Regional Air Pollution Study
Data Base
13B
12A
14B
09 B
13. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report I
UNCLASSIFIED
21. NO. OF PAGES
138
20. SECURITY CLASS (This page)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
130
-------
<* .'
------- |